26895 world development report 2004 Making ServicesWork for Poor People (c) The International Bank for Reconstruction and Development / The World Bank (c) The International Bank for Reconstruction and Development / The World Bank world development report 2004 Making ServicesWork for Poor People A Copublication of the World Bank and Oxford University Press (c) The International Bank for Reconstruction and Development / The World Bank © 2003 The International Bank for Reconstruction and Development/The World Bank 1818 H Street, N.W. Washington, D.C. 20433 Telephone 202-473-1000 Internet www.worldbank.org E-mail feedback@worldbank.org All rights reserved. 1 2 3 4 05 04 03 A copublication of the World Bank and Oxford University Press. Cover and interior design: Susan Brown Schmidler. Cover photographs, from left to right: Nurse in Rwanda showing a newborn infant; © David Turnley. A street child in New Delhi drinks water from a tap; © Reuters NewMedia/CORBIS. Children learning in an elementary school in Myanmar; © Chris Lisle/CORBIS. This volume is a product of the staff of the World Bank. The �ndings, interpretations, and conclusions expressed herein do not necessarily reflect the views of the Board of Executive Directors of the World Bank or the governments they represent. The World Bank does not guarantee the accuracy of the data included in this work. The bound- aries, colors, denominations, and other information shown on any map in this work do not imply any judgment on the part of the World Bank concerning the legal status of any territory or the endorse- ment or acceptance of such boundaries. Rights and Permissions The material in this work is copyrighted. Copying and/or transmitting portions or all of this work without permission may be a violation of applicable law. The World Bank encourages dissemination of its work and will normally grant permission promptly. For permission to photocopy or reprint any part of this work, please send a request with complete information to the Copyright Clearance Center, Inc., 222 Rosewood Drive, Danvers, MA 01923, USA, telephone 978-750-8400, fax 978-750-4470, www.copyright.com. All other queries on rights and licenses, including subsidiary rights, should be addressed to the Of�ce of the Publisher, World Bank, 1818 H Street NW, Washington, DC 20433, fax 202-522-2422, e-mail pubrights@worldbank.org. ISBN 0-8213-5537-6 (clothbound) ISBN 0-8213-5468-X (paperback) ISSN 0163-5085 (c) The International Bank for Reconstruction and Development / The World Bank Contents Acknowledgments xiii Abbreviations and Data Notes xiv Foreword xv Overview 1 The problem 2 The framework of relationships—between clients, providers, and policymakers 6 What not to do 10 What can be done? 12 1 Services can work for poor people but too often they fail 19 Outcomes are substantially worse for poor people 20 Affordable access to services is low—especially for poor people 20 Quality—a range of failures 22 Making services work to improve outcomes 26 spotlight on Progresa 30 2 Governments should make services work 32 A public responsibility 32 Growth, though essential, is not enough 35 More public spending alone is not enough 35 Technical adjustments without changes in incentives are not enough 40 Understanding what works and why—to improve services 42 spotlight on Kerala and Uttar Pradesh 44 3 The framework for service provision 46 An analytical framework: actors and accountabilities 47 Why establishing relationships of accountability is so complex 52 Successes and failures of the public sector and the market 54 From principles to instruments 58 Reforming institutions to improve services for poor people will be dif�cult 60 spotlight on Uganda 62 v (c) The International Bank for Reconstruction and Development / The World Bank vi WORLD DEVELOPMENT REPORT 2004 4 Clients and providers 64 When will strengthening the client-producer link matter most? 64 Increasing client power through choice 66 Increasing consumer power through participation 70 Client power in eight sizes 74 spotlight on the Bamako Initiative 76 5 Citizens and politicians 78 Citizen voice and political accountability 78 The politics of providing public services to poor people 81 Beyond the ballot box: citizen initiatives to increase accountability 85 Information strategies to strengthen voice 86 Decentralization to strengthen voice 89 Citizen voice in eight sizes 90 spotlight on the Kecamatan Development Program 92 spotlight on Norway and Estonia 94 6 Policymakers and providers 95 Compacts, management, and the “long route� of accountability 95 Increasing accountability: separating the policymaker from the provider 98 Limits to accountability 99 Overcoming the limits 100 Provider incentives in eight sizes 106 Scaling up, scaling back, and wising up 108 spotlight on Cambodia 109 7 Basic education services 111 Common problems of service provision 111 For higher-quality systems, strengthen the relationships of accountability 113 Citizens and clients, politicians and policymakers: voice 114 Policymakers and organizational providers: compacts 117 Organizational and frontline providers: management 124 Client power 124 Getting reform going 128 spotlight on Educo 131 8 Health and nutrition services 133 The health of poor people 134 Market failures and government failures 136 Strengthening client power 143 Strengthening poor citizens’ voice 146 (c) The International Bank for Reconstruction and Development / The World Bank Contents vii Compacts: provider incentives to serve the poor 149 Six sizes �t all? 154 spotlight on Costa Rica and Cuba 157 9 Drinking water, sanitation, and electricity 159 The state of water and sanitation services 159 Infrastructure and the accountability framework for service delivery 160 Urban water networks 164 Rural areas: network and non-network systems 171 Sanitation 173 Electricity 175 Moving the reform agenda forward 176 spotlight on Johannesburg 178 10 Public sector underpinnings of service reform 180 Strengthening the foundations of government 180 Spending wisely 181 Decentralizing to improve services 185 Making, managing, and implementing good policies 191 Curbing corruption in service delivery 195 Managing transitions: overcoming reform hurdles 198 Evaluating and learning 199 spotlight on Ceará 201 11 Donors and service reform 203 Aid and accountability 203 Strengthen—don’t weaken—the compact 204 Let provider organizations manage 206 Increase client power 208 Promote voice 209 Align aid delivery with service delivery 211 Why reforming aid is so dif�cult 216 Bibliographical Note 218 Endnotes 219 References 228 Selected World Development Indicators 2004 249 (c) The International Bank for Reconstruction and Development / The World Bank viii WORLD DEVELOPMENT REPORT 2004 Boxes 1 The eight Millennium Development Goals 2 2 Services—a public responsibility 3 3 Spotlight on “spotlights� 16 1.1 Who are “poor people?� 20 1.2 HIV/AIDS is killing teachers 23 1.3 School services for girls are not in high demand in Dhamar Province, Yemen 25 Crate 1.1 Determinants of health and education outcomes—within, outside, and across sectors 27–29 2.1 Most governments take responsibility for health and education—often appealing to human rights 34 2.2 The Fast-Track Initiative—providing assistance for credible national education strategies 36 2.3 Why it’s so hard to “cost� the Millennium Development Goals 41 3.1 A glossary for this Report 48 3.2 The many meanings of accountability 51 3.3 Creating conditions of accountability: the police 54 3.4 The “Progressive Era�: creation of modern bureaucracy 55 3.5 Seeking services in Egypt 56 3.6 Health care in Central Asia and the Caucasus: the long and short of it 59 4.1 The private sector is preferred in Andhra Pradesh, India 67 4.2 Bribery in Eastern Europe 68 4.3 Payment and accountability 69 4.4 No blanket policy on user fees 71 5.1 Why are public health and education services so dif�cult to get to poor people? 82 5.2 The “Curley effect� 83 5.3 Better to build rural schools than to run them well in Pakistan 84 5.4 Follow the public’s money 86 5.5 Down to earth: information technology improves rural service delivery 87 6.1 A good doctor is hard to �nd 96 6.2 Provider discretion can hurt the poor 97 6.3 Bribery hurts the poor 98 6.4 Learning to regulate 99 6.5 Be careful what you wish for—part 1 100 6.6 Be careful what you wish for—part 2 101 6.7 Incentive pay works for speci�c health interventions 102 6.8 NGOs can be more flexible than government 104 6.9 Is the GATS a help or a hindrance? 105 7.1 The dismal state of teacher training in Pakistan circa 1990 114 7.2 Test-based accountability—nothing new under the sun 119 7.3 School-based performance awards in Chile 121 7.4 Two large-scale cross-national assessments of learning 122 (c) The International Bank for Reconstruction and Development / The World Bank Contents ix 7.5 Randomized experiments in Busia district, Kenya 123 7.6 School improvement in Cambodia 125 7.7 Alternate routes to basic education in Ethiopia 126 7.8 Education reform and teachers’ unions in Latin America 129 8.1 Ethnicity and health 134 8.2 Making health services work for poor people in the Islamic Republic of Iran 135 8.3 The changing mix of cure and care: who treats what, and where? 139 8.4 Buying results to reach the Millennium Development Goals 142–143 8.5 Vouchers for sex workers in Nicaragua 145 8.6 Making health insurance work for poor people 146 8.7 The government as active purchaser of health outcomes through strategic contracting 149 8.8 The risks of capitation payments 150 8.9 Modulated payments for providers according to income criteria 151 8.10 The human resource crisis in health services 154 8.11 Developing a professional ethos in midwifery 154 9.1 Clientelism in service delivery 163 9.2 Decentralization and the water industry—in history 164 9.3 Trends in private participation: water, sewerage, and electricity 166 9.4 Private participation—in history 167 9.5 Private participation in water and sanitation can save poor people’s lives, and money 168 9.6 Charging for water—in history 170 9.7 Fighting arsenic by listening to rural communities 173 9.8 Are pipes and wires different? 175 10.1 The impact of Argentina’s crisis on health and education services 181 10.2 The case of the missing money: public expenditure tracking surveys 185 10.3 Decentralization as a political imperative: Ethiopia 187 10.4 Many roads to decentralization: Latin America 187 10.5 Building local capacity: the role of the center 191 10.6 “Yes, Minister� 192 10.7 Managing the thorny politics of pro-poor service delivery reforms 198 10.8 Ready for results? 200 11.1 The debate over global funds: Uganda 205 11.2 Social Investment Fund: Jamaica 209 11.3 Donors support democratic governance 210 11.4 Donors support transparent budget processes: Tanzania 211 11.5 Why aid agencies focus on inputs 212 11.6 A case for harmonization in Bolivia 213 11.7 Linking budget support to performance 215 11.8 Pooling knowledge transfers 216 (c) The International Bank for Reconstruction and Development / The World Bank x WORLD DEVELOPMENT REPORT 2004 Figures 1 Progress in human development: off track 2 2 More public spending for the rich than for the poor 4 3 Water, water everywhere, nor any drop to drink 5 4 The framework of accountability relationships 6 5 It paid to vote for PRI 7 6 Increased public spending is not enough 11 7 Eight sizes �t all? 14 1.1 Child mortality is substantially higher in poor households 20 1.2 The poor are less likely to start school, more likely to drop out 21 1.3 Water, water everywhere, nor any drop to drink 23 2.1 National income and outcomes are strongly associated, especially in low-income countries 35 2.2 Changes in public spending and outcomes are only weakly related: schooling 37 2.3 Changes in public spending and outcomes are only weakly related: child mortality 37 2.4 The association between outcomes and public spending is weak, when controlling for national income 38 2.5 Richer people often bene�t more from public spending on health and education 39 2.6 The dominant share of recurrent spending on education goes to teachers (selected Sub-Saharan countries) 40 3.1 The relationships of accountability have �ve features 47 3.2 Key relationships of power 49 4.1 Client power in the service delivery framework 65 4.2 Eight sizes �t all 75 5.1 Voice in the service delivery framework 79 5.2 Democracy’s century 81 5.3 It paid to vote for PRI 85 5.4 Eight sizes �t all 91 6.1 Compact and management in the service delivery framework 95 6.2 Eight sizes �t all 107 7.1 Poor children: less likely to start school, more likely to drop out 112 7.2 Fifteen-year-olds in Brazil and Mexico perform substantially worse on standardized tests than students in OECD countries 113 7.3 Increases in test scores per dollar spent on different inputs 116 7.4 School success depends on more than spending per student 120 7.5 Centralized exams have a strong impact on student performance 120 7.6 In Chile, good schools service students from every level of socioeconomic status 121 8.1 Reaching the MDGs in health: accelerate progress 134 8.2 Reaching the MDGs in health: focus on poor households 135 8.3 Poor people use high-impact services less 136 8.4 Richer groups do well in absolute terms 136 8.5a Poor women do not know much about HIV 137 8.5b Husbands say no to contraception 137 (c) The International Bank for Reconstruction and Development / The World Bank Contents xi 8.6 A public responsibility, but private spending matters 137 8.7 The public-private mix differs between poor and rich, and among interventions 138 8.8 Making health services easier to deliver, through standardization and empowerment 140 8.9 Community-managed health services increase utilization and reduce spending 144 8.10 High spending does not ensure more equitable immunization 150 8.11 Citizens exert power on both providers and purchasers 151 8.12 Six sizes �t all 155 9.1 Little progress in access to improved water and sanitation, 1990 and 2000 160 9.2 24-hour water: a pipe dream 160 9.3 Water and sanitation by poorest and richest �fths 161 9.4 Alternative sources of water: poor people pay more 161 9.5 Accountability in infrastructure services 162 10.1 Strengthening public sector foundations for service delivery requires coordinating multiple compact relationships 180 10.2 Subnational shares of expenditures vary considerably 186 10.3 Decentralization and the service delivery framework 188 10.4 The anatomy of policy mismanagement at the top 192 10.5 Working to keep citizens educated, healthy, and safe 193 10.6 No straight roads to success: sequencing budget reforms 194 10.7 From weak basics to strong foundations in public sector institutional reforms 195 10.8 Many forces at play in curbing corruption in service delivery 197 11.1 The feedback loop between bene�ciaries and donor country taxpayers is broken 204 11.2 Donor fragmentation: on the rise 206 11.3 Bureaucratic quality declines with donor fragmentation in Sub-Saharan Africa 208 Tables 1 Economic growth alone is not enough to reach all the Millennium Development Goals 3 1.1 The nearest school or health center can be quite far 22 1.2 Staff are often absent 24 1.3 Absence rates vary a lot—even in the same country 24 2.1 Public expenditures on health and education: large but varied 33 3.1 Organizational providers take a variety of ownership and organizational structures 50 3.2 Examples of discretionary and transaction-intensive services 53 3.3 Modern institutions took a long time to develop 61 5.1 Pro-poor and clientelist service environments when the average citizen is poor 80 7.1 In Madagascar, at higher levels of education unit costs are much higher and participation of the poor much lower 117 7.2 Schools account for only a small part of variance in student learning outcomes (percent) 118 7.3 Autonomy and outcome in Merida, República Bolivariana de Venezuela, in the mid-1990s 125 8.1 Selected examples of obstacles for the delivery of health and nutrition services to the poor 141 (c) The International Bank for Reconstruction and Development / The World Bank xii WORLD DEVELOPMENT REPORT 2004 8.2 Affordability remains a problem for the poor 144 8.3 How do we know whether poor people’s voices have been heard? 147 10.1 Fallible markets, fallible governments, or both? 182 10.2 Decentralization is never simple 189 10.3 Walk before you run 196 11.1 So many donors . . . 207 (c) The International Bank for Reconstruction and Development / The World Bank Acknowledgments This Report has been prepared by a team led by Shantayanan Devarajan and Ritva Reinikka, and comprising Junaid Ahmad, Stephen Commins, Deon Filmer, Jeffrey Hammer, Lant Pritch- ett, Shekhar Shah, and Agnès Soucat, with additional contributions by Nazmul Chaudhury. The team was assisted by Claudio E. Montenegro and Manju Rani. Bruce Ross-Larson and Meta de Coquereaumont were the principal editors. The work was carried out under the gen- eral direction of Nicholas H. Stern. The team was advised by a panel of experts led by Emmanuel Jimenez and comprising Martha Ainsworth, Abhijit Banerjee, Timothy Besley, John Briscoe, Anne Case, Luis Crouch, Angus Deaton, David G. de Groot, Anil Deolalikar, Esther Duflo, Vivien Foster, Anne-Marie Goetz, Jonathan D. Halpern, Joel Hellman, Charles Humphreys, R. Mukami Kariuki, Elizabeth King, Michael Kremer, Kenneth Leonard, Maureen Lewis, Benjamin Loevinsohn, Michael Mer- taugh, Allister Moon, Howard Pack, Samuel Paul, Sanjay Pradhan, Michael Walton, and Dale Whittington. Many others inside and outside the World Bank provided helpful comments, wrote back- ground papers and other contributions, and participated in consultation meetings. These con- tributors and participants are listed in the Bibliographical Note. The Development Data Group contributed to the data appendix and was responsible for the Selected World Development Indicators. Much of the background research was supported by several generous trust fund grants from the UK Department for International Development, and the Dutch, Finnish and Norwegian Governments. The team undertook a wide range of consultations for this Report, which included work- shops in Berlin, Brussels, Cairo, Colombo, Dhaka, Geneva, Havana, Helsinki, Kampala, New Delhi, Paris, Pretoria, Tokyo, Washington D.C., and a series of video conferences with partici- pants from Africa, East and South Asia, Europe, Latin America, and the Middle East. The par- ticipants in these workshops and video conferences included researchers, government of�cials, and staff of nongovernmental and private-sector organizations. Rebecca Sugui served as executive assistant to the team; Leila Search as program assistant and technical support; and Endy Shri Djonokusomo and Ofelia Valladolid as team assistants. Evangeline Santo Domingo served as resource management assistant. Book design, editing, and production were coordinated by the Production Services Unit of the World Bank’s Of�ce of the Publisher, under the supervision of Susan Graham and Ilma Kramer. xiii (c) The International Bank for Reconstruction and Development / The World Bank Abbreviations and Data Notes Abbreviations ACE Community Education Association MWSS Metro-Manila Waterworks and Sewerage (El Salvador) System BCG Bacillus Calmette-Guérin NGO Nongovernmental organization BRAC Bangladesh Rural Advancement Committee ODA Of�cial development aid DAC Development Assistance Committee (OECD) OECD Organisation for Economic Co-operation DPT Diphtheria-pertussis-tetanus and Development Educo Educación con Participación de la PRI Institutional Revolutionary Party (Mexico) Comunidad (Education with the Progresa El Programa de Educación, Salud y Participation of Communities, El Salvador) Alimentación de México (Education, Health, EPRDF Ethiopia People’s Revolutionary Democratic and Nutrition Program of Mexico) Front PRONASOL Programa Nacional de Solidaridad (National GDP Gross domestic product Solidarity Program of Mexico) GNI Gross national income REB Rural electricity board HIPC Heavily indebted poor country SIDA Swedish International Development Agency ICRG International Country Risk Guide TIMSS Third International Mathematics and Science IDT Impres Desa Tertinggal (Indonesia) Study KDP Kecamatan Development Program (Indonesia) USAID U.S. Agency for International Development LICUS Low-income country under stress VERC Village education resource center MKSS Mazdoor Kisan Shakti Sanghathan WHO World Health Organization (India) WSP Water and Sanitation Program Data Notes The countries included in regional and income groupings in other status of a territory. The term developing countries this Report are listed in the Classi�cation of Economies table includes low- and middle-income economies and thus may at the end of the Selected World Development Indicators. include economies in transition from central planning, as a Income classi�cations are based on GNP per capita; thresh- matter of convenience. The term developed countries or olds for income classi�cations in this edition may be found industrial countries may be used as matter of convenience to in the Introduction to Selected World Development Indica- denote the high-income economies. tors. Group averages reported in the �gures and tables are Dollar �gures are current U.S. dollars, unless otherwise unweighted averages of the countries in the group unless speci�ed. Billion means 1,000 million; trillion means 1,000 noted to the contrary. billion. The use of the word countries to refer to economies implies no judgment by the World Bank about the legal or xiv (c) The International Bank for Reconstruction and Development / The World Bank Foreword We enter the new millennium with great hopes. For the �rst time in human history, we have the possibility of eradicating global poverty in our lifetime. One hundred and eighty heads of state signed the Millennium Declaration in October 2000, pledging the world to meeting the Millennium Development Goals by 2015. In Monterrey, Mexico, in the spring of 2002, the world’s nations established a partnership for increasing external assistance, expanding world trade, and deepening policy and institutional reforms to reach these goals. Foreign aid, which declined during the 1990s, has begun to increase again. But the �rst few years of the 21st century bring heightened challenges. HIV/AIDS and other diseases, illiteracy, and unclean water threaten to dash the hopes of millions, possibly billions, of people that they might escape poverty. Tragically, conflict has undermined devel- opment in many countries. Peace and development go hand in hand. And even as we learn how to make development assistance more effective, aid continues to be criticized for not being effective enough. This year’s World Development Report, the 26th in the World Bank’s flagship series, helps to re-ignite and reinforce our hopes by confronting these challenges. Development is not just about money or even about numerical targets to be achieved by 2015, as important as those are. It is about people. The WDR focuses on basic services, particularly health, educa- tion, water, and sanitation, seeking ways of making them work for poor people. Too often, services fail poor people. These failures may be less spectacular than �nancial crises, but their effects are continuing and deep nonetheless. The report shows that there are powerful examples of services working for poor people. Services work when they include all the peo- ple, when girls are encouraged to go to school, when pupils and parents participate in the schooling process, when communities take charge of their own sanitation. They work when societies can curtail corruption—which hurts poor people more than it hurts the better off—particularly when it hits basic health services, which poor people need desperately. They work when we take a comprehensive view of development—recognizing that a mother’s education will help her baby’s health, that building a road or a bridge will enable children to go to school. Services work especially well when we recognize that resources and their effective use are inseparable. More effective use makes additional resources more productive—and the argu- ment for aid more persuasive. External resources can provide strong support for changes in policy and practice that can bring more effective use. This is how we can scale up to achieve the Millennium Development Goals. To improve service delivery, the WDR recommends institutional changes that will strengthen relationships of accountability—between policymakers, providers, and citizens. These changes will not come overnight. Solutions must be tailored not to some imaginary “best practice� but to the realities of the country or the town or the village. One size will not xv (c) The International Bank for Reconstruction and Development / The World Bank xvi WORLD DEVELOPMENT REPORT 2004 �t all. But I am convinced that this new way of thinking about service delivery, and indeed about development effectiveness, will bear fruit, particularly when matched with adequate resources and a desire to assess what works and what does not, and to decide what must be scaled up and, indeed, what must be scaled down. In short, this year’s WDR is central to the World Bank’s two-pronged strategy for devel- opment—investing in and empowering people, and improving the climate for investment. Next year’s WDR will focus on the second of these. Together, these reports form part of the World Bank’s contribution to meeting the challenge the global community has set for itself—to eradicate poverty in our lifetime. James D. Wolfensohn (c) The International Bank for Reconstruction and Development / The World Bank Overview T oo often, services fail poor ization to local governments, community people—in access, in quantity, in participation, and direct transfers to house- quality. But the fact that there are holds. There have been spectacular suc- strong examples where services do cesses and miserable failures. Both point to work means governments and citizens can the need to strengthen accountability in do better. How? By putting poor people at three key relationships in the service deliv- the center of service provision: by enabling ery chain: between poor people and them to monitor and discipline service providers, between poor people and policy- providers, by amplifying their voice in poli- makers, and between policymakers and cymaking, and by strengthening the incen- providers. Foreign-aid donors should rein- tives for providers to serve the poor. force the accountability in these relation- I go to collect water four times a Freedom from illness and freedom from ships, not undermine it. day, in a 20-litre clay jar. It’s hard illiteracy—two of the most important ways Increasing poor clients’ choice and partic- work! . . . I’ve never been to school poor people can escape poverty—remain ipation in service delivery will help them as I have to help my mother with elusive to many. To accelerate progress in monitor and discipline providers. Raising her washing work so we can earn human development, economic growth is, poor citizens’ voice, through the ballot box enough money. . . . Our house of course, necessary. But it is not enough. and widely available information, can doesn’t have a bathroom. . . . If I could alter my life, I would really Scaling up will require both a substantial increase their influence with policymakers— like to go to school and have more increase in external resources and more and reduce the diversion of public services to clothes. effective use of all resources, internal and the non-poor for political patronage. By Elma Kassa, a 13-year-old girl external. As resources become more produc- rewarding the effective delivery of services from Addis Ababa, Ethiopia tive, the argument for additional resources and penalizing the ineffective, policymakers becomes more persuasive. And external can get providers to serve poor people better. resources can provide strong support for Innovating with service delivery arrange- changes in practice and policy to bring ments will not be enough. Societies should about more effective use. The two are com- learn from their innovations by systemati- plementary—that is the essence of the cally evaluating and disseminating informa- development partnership that was cemented tion about what works and what doesn’t. in Monterrey in the spring of 2002. Only then can the innovations be scaled up This Report builds an analytical and to improve the lives of poor people around practical framework for using resources, the world. whether internal or external, more effec- The challenge is formidable, because tively by making services work for poor making services work for poor people people. We focus on those services that involves changing not only service delivery have the most direct link with human arrangements but also public sector institu- development—education, health, water, tions. It also involves changing the way much sanitation, and electricity. foreign aid is transferred. As governments, Governments and citizens use a variety citizens, and donors create incentives for of methods of delivering these services— these changes, they should be selective in the central government provision, contracting problems they choose to address. They out to the private sector and nongovern- should be realistic about implementation mental organizations (NGO)s, decentral- dif�culties. And they should be patient. 1 (c) The International Bank for Reconstruction and Development / The World Bank 2 WORLD DEVELOPMENT REPORT 2004 The problem world is off track in reaching the goals for Poverty has many dimensions. In addition to primary education, gender equality, and low income (living on less than $1 a day), child mortality. illiteracy, ill health, gender inequality, and To reach all of these goals, economic environmental degradation are all aspects of growth is essential. But it will not be Figure 1 Progress in human being poor. This is reflected in the Millen- enough. The projected growth in per capita development: off track nium Development Goals, the international GDP will by itself enable �ve of the world’s People living on less than $1 a day community’s unprecedented agreement on six developing regions to reach the goal for Percent 30 the goals for reducing poverty (box 1). The reducing income poverty (table 1). But that multidimensional nature of poverty is also growth will enable only two of the regions to reflected in the World Bank’s two-pronged achieve the primary enrollment goal and 20 strategy for development—investing in peo- none of them to reach the child mortality ple and improving the investment climate. goal. If the economic growth projected for 10 That �ve of the eight goals and one of the two Africa doubles, the region will reach the prongs of the strategy for development con- income poverty goal—but still fall short of 0 cern health and education signals how central the health and education goals. In Uganda, 1990 1995 2000 2005 2010 2015 human development is to human welfare. despite average annual per capita GDP But progress in human development has growth of 3.9 percent in the past decade, Primary school completion rate lagged behind that in reducing income child mortality is stagnating—and only Percent 100 poverty (�gure 1). The world as a whole is partly due to the AIDS epidemic.2 on track to achieve the �rst goal—reducing Because growth alone will not be enough by half the proportion of people living on to reach the goals, the international com- 90 less than $1 a day—thanks mainly to rapid munity has committed itself—in a series of economic growth in India and China, where recent meetings in Monterrey, Doha, and 80 many of the world’s poor live.1 But the Johannesburg—to greater resource trans- 70 1990 1995 2000 2005 2010 2015 BOX 1 The eight Millennium Development Goals Ratio of girls to boys in primary and secondary school With starting points in 1990, each goal is to be 5. Improve maternal health Girls as a percent of boys reached by 2015: 100 Reduce by three-quarters the maternal mortality 1. Eradicate extreme poverty and hunger ratio. 95 Halve the proportion of people living on less 6. Combat HIV/AIDS, malaria, and other diseases than one dollar a day. Reverse the spread of HIV/AIDS. 90 Halve the proportion of people who suffer from 7. Ensure environmental sustainability hunger. Integrate sustainable development into country 85 2. Achieve universal primary education policies and reverse loss of environmental Ensure that boys and girls alike complete resources. 80 primary schooling. Halve the proportion of people without access 1990 1995 2000 2005 3. Promote gender equality and empower to potable water. women Signi�cantly improve the lives of at least 100 mil- Under-five mortality rate lion slum dwellers. Eliminate gender disparity at all levels of educa- Deaths per 1,000 live births tion. 8. Develop a global partnership for 100 development 4. Reduce child mortality 80 Reduce by two-thirds the under-�ve mortality Raise of�cial development assistance. rate. Expand market access. 60 Three points about the Millennium Development Goals: First, to be enduring, success in reaching the goals 40 must be based on systemwide reforms to support progress. Second, focusing on these outcomes does not 20 imply focusing on education and health services alone. Health and education outcomes depend on too many other factors for that to work—everything from parents’ knowledge and behavior, to the ease and safety of 0 reaching a health clinic or school, or the technology available for producing outcomes (see crate 1.1).Third, in 1990 1995 2000 2005 2010 2015 countries that have already achieved universal primary completion or low infant and maternal mortality rates, Note: Blue line is the trend line to reach the the spirit of the Millennium Development Goals—time-bound, outcome-based targets to focus strategies— Millenium Development Goal. The red line remains important. shows the actual progress to date. Source: www.developmentgoals.org. (c) The International Bank for Reconstruction and Development / The World Bank Overview 3 Table 1 Economic growth alone is not enough to reach all the Millennium Development Goals People living on less than $1 a day Primary school completion rate Under-�ve mortality Annual average Target 2015 growth Target 2015 growth Target 2015 growth GDP per capita (percent) alone (percent) (percent) alone (percent) (per 1,000 alone (per growth births) 1,000 births) 2000–2015* (percent per year) East Asia 5.4 14 4 100 100 19 26 Europe and Central Asia 3.6 1 1 100 100 15 26 Latin American and the Caribbean 1.8 8 8 100 95 17 30 Middle East and North Africa 1.4 1 1 100 96 25 41 South Asia 3.8 22 15 100 99 43 69 Africa 1.2 24 35 100 56 59 151 *GDP growth projections from World Bank (2003a). Note: Elasticity assumed between growth and poverty is –1.5; primary completion rate is 0.62; under-�ve mortality is –0.48. Sources: World Bank (2003a), Devarajan (2002). fers by developed countries and better poli- Services are failing poor people cies and institutions in developing coun- in four ways tries. The level of resource transfers is dif�- How do we know that these services are fail- cult to calculate precisely. Some estimates ing poor people? First, while governments are converging around a �gure of $40 bil- devote about a third of their budgets to lion to $60 billion a year in additional for- health and education, they spend very little of eign aid—so long as the money is accompa- it on poor people—that is, on the services nied by policy and institutional reforms to poor people need to improve their health and enhance the productivity of domestic and education. Public spending on health and external resources.3 education is typically enjoyed by the non- Focusing on the human development poor (�gure 2). In Nepal 46 percent of educa- goals, this Report describes the reforms in tion spending accrues to the richest �fth, only services needed to achieve them. Ensuring 11 percent to the poorest. In India the richest basic health and education outcomes is the �fth receives three times the curative health responsibility of the state (box 2). But many care subsidy of the poorest �fth.4 Even governments are falling short on their oblig- though clean water is critical to health out- ation, especially to poor people. In Armenia comes, in Morocco only 11 percent of the and Cambodia, child mortality rates for the poorest �fth of the population has access to poorest �fth of the population are two to three times those for the richest �fth. Only about 60 percent of the adolescents in the BOX 2 Services—a public responsibility poorest �fth of the population in the Arab By �nancing, providing, or regulating the rights.The Universal Declaration of Human Republic of Egypt and Peru have completed services that contribute to health and edu- Rights asserts an individual’s right to “a stan- primary school, while all those from the cation outcomes, governments around the dard of living adequate for the health and richest �fth have. world demonstrate their responsibility for well-being of himself and of his family, To meet this responsibility, governments the health and education of their people. including . . . medical care . . . [and a right to Why? First, these services are replete with education that is] . . . free, at least in the ele- and citizens need to make the services that market failures—with externalities, as when mentary and fundamental stages.� No mat- contribute to health and education—water, an infected child spreads a disease to play- ter how daunting the problems of delivery sanitation, energy, transport, health, and edu- mates or a farmer bene�ts from a may be, the public sector cannot walk away cation—work for poor people. Too often, neighbor’s ability to read. So the private sec- from health and education.The challenge is tor, left to its devices, will not achieve the to see how the government—in collabora- these services are failing. Sometimes, they are level of health and education that society tion with the private sector, communities, failing everybody—except the rich, who can desires. Second, basic health and basic edu- and outside partners—can meet this funda- opt out of the public system. But at other cation are considered fundamental human mental responsibility. times, they are clearly failing poor people. (c) The International Bank for Reconstruction and Development / The World Bank 4 WORLD DEVELOPMENT REPORT 2004 Figure 2 More public spending for the rich than for the poor Share of public spending that accrues to the richest and poorest �fths All health spending Primary health All education spending Primary education Guinea Nepal 1994 1996 Percent richest Armenia Kosovo quintile 1999 2000 Percent poorest Ecuador Nicaragua quintile 1998 1998 India Cambodia 1995/96 1996/97 ˆ d’Ivoire Cote Brazil 1997 1995 (NE&SE) Madagascar Morocco 1993 1998/99 Bangladesh Mexico 2000 1996 Bulgaria Kenya 1995 1992 Costa Rica Romania 1992 1994 0 20 40 0 20 40 0 20 40 0 20 40 Percent Percent Percent Percent Source: Compiled from various sources by World Bank staff. safe water, while everybody in the richest �fth By no means do all frontline service does (�gure 3). providers behave this way. Many, often the Second, even when public spending can majority, are driven by an intrinsic motivation be reallocated toward poor people—say, by to serve. Be it through professional pride or a shifting to primary schools and clinics—the genuine commitment to help poor people (or money does not always reach the frontline both), many teachers and health workers service provider. In the early 1990s in deliver timely, ef�cient, and courteous ser- Uganda the share of nonsalary spending on vices, often in dif�cult circumstances— primary education that actually reached collapsing buildings, overflowing latrines— primary schools was 13 percent. This was and with few resources—clinics without the average: poorer schools received well drugs, classes without textbooks.8 The chal- below the average.5 lenge is to reinforce this experience—to repli- Third, even if this share is increased—as cate the professional ethics, intrinsic motiva- the Ugandans have done—teachers must be tion, and other incentives of these providers present and effective at their jobs, just as doc- in the rest of the service work force. tors and nurses must provide the care that The fourth way services fail poor people is patients need. But they are often mired in a the lack of demand. Poor people often don’t system where the incentives for effective ser- send their children to school or take them to a vice delivery are weak, wages may not be clinic. In Bolivia 60 percent of the children paid, corruption is rife, and political patron- who died before age �ve had not seen a for- age is a way of life. Highly trained doctors sel- mal provider during the illness culminating dom wish to serve in remote rural areas. in their death. Sometimes the reason is the Since those who do serve there are rarely poor quality of the service—missing materi- monitored, the penalties for not being at als, absent workers, abusive treatment. At work are low. A survey of primary health care other times it is because they are poor. Even facilities in Bangladesh found the absentee when the services are free, many poor rural rate among doctors to be 74 percent.6 When families cannot afford the time it takes to present, some service providers treat poor travel the nearly 8 kilometers to the nearest people badly. “They treat us like animals,� primary school in Mali or the 23 kilometers says a patient in West Africa.7 to the nearest medical facility in Chad.9 (c) The International Bank for Reconstruction and Development / The World Bank Overview 5 Weak demand can also be due to cultural Figure 3 Water, water everywhere, nor any drop to drink factors, notably gender. Some parents refuse Percent of households who use an improved water to send their daughters to school. Husbands source (selected countries) have been known to prevent their wives from Poorest Richest going to clinics—even for deliveries. And the fifth fifth social distance between poor people and ser- Ethiopia 2000 vice providers (70 percent of nurses and mid- Morocco 1995 wives in rural Niger had been raised in the Guinea 1999 city) is often a deterrent. Cambodia 2000 Alternative service delivery Kazakhstan 1999 arrangements Nicaragua 1998 Ensuring access to basic services such as health, education, water, energy, and sanitation Indonesia 1997 is a public responsibility today, but it has not Tanzania 1999 always been. Nor do governments discharge Philippines 1998 this responsibility solely through central- government provision. Throughout history Brazil 1996 and around the world, societies have tried dif- India 1998–99 ferent arrangements—with mixed results. Uzbekistan 1996 • Some governments contract services 0 20 40 60 80 100 out—to the private sector, to NGOs, Source: World Bank staff. even to other public agencies. In the aftermath of a civil war Cambodia in- troduced two forms of contracting for government delivery of infrastructure in the delivery of primary health care South Africa improved service provision (“contracting out� whole services and in a short time.11 But decentralizing “contracting in� some services). Ran- social assistance in Romania weakened domly assigning the arrangements the ability and incentives of local coun- across 12 districts (to avoid systematic cils to deliver cash transfers to the poor.12 bias), it found that health indicators, as The program is now being recentralized. well as use by the poor, increased most in • Responsibility is sometimes transferred to the districts contracting out.10 Whether communities—or to the clients themselves. this can be scaled up beyond 12 districts El Salvador’s Community-Managed Schools in Cambodia is worth exploring. Program (Educo) gives parents’ associations • Governments also sell concessions to the the right to hire and �re teachers. That, plus private sector—in water, transport, elec- the monthly visits to the schools by the par- tricity—with some very good and some ents’ associations, has reduced teacher—and very bad results. Privatizing water in student—absenteeism, improving student Cartagena, Colombia, improved services performance. and access for the poor. A similar sale in • Still other programs transfer resources and Tucuman, Argentina, led to riots in the responsibility to the household. Mexico’s streets and a reversal of the concession. Education, Health, and Nutrition Program • Some societies transfer responsibility (Progresa) gives cash to families if their (for �nancing, provision, and regula- children are enrolled in school and they tion) to lower tiers of government. regularly visit a clinic. Numerous evalua- Again, the record has varied—with tions of the program show consistently potentially weaker capacity and greater that it increased school enrollment (eight political patronage at the local level and percentage points for girls and �ve for the reduced scope for redistribution boys at the secondary level) and improved sometimes outweighing the bene�ts children’s health (illness among young from greater local participation. Local- children fell 20 percent).13 (c) The International Bank for Reconstruction and Development / The World Bank 6 WORLD DEVELOPMENT REPORT 2004 The framework of relationships— Consider the �rst of the two relationships between clients, providers, along the long route—the link between poor people and policymakers or politicians (�g- and policymakers ure 4). Poor people are citizens. In principle, To help understand the variety of experiences they contribute to de�ning society’s collective with traditional and alternative service deliv- objectives, and they try to control public ery arrangements, the service delivery chain action to achieve those objectives. In practice, can be unbundled into three sets of actors, this does not always work. Either they are and the relationships between them exam- excluded from the formulation of collective ined (�gure 4). Poor people—as patients in objectives or they cannot influence public clinics, students in schools, travelers on buses, action because of weaknesses in the electoral consumers of water—are the clients of ser- system. Free public services and “no-show� vices. They have a relationship with the front- jobs are handed out as political patronage, line providers, with schoolteachers, doctors, with poor people rarely the bene�ciaries. bus drivers, water companies. Poor people Even if poor people can reach the policy- have a similar relationship when they buy maker, services will not improve unless the something in the market, such as a sandwich policymaker can ensure that the service (or a samosa, a salteña, a shoo-mai). In a provider will deliver services to them. In competitive-market transaction, they get the Cambodia, policymakers were able to specify “service� because they can hold the provider the services required to the NGOs with accountable. That is, the consumer pays the whom they contracted. But for many ser- provider directly; he can observe whether or vices, such as student learning or curative not he has received the sandwich; and if he is care, the policymaker may not be able to dissatis�ed, he has power over the provider specify the nature of the service, much less with repeat business or, in the case of fraud, impose penalties for underperformance of with legal or social sanctions. the contract. Teacher and health-worker For the services considered here—such as absenteeism is often the result. health, education, water, electricity, and Given the weaknesses in the long route of sanitation—there is no direct accountability accountability, service outcomes can be of the provider to the consumer. Why not? For improved by strengthening the short route— various good reasons, society has decided that by increasing the client’s power over the service will be provided not through a providers. School voucher schemes (Colom- market transaction but through the govern- bia’s PACES) or scholarships (Bangladesh’s ment taking responsibility (see box 2). That is, Female Secondary School Assistance Pro- through the “long route� of accountability— gram, in which schools receive a grant based by clients as citizens influencing policymak- on the number of girls they enroll) enable ers, and policymakers influencing providers. clients to exert influence over providers When the relationships along this long route through choice. El Salvador’s Educo program break down, service delivery fails (absentee and Guinea’s revolving drug scheme (where teachers, leaking water pipes) and human co-payments inspired villagers to stop theft) development outcomes are poor. are ways for client participation to improve service provision.14 Turn now to a closer look at the individual Figure 4 The framework of accountability relationships relationships in the service delivery chain— why they break down, how they can be Policymakers strengthened. Citizens and politicians/ policymakers—stronger voice Poor citizens have little clout with politicians. In some countries the citizenry has only a Poor people Providers weak hold on politicians. Even if there is a well-functioning electoral system, poor peo- (c) The International Bank for Reconstruction and Development / The World Bank Overview 7 ple may not be able to influence politicians Figure 5 It paid to vote for PRI PRONASOL expenditures according to party in municipal about public services: they may not be well government informed about the quality of public services Average expenditures per capita (and politicians know this); they may vote (real 1995 pesos) along ethnic or ideological lines, placing less 400 weight on public services when evaluating PRI politicians; or they may not believe the candi- 300 dates who promise better public services— PRD because their term in of�ce is too short to Other 200 deliver on the promise—and they may vote instead for candidates who provide ready PAN 100 cash and jobs. As a result, public services often become 0 the currency of political patronage and clien- 1989 1990 1991 1992 1993 1994 telism. Politicians give “phantom� jobs to Note: PRI = Instituational Revolutionary Party; PRD = Party of the teachers and doctors. They build free public Democratic Revolution; schools and clinics in areas where their sup- PAN = National Action Party. Source: Estévez, Magaloni, and Diaz-Cayeros (2002). porters live. Former Boston mayor James Curley strengthened his political base by con- centrating public services in the Irish mortality dramatically, and achieved nearly Catholic areas while denying them to the universal primary enrollment. To be sure, in Protestants, who eventually moved to the China, cases during the earliest phase of the suburbs.15 outbreak of severe acute respiratory syn- In 1989 Mexico introduced PRONASOL drome in 2002 were not openly reported— (Programa Nacional de Solidaridad, or thus making its further spread almost National Solidarity Program), a poverty alle- inevitable. And Cubans, who had high levels viation program that spent 1.2 percent of of health and education in the 1950s, remain GDP annually on water, electricity, nutrition, poor on other dimensions.17 and education construction in poor commu- The lesson seems to be that the citizen- nities. Assessments of the six-year program policymaker link is working either when citi- found that it reduced poverty by only about zens can hold policymakers accountable for 3 percent. Had the budget been distributed public services that bene�t the poor or when to maximize its impact on poverty, the the policymaker cares about the health and expected decline would have been 64 per- education of poor people. These politics are cent. It would have been 13 percent even “pro-poor.� with an untargeted, universal proportional What can be done when the politics are transfer to the whole population. The reason not pro-poor? Societies can still introduce Policymakers becomes apparent when one examines the various intermediate elements to make pub- political af�liation of communities that lic institutions more accountable. Participa- received PRONASOL spending. Municipali- tory budgeting in Porto Allegre, Brazil, ties dominated by the Institutional Revolu- started as a means for the citizens to partici- tionary Party (PRI), the party in power, pate in budget formulation and then to hold received signi�cantly higher per capita trans- the municipal government accountable for Poor people fers than those voting for another party (�g- executing the budget. ure 5).16 Perhaps the most powerful means of Just as a well-functioning democracy does increasing the voice of poor citizens in poli- not guarantee that poor people will bene�t cymaking is better information. When the from public services, some one-party states government of Uganda learned that only 13 get good health and education outcomes— percent of recurrent spending for primary even among the poor. Cuba has among the education was arriving in primary schools, it best social indicators in Latin America—at a launched a monthly newspaper campaign on much lower income than its peers, such as the transfer of funds. That campaign galva- Chile and Costa Rica. China reduced infant nized the populace, inducing the government (c) The International Bank for Reconstruction and Development / The World Bank 8 WORLD DEVELOPMENT REPORT 2004 to increase the share going to primary schools control over providers, and health and educa- (now over 80 percent) and compelling school tion services collapsed. principals to post the entire budget on the Solving the problem requires mentally, schoolroom door. and sometimes physically, separating the The media can do much to disseminate policymaker from the provider—and think- information about public services. Higher ing of the relationship between the two as a newspaper circulation in Indian districts is compact. The provider agrees to deliver a associated with better local-government per- service, in return for being rewarded or formance in distributing food and drought penalized depending on performance. The relief.18 The more people who can read, the compact may be an explicit contract with a stronger the influence of the media. In Ker- private or nonpro�t organization—or ala, India, this led to a virtuous cycle of liter- between tiers of government, as in Johan- acy leading to better public services, which nesburg, South Africa.21 Or it could be raised literacy even more.19 implicit, as in the employment agreements But information is not enough. People of civil servants. must also have the legal, political, and eco- Separating the policymaker from the nomic means to press demands against the provider is not easy, for those who bene�t government. Most citizens in Uttar Pradesh, from the lack of separation may resist it. India, know that government services are Teachers’ unions in Uttar Pradesh, India, dismal, and know that everyone else knows blocked an attempt to put teacher hiring, that—and yet most do not feel free to �ring, and attendance under the control of complain.20 the village panchayat. On the other hand, health professionals in Brazil participated in Policymakers and providers— a national coalition that prepared the plan stronger compacts for health reforms and municipal health Strengthening poor people’s voice can councils.22 The separation usually happens make policymakers want to improve ser- because of a �scal crisis (Johannesburg), a vices for the poor. But they still may not be major political change (decentralization in able to. Well-intentioned policymakers Latin America), or a legacy of history (pub- often cannot offer the incentives and do the lic regulation of water providers in the monitoring to ensure that providers serve Netherlands). the poor. The absenteeism of teachers, the Even with a separation of policymaker rude treatment of patients, and the siphon- and provider, the compacts cannot be too ing of pharmaceuticals are symptoms of explicit. It is dif�cult to specify precisely what the problem. the schoolteacher should do at every point in Even in the private sector, where the the day. Too much speci�city can lead to incentives presumably are better aligned, per- inflexibility. Parisian taxi drivers, to make a Policymakers formance is not much better—for the same point about excessive regulations, sometimes reasons that private markets are not the solu- meticulously follow the rules in the Code de la tion to these problems in the �rst place. Pri- route—slowing traf�c in the French capital to vate providers fail to reach the very poor. a snail’s pace.23 Weak regulation leads to poor-quality health Since the contract cannot be fully speci- services in India’s private sector. Ineffectively �ed, policymakers look to other means of Providers privatizing water incites riots in the streets of eliciting pro-poor services from providers. Cochabamba. One way is to choose providers who have an In the former Soviet Union, state and intrinsic motivation to serve the poor. A party control over providers ensured compli- study of faith-based health care providers in ance with delivery norms for free services. Uganda estimates that they work for 28 per- Services worked, and levels of health status, cent less than government and private for- particularly for the poorer Central Asian pro�t staff, and yet provide a signi�cantly republics, were much higher than for other higher quality of care than the public sec- countries at their level of income. But the tor.24 Another way is to increase incentives breakup of the Soviet Union weakened state to serve the poor or work in underserved (c) The International Bank for Reconstruction and Development / The World Bank Overview 9 areas. But one study of Indonesia shows These are but symptoms of the larger prob- that it would require multiples of current lem: many service delivery arrangements pay levels to get doctors to live in West neglect the role of clients, especially poor Papua, for instance (where the vacancy rate clients, in making services work better. is 60 percent).25 A third way is to solicit bids Clients can play two roles in strengthening for services and use the competition in the service delivery. First, for many services, bidding process to monitor and discipline clients can help tailor the service to their providers. Many water concessions are man- needs, since the actual mix cannot be speci�ed aged this way. A recent innovation in Mad- in advance. In some parts of Pakistan, girls are hya Pradesh, India, allows NGOs to compete more likely to attend school if there is a female for concessions to primary schools, with teacher. The construction of separate latrines payments conditional on higher test scores for girls has had a strong effect on girls’ enroll- based on independent measurement. ment in primary schools. When the opening As with the citizen-politician relationship, hours of health clinics are more convenient a critical element in the policymaker- for farmers, visits increase. Second, clients can provider relationship is information. The be effective monitors of providers, since they policymaker can specify a contract based only are at the point of service delivery. The major on what he can observe—on what informa- bene�t of Educo came from the weekly visits tion is available. There has to be a method for of the community education association to monitoring providers and for having that schools. Each additional visit reduced student information reach the policymaker. New absenteeism (due to teacher absenteeism) by 3 technologies, including e-government, can percent.28 make this easier.26 How can the role of clients in revealing So can some ingenious methods using demand and monitoring providers be human beings. When Ceará, Brazil, hired a strengthened? By increasing poor people’s cadre of district health workers, the govern- choice and participation in service delivery. ment sent their names to the applicants who When clients are given a choice among ser- were not selected, inviting them to report any vice providers, they reveal their demand by problems with service in the health clinics. “voting with their feet.� Female patients who More fundamentally, these output-based feel more comfortable with female doctors incentive schemes require rigorous program can go to one. The competition created by evaluation, so that the policymaker knows client choice also disciplines providers. A and understands what is working and what doctor may refuse to treat lower-caste isn’t. Evaluation-based information, impor- patients, but if he is paid by the number of tant not only for monitoring providers, also patients seen, he will be concerned when the enables the rest of the world to learn about waiting room is empty. Reimbursing schools service delivery. based on the number of students (or female students) they enroll creates implicit compe- Clients and providers—more tition among schools for students, increasing Poor people Providers choices, more participation students’ choice. Given the dif�culties in strengthening the long School voucher programs—as in Ban- route of accountability, improving the short gladesh, Chile, Colombia, Côte d’Ivoire, and route—the client-provider relationship— Czech Republic—are explicitly aimed at deserves more consideration. There is no improving education quality by increasing question that this relationship is broken for parents’ choices. The evidence on these hundreds of millions of poor people. Voices schemes is mixed, however. They seem to of the Poor and other surveys point to the have improved student performance among helplessness that poor people feel before some groups. But the effects on the poor are providers—nurses hitting mothers during ambiguous because universal voucher childbirth, doctors refusing to treat patients of schemes tend to increase sorting—with a lower caste.27 Unlike most private providers, richer students concentrating in the private public water companies funded through bud- schools.29 When the voucher is restricted to getary transfers often ignore their customers. poor or disadvantaged groups, the effects are (c) The International Bank for Reconstruction and Development / The World Bank 10 WORLD DEVELOPMENT REPORT 2004 better.30 The Colombia program showed between policymaker and provider. In their lower repetition rates and higher perfor- zeal to get services to the poor, donors often mance on standardized tests for students par- bypass one or more of these relationships. ticipating in the scheme—with the effect for The typical mode of delivering aid—a girls higher than that for boys.31 Even in net- project—is often implemented by a separate work systems such as urban water provision, unit outside the compact, bypassing the rela- it is possible to give poor communities tionship between policymakers and pro- choice—by allowing the poor to approach viders. The project is typically �nanced by independent providers, introducing flexibil- earmarked funds subject to donor-mandated ity in service standards such as lifeline rates, �duciary requirements. It and other donor and so on. initiatives, including global “funds,� bypass When there is no choice of providers, the citizen-policymaker relationship where increasing poor people’s participation in ser- the budget is concerned. To be sure, when the vice provision—giving them the ability to existing relationship is dysfunctional, it may monitor and discipline the provider, for be necessary to go around it. But the cases example—can achieve similar results. Clients where the bene�ts outweigh the costs are can play the role of monitors since they are probably fewer than imagined. present at the point of service. But they need Recognizing the gap between ends and to have an incentive to monitor. means, some donors and recipients try to use In Bangladesh, thanks to reduced import foreign aid to strengthen, not weaken, the tariffs, households were able to purchase links in the service delivery chain. One tubewells that tapped ground sources—shal- approach is to incorporate donor assistance low aquifers—for drinking water. Unfortu- in the recipient’s budget, shifting to the recip- nately, no one arranged for the monitoring of ient’s accountability system. In Uganda assis- water quality—a public good—so the arsenic tance from Germany, Ireland, the Nether- in the water went undetected. If the stakes are lands, Norway, United Kingdom, and the high enough, communities tackle the prob- World Bank is all part of the country’s bud- lem. When the Zambian government intro- get, the outcome of a coordinated and partic- duced a road fund �nanced by a charge on ipatory process. trucks, truck drivers took turns policing a Another approach is for donors to pool bridge crossing to make sure that overloaded their assistance in a single “pot� and to har- trucks did not cross. Of course such co- monize their �duciary standards around that payments or user fees reduce demand—and of the rest of the government. The sectorwide so should not be used when the demand approach to health, education, transport, and effects outweigh the increase in supply, as in other sectors is a step in this direction. Possi- primary education. But for water, electricity, bly the biggest payoff comes when donors and other services whose bene�ts are enjoyed help generate knowledge—as when donor- mainly by the user, charging for them has the �nanced impact evaluation studies reveal added bene�t of increasing the consumer’s what works and what doesn’t in service deliv- incentive to monitor the provider. Farmers in ery, or when donors pool technical assistance Andhra Pradesh, India, are �nding that, when resources at the retail level, as in the multi- they pay for their water, the irrigation depart- donor Water and Sanitation Program. ment becomes more accountable to them. In Knowledge is essential to scaling up service the words of one farmer, “We will never allow delivery. Although it emerges locally, it is a the government to again give us free water.�32 global public good—precisely what aid is designed to �nance. Donors and recipients— strengthening accountability, What not to do not undermining it The picture painted so far of the dif�culties in Improving service outcomes for poor people government-led service delivery may lead requires strengthening the three relationships some to conclude that government should in the chain—between client and provider, give up and leave everything to the private between citizen and policymaker, and sector. That would be wrong. If individuals (c) The International Bank for Reconstruction and Development / The World Bank Overview 11 are left to their own devices, they will not Figure 6 Increased public spending is not enough provide levels of education and health that Under-five mortality rate, 2000* they collectively desire (see box 2). Not only 150 is this true in theory, but in practice no 100 country has achieved signi�cant improve- ment in child mortality and primary educa- 50 tion without government involvement. Fur- thermore, as mentioned earlier, private 0 sector or NGO participation in health, edu- –50 cation, and infrastructure is not without problems—especially in reaching poor peo- –100 ple. The extreme position is clearly not desirable. –150 –150 –100 –50 0 50 100 150 Some aid donors take a variant of the Per capita public spending on health, 1990s average* “leave-everything-to-the-private-sector� posi- *Public spending and child mortality are given as the percent tion. If government services are performing deviation from rate predicted by GDP per capita. so badly, they say, why give more aid to those Note: For the regression line shown, the coef�cient is –0.148 and the t-statistic is 1.45. governments? That would be equally wrong. Source: GDP per capita and public spending data, World Development Indicators database; under-5 mortality, UNICEF. There is now substantial research showing that aid is productive in countries with good policies and institutions, and those policies the non-poor, much of it fails to reach the and institutions have recently been improv- frontline service provider, and service ing.33 The reforms detailed in this Report providers face weak incentives to deliver ser- (aimed at recipient countries and aid agen- vices effectively. cies) can make aid even more productive. Linked to the “simply-increase-public- When policies and institutions are improv- spending� approach is one that advocates for ing, aid should increase, not decrease, to real- more foreign aid without accompanying ize the mutually shared objective of poverty measures to improve the productivity of for- reduction, as speci�ed in the Millennium eign aid. This can be just as misleading—and Development Goals. not just for the same reasons that simply At the same time, simply increasing public increasing public expenditure is misleading. spending—without seeking improvements in Sometimes the modes of delivering foreign the ef�ciency of that spending—is unlikely to aid, by undermining rather than strengthen- reap substantial bene�ts. The productivity of ing service delivery in the recipient country, public spending varies enormously across can reduce the productivity of public spend- countries. Ethiopia and Malawi spend ing in the medium run. roughly the same amount per person on pri- Finally, when faced with disappointing mary education—with very different out- health and education outcomes, especially comes. Peru and Thailand spend vastly differ- for poor people, it is tempting to recom- ent amounts—with similar outcomes. mend a technical solution that addresses the On average, the relationship between proximate cause of the problem. Why not public spending on health and education give vitamin A supplements, de-worm and the outcomes is weak or nonexistent. A schoolchildren, and train teachers better? simple scatter plot of spending and out- Why not develop a “minimum package� of comes shows a clear line with a signi�cant health interventions for everybody? slope—because richer countries spend more Although each intervention is valuable, rec- on health and education and have better out- ommending them alone will not address comes. But controlling for the effect of per the fundamental institutional problems capita income, the relationship between that precluded their adoption in the �rst public spending on health and under-�ve place.34 Lack of knowledge about the right mortality rates is not statistically signi�cant technical solution is probably not the bind- (�gure 6). That is not surprising: most public ing constraint. What is needed is a set of spending on health and education goes to institutional arrangements that will give (c) The International Bank for Reconstruction and Development / The World Bank 12 WORLD DEVELOPMENT REPORT 2004 policymakers, providers, and citizens the Pro-poor or clientelist politics? incentives to adopt the solution and adapt it How much is the political system in the to local conditions. country geared toward pro-poor public ser- vices—and how much does it suffer from What can be done? clientelist politics and corruption? This is the The varied experience with traditional and most dif�cult dimension for an outside actor, innovative modes of service delivery clearly such as a donor, to address: the recipient of shows that no single solution �ts all services the advice may also be the source of the prob- in all countries. The framework of account- lem. And politics do not change overnight. ability relationships explains why. In differ- Even so, at least three sets of policy instru- ent sectors and countries, different relation- ments can be deployed where the politics are ships need strengthening. In education the more clientelist than pro-poor. biggest payoff may come from strengthen- ing the client-provider link, as with vouch- • First is choosing the level of government ers in Colombia or scholarships for girls in responsible for the service. Countries dif- Bangladesh. But that may not be so in fer in the patronage politics and capabili- immunization campaigns. ties of different tiers of government— Furthermore, poor people are often and this should inform the service trapped in a system of dysfunctional service- delivery arrangement. delivery relationships. Making just one link • Second, if politicians are likely to capture more effective may not be enough—it may the rents from free public services and even be counterproductive—if there are seri- distribute them to their clients, an ous problems elsewhere in the service deliv- arrangement that reduces the rents may ery chain. In water or curative health care, leave the poor better off. This might tightening the policymaker-provider link include transparent and publicly known could make providers respond more to the rules for allocation, such as per-student demands of their superiors—and less to grants to schools, or conditional transfers their poor clients. Relying on user groups, to households, as in Progresa. In some often generously funded by donors, may cases it may include fees to reduce the inhibit the development of genuinely demo- value of the politicians’ distribution deci- cratic local governments. Finally, countries, sions. India’s power sector was nationally and regions within countries, vary enor- owned and run because it was a network mously in the conditions that make service (and therefore not amenable to head-to- innovations work. A failed state mired in head competition). But the huge rents conflict will be overstretched in resources from providing subsidized electricity and institutional capacity, and able to man- have been diverted to people who are not age only certain interventions. Countries poor—all within a parliamentary democ- with high prevalence of HIV/AIDS will racy. Reducing those rents by raising require short- and long-term adaptations of power tariffs or having the private sector the service delivery systems. provide electricity, even if it violates the Does this mean there are no general principles of equity—they are already lessons about making services work for violated in the existing system—may be yes poor people? No. The experience with ser- the only way of improving electricity ser- vice delivery, viewed through the lens of vices to the poor. this Report, suggests a constellation of • Third, better information—that makes solutions, each matching various charac- citizens more aware of the money allo- Pro-poor? teristics of the service and the country or cated to their services, the actual condi- region. While no one size �ts all, perhaps tions of services, and the behavior of eight sizes do. Even eight may be too few, policymakers and providers—can be a which is why some of the “sizes� are powerful force in overcoming clientelist no adjustable, like waistbands. politics. The role of a free and vibrant The eight sizes can be arrived at by press and improving the level of public answering a series of questions. discourse cannot be overstated. (c) The International Bank for Reconstruction and Development / The World Bank Overview 13 Homogeneous or heterogeneous status or parental involvement. More easily clients? monitored are immunizations and clean yes The answer to this question depends on the latrines—all measurable by a quantitative, service. Students with disabilities have special observable indicator. Homogeneous? needs for quality education but not for Of course it depends on who is doing the immunization. Heterogeneity is also de�ned monitoring. Parents can observe whether the teacher is in attendance, and what their chil- no by regional or community preferences. yes Whether a girl goes to school may depend on dren are learning, more easily than some cen- whether there are separate latrines for boys tral education authority. Better management and girls. If that depends on local preferences, information systems and e-government can the village should have a say in design. Previ- make certain services easier to monitor. And monitoring costs can be reduced by judicious Pro-poor? ously homogeneous societies, such as Sweden and Norway, are changing with increased choice of providers—such as some NGOs, immigration. They are giving more discre- which may be trustworthy without formal tion to local communities in tailoring the monitoring. In short, the dif�culty of moni- no education system to suit the linguistic abili- toring is not �xed: it can vary over time and yes ties of their members. with policies. The more that people differ in their Eight sizes �t all Homogeneous? desires, the greater the bene�ts from decen- tralizing the decision. In the most extreme Now examine different combinations of case—when individual preferences matter— these characteristics, to see which service no the appropriate solution will involve individ- delivery arrangement would be a good �t— ual choices of service (if there is the possibil- and which would be a mis�t (�gure 7). To be ity of competition) and such interventions as sure, none of the characteristics can be easily cash transfers, vouchers, or capitation pay- divided into such clean categories, because ments to schools or medical providers. If countries and services lie on a continuum. there are shared preferences, as in education, Even so, by dividing the salient characteris- or free-rider problems as in sanitation, the tics, and looking at various combinations, the community is the correct locus of decision- “eight sizes �t all� approach can be applied to making. The appropriate policy will then the considerations spelled out earlier. involve local-government decisions in a decentralized setting—or depending on Central government �nancing with con- political realities, community decisions (as tracting (1). In a favorable political context, for social investment funds) and user groups with agreement on what government should (such as parents in school committees). do, an easy-to-monitor service such as immunization could be delivered by the Easy or hard to monitor? public sector, or �nanced by the public sec- Services can be distinguished by the dif�culty tor and contracted out to the private or non- of monitoring service outputs. The dif�culty pro�t sector, as with primary health centers depends on the service and on the institu- in Cambodia.35 Infrastructure services could tional capacity of government to do the mon- be managed by a national utility or provided itoring. At one extreme are the services of by the private sector with regulatory over- teachers in a classroom or doctors in a clinic. sight. Both transactions allow much discretion by Note that the particular con�guration in the provider that cannot be observed easily. A which this arrangement will work is special. doctor has much more discretion in treating In the developed countries there is much a patient than an electrician switching on a discussion of a set of reforms, started in New power grid. And it is dif�cult to know when Zealand, that involve greater use of explicit high-quality teaching or health care is being contracts—either from the government to provided. It may be possible to test students. the private sector, or from central ministries But test scores tell very little about the to the ministries responsible for speci�c ser- teacher’s ability or effort, since they depend at vices. The New Zealand reforms are justi�ed least as much on students’ socioeconomic by a well-established public sector ethos, (c) The International Bank for Reconstruction and Development / The World Bank 14 WORLD DEVELOPMENT REPORT 2004 Figure 7 Eight sizes �t all? tracts are dif�cult to write or enforce—but the politics are pro-poor and clients homo- 1 Central government geneous, the traditional, centralized public yes financing with contracting Easy sector is the appropriate delivery system. to The French education system, which admin- monitor? no 2 isters a uniform service centrally, is one of yes Central government provision the best examples.37 But too many countries fall into the trap of thinking that just Homogeneous ? because the service is dif�cult to monitor, it 3 Local government financing with must be delivered by the government. When no yes contracting students are heterogeneous, when the poli- Easy yes to tics of the country are not geared toward monitor? poor people, government control of the no 4 Local government provision education system—with no participation by students, parents, or local communities— Pro-poor? can leave the poor worse off. 5 Client power—experiment with yes contracts Local government �nancing with contracting Easy to (3). With heterogeneous preferences, local no monitor? governments should be involved in services. no 6 Client power—experiment with yes When local politics are pro-poor (but self-monitoring providers national politics aren’t), local governments Homogeneous ? could be more reliable �nanciers of services, 7 Client power—experiment with and vice versa. Easily monitored services such no yes community control, vouchers as water or electricity can be contracted out to Easy to public or private utilities, as in Johannesburg. monitor? no 8 Client power—imitate market Local government (or deconcentrated cen- tral government) provision (4). For dif�cult- to-monitor services, such as education (for quality), management responsibility might reasonable management information sys- be ceded to parent groups when the politics tems, and supporting institutions, including are conducive, as in the Educo program. legal systems, to allow contract enforcement. Giving clients a choice through vouchers These features increase the “monitorability� enables them to express their heterogeneous of certain services by reducing the gap preferences. And the competition created by between contracted and realized outcomes. clients having a choice may improve service These preconditions do not exist in many quality—as with water vouchers in Chile or developing countries, so the template of these sanitation vouchers in Bangladesh. reforms cannot be used mechanically.36 If there is no good legal system and the civil ser- Client power (5, 6, 7, 8). When publicly vice is subject to bribes (a form of clientelist �nanced services are subject to capture—the politics), private sector contracts might be a politics are not pro-poor—the best thing to major source of corruption. In these coun- do is to strengthen the client’s power as tries, government should perhaps be even much as possible. But that can be dif�cult. more output-oriented—not as a means of Even means-tested voucher schemes or sub- tweaking a well-functioning system but as a sidies could be diverted to the non-poor. way of getting the system to provide much Transparent, rule-based programs, such as greater improvements in services and gener- Progresa in Mexico, are needed to make it ating new information. dif�cult to hide middle-class capture. In services such as water and electricity, Central government provision (2). When the governments intervene to regulate monopoly service is dif�cult to monitor—explicit con- providers and protect the poor—and not (c) The International Bank for Reconstruction and Development / The World Bank Overview 15 because there are signi�cant externalities. So In Senegal—a stable democracy—the reforms separating the policymaker from the provider, in education, including those that strengthen and making the provider accountable to the client-provider links, would go through the client through prices, can strengthen client government (to strengthen the policymaker- power and lead to better results. Poor people provider links as well). In the Democratic can be protected from high prices if charges Republic of Congo—where conflict has sig- rise with use (with an initial, free amount). ni�cantly weakened the state—ways should Allowing small, independent water providers be found to empower communities to to compete with the local monopoly can also improve education services—even if it means discipline provision and keep prices down. bypassing government ministries in the short But prices—without accompanying subsi- to medium term. Social funds and commu- dies or transfers to poor people—cannot be nity-driven development are examples. They used to strengthen client power in education can be effective in improving service out- because of the externalities in primary educa- comes, but concerns about their sustainability tion. A market-based allocation would not be and scalability—and whether they crowd out in society’s interest. The same applies to the growth of local government capacity— health services with externalities, such as should not be overlooked. immunization. In curative health care, the asymmetry of information between client History. The country’s history can also have and provider makes strengthening client a bearing on which service delivery arrange- power problematic. Better information on ments are likely to succeed. Until the 19th preventive care or on how to choose medical century, the education systems of Britain providers (possibly disseminated by non- and France were private and the church was pro�t organizations) can ameliorate the the dominant provider. The government problem. In extreme cases, it may be that only had an incentive to develop an oversight community groups or altruistic nonpro�ts mechanism to ensure that the schools can effectively provide these services to poor taught more than just religion. That proved people.38 valuable when education was nationalized These service delivery arrangements repre- in these countries: the systems continued to sent efforts to balance problems with the long run with strong regulatory oversight. route of accountability (clientelist politics, Water providers in the Netherlands started hard-to-monitor services) with the short as private companies, making the concept of route. The reason societies choose the long water as an economic good, and charging for route is that there are market failures or it, acceptable. When the system was shifted to concerns with equity that make the tradi- municipal ownership, pricing remained. Even tional short route—consumers’ power over if the Dutch never introduce private participa- providers—inadequate. But the “government tion in water, they have achieved the separa- failures� associated with the long route may be tion between policymaker and provider. In so severe that, in some cases, the market solu- sum, a country’s history can generate the tion may actually leave poor people better off. incentives for certain institutions to develop— and those institutions can make the difference Eight sizes �t all with adjustable in whether a particular service arrangement waistbands succeeds or fails. The foregoing simpli�ed scheme captures only part of the story. At least two features are Sectoral service reforms left out. What do these conclusions tell us about the reform agenda in individual sectors? In edu- Failed states. Countries where the state is fail- cation there is a tradeoff between the need ing (often countries in conflict) need service for greater central authority to capture soci- delivery arrangements different from those etywide bene�ts, such as social cohesion, where the state is fairly strong. Primary school and the need for greater local influence completion rates in Senegal and the Democ- because student learning is dif�cult to ratic Republic of Congo are about 40 percent. monitor at the central level. The tradeoff is (c) The International Bank for Reconstruction and Development / The World Bank 16 WORLD DEVELOPMENT REPORT 2004 sharper when the concern is the quality of education rather than the quantity. In BOX 3 Spotlight on “spotlights� Indonesia centralized public delivery of In addition to the usual assortment of boxes and education has enrolled children in schools, examples to illustrate particular points, this but it has been less successful in teaching Report contains 11 “spotlights� that appear them valuable skills. To increase the quality between chapters. Each spotlight describes a particularly important service delivery innova- of education, therefore, reforms should tion or experience.The purpose of these spot- concentrate on increasing the voice and lights is to tell the story behind these participation of clients—but not neglect innovations or experiences, and provide a bal- the importance of central government over- anced perspective on the evidence. sight. In practical terms, this would call for more community management of schools and demand-side subsidies to poor people, ferent from that in education and health, and but with continuing stress on nationally so should be the policy responses. The main determined curricula and certi�cation. reason for government involvement in water Governments intervene in health to con- and energy provision is that those services are trol communicable diseases, protect poor provided through networks, so direct compe- people from impoverishing health expendi- tition is not possible. Governments also inter- tures, and disseminate information about vene to ensure access by poor people to these home-based health and nutrition practices. services. So the role of government is to regu- Each of these activities is different, yet they late and in some cases subsidize production are often provided by the same arrangement, and distribution. There are few advantages to such as a central government public health the government’s providing the service itself, system. They should be differentiated. which explains why the past decade has seen many privatizations, concessions, and the like • Information about hand washing, exclu- in water and energy. sive breastfeeding, and nutrition can be Whether delivered by a private or public delivered (and even �nanced) by NGOs company, the service needs to be regulated. and other groups, delivery that works Who that regulator is will determine service best when reinforced by the community. outcomes. At the very least, when the com- • Outreach services, such as immuniza- pany is public, the regulator should be sepa- tions, can be contracted out but should rate from the provider (when the policy- be publicly �nanced. maker and provider are indistinguishable, • Clinical care is the service the client is making this separation is all the more dif�- least able to monitor, but the case in cult). The situation is worse when water or which government failures might swamp energy is subsidized, because the sizable rents market failures. Where the politics are from this subsidy—the bene�ts of below- extremely pro-rich, even public �nanc- market-rate services—can be captured by ing of these services (with private provi- politicians, who use them to curry favor with sion) can be counterproductive for poor their rich clients rather than the poor. people. The non-poor can capture this Sanitation is different because individuals �nancing, leaving no curative services can offload their refuse onto their neighbors. for the poor—and no room in the bud- So subsidies to individual households will get for public health services. Strength- not solve the collective action problem. ening client power, through either Instead, using community-level subsidies, demand-side subsidies or co-payments, and giving communities the authority to can improve matters for poor people, allocate them, puts the locus of authority even if there is asymmetric information where the external effects of individual between client and provider. behavior can be contained. In the infrastructure sectors—such as Scaling up water, sanitation, transport, and energy—the How can all these reforms be scaled up so rationale for government intervention is dif- that developing countries will have a chance (c) The International Bank for Reconstruction and Development / The World Bank Overview 17 of meeting the Millennium Development countries what works and what doesn’t. They Goals? First, as noted at the beginning of this are global public goods—which might explain Report, additional resources—external and why they are so scarce.39 If these evaluations internal—will be needed to capitalize on are global public goods, the international com- these reforms. Second, these reforms must be munity should �nance them. One possibility embedded in a public sector responsible for would be to protect the 1.5 percent of World ensuring poor people’s access to basic ser- Bank loans that is supposed to be used for vices. This means that the sectoral reforms evaluation (but rarely is), so that this sum— must be linked to ongoing (or nascent) pub- about $300 million a year—could be used to lic sector reforms in such areas as budget administer rigorous evaluations of projects management, decentralization, and public and disseminate the results worldwide. administration reform. It also means that a In addition to creating and disseminating well-functioning public sector is a crucial information, other reforms to improve service underpinning of service delivery reform. In delivery will require careful consideration of the same vein, there should be reform in the particular setting. There is no silver bullet donor practices—such as harmonizing pro- to improve service delivery. It may be known cedures and making more use of budget how to educate a child or stop an infant from support—to strengthen recipient countries’ dying. But institutions are needed that will efforts to improve service outputs. educate a generation of children or reduce Third, a recurring theme in this Report is infant mortality by two-thirds. These do not what information can do—as a stimulant for crop up overnight. Nor will a single institu- public action, as a catalyst for change, and as tional arrangement generate the desired an input for making other reforms work. results. Everything from publicly �nanced Even in the most resistant societies, the cre- central government provision to user- ation and dissemination of information can �nanced community provision can work (or be accelerated. Surveys of the quality of ser- fail to work) in different circumstances. vice delivery conducted by the Public Affairs Rather than prescribe policies or design Centre in Bangalore, India, have increased the optimal institution, this Report public demand for service reform. The sur- describes the incentives that will give rise to veys have been replicated in 24 Indian states. the appropriate institution in a given con- The public expenditure tracking survey in text. Decentralization may not be the opti- Uganda is another example, as is the Probe mal institutional design. But it may give report on India’s education system. local governments the incentives to build Beyond surveys, the widespread and sys- regulatory capacity that, in turn, could make tematic evaluation of service delivery can have water and energy services work better for a profound effect on progress toward the Mil- poor people. NGO service provision might lennium Development Goals. Evaluations be effective in the medium run, as it has based on random assignments, such as Mex- been in education in Bangladesh. But the ico’s Progresa, or other rigorous evaluations incentives it creates for the public sector to give con�dence to policymakers and the pub- stay out of education make it much harder lic that what they are seeing is real. Govern- to scale up or improve quality—as Ban- ments are constantly trying new approaches gladesh is discovering today. Many of these to service delivery. Some of them work. But institutions cut across the public sector— unless there is some systematic evaluation of budgetary institutions, intergovernmental these programs, there is no certainty that they relations, the civil service—which reinforces worked because of the program or for other the notion that service delivery reform reasons. Based on the systematic evaluations should be embedded in the context of public of Progresa, the government has scaled up the sector reform program to encompass 20 percent of the In addition to looking for incentives to Mexican people. generate the appropriate institutions, gov- The bene�ts of systematic program evalua- ernments should be more selective in what tion go beyond the program and the country. they choose to do. The experience with ser- These evaluations tell policymakers in other vice delivery teaches us the importance of (c) The International Bank for Reconstruction and Development / The World Bank 18 WORLD DEVELOPMENT REPORT 2004 implementation. Singapore and Nigeria rise to appropriate institutions, that we (both former British colonies) have similarly need to be more realistic about implemen- designed education systems. But in imple- tation in choosing among options—all mentation, the outcomes, especially for poor imply that these reforms will take time. people, could not be more different. Govern- Even if we know what is to be done, it may ments and donors often overlook implemen- be dif�cult to get it done. Despite the urgent tation dif�culties when designing policies. needs of the world’s poor people, and the There may be bene�ts to having the central many ways services have failed them, quick government administer schools (such as results will be hard to come by. Many of the social cohesion). But the problems with cen- changes involve fundamental shifts in tral provision of a hard-to-monitor activity power—something that cannot happen such as primary education are so great, espe- overnight. Making services work for poor cially among heterogeneous populations, that people requires patience. But that does not the government should rethink its position of mean we should be complacent. Hubert centrally controlled schools. Selectivity is not Lyautey, the French marshal, once asked his just about choosing from the available design gardener how long a tree would take to options—it is about choosing with an eye reach maturity. When the gardener toward options that can be implemented. answered that it would take 100 years, Mar- That there is no silver bullet, that we shal Lyautey replied, “In that case, plant it should be looking for incentives that give this afternoon.� (c) The International Bank for Reconstruction and Development / The World Bank Services can work for poor people but too often they fail 1 With seven other children to care for, Maria’s outcomes. They are often inaccessible or pro- mother, Antonia Souza Lima, explained that she hibitively expensive. But even when accessi- could not afford the time—an hour-and-a-half ble, they are often dysfunctional, extremely chapter walk—or the 40-cent bus fare to take her listless low in technical quality, and unresponsive to baby to the nearest medical post. Maria seemed destined to become one of the 250,000 Brazilian the needs of a diverse clientele. In addition, children who die every year before turning 5. But innovation and evaluation—to �nd ways to in a new effort to cut the devastating infant mor- increase productivity—are rare. tality rate here, a community health worker Many services contribute to improving recently started to walk weekly to the Lima house- hold, bringing oral rehydration formula for Maria human welfare, but this Report focuses on and hygiene advice for her mother, who has a tele- services that contribute directly to improving vision set but no water �lter. Once a month, the health and education outcomes—health ser- 7,240 workers in the Ceará health program enter vices, education services, and such infrastruc- the homes of four million people, the poor major- ture services as water, sanitation, and energy. ity of a state where most people’s incomes are less than $1 a day. Erismar Rodrigues de Lima, a “Services� include what goes on in schools, neighbor of the Limas, listened intently to instruc- clinics, and hospitals and what teachers, tions on �ltering drinking water. “I am the �rst nurses, and doctors do. They also include member of my family to ever receive prenatal how textbooks, drugs, safe water, and elec- care,� said the 22-year-old woman, who is expect- tricity reach poor people, and what informa- ing a baby in June. tion campaigns and cash transfers can do to From the New York Times40 enable poor people to improve outcomes I go to collect water four times a day, in a 20-litre directly. Much of all this is relevant for other clay jar. It’s hard work! . . . I’ve never been to sectors, such as police services, so the Report school as I have to help my mother with her wash- features examples from those sectors as well. ing work so we can earn enough money. I also have to help with the cooking, go to the market to Just how bad can services be? Testimonies buy food, and collect twigs and rubbish for the show that they can be very bad. In Adaboya, cooking �re. Our house doesn’t have a bathroom. I Ghana, poor people say that their “children wash myself in the kitchen once a week, on Sun- must walk four kilometers to attend school day. At the same time I change my clothes and because, while there is a school building in wash the dirty ones. When I need the toilet I have to go down to the river in the gully behind my the village, it sits in disrepair and cannot be house. I usually go with my friends as we’re only used in the rainy season.�42 In Potrero Sula, El supposed to go after dark when people can’t see us. Salvador, villagers complain that “the health In the daytime I use a tin inside the house and post here is useless because there is no doctor empty it out later. If I could alter my life, I would or nurse, and it is only open two days a week really like to go to school and have more clothes. until noon.�43 A common response in a client Elma Kassa, a 13-year-old girl from Addis Ababa, Ethiopia41 survey by women who had given birth at rural health centers in the Mutasa district of Citizens and governments can make services Zimbabwe is that they were hit by staff dur- that contribute to human development work ing delivery.44 better for poor people—and in many cases This chapter illustrates many types of they have. But too often services fail poor failures—inaccessible or unaffordable ser- people. Services are failing because they are vices, and various shortfalls in quality—using falling short of their potential to improve testimonials from poor people, compilations 19 (c) The International Bank for Reconstruction and Development / The World Bank 20 WORLD DEVELOPMENT REPORT 2004 of data from several countries, and in-depth school.46 These countries are not special studies. The chapter also provides examples cases. Worldwide more than 100 million chil- of services that are working for poor people. dren of primary school age are not in pri- Learning from success and understanding the mary school.47 Almost 11 million children, sources of failure require a framework for roughly the population of Greece or Mali, die analysis. Chapters 3 to 6 of the Report present before their �fth birthday.48 and develop that framework; Chapters 7 to Most countries have rich-poor differen- 11 consider options and issues for reform. tials in education or health outcomes. This is Figure 1.1 Child mortality is not necessarily evidence of services failing substantially higher in poor households Outcomes are substantially poor people—there are many determinants Deaths per 1,000 live births worse for poor people of outcomes (see crate 1.1 at the end of this Central African Republic 1994–95 Just how bad are outcomes? Rates of illness chapter).49 Comparing outcomes for richer 200 and death are high—and rates of school and poorer people within countries high- 189 enrollment, completion, and learning are lights two things. First, it shows the absolutely 150 low—especially for poor people (box 1.1). In bad outcomes among the poor—for exam- Poorest fifth Cambodia under-�ve mortality is 147 per ple, in Bolivia 143 children of every 1,000 100 1,000 births among the poorest �fth of the from the poorest quintile died before their population; in Armenia it is 63 (�gure 1.1).45 �fth birthday, and in Niger fewer than 10 per- Richest fifth 50 Many children are unlikely to complete even cent of adolescents from the poorest quintile primary schooling. Among adolescents 15 to completed grade 6. Second, within-country 0 comparisons give a sense of the possible— 19 years old in Egypt, only 60 percent in the Bolivia 1997 poorest �fth have completed the �ve years of that is, speci�c goals already being reached 200 primary school (�gure 1.2). In Peru only 67 within a country. percent of youths in the poorest �fth have 150 143 �nished the six-year primary cycle, even Affordable access to services is though almost all started school. In both low—especially for poor people 100 countries nearly all adolescents in the richest In many of the poorest countries, access to 50 �fth of the population completed primary schools, health clinics, clean water, sanitation 0 Cambodia 2000 BOX 1.1 Who are “poor people�? 200 De�ning who is “poor� is always a dif�cult proposi- latter method are typically referred to as “asset� or tion because there are several concepts of poverty. “wealth� quintiles (since asset ownership and hous- 150 147 Perhaps most familiar is the one used to identify the ing characteristics are arguably reflections of a poor in sample surveys in low-income countries: that household’s wealth).51 is based on a composite measure of total household But poverty based on consumption,“wealth,� or 100 consumption per member (with adjustments for an alternative derived from income, is not the only household size and composition).50 “Poor people� social disadvantage that creates dif�culties in the 50 are then de�ned as those living in households below demand for and provision of services. Gender can a particular threshold of this measure of consump- exclude women from both household and public 0 tion, such as below $1 or $2 a day, or below a nation- demands for better services. In many countries eth- ally de�ned level. nicity or other socially constructed categories of dis- Armenia 2000 An alternative approach divides the population advantage are important barriers. People with physi- 200 into various groups, for example, quintiles, according cal and mental disabilities are often not to a ranked ordering of the measure.The poorest accommodated by education and health services. 150 quintile or poorest �fth, for example, is the 20 per- Even broader concepts of poverty are relevant to cent of people who live in households with the low- effective services.“Poor people� include people expe- est values of the consumption measure. riencing any of the many dimensions of poverty— 100 Many surveys, including some used in this and those vulnerable or at risk of poverty—in low- 63 Report, do not include consumption data, which income and lower middle-income countries.52 So 50 are dif�cult to collect. One approach to assigning poor people can be seen as the “working class,� or people to quintiles is to aggregate indicators of a “popular� in Spanish, or simply just “not rich.� Even in 0 household’s asset ownership and housing charac- middle-income countries the “poor� includes a large Under age 1 Under age 5 teristics into an index, and then to rank households part of the population: much of the population can- according to this index. To distinguish these not insulate itself from the consequences of failures Note: Fifths based on asset index quintiles. approaches in this Report, quintiles based on the of public services. Source: Analysis of Demographic and Health Survey data. (c) The International Bank for Reconstruction and Development / The World Bank Services can work for poor people but too often they fail 21 facilities, rural transport, and other services is and hire more teachers. Primary enrollment Figure 1.2 The poor are less likely to start school, more likely to drop out limited. For children in Aberagerema village doubled between 1973 and 1986, reaching 90 Percent of 15- to 19-year-olds who in Papua New Guinea, the nearest school is in percent—though the story on quality is less have completed each grade or higher Teapopo village, an hour away by boat, two positive.58 Despite a limited budget El Sal- Niger 1998 hours by canoe.53 This is not unusual: the vador expanded access to schooling in poor 100 average travel time to the nearest school in rural communities after a civil war in the that country is one hour.54 The availability of 1980s by using innovative institutional 80 services varies dramatically across countries. arrangements (see Educo spotlight). 60 Typically, however, poor people need to travel The exact relationship between use of ser- Richest fifth substantial distances to reach health and edu- vices and prices or family income varies, but 40 cation services—and often much longer dis- for poor people, lower incomes and higher 20 6.4% tances than richer people in the same coun- prices are associated with less use.59 Poor peo- Poorest fifth 0 try. In rural Nigeria children from the poorest ple spend a lot of their money on services: 75 1 2 3 4 5 6 7 8 9 �fth of the population need to travel more percent of all health spending in low-income Grade than �ve times farther than children in the countries is private, 50 percent in middle- India 1998–99 richest �fth to reach the nearest primary income countries.60 Based on government 100 school, and more than seven times farther to sources, these broad aggregates are probably 80 reach the nearest health facility (table 1.1). underestimates, hiding the heavier burden on And traveling theses distances can be hard. In poor people. And poor people often need to 60 Lusikisiki village, South Africa, it may be nec- pay more for the same goods. For example, 40 essary to hire neighbors to carry a sick person poor people often pay higher prices to water 36% 20 uphill to even reach the nearest road, which sellers than the better-off pay to utilities may be inaccessible during the rainy season.55 (chapter 9). In Ghana the approximate price 0 On top of this, staff are getting rarer in paid per liter for water purchased by the 1 2 3 4 5 6 7 8 9 Grade some parts of the world. There is mounting bucket was between 5 and 16 times higher Egypt 2000 evidence that AIDS is reducing the pool of than the charge for public supply, even 100 people able to become teachers or health pro- though women and children often had to fessionals (box 1.2), and international labor walk a long distance to purchase the water. In 80 markets are making it hard to keep quali�ed Pune, India, low-income purchasers of water 60 medical staff in poor countries (chapters 6 paid up to 30 times the sale price of the 60% and 8). metered water that middle- and upper- 40 Coverage of other services is also far from income households used.61 20 universal. More than a billion people world- The poor also lack the collateral needed to 0 wide have no access to an improved water formally borrow to pay for expensive services 1 2 3 4 5 6 7 8 9 source, and 2.5 billion do not have access to for which they lack insurance, and therefore Grade improved sanitation. In Africa only half the resort to informal moneylenders who charge Peru 2000 rural population has access to improved very high interest rates. If this �nancing 100 water or improved sanitation. In Asia only 30 channel is unavailable, they use more expen- 80 percent of the rural population has access to sive traditional or private providers who 67% improved sanitation.56 Again, there are large often provide more flexibility in the terms of 60 variations across and within countries. In payment.62 40 Cambodia 96 percent of the richest �fth of This need not be. In Egypt making health the population has access to an improved insurance available to school children in the 20 drinking water source, but just 21 percent of early 1990s almost doubled the probability of a 0 the poorest �fth does (�gure 1.3). In health facility visit among the poorest �fth of 1 2 3 4 5 6 7 8 9 Morocco in 1992, 97 percent of the richest the population, substantially reducing the Grade �fth of the population had access to an rich-poor gap.63 In Mexico an innovative pro- Notes: The grade number boldfaced denotes the end of the primary cycle. Fifths based on asset improved water source, but just 11 percent of gram—Progresa—provided parents with cash index quintiles. the poorest �fth did. In Peru the correspond- transfers if they attended health education lec- Source: Analysis of Demographic and Health Survey data. ing shares are 98 percent and 39 percent.57 tures (where they also received nutrition sup- This need not be. Indonesia expanded plements), and family members got regular access to primary education in the mid-1970s medical checkups. The impact of this combi- by using its oil windfall to build new schools nation of higher income and facility visits was (c) The International Bank for Reconstruction and Development / The World Bank 22 WORLD DEVELOPMENT REPORT 2004 Table 1.1 The nearest school or health center can be quite far Mean distance to nearest facility in rural areas among the poorest and richest wealth quintiles in 19 developing countries GNI per Distance to the nearest Distance to the nearest capita primary school (kilometers) medical facility (kilometers) Poorest Richest Ratio Poorest Richest Ratio �fth �fth �fth �fth Bangladesh 1996–97 374 0.2 0.1 1.6 0.9 0.7 1.3 Benin 1996 395 1.5 0.0 — 7.5 2.8 2.7 Bolivia 1993–94 1004 1.2 0.0 — 11.8 2.0 6.0 Burkina Faso 1992–93 336 2.9 0.8 3.9 7.8 2.6 3.0 Central African Republic 1994–95 819 6.7 0.8 8.9 14.7 7.7 1.9 Cameroon 1991 611 2.6 0.7 3.8 7.0 5.4 1.3 Chad 1998 250 9.9 1.3 7.6 22.9 4.8 4.8 Côte d’Ivoire 1994 788 1.4 0.0 — 10.5 3.4 3.1 Dominican Rep. 1991 1261 0.6 0.4 1.3 6.3 1.3 5.0 Haiti 1994–95 336 2.2 0.3 6.4 8.0 1.1 7.2 India 1998–99 462 0.5 0.2 2.3 2.5 0.7 3.6 Madagascar 1992 303 0.6 0.3 1.8 15.5 4.7 3.3 Mali 1995–96 281 7.9 5.2 1.5 13.6 6.7 2.0 Morocco 1992 1388 3.7 0.3 13.1 13.5 4.7 2.9 Niger 1998 217 2.2 1.5 1.5 26.9 9.7 2.8 Nigeria 1999 266 1.8 0.3 5.5 11.6 1.6 7.1 Senegal 1992–93 933 3.8 2.3 1.7 12.8 10.0 1.3 Tanzania 1991–92 224 1.2 0.6 1.9 4.7 3.0 1.6 Uganda 1995 290 1.4 0.9 1.5 4.7 3.2 1.5 Zimbabwe 1994 753 3.0 3.5 0.8 8.6 6.3 1.4 Note: Gross national income (GNI) per capita is that at the time of the survey, expressed in 2001 dollars. Medical facility encompasses health centers, dispensaries, hospitals, and pharmacies. Although some of these data are a bit dated, they are the latest that were collected in a consistent manner across these countries. The situation in some countries may be different today. Source: Analysis of Demographic and Health Survey data. signi�cant: illnesses among children under �ve Punjab, Pakistan, only about 5 percent of sick fell by about 20 percent (see spotlight). children were taken for treatment to rural primary health care facilities; half went to Quality—a range of failures private dispensaries, and the others to private Lack of access and unaffordability are just two doctors.66 When quality improves, the ways services fail. In low- and middle-income demand for services increases—even among countries alike, if services are available at all poor clients.67 they are often of low quality. So, many poor people bypass the closest public facility to go Services are often dysfunctional to more costly private facilities or choose bet- Ensuring that positions are �lled, that staff ter quality at more distant public facilities. An report for work, and that they are responsive in-depth study of the Iringa district in Tanza- to all their clients is a major challenge. The nia, a poor rural area, showed that patients more skilled the workers, the less likely they bypassed low-quality facilities in favor of are to accept a job as a teacher or a health those offering high-quality consultations and worker in a remote area. A recent study in prescriptions, staffed by more knowledgeable Bangladesh found 40 percent vacancy rates physicians, and better stocked with basic sup- for doctor postings in poor areas.68 In Papua plies.64 A study in Sri Lanka found similar New Guinea, with a substantial percentage of behavior, with patient demand for quality teaching positions un�lled, many schools varying with the severity of the illness.65 closed because they could not get teachers.69 One result: underused publicly funded Incentive payments might encourage profes- clinics. In the Sheikhupura district of rural sionals to work in remote areas, but they can (c) The International Bank for Reconstruction and Development / The World Bank Services can work for poor people but too often they fail 23 Figure 1.3 Water, water everywhere, nor any drop to drink BOX 1.2 HIV/AIDS is killing Percent of households who use an improved water source, poorest and richest �fths teachers Poorest fifth Richest fifth Ethiopia 2000 Many countries lack reliable data on AIDS-related Madagascar 1997 deaths and HIV prevalence among teachers, but Chad 1998 the available information suggests rising teacher Cameroon 1998 mortality in the presence of HIV/AIDS.For example: Morocco 1995 In the Central African Republic 85 percent of Guinea 1999 teachers who died between 1996 and Mozambique 1997 1998 were HIV-positive. On average they Rwanda 1992 died 10 years before they were due to Kenya 1998 retire. Zambia 1996–97 In Zambia 1,300 teachers died in the �rst 10 Cambodia 2000 months of 1998, compared with 680 Senegal 1997 teachers in 1996. Central African Rep. 1994–95 In Kenya teacher deaths rose from 450 in Haiti 1994–95 1995 to 1,500 in 1999 (reported by the Niger 1998 Teaching Service Commission), while in Togo 1998 one of Kenya’s eight provinces 20 to 30 Yemen 1991–92 teachers die each month from AIDS. Nigeria 1999 HIV-positive teachers are estimated at more Ghana 1998 than 30 percent in parts of Malawi and Peru 2000 Uganda, 20 percent in Zambia, and 12 Burkina Faso 1999 percent in South Africa. Benin 1996 Sources: Coombe (2000), Gachuhi (1999), Kelly Mali 1995–96 (1999), Kelly (2000), UNAIDS (2000), World Bank Uganda 2000 (2002h). Indonesia 1997 Nicaragua 1998 Côte d’Ivoire 1998–99 Turkey 1998 be expensive. A study in Indonesia estimated Bolivia 1997 that doctors would need to be paid several Namibia 1992 times their current salaries to induce them to Tanzania 1999 go to the most remote areas.70 Colombia 2000 Even when positions are �lled, staff Zimbabwe 1999 absence rates can be high. In random visits to Dominican Rep. 1996 Kazakhstan 1999 200 primary schools in India, investigators Malawi 2000 found no teaching activity in half of them at Philippines 1998 the time of visit.71 Recent random samples of Kyrgyz Rep. 1997 schools and health clinics in several develop- Nepal 2001 ing countries found absence rates over 40 per- Brazil 1996 Pakistan 1990–91 cent, with higher rates in remote areas and for India 1998–99 some kinds of staff—although there is wide Armenia 2000 variation within countries (tables 1.2 and Comoros 1996 1.3). Earlier studies have found similar results. Egypt 2000 Up to 45 percent of teachers in Ethiopia were Uzbekistan 1996 Bangladesh 1996–97 absent at least one day in the week before a Guatemala 1999 visit—10 percent of them for three days or 0 20 40 60 80 100 more.72 Health workers in rural health centers Percent of households in Honduras worked only 77 percent of the Note: The poorest �fth in one country may correspond to the standard of living in the middle �fth in possible days in the week before a visit.73 In another country. Within-country inequalities reflect inequality in access to water and in the wealth index used to construct quintiles. An “improved� water source, as de�ned by UNICEF, provides ade- rural Côte d’Ivoire only 75 percent of doctors quate quality and quantity of water (that is, a household connection or a protected well, not an unpro- were in attendance on the day before a visit.74 tected well or bottled water). Source: Analysis of Demographic and Health Survey data. Staff alone cannot ensure high-quality services. They also need the right materials— books in schools, drugs in clinics. Studies in Ghana and Nigeria in the early 1990s found (c) The International Bank for Reconstruction and Development / The World Bank 24 WORLD DEVELOPMENT REPORT 2004 Table 1.2 Staff are often absent that about 30 percent of public clinics lacked and the Russian Federation—more than 90 Absence rates among teachers and health care workers in public drugs.75 A quarter of rural clinics in Côte percent in Armenia.82 facilities (percent) d’Ivoire had no antibiotics.76 By itself, the Corruption hurts patients elsewhere too. Primary Primary availability of drugs in a health facility is an For example, studies based on data from the schools health ambiguous measure of quality: stockouts mid-1990s found that informal payments facilities could be caused by high demand. But when substantially increased the price of health ser- Bangladesh — 35 medicines are lacking in clinics and available vices in Guinea and Uganda.83 A recent Ecuador 16 — on the black market, as is often the case, some- review of case studies in Latin America found India* 25 43 thing is amiss. Educational materials are simi- widespread corruption in hospitals, ranging Indonesia 18 42 larly lacking in schools. In Nepal a study found from theft and absenteeism to kickbacks for Papua as many as six students sharing local-language procurement.84 Villagers in one North New textbooks. In Madagascar textbooks had to be African country where people are covered by Guinea 15 19 shared by three to �ve students, and only half “free medical care� reported in a discussion Peru 13 26 the classrooms had a usable chalk board.77 group that “there isn’t a single tablet in the Zambia 17 — When staff report to work—as many do clinic and the doctor has turned it into his Uganda 26 35 conscientiously—and when complementary private clinic.�85 *Average for 14 states. inputs are available, service quality will suffer Again, this need not be. In Benin cost- if facilities are inadequate or in disrepair. sharing in health clinics—in line with the Conditions can be horri�c. An account of a Bamako Initiative—and revolving drug- school in north Bihar in India describes class- funds increased the availability of drugs in rooms “. . . close to disintegration. Six chil- clinics that previously provided services free dren were injured in three separate incidents but almost never had any drugs. Use when parts of the building fell down, and increased in all the clinics that introduced Table 1.3 Absence rates vary a lot— even in the same country even now there is an acute danger of terminal these measures (see spotlight).86 Innovative Absence rates among teachers and collapse. . . . The playground is full of muck arrangements can encourage teachers to health care workers in public facilities and slime. The overflowing drains could eas- report for work. In Nicaragua between 1995 in different states of India (percent) ily drown a small child. Mosquitoes are and 1997 teacher attendance increased by Primary Primary schools health swarming. There is no toilet. Neighbors com- twice as much in primary schools that were facilities plain of children using any convenient place granted autonomy as in state schools man- Andhra Pradesh 26 — to relieve themselves, and teachers complain aged through the bureaucratic system.87 In of neighbors using the playground as a toilet India a large-scale basic education program Assam 34 58 in the morning.� 78 The same study in India in the 1990s doubled the toilets and drinking Uttar Pradesh 26 42 found that half the schools visited had no water facilities in schools in districts where it Bihar 39 58 drinking water available. In rural areas of was implemented. Stakeholders can mobilize Uttar Anchal 33 45 Bangladesh and Nepal a study found an aver- to reduce corruption. Public sector unions Rajasthan 24 39 age of one toilet for 90 students, half of them have organized an anticorruption network Karnataka 20 43 not usable.79 In Pakistan there were no sepa- (UNICORN) that is supporting national ini- West Bengal 23 43 rate toilet facilities for girls in 16 percent of tiatives to protect whistleblowers. Gujarat 15 52 schools visited in one study.80 Haryana 24 35 Another problem is corruption in various The technical quality of services is Kerala 23 — forms. Teachers and principals might solicit often very low Punjab 37 — bribes to admit students or give better grades, Services also fail poor people when technical Tamil Nadu 21 — or they might teach poorly to increase the quality is low—that is, when inputs are com- Orissa 23 35 demand for private tuition after hours. Sur- bined in ways that produce outcomes in inef- veys in 11 Eastern and Central European �cient, ineffective, or harmful ways. For Notes for tables 1.2 and 1.3: The absence rate is the percentage of staff who are supposed countries found that the health sector was example, health workers with low skills give to be present but are not on the day of an 81 unannounced visit. It includes staff whose considered one of the most corrupt. Of�- the wrong medical advice or procedure, or absence is “excused� and “not excused� and cially only 24 percent of health spending in schools use ineffective teaching methods. so includes, for example, staff in training, per- forming nonteaching “government� duties, as Europe and Central Asia is estimated to be Gross inef�ciency was identi�ed as the reason well as shirking. private, but this fails to include the informal for soaring expenditures in a hospital in the — indicates data not available. Sources for tables 1.2 and 1.3: Chaudhury and payments—gratuities and bribes—that many Dominican Republic.88 A multicountry study others (2003), Habyarimana and others (2003), patients pay. More than 70 percent of patients of health facilities in the mid-1990s found and NRI and World Bank (2003). Data should be considered preliminary. (c) The International make Bank for Reconstruction such payments in Azerbaijan, Developmentshockingly andPoland, Bankcases of proper assessments of / The World low Services can work for poor people but too often they fail 25 diarrhea in children under �ve, and even often starts at 8 a.m. while girls are still fetch- fewer cases correctly treated or advised. For ing water, and school holidays are at odds example, in Zambia only 30 percent of cases with local market dates. were correctly assessed, and only 19 percent The “social distance� between providers correctly rehydrated.89 Another study in and their clients can be large. In Niger, a Egypt found only 14 percent of acute cases of mainly rural country, a study found 43 per- diarrhea were treated appropriately with oral cent of the parents of nurses and midwives rehydration salts.90 A recent study in Benin were civil servants, and 70 percent of them found that one in four sick children received had been raised in the city. All of them went to unnecessary or dangerous drugs from health work by car—a rarity in that country.99 Sad workers.91 In India the contamination of consequences of the social distance between injection needles used by registered medical providers and clients are not hard to �nd. In practitioners was alarmingly widespread.92 Egypt participants in a discussion group com- Even though technical quality is more dif- plained about the attitude of staff at the local �cult to identify in basic education, some rural hospital, with one respondent summing indicators raise alarm. For example, spending up the experience: “They have their noses up is ineffectively allocated, with substantially in the air and neglect us.�100 In South Africa a more going to teacher salaries relative to other focus group member comments about a pri- factors that would be more ef�cient.93 Or time mary health care provider: “Sometimes I feel is misspent: in �ve Middle Eastern and North as if apartheid has never left this place. . . . African countries primary school students They really have a way of making you feel like spend only about 65 percent of the potential you are a piece of rubbish.�101 time actually on task.94 In Indonesia �rst and Services must be relevant—�lling a per- second grade students of�cially spend only ceived need—or there will be little demand 2.5 hours a day in school, and absences and for them (box 1.3). If primary schools teach classroom time spent on administrative tasks skills relevant only for secondary school—and reduced time spent learning even further.95 not for life outside of school—only children from richer families who expect to continue Services are not responsive to clients to the secondary level will deem it worthwhile Services also fail in the interaction between to complete primary school. In Ghana one provider and client. Clients are diverse: they respondent claimed: “School is useless: chil- differ by economic status, religion, ethnicity, dren spend time in school and then they’re gender, marital status, age, social status, caste. unemployed and haven’t learned to work on They may also differ in the constraints on the land.�102 In India one component of an their time, their access to information and integrated childhood development program social networks, or their civic skills and ability failed when bene�ciaries rejected the food to act collectively. The inequalities between these groups are mirrored in the relationship between clients and providers.96 In India dis- tricts with a higher proportion of lower castes BOX 1.3 School services for girls are not in high demand and some religious groups have fewer doctors in Dhamar Province, Yemen and nurses per capita, and health outreach “At the back of the classroom of 40 boys sat ceptable for them either to learn alongside workers are less likely to visit lower-caste and 2 girls. . . .What did the girls want to be when boys or to walk to class in the street.“ poor households.97 Clients report that they they grew up? ‘A teacher,’ one said. ’A doc- In Yemen girls make up about one in value health facilities that are open at conve- tor,’ said the other. But less than a quarter of three students at the primary level, one in the women in Yemen are literate, and they nient times, with staff who treat them with four at the secondary level. More than 75 must follow the path of the traditional vil- respect. In El Salvador infrequent and incon- lage women, who usually marry in their percent of women over 15 are illiterate, compared with 35 percent of men. Girls’ venient operating hours greatly reduced the teens and have an average of 10 children. In education is not the only problem, however. use of health posts. According to focus group the countryside of Dhamar Province, one of The net enrollment rate for boys is only 75 the country’s poorest, there are few profes- respondents: “Health posts operate only sional activities for anyone, much less for percent at the primary level, 70 percent at twice a week. Waiting time is three hours on the secondary level. women. Besides, most parents won’t let their average. Only those who arrive by 8 a.m. get a daughters go to school—deeming it unac- Sources: Mayer (1997), World Bank (2002g). consultation.�98 In Sub-Saharan Africa school (c) The International Bank for Reconstruction and Development / The World Bank 26 WORLD DEVELOPMENT REPORT 2004 grain supplied. Eventually the program mental design, it is possible to learn about changed what it offered to match varying systems and to innovate. For example, the preferences in different parts of the country. Probe study in India documented a variety of And again, this need not be. In the Nioki shortcomings of the quality of primary area of Zaire (now the Democratic Republic schools. The widely publicized results con- of Congo), where the use of health services tributed to mobilizing support for reform.106 declined substantially between 1987 and 1991, it increased in clinics with nurses who had good interpersonal skills.103 Among Making services work indigenous peoples in Bolivia, Ecuador, to improve outcomes Guatemala, Mexico, Paraguay, and Peru, pro- Many of the examples discussed so far moting bilingual and intercultural education describe failures in the public sector’s provi- contributed to improved schooling out- sion of services, but they are not the only comes.104 An innovative public health cam- story. The 20th century has seen enormous paign among sex workers in Sonagachi, improvements in living standards. Life India, trained “peer educators� to pass infor- expectancy has improved dramatically in mation to their co-workers. Disseminating nearly every country. The expansion of information in this way resulted in more con- schooling has been similarly remarkable. In dom use and substantially less HIV infection nearly every country illiteracy has been than in other cities. The approach had knock- reduced dramatically, enrollment rates have on effects as well: sex workers organized a gone up, and the average schooling of the union and effectively lobbied for legalization, population has more than doubled. Civil ser- reduction in police harassment, and other vice bureaucracies providing good services rights.105 have been integral elements of those suc- cesses. In many settings staff must overcome Little evaluation, little innovation, major obstacles—including threats to their stagnant productivity own safety—in order to teach children or In most settings there are few evaluations of provide care to the sick. new interventions, and so no effective inno- What do services that work look like? Safe vation and improvement in the productivity and pleasant schools with children learning of services. Evaluating innovative service to read and write. Primary clinics with health arrangements—such as new forms of workers dispensing proper advice and medi- accountability—is rarer still. If systems don’t cine. Water networks distributing safe and build in ways of learning about how to do dependable water. Direct subsidies to poor things better, it should be no surprise when children and their families encouraging they stagnate. Relying on research from other demand. Services that are accessible, afford- countries, while useful, is not enough. Find- able, and of good quality—helping to ing out how a particular intervention works improve outcomes for poor people. in each country setting is crucial, since his- Governments take on a responsibility to tory, politics, and institutions determine what make services work in order to promote works, what doesn’t, and why. health and education outcomes. Chapter 2 Once again, this need not be. Although addresses the reasons for this responsibility, rarely carried out, some programs have tried dwelling on three seemingly straightforward to incorporate evaluation components to ways to discharge it: relying solely on eco- learn about the program. Mexico’s Progresa nomic growth, allocating public spending, or explicitly included randomization and evalu- applying technical �xes. None of them is ation in its design. The results of the evalua- enough by itself. Making services work tion—well documented and disseminated in requires improving the institutional arrange- the media—helped solidify support for the ments for producing them. Chapters 3 to 6 of program. They showed what was most effec- this Report develop a framework for analyz- tive, contributing to the program’s extension ing those arrangements. Chapters 7 to 11 to a large part of the country’s poor people apply the framework and draw lessons for (see spotlight). But even without an experi- governments and donors. (c) The International Bank for Reconstruction and Development / The World Bank Services can work for poor people but too often they fail 27 C R AT E 1 . 1 Determinants of health and education outcomes—within, outside, and across sectors Health and education outcomes are returns. But the returns might vary for differ- income by working inside or outside the determined by more than the availability and ent people, such as lower expected earnings home (looking after siblings, working on the quality of health care and schooling. Better for women or for ethnic minorities. In these family farm).The value of this contribution is nutrition helps children learn. Better refrigera- cases one would expect different levels of forgone if they spend substantial time in tion and transport networks help keep medi- investment: different desired levels of school- school. cines safe. Many factors determine outcomes ing, for instance. A crucial element of The total cost of illness includes days of on both the demand and the supply sides, demand is the degree to which individuals work lost recovering, seeking care, or looking linked at many levels.The demand for health rather than society reap the rewards. Goods after the ill. Richer families can cope better and education is determined by individuals with large positive externalities—in the with these costs, which leads to a direct asso- and households weighing the bene�ts and extreme, public goods—will be demanded at ciation between income and outcomes. In costs of their choices and the constraints they less than the socially optimal level. addition, better health and education are face.The supply of services that affect health What are the costs? There are direct often valued in themselves. As incomes and education outcomes starts with global costs: user fees, transport costs, textbook increase families demand more of them, technological knowledge and goes all the way fees, drug costs. Some of these can be borne which again results in an association to whether teachers report for work and com- by families—though not all families. Coping between income and outcomes. munities maintain water pumps. mechanisms for those that cannot are often The production of health and education hard to use. For example, the lack of insur- depends on the knowledge and practices of Demand: individuals and households ance markets can make it hard to absorb adults in households.This works through Bene�ts and costs determine how much an the �nancial burden of sudden illness. Or both the demand for human capital and the individual invests in education or health. the inability to borrow against future earn- generation of outcomes. A review of four What are the bene�ts? Higher levels of edu- ings can make it hard to borrow for school- hygiene interventions that targeted hand cation and health are associated with higher ing investments. washing in poor countries found 35 percent productivity—and higher earnings. Investing Indirect costs can also be large. For exam- less diarrhea-related illness among children in human capital is a way to get those ple, children often contribute to household who received the interventions. And factors The determinants of supply and demand operate through many channels Policies, capacity, Health, nutrition, Households and technical know-how, education sectors individuals politics Behaviors and actions • Service price, accessibility, and quality • Global knowledge • Financing arrangements • Health: preventive care, care-seeking for illness, • National macro-, feeding practices, sector-, and micro-level sanitary practices, . . . policies • Education: enrollment • Technical capacity to and school participation, implement policies Related sectors Outcomes learning outside of school, . . . • Child mortality • Governance; politics • Availability, prices, and accessibility and patronage; political of food, energy, roads . . . Constraints capacity; and incentives • Other infrastructure • Child nutrition to implement policies • Environment • Income • School completion/ • Wealth learning, achievement • Education and knowledge Local context • Local government and politics • Community institutions • Cultural norms (e.g., women’s status) • Social capital Supply Demand (continued) (c) The International Bank for Reconstruction and Development / The World Bank 28 WORLD DEVELOPMENT REPORT 2004 C R AT E 1 . 1 Determinants of health and education outcomes—within, outside, and across sectors (continued) in the home complement schooling: books but for a school snack program there was level impact everyone’s health. In Peru the and reading at home contribute to literacy. sharing in poorer families. sanitation investments of a family’s neighbors Investments in the human capital of chil- Parents’ education has intergenerational were associated with better nutritional status dren are sensitive to the allocation of power effects on the health, nutritional status, and for that family’s children. within households: families in which the bar- schooling of their children. Adult female edu- The use of safe energy sources affects gaining power of women is stronger tend to cation is one of the most robust correlates of both health and education. Indoor air pollu- invest more in health and education. A study child mortality in cross-national studies, even tion—from using dirty cooking and heating in Brazil found that demand for calories and controlling for national income. Similarly, fuels—hurts child health. One review of protein was up to 10 times more responsive mother’s education is a strong determinant studies found that the probability of respi- to women’s than men’s income.This result, of lower mortality at the household level, ratory illness, or even death, was between strongest in societies that proscribe women’s though the relationship weakens when other two and �ve times higher in houses where roles, tends to affect girls more than boys. household and community socioeconomic exposure to indoor air pollution was high. A More generally, the roles and responsibili- characteristics are controlled for. A large part study in Guatemala found birth weights 65 ties of different household members can of this effect might not be general education grams lower among newborns of women affect how investments are made. A woman but speci�c health knowledge, perhaps who used wood as a domestic cooking fuel. in Egypt says:“We face a calamity when my acquired using literacy and numeracy skills Coping with the cold, in cold climates, husband falls ill. Our life comes to a halt until learned in school, as a study in Morocco affects health and imposes substantial he recovers.� Her husband’s earnings are cru- found.The effects can also be interspatial: a direct and indirect costs on households. cial for sustaining the family. Since productiv- study in Peru found that the education of a Education is affected as well: schools have ity is related to illness, households respond. mother’s neighbors signi�cantly increases to close when there is not enough heat, and In Bangladesh a study found that household the nutritional status of her children. Parents’ it is hard to imagine that working on members who engaged in more strenuous education is similarly associated with the schoolwork at home is an option when activities received more nutritious food. schooling of their children, though the mag- indoor temperatures are below freezing. Daughters’ education might be less valuable nitude of the effect—and the relative roles of to parents if sons typically look after them in mother’s and father’s education—vary sub- Supply: global developments their old age, so parents might be less willing stantially across countries. At any given income, health and education to send girls to school. Access to—and use of—safe water, as outcomes have been improving. A continu- well as adequate sanitation, have direct ing trend in improvements in health going Demand: links between sectors at effects on health status. Hand washing is a back several decades is interpreted as individual and household levels powerful health practice, but it requires suf�- advances in technologies and leaps in Health and nutritional status directly affects cient quantities of water. An eight-country knowledge about health and hygiene. More a child’s probability of school enrollment and study found that going from no improved recently, at a national income of $600 per capacity to learn and succeed in school. Mal- water to “optimal� water was associated with capita, predicted child mortality fell from nutrition among children was associated a 6-percentage-point reduction in the preva- 100 per 1,000 births to 80—a full 20 with signi�cant delays in school enrollment lence of diarrhea in children under three percent reduction—between 1990 and in Ghana. Improving child health and nutri- years of age (from a base of 25 percent) in 2000. If this association were sustained, tion at the pre-primary level has long-term households without sanitation. Nutritional major headway would be made toward the impacts on development. A study in the status was likewise associated with access to Millennium Development Goal through Philippines found that a one-standard-devia- improved water. But not all studies �nd these changes alone. Major breakthroughs tion increase in early-age child health strong associations between water source in immunizations for malaria—or HIV— increased subsequent test scores by about a and better health. could have a huge impact on mortality at third of a standard deviation. The water source is only part of the story: all income levels. Improving the health and nutritional sta- in Bangladesh water accessed through tube- Recent years have seen major develop- tus of students positively affects school wells—an “improved� source— is frequently ments in global funding for health and edu- enrollment and attendance. A longitudinal contaminated with arsenic. One study found cation expenditures. Debt relief through the study in Pakistan found that a one-third of a that arsenic levels higher than the World Heavily Indebted Poor Countries initiative is standard deviation increase in child height Health Organization’s maximum acceptable tied to increases in expenditures on these increased school enrollment by 19 percent- level are associated with twice the level of sectors. New assistance, delivered through age points for girls and 4 percentage points diarrhea in children under age six. Extremely multisectoral products such as Poverty for boys. An evaluation of school-based mass high levels of arsenic are associated with Reduction Support Credits, requires explicit treatment for deworming in rural Kenya shorter stature among adolescents. strategies for human development found that student absenteeism fell by a The same eight-country study mentioned investments. Global funds for health and the quarter—but test scores did not appear to above found that going to “optimal� “Fast-Track Initiative� for education are inter- be affected. Improving nutrition is not as sim- sanitation from none was associated with a national pledges to support initiatives in the ple as supplementary feeding at school: 10-percentage-point drop in recent diarrhea sectors (chapter 11). Easing �nancial households can reallocate resources with the in households with no improved water constraints goes hand in hand with using effect of “sharing� that food. A study in the source. As in education, there are spillover resources effectively to support services that Philippines found no such sharing in general, effects: sanitation practices at the community work for poor people. (c) The International Bank for Reconstruction and Development / The World Bank Services can work for poor people but too often they fail 29 C R AT E 1 . 1 Determinants of health and education outcomes—within, outside, and across sectors (continued) Supply: national resources thousands did not. Wars, including civil wars, Supply: services and their �nancing National income is strongly associated with lead to “lost generations� of undernourished Services themselves are important. Inacces- child mortality and primary school comple- and undereducated children.These de�cien- sible or poor-quality services raise the tion. Income and health and education out- cies are dif�cult—if not impossible—to make effective price of health care and schooling, comes build on each other. More income up for. When children have been out of which results in higher mortality and lower leads to better human development school for a long time, it is hard to return. educational achievement. Poor-quality outcomes, and better health and education And bad health and poor nutrition at an early schools deter enrollment and reduce can lead to increased productivity and better age affect children throughout their lives. attainment and achievement, especially incomes. Studies that have tried to disentan- Periods of national economic and social among children of poor families. Health gle these relationships typically �nd income crisis can result in bad health and education clinics where the technical skills of staff are to be a robust and strong determinant of outcomes.This is clear in Russia’s recent his- so bad as to be dangerous will lead to outcomes. tory: adult mortality has increased dramati- higher mortality. Lack of water will signi�- National endowments are also a strong cally over the past 10 years. Sustained cantly hurt child health. determinant. Geography and climate some- economic depression can severely compro- Financing arrangements matter. Absorb- times make it tougher to overcome health mise children’s health and have cascading ing the burden of unpredictable large expen- problems. For example, areas conducive to effects on subsequent development and ditures through health insurance can reduce mosquito survival have great dif�culty in com- learning.The evidence of shorter-term eco- impoverishment, which in turn will affect bating malaria—and widely dispersed popu- nomic crises is more mixed. In middle- outcomes. Financing primary schooling lations are dif�cult to serve through income environments school enrollments might seem relatively minor: direct costs are traditional school systems. might increase as the opportunity cost of typically small. Even so, a lack of access to The performance of public expenditure in time for young people falls. Even in Indone- credit has been found to be associated with producing outcomes varies substantially across sia, a relatively poor country, the deep lower school enrollment. Borrowing to pay countries.There are large differences in economic and social crisis of the late 1990s the direct costs of primary school is almost achievements at similar levels of expenditure had smaller impacts on outcomes than ini- unheard of, but there could be second-round and similar achievements with very large differ- tially feared.This was partly because broad effects if the lack of access to credit means ences in expenditures—conditional on income. social safety nets were rapidly put in place. that families need children to engage in Spending more through the public sector is not home production. Supply: the local context of government always associated with improved outcomes. and communities Supply: services working together to This is not to say that spending cannot be help- Decentralization can be a powerful tool for produce outcomes ful—but the way resources are used is crucial moving decisionmaking closer to those Links among services are critical. Vaccines to their effectiveness. affected by it. Doing so can strengthen the can become less effective, ineffective, or Supply: political, economic, links and accountability between policymak- even dangerous if they get too hot, freeze, and policy context ers and citizens—local governments are or are exposed to light. The ability to trans- Governance affects the ef�ciency of expendi- potentially more accountable to local port and store vaccines properly thus tures: in corrupt settings money that is osten- demands. It can also strengthen them determines the success of immunization sibly earmarked for improving human devel- between policymakers and providers—local campaigns. In cold climates schools and opment outcomes is diverted. Staffs governments are potentially more able to health facilities often need to close because ostensibly delivering services do not. But the monitor providers. But local governments of the lack of heating, and dependable effects of poor governance can be deeper. should not be romanticized. Like national energy sources can directly affect health Famines are caused as much by human fac- governments they are vulnerable to and education outcomes. The accessibility tors as by nature. And the repercussions run capture—and this might be easier for local of services can be a deterrent to their use: across national borders. For example, a elites on a local scale. roads and adequate transport contribute to drought combined with misguided policies Community-level institutions, shaped by the total cost of using a service. Since the and bad governance in Zimbabwe resulted cultural norms and practices, can facilitate expected return to education determines in a regional food shortage. or hinder an environment for improving the bene�ts of schooling, labor markets Managing public expenditures can be a outcomes. A review of safe-water projects in that are not fundamentally distorted (for critical link in ensuring that allocated expen- Central Java, Indonesia, associates success example, through discriminatory practices ditures get put to uses that improve with greater social capital. In Rajasthan, toward marginalized groups) can outcomes.“Cash budgeting� in Zambia led to India, manifestations of “mutually bene�cial contribute to higher education unpredictable social service spending and collective action� were associated with achievement. Services therefore need to deep cuts in spending on rural infrastructure. watershed conservation and development work together to promote improved Conflict leaves long-lived scars on health activities more generally. A broader review outcomes. and education. Children in war-torn countries of the literature suggests that participatory are hard to �nd, hard to get into school, and approaches to implementing projects are hard to keep in school. During Sierra Leone’s more successful in communities with less recent civil war, tens of thousands of children economic inequality and less social and eth- attended primary school but hundreds of nic heterogeneity. Source: Sources are detailed in Filmer (2003a). (c) The International Bank for Reconstruction and Development / The World Bank spotlight on Progresa Conditional cash transfers to reduce poverty in Mexico Progresa, the Education, Health, and Nutrition Program of Mexico, transferred money directly to families on the condi- tion that family members went for health checkups, mothers went for hygiene and nutrition information sessions, and children attended school. By documenting success through rigorous evaluation, the program has improved, scaled up, and taught others.107 When President Vincente Fox was elect- Almost 60 percent of program transfers W hen Ernesto Zedillo became Mexico’s president in 1995, a �fth of the population could not afford the minimum daily nutritional requirements, 10 million Mexicans lacked ed, his government embraced the program, built on it using the evaluation results, expanded it to urban areas, and renamed it Oportunidades. By the end of 2002 the pro- went to households in the poorest 20 per- cent of the national income distribution and more than 80 percent to the poorest 40 per- cent. This is impressive. The median target- gram had about 21 million bene�ciaries— ing effectiveness in 77 safety net programs access to basic health services, more than roughly a �fth of the Mexican population. from around the world was to have 65 per- 1.5 million children were out of school, cent of bene�ts go to the poorest 40 percent and student absenteeism and school deser- Designing a comprehensive (according to one recent study). tions were three times higher in poor and program Even with careful targeting and moni- remote areas than in the rest of the coun- toring, the program’s administrative costs try. The country had a history of unpro- Children over seven were eligible for educa- were less than 9 percent of total costs—sub- ductive poverty alleviation programs. tion transfers. Bene�ts increased by grade stantially lower than earlier poverty allevia- Worse, the 1994–95 economic crisis left the (since opportunity costs increase with age) tion efforts in Mexico. Despite its initial government with even fewer resources— and were higher for girls in middle school, large scale, the program did not cover all and greater demands, as more people were to encourage their enrollment. To retain the the poor, particularly in urban areas. falling into poverty. bene�ts, children needed to maintain an 85 The administration decided that a new percent attendance record and could not approach to poverty alleviation was need- Boosting enrollments repeat a grade more than once. ed. The Education, Health, and Nutrition Eligible families could also receive a Girls’ enrollment in middle school rose from Program of Mexico, called Progresa, intro- monthly stipend if members got regular med- 67 percent to around 75 percent, and boys’ duced a set of conditional cash transfers to ical checkups and mothers attended monthly from 73 percent to 78 percent. Most of the poor families—if their children were nutrition and hygiene information sessions. increase came from increases in the transi- enrolled in school and if family members Households with children under three could tion from primary to middle school (�gure visited health clinics for checkups and also receive a micronutrient supplement. 1). The program worked primarily by keep- nutrition and hygiene information. The transfers went to mothers, who ing children in school, not by encouraging The program was intended to remedy were thought more responsible for caring those who had dropped out to return. It also several shortcomings of earlier programs. for children. The program imposed a helped reduce the incidence of child labor. First, it would counter the bias in poor fam- monthly ceiling of $75 per family. In 1999 ilies toward present consumption by bol- the average monthly transfer was around stering investment in human capital. $24 per family, nearly 20 percent of mean Figure 1 Higher school retention, more transitions from primary to middle school Second, it would recognize the interdepen- household consumption before the pro- Expected grade completion before and after dencies among education, health, and gram. Transfers were also inflation-indexed intervention, and with and without Progresa nutrition. Third, to stretch limited every six months (today the maximum is Percent resources, it would link cash transfers to $95 and the average is $35). 100 household behavior, aiming at changing Highly centralized, the program has just Post-intervention: Progresa attitudes. Fourth, to reduce political inter- one intermediary between program of�- 80 ference, the program’s goals, rules, require- cials and bene�ciaries—a woman commu- ments, and evaluation methods would be nity promoter chosen by a general assembly 60 widely publicized. of households in targeted communities. She The program has been rigorously evalu- can also liaise between bene�ciaries and 40 ated, and evaluators have exploited the ran- education and health providers. Post-intervention: Non-Progresa domized way the program was rolled out. By the end of 1999 the program covered 20 The results have been impressive. To some 2.6 million rural families—about 40 emphasize the apolitical nature of the pro- percent of rural families and a ninth of all 0 gram, the government suspended the families in Mexico. The program budget 1 2 3 4 5 6 7 8 growth of the program for six months prior was almost $780 million, or 0.2 percent of Grade to the election in 2000—to show that gross domestic product and 20 percent of Note: Among children who enroll. Progresa was not a political tool. the federal poverty alleviation budget. Source: Schultz (2001). (c) The International Bank for Reconstruction and Development / The World Bank Spotlight on Progressa 31 Labor force participation decreased by about Figure 2 Improving child health The evaluation design captures the 20 percent for boys. Still, a substantial num- Percentage of children reported to have had an many determinants of outcomes. But it has illness. ber of children from poor families continue limitations. Policymakers would bene�t Percent to combine work with school. from knowing how the program could be 50 The impacts on learning are less clear. Age 0–2: Non-Progresa manipulated to improve impacts. For Teachers report improvements, attributing example, what is the impact of condition- 40 Age 0–2: Progresa them to better attendance, student interest, ing the transfers rather than giving pure and nutrition. But a study conducted one unconditioned transfers? In addition, year after the program started found no dif- 30 households in the control group might have ference in test scores. been affected by the intervention or by 20 knowing that they might receive it in the Improving nutrition and health Age 3–5: Non-Progresa future, an effect that would muddy the The program helped reduce the incidence 10 comparisons. of low height for age among children one to Evaluations can address these issues, but three years old. (Before the program stunt- Age 3–5: Progresa the complexity (and expense) increases 0 ing was very high, at 44 percent.) Annual 0 5 10 15 20 substantially. Alternative approaches that mean growth in height was 16 percent for rely on modeling—imposing additional Time since intervention (months) children covered by the program. On aver- assumptions on the analysis—might be Note: Age at start of intervention. age, height increased by 1–4 percent, and Source: Gertler (2000). necessary. Such analyses are currently weight by 3.5 percent. These gains were underway. achieved despite evidence that some house- holds did not regularly receive nutrition cation impacts would increase them by 8 Evidence makes the difference supplements and that supplements were percent. A general equilibrium analysis of A conditional cash transfer program can be often “shared� with older children. Part of Progresa found that the welfare impact was a powerful way of promoting education, the effect can be attributed to spending 60 percent higher than that of the highly health, and nutritional outcomes on a more on food and to consuming more distortionary food subsidies that Mexico massive scale. The success of the Progresa nutritious food, as recommended by the used previously. program has led to similar programs, nutrition information sessions. There were especially in other Latin American coun- also positive spillover effects for nonbene�- Evaluating impacts tries (Colombia, Honduras, Jamaica, and ciaries in the same community. Progresa was unusual in integrating evalua- Nicaragua). The program substantially increased tion from the beginning, enabling it to Evaluation was not an afterthought. It preventive health care visits. Visits by preg- assess impacts fairly precisely. To ensure continually fed back into improving pro- nant women in their �rst trimester rose 8 political credibility, the evaluation was con- gram operations. And its rigor increased percent, keeping babies and mothers tracted out to a foreign-based research con�dence in the validity of assessments of healthier. Illnesses dropped 25 percent group, the International Food Policy the program’s effects. among newborns and 20 percent among Research Institute. Evaluation was important for domestic children under �ve (�gure 2). The preva- Phasing in communities in a random and international political and economic lence of anemia in children two to four fashion—required for budgetary purpos- support—and thus contributed to pro- years old declined 19 percent. Adult health es—allowed the construction of 186 con- gram sustainability. Unlike previous pro- improved too. trol and 320 treatment groups. Having the grams, this one was not abandoned after a control groups enabled evaluators to “wash change in government. Clear and credible Reducing poverty out� confounding factors, including time evidence of large bene�ts for the country’s The program is not only raising incomes trends and shocks (economic and climatic). poor contributed to maintaining the temporarily, it should help raise future pro- Eventually all control communities were integrity of the program’s design (albeit ductivity and earnings of the children ben- incorporated in the program. Both quanti- with a name change). It also made it easier e�ting. Modeling exercises �nd that nutri- tative and qualitative evaluations were con- to get support from the Inter-American tional supplements alone would boost life- ducted, the latter using semistructured Development Bank for a major expansion time earnings by about 3 percent and edu- interviews, focus groups, and workshops. of the program. (c) The International Bank for Reconstruction and Development / The World Bank Governments should make services work 2 The responsibility that governments take on sic welfare economics arguments for govern- for basic health and education can be dis- ment intervention, political economy reasons charged in many ways—among them, foster- for intervention in key social sectors, appeals chapter ing economic growth, increasing public to fundamental human rights. Governments spending, and applying technical interven- demonstrate their responsibility by �nanc- tions. Each can contribute to better outcomes. ing, providing, or regulating the services that But if they are not supporting services that contribute to health and education out- work—services that result from effective insti- comes. The services come in many shapes tutional arrangements—they will not make a and sizes: building and staf�ng schools, sub- large sustainable difference. Making services sidizing hospitals, regulating water and elec- work requires changing the institutional rela- trical utility companies, building roads, pro- tionships among key actors. Subsequent chap- viding cash transfers to individuals and ters of this Report develop and apply a frame- households. Making these services work work to understand how and why those means that governments are meeting their relationships play out for different services. responsibility. Economic growth, though a major determinant of human development out- Public spending comes, would need to be substantially faster This responsibility is often reflected in gov- than it has been in most countries to make ernment spending. Health and education dramatic improvements through that chan- alone account for about a third of aggregate nel alone. Public spending makes improve- government spending, with the average ments possible, but the improvements will slightly lower in poorer countries and regions fall short if spending fails to reach poor (table 2.1). But there are wide variations people—either because it goes for things across countries, even within the same the poor do not use or because it is diverted region. Health and education spending along the way—or if services are not made accounted for 13 percent of public spending more productive. Applying technical inter- in Sierra Leone in 1998 but 34 percent in ventions—combining inputs to produce Kenya—18 percent in Estonia in 1997 but 59 outputs and outcomes more effectively—is percent in Moldova in 1996. Social security also important. But simply adjusting inputs and welfare spending, much of it directed to without reforming the institutions that pro- improving health and education, typically duce inef�ciencies will not lead to sustain- makes up another 10–20 percent of aggregate able improvements. spending.108 Governments contribute a large share of A public responsibility the �nancing for schools and clinics. Wages Governments—and the societies they and salaries on average account for 75 per- represent—often see improving outcomes in cent of recurrent public spending on educa- health and education as a public responsibil- tion—and often for almost all the spending ity. They are supported in this by the interna- (96 percent in Kenya).109 Most teachers and tional endorsement of the Millennium many health workers are civil service employ- Development Goals (see Overview). A variety ees. Salaries aside, government subsidies can of reasons lie behind this responsibility: clas- make up a large share of a facility’s budget. 32 (c) The International Bank for Reconstruction and Development / The World Bank Governments should make services work 33 Public provision Table 2.1 Public expenditures on health and education: large but varied Education and health spending as a share of government expenditures and as a share of GDP, In education, health, water, and electricity the in 2000 or latest year available (percent) public sector is a major provider (if not a Share of public expenditures Share of GDP monopoly) as well as a funder of services. The Average Minimum Maximum Average Minimum Maximum Indonesian government operates more than East Asia and 150,000 primary schools and 10,000 junior Paci�c 27 12 53 6 2 11 secondary schools that cover 85 percent and Europe and 60 percent of the respective enrollments.110 Central Asia 31 18 59 10 4 17 The Ugandan government operated 1,400 Latin America primary level facilities and close to 100 hospi- and Caribbean 33 14 52 8 4 13 tals in 1996.111 The Indian public sector runs Middle East and North Africa 23 13 39 7 4 12 almost 200,000 primary health facilities and South Asia 21 16 25 5 4 8 15,000 secondary and tertiary facilities.112 But wide public provision does not always trans- Sub-Saharan Africa 25 13 34 7 2 12 late into substantial use. In Uganda govern- ment health facilities handled just 40 percent Low-income of treatments sought in facilities.113 In India, countries 25 12 59 6 2 17 even with the huge organization of public Middle-income health facilities, the private sector accounts for countries 29 13 53 8 4 14 80 percent of outpatient treatments and High-income almost 60 percent of inpatient treatments.114 countries 33 20 56 11 3 15 Note: Of the 135 countries included, 52 have data for 2000, 8 for 2001, 30 for 1999, 17 for 1998. The remaining 28 have data from earlier in the 1990s. Reasons for public responsibility Source: World Development Indicators database. Economics gives two rationales for public responsibility. First, because of market fail- ures, the amount of services produced and These market failures call for government consumed would be less than optimal from intervention, but they do not necessarily call society’s standpoint without government for public provision: it could well be that the intervention. Market failures can be external- proper role is �nancing, regulation, or infor- ities. The fact that an immunized child mation dissemination. reduces the spread of disease in society is an The second economics justi�cation for incentive to immunize more children. Basic public responsibility is equity. Improving education might bene�t others besides the health and education outcomes for poor peo- graduate, another externality. Individuals ple, or reducing the gaps in outcomes between have little incentive to build and maintain the poor people and those who are better off, is roads that are crucial to promoting access to often considered a responsibility of govern- services, but communities and societies do. ment. There are a variety of social justice rea- “Public goods� (goods that, once produced, sons behind this. Some see this responsibility cannot be denied to anyone else and whose as rooted in the belief that basic education consumption by one person does not dimin- and basic health are fundamental human ish consumption by others) are an extreme rights (box 2.1). The United Nations Univer- form of market failure. Mosquito control in a sal Declaration of Human Rights asserts an malaria-endemic area is an example. There is individual’s right to “a standard of living ade- no market incentive to produce public goods, quate for the health and well-being of himself so government intervention is required. and of his family, including food, clothing, Other market failures relate to imperfect housing and medical care� and a right to edu- information. Different information about cation that is compulsory and “free, at least in individuals’ risk of illness can lead to a break- the elementary and fundamental stages.�116 down in the market for health insurance. Subsequent international accords have Lack of knowledge about the bene�ts of hand expanded the set of health and education washing or of education can lead to less than rights.117 Many national constitutions have desirable investment and consumption.115 guarantees for health and education. (c) The International Bank for Reconstruction and Development / The World Bank 34 WORLD DEVELOPMENT REPORT 2004 BOX 2.1 Most governments take responsibility for health and education—often appealing to human rights Debates on health care and education in devel- practical implications of approaches based on More generally, the approaches overlap on oping countries often appeal to human rights. rights complement welfare economics. many of their practical policy consequences. Rooted in the broader context of social justice, An approach based on rights emphasizes Both are skeptical that electoral politics and the these rights are set forth in the Universal Decla- equality in dignity and equality of opportunity. market provide enough accountability for effec- ration of Human Rights (1948) as well as other It highlights the need to look at outcomes for all tive and equitable provision of health and edu- international conventions, such as the Interna- individuals and groups, especially the legally cation services—so there is a need for govern- tional Covenant on Economic and Social Rights and socially disadvantaged. It makes explicit a ment and community involvement. An (1966). Several international and bilateral agen- consideration that economics incorporates with economics approach to making services work— cies have endorsed a human rights orientation. dif�culty: many psychological repercussions to such as the one in this Report—is informed by In addition, the constitutions and laws of many poverty result in poor people’s inability to avail the guidance on participation and countries include references to rights to educa- themselves of health care and education empowerment that international human rights tion and health care (a review of constitutional services, even when such services are available. instruments provide. In addition, rights reinforce rights in 165 countries with written Welfare economics provides tools for assess- poor people’s claims on resources overall and constitutions found that 116 referenced a right ing priorities and possibilities for intervening on those allocated for basic services in particu- to education and 73 a right to health care; 95 when budgets are limited —and offers a metric lar—key elements of the effective “voice� of stipulated free education and 29 free health for doing so. Several aspects of economic analy- poor people discussed here. care for at least some services to some groups). sis provide instruments for implementing rights, With education and health central to rights, the complementing a rights-based approach. Source: Gauri (2003). The notion of health and education as public education system focused on national- basic human rights provides a strong basis for ism after the Meiji restoration in Japan.120 public responsibility, but ambiguities remain. Much of this involvement is high-minded: Does a right to medical care imply that gov- a coherent public education system probably ernment must provide it or even �nance it? contributes to social cohesion, particularly The human rights to “periodic holidays with important in fractionalized societies.121 Post- pay� or “equal work for equal pay,� as men- colonial states embraced public provision of tioned in the United Nations declaration, are education as a strategy for nation building. generally not interpreted to imply govern- But public provision can also be the rational ment subsidies.118 Although free elementary manifestation of a state’s desire to inculcate a education is asserted as a right, parents also particular set of beliefs. Tanzania’s 1967 edu- have a right “to choose the kind of education cation reforms were wrapped up with Ujama that shall be given to their children�—sug- and African Socialism. Indonesia’s mass edu- gesting that universal public provision is not cation campaign was closely tied to nation- required.119 Social equity and fundamental building and national ideology codi�ed in human rights suggest a responsibility for gov- pancasila—principles whose teaching was ernment but leave open the ways of discharg- enforced in every school until the fall of the ing that responsibility. Importantly, enshrin- New Order government. ing these notions as rights legitimizes the Beyond nation building and social cohe- demands of citizens—especially poor citi- sion, services operate fully in the political zens—that government take responsibility realm: free education and free health care are for making services work. electoral rallying cries in many countries, Market failures and social justice are nor- popular with many voters. In 1997 Uganda’s mative justi�cations for public responsibility— President Museveni campaigned on a plat- they describe why governments should be form of free universal primary education. involved. They do not always give much guid- The message was extremely popular—he ance on how. Why governments actually get won—and within a short time of�cial enroll- involved provides insight on how public ments nearly doubled (see spotlight).122 responsibility is discharged. Education has long Uganda is not unique: many politicians iden- been a battleground for beliefs, ideas, and val- tify themselves with their stance toward pub- ues. The late 19th and early 20th centuries offer lic provision of services. But success is hard: many stories of this battle, from the movement few politicians have been able to transform for secular primary education in France to a these political platforms into outcomes. Ser- (c) The International Bank for Reconstruction and Development / The World Bank Governments should make services work 35 vices operate in the political realm in yet Figure 2.1 National income and outcomes are strongly associated, especially in low-income countries another way. Many politicians use jobs in the large bureaucracies associated with services Under-five mortality rate, 2000 (log) Primary school completion rate, 1999 (log) 6 6 to reward supporters or to build power. Malawi Cambodia Growth, though essential, 5 5 Nigeria is not enough Madagascar 4 Brazil 4 Given the responsibility to promote education Vietnam and health outcomes, what can governments 3 3 Madagascar do? One approach is hands-off: rely solely on economic growth since higher national 2 2 income is strongly associated with lower child Malaysia mortality and higher primary school comple- 1 1 tion (crate 1.1 and �gure 2.1). Among low- 4 6 8 10 4 6 8 10 income countries, 10 percent more income GDP per capita (log) GDP per capita (log) per capita is associated, on average, with a 6.6 Note: The GDP per capita is based on a 1990s average, 1995 U.S. dollars. Lines show outcomes as predicted by a non- percent lower child mortality rate and a 4.8 linear function of GDP per capita. Source: GDP per capita data from World Development Indicators database; under-�ve mortality from UNICEF (2002); percent higher primary school completion primary completion rates from Bruns, Mingat, and Rakatomalala (2003). rate. Among middle-income countries 10 per- cent more income per capita is associated with 7.7 percent less mortality but little improve- tion can be expected from income growth ment in primary completion. alone? Cutting child mortality by two-thirds At low levels of income relatively small dif- between 1990 and 2015 (one of the Millen- ferences in per capita income can mean big dif- nium Development Goals) means reducing it ferences in outcomes. Per capita income was by 4.4 percent a year. Low-income countries only about $90 higher in Madagascar than in would need sustained per capita income Malawi in the 1990s, but there were almost 50 growth of 6.7 percent a year to reduce mor- fewer child deaths per 1,000 births in Madagas- tality by two-thirds by 2015. Senegal would car in 2000.123 The association between income have to boost per capita income from about and health and education outcomes works $650 to $3,500—close to the level in Panama. both ways: more income leads to lower mortal- Brazil would need an increase from almost ity and more children completing primary $5,000 to $20,000—close to the per capita school; better health and education can lead to income in New Zealand.126 higher productivity and incomes. Studies have Similarly, achieving universal primary tried to disentangle these relationships, and school completion through income alone they typically still �nd income to be a robust would require massive economic growth. In and strong determinant of outcomes.124 Mauritania, where primary school comple- But income is not the whole story: at any tion was 46 percent in 1990, per capita income given income there are wide variations in growth would need to average 6.5 percent a achievement. With average incomes of just year. So while income and outcomes are under $300 per capita in the 1990s, Vietnam strongly associated, especially in low-income had a child mortality rate of about 40 per countries, reaching the Millennium Develop- 1,000 in 2000 and Cambodia of 120. At per ment Goals will require dramatically high— capita incomes around $4,000 in the 1990s, perhaps unrealistically high—growth rates if Malaysia had a child mortality rate of about growth is the only channel for achieving the 12 per 1,000 in 2000 and Brazil of just less goals. Policies that do more than increase than 40. Similarly, Madagascar and Nigeria growth are required.127 both had per capita incomes close to $300 in the 1990s, but by the end of the decade the More public spending alone is primary completion rate was 26 percent in not enough Madagascar and 67 percent in Nigeria.125 If growth is not enough, what else can gov- How much reduction in child mortality ernments do to improve outcomes? One and improvement in primary school comple- approach is to spend more. Increasing public (c) The International Bank for Reconstruction and Development / The World Bank 36 WORLD DEVELOPMENT REPORT 2004 BOX 2.2 The Fast-Track Initiative—providing assistance for credible national education strategies With more than 100 million children not in pri- tries.With impressive speed countries have Some problems for donors mary school, the “Fast-Track Initiative� (FTI) was ensured that their sector plans meet the new tests Despite some progress donor procedures are launched in June 2002 to accelerate progress for credibility and sustainability. And the donors not yet harmonized, and much �nancing toward Education for All in low-income have increased resources for FTI countries, remains fragmented. Some donor assistance countries. Under the FTI national education seconded staff to an international secretariat for under the FTI continues to be input-driven, sub- plans are assessed against an indicative frame- the initiative (in the World Bank), and agreed on ject to a “donor discount,� with resources work of policy benchmarks, prospects for scal- FTI operating principles and guidelines. earmarked for contractors in donor countries ing up, and allowances for flexibility and learn- More generally, the FTI has: rather than providing flexible support for core ing by doing.To ensure that education goals are expenditures. • Raised the political pro�le of Education for embedded in an overall national strategy and Too much aid still flows to historically All, and increased awareness of the need for consistent with countries’ medium-term expen- preferred countries, rather than good perform- faster progress to reach education goals. diture framework, a criterion for FTI eligibility is ers. Although the donors have mobilized addi- a national commitment to a formal Poverty • Sharpened developing countries’ focus on tional funding for FTI countries case by case, Reduction Strategy. primary school completion and quality (not there remain some “donor orphans.�Without The FTI supports countries in addressing key just coverage) and on the importance of get- pooled funding to support these countries, the policy, capacity, data, and �nancing constraints ting policies right. FTI will not be able to deliver on the donors’ to universal primary completion by 2015, net • Brought �eld-based donors into a uni�ed pol- commitment that “no country with a credible intake into �rst grade of 100 percent of girls and icy dialogue with governments, improving plan for Education for All will be thwarted for boys by 2010, and improved learning outcomes. coordination. lack of external support.�The momentum of FTI An initial group of 23 countries was invited to • Mobilized more resources for primary educa- could easily be lost if a fundamental principle of join the initiative, and all accepted. tion (a 60 percent increase in of�cial develop- the compact—assistance supporting effective The FTI was inspired by the Monterrey Con- ment assistance commitments to the �rst FTI policies—is not honored. sensus—that better results accrue when devel- countries). The FTI is a major part of international opment support is targeted to countries that responses—including the G8 process, the Mon- But the experience with FTI has also high- accept clear accountability for results and adopt terrey Consensus, and the New Economic Part- lighted some obstacles. At the country level appropriate policy reforms.The FTI was nership for African Development—to provide these include dif�culties in ensuring that conceived as a process for countries with sound momentum for universal primary completion by resources reach the service delivery level; a need education policies, embedded in an agreed-on 2015, perhaps the most achievable of the Mil- to consider a variety of service delivery macroeconomic framework, to receive added lennium Development Goals. Success will modes—including community-run schools, support from donors. require that developing countries pay attention NGO-run schools, and faith-based schools—and to policy reform and human and �nancial Clear impacts—and obstacles the complexities of public support to this range resources. It will also require that donors coordi- Less than a year into the process the FTI has had of providers; the need to make dif�cult reforms nate better and honor their side of the bargain some clear positive impacts. First, it has demon- to increase ef�ciency and ensure sustainability; by assisting performing countries. strated that the new framework of mutual and the need for better data systems to support accountability is accepted by developing coun- “real-time� tracking of education results. Source: FTI secretariat. spending can be a critical part of promoting Malawi increased by $8 per child of pri- improvements in health and education. For mary school age.128 In Ethiopia primary example, it may be necessary to spend more school completion stagnated, going from on interventions to reduce mortality or on 22 percent in 1990 to 24 percent in 1999, education reforms that underpin increases while in Malawi it rose from 30 percent in primary completion rates—and part of to 50 percent (�gure 2.2). this spending might require international • Per capita public spending on health fell assistance (box 2.2). But the large variation between $1 and $5 in Côte d’Ivoire and in the effectiveness of using funds makes it Haiti from the 1980s to the 1990s: child hard to �nd a consistent relationship mortality worsened substantially in Côte between changes in spending and out- d’Ivoire but improved in Haiti—though comes—highlighting the importance of remaining high (�gure 2.3). spending money well. Just how variable is the association • Thailand increased public spending on primary schooling more than Peru did, between public spending and outcomes? A yet primary school completions fell in glimpse at a handful of countries provides an Thailand and increased in Peru. indication. • Public spending on health diverged in • Between the 1980s and 1990s total public Mexico and Jordan, yet reductions in spending on education in Ethiopia and child mortality were similar. (c) The International Bank for Reconstruction and Development / The World Bank Governments should make services work 37 Figure 2.2 Changes in public spending and income, public spending and outcomes are outcomes are only weakly related: schooling only weakly associated (�gure 2.4)—both Public spending on education Primary school completion substantively (in the sense that the correla- Dollars Percent tion is small) and statistically (in the sense 25 60 Malawi Malawi that the correlation is indistinguishable from 20 zero).133 With similar changes in spending 15 40 associated with different changes in out- Ethiopia comes, it should come as no surprise that the 10 Ethiopia 20 cross-country association is so weak. 5 Why does public spending 0 0 have different impacts? 1000 100 Peru Deeper analysis of the relation between Thailand public spending and child mortality �nds 800 90 Thailand results varying from statistical signi�cance 600 80 to insigni�cance—for four main reasons. Peru First, some countries might spend more 400 70 because they need to spend more to remedy 200 60 urgent underlying health problems. The 0 50 resulting cross-sectional association would 1980s 1990s 1990 1999 be uninformative since more spending would appear to be associated with worse Note: “Spending� refers to total annual public spending on educa- tion per child of primary school age, in 1995 U.S. dollars averaged outcomes. Using statistical techniques that for the 1980s and the 1990s. Primary school completion rates are calculated on the basis of 6 years in Ethiopia (primary plus two exploit the variation in spending that years lower secondary), 8 years in Malawi, 6 years in Thailand, depends on factors unrelated to mortality and 6 years in Peru. Source: Spending data from World Development Indicators data- base; primary school completion data from Bruns, Mingat, and Rakatomalala (2003). Figure 2.3 Changes in public spending and outcomes are only weakly related: child mortality For each country there is a story about Public spending on health Under-five mortality why public spending contributed to improv- Dollars Per thousand ing outcomes or why it did not. That is the 60 200 crux: the effectiveness of public spending 180 varies tremendously.129 In-depth studies con- Côte d‘Ivoire 40 �rm this variability—for example, an analysis 160 of Malaysia over the late 1980s found little 140 association between public spending on doc- 20 Côte d‘Ivoire tors and infant or maternal mortality.130 A Haiti 120 Haiti major improvement in the incidence of pub- 0 100 lic education spending on poor people in South Africa has been slow to translate into 100 100 better outcomes.131 But an impact evaluation 80 of the expansion of public school places in 80 Mexico Indonesia in the 1970s found a signi�cant 60 positive impact on school enrollments.132 40 Jordan Another way to look at the impact of pub- 60 Mexico lic spending is in a cross-section of countries. Jordan 20 In general, countries that spend more public 40 0 resources on health have lower child mortal- 1980s 1990s 1990 2000 ity, and countries that spend more on educa- Note: “Spending� refers to total annual per capita public spending tion have higher completion rates. But this on health in 1995 dollars averaged for the 1980s and the 1990s. Source: Spending data for 1990s from World Development Indica- association is driven largely by the fact that tors database. For Jordan and Côte d’Ivoire, spending data for the 1980s are from World Bank sources. For Haiti and Mexico, spend- public spending increases with national ing data for the 1980s are from Govindaraj, Murray, and Gnanaraj income. After controlling for national (1995). Child mortality data are from UNICEF (2002). (c) The International Bank for Reconstruction and Development / The World Bank 38 WORLD DEVELOPMENT REPORT 2004 Figure 2.4 The association between outcomes and public spending is weak, when controlling sample might yield no association. The num- for national income ber of countries in studies that have Under-five mortality rate*, 2000 Primary school completion rate*, 1999 addressed this question varies dramatically— 150 150 from 22 to 116—so different results should 100 100 not be surprising.138 Second, the speci�cation in the analysis 50 50 can change the assessment of the result. For 0 0 example, controlling for adult literacy in addition to income in the associations illus- –50 –50 trated in �gure 2.4 yields an association and a signi�cance level that are even closer to zero. –100 –100 The message from these studies is not that –150 –150 public funding cannot be successful. It is that –150 –100 –50 0 50 100 150 –150 –100 –50 0 50 100 150 commitment and appropriate policies, Public spending per capita on Public spending per child on backed by public spending, can achieve a lot. health, 1990s average* education, 1990s average* Infant mortality was high in Thailand in 1970 * Public spending, child mortality rate, and primary school completion are given as the percent deviation from rate pre- at 74 per 1,000 births, and the use of commu- dicted by GDP per capita. Note: For the under-�ve mortality regression, the coef�cient is –0.148 and the t-statistic is 1.45. For the primary comple- nity hospitals and health centers was low, in tion regression, the coef�cient is 0.157 and the t-statistic is 1.70. part because quality was low. But the govern- Source: GDP per capita and public spending data, World Development Indicators database; under-�ve mortality, UNICEF; primary completion rate, Bruns, Mingat, and Rakatomalala (2003). ment’s commitment to reduce infant mortal- ity was strong. Health planners took stock, outcomes still produces an insigni�cant rela- analyzing information on service use and tionship.134 Those techniques are not univer- from household surveys. Thailand doubled sally accepted, however, and it is possible that real per capita public spending on health governments are adjusting what they spend in between the early 1970s and the mid-1980s. response to underlying health conditions. But it also did more. It built facilities in Second, spending may affect different remote areas, directed more services to poor groups in society differently. Public spending areas and poor people, improved staff train- could affect child mortality among poor fam- ing, provided incentives for doctors to locate ilies without having a large overall impact. in remote areas, and promoted community Studies allowing for this effect have found a involvement in managing health care deliv- stronger association between spending and ery. The oversight of doctors was strength- outcomes for poor people—but the result is ened. And the authority for various programs only weakly signi�cant and not robust.135 was devolved to the provincial level, freeing A third strand of the research on this issue the central Ministry of Public Health to con- focuses on the composition of spending: does centrate on planning, coordination, and tech- more spending on primary rather than ter- nical support. By 1985 infant mortality had tiary health activities have a different impact fallen to 42 per 1,000 births, and today it is 28 on mortality? The cross-national statistical per 1,000.139 Similar stories are playing out 20 evidence is weak.136 A fourth strand investi- years later in other parts of the developing gates factors that might modulate the effec- world. tiveness of public spending. It �nds that cor- ruption, governance, or urbanization might Public spending on services fails play a role, but the results are inconsistent to reach poor people from one analysis to another.137 Most poor people do not get their fair share Two methodological issues are important of public spending on services, let alone the for interpreting these analyses. First, the sam- larger share that might be justi�ed on ple of countries in a study affects the results. equity grounds. Public expenditure inci- A sample of a few countries that have spent a dence analysis—matching who uses pub- lot and achieved a lot—and a few countries licly �nanced services with how much gov- that have spent little and achieved little—will ernments spend per user—provides a yield a signi�cant association between more snapshot of who bene�ts from government spending and lower mortality. A different spending. Results typically show that the (c) The International Bank for Reconstruction and Development / The World Bank Governments should make services work 39 Figure 2.5 Richer people often bene�t more from public spending on health and education Share of public spending on health and education going to the richest and poorest �fths All health spending Primary health only All education spending Primary education only Guinea 1994 Nepal 1996 Richest fifth India (UP) 1996 Guinea 1994 Armenia 1999 Madagascar 1994 Poorest fifth Ecuador 1998 Kosovo 2000 Ghana 1994 FYR Macedonia 1996 India 1996 Tanzania 1994 Côte d‘Ivoire 1995 South Africa 1994 Madagascar 1993 Côte d‘Ivoire 1995 Tanzania 1993 Nicaragua 1998 Indonesia 1990 Lao PDR 1993 Vietnam 1993 Guyana 1993 Bangladesh 2000 Bangladesh 2000 Bulgaria 1995 Uganda 1992–93 Kenya (rural) 1992 Indonesia 1989 Sri Lanka 1996 Cambodia 1997 Nicaragua 1998 Pakistan 1991 South Africa 1994 Armenia 1996 Colombia 1992 Kyrgyz Rep. 1993 Costa Rica 1992 Kazakhstan 1996 Honduras 1995 Brazil (NE&SE)1997 Argentina 1991 Malawi 1995 Tajikistan 1999 Ecuador 1998 Moldova 2001 Morocco 1999 Brazil (NE&SE) 1997 Peru 1994 Georgia 2000 Yemen 1998 Guyana 1994 Azerbaijan 2001 0 20 40 Vietnam 1998 0 20 40 Mexico 1996 Percent Percent Panama 1997 Note: Figure reports most recent available data. Source: Filmer 2003b. Kenya 1992 Ghana 1992 Costa Rica 2001 Romania 1994 Jamaica 1998 Colombia 1992 Mauritania 1996 0 20 40 0 20 40 Percent Percent poorest �fth of the populace receives less Armenians got almost 30 percent of the than a �fth of education or health expendi- bene�t of public spending on primary edu- tures, while the richest �fth receives more. cation in 1999. But not all spending on pri- In Ghana for example, the poorest �fth mary services is pro-poor. While public received only 12 percent of public expendi- spending on primary health care tends to be tures on health in 1994, whereas the richest more pro-poor than overall spending, it �fth received 33 percent (�gure 2.5). does not always disproportionately reach One reason for this imbalance is that the poor. The poorest �fth of the populace spending is skewed to services dispropor- in Côte d’Ivoire bene�ted from only 14 per- tionately used by richer people. Public cent of public spending on primary health spending on primary education tends to facilities in 1995 (compared with 11 percent reach poor people. The poorest �fth of from all health spending).140 (c) The International Bank for Reconstruction and Development / The World Bank 40 WORLD DEVELOPMENT REPORT 2004 Orienting public spending toward services hard to associate a cost with achieving any used by poor people helps, but it does not target improvement in outcomes, as with the help unless the spending reaches the frontline, Millennium Development Goals (box 2.3). where it bene�ts poor people. A study in Uganda found that in the early 1990s only 13 percent of government primary education Technical adjustments capitation grants made it to the intended des- without changes in incentives tination, primary schools. The rest went to are not enough purposes unrelated to education or to private If more public money is spent on services— gain. Poor students suffered disproportion- and more of that money is spent on services ately, as schools catering to them received even used by poor people and makes it to the smaller shares of the grants (see spotlight).141 intended school or clinic—how the money is The story in health is the same. Drugs used still determines its ef�cacy. Consider intended for health clinics often never get recurrent spending on education in Sub- there. In the mid-1980s more than 70 percent Saharan Africa. Of 18 populous Sub-Saharan of the government’s supply of drugs disap- African countries with data, most spend sub- peared in Guinea.142 Studies in Cameroon, stantially more than the recommended 66 Tanzania, and Uganda estimated that 30 per- percent on teachers (�gure 2.6).146 And this cent of publicly supplied drugs were misap- isn’t just a central government phenomenon. propriated—in one case as much as 40 per- In Nigeria wages account for about 90 per- cent were “withdrawn for private use.�143 cent of local government recurrent expendi- tures on primary education. No one would Private and public sectors interact deny that teachers are a key part of the Public spending has trouble creating quality schooling process and that paying them ade- services and reaching poor people. So why be quately is important. But if there is no money surprised that spending is only weakly associ- left to pay for other important inputs, such as ated with outcomes? But there is another rea- textbooks, learning will suffer. son for the weak association: private and Why does such a large share of education public sectors interact, and what matters is spending go to teachers? Spending on teach- Figure 2.6 The dominant share of the net impact on the use of services. Increas- ers is the result of balancing technical issues recurrent spending on education goes to teachers (selected Sub-Saharan ing public provision may simply crowd out, with political jockeying by parents, teachers, countries) in whole or in part, equally effective services the rest of the civil service, and advocates of obtained from nongovernment providers. spending priorities outside of the education Kenya 96% This works through two channels. First, sector. Spending on other inputs often loses D.R. Congo 90% Tanzania 89% individual demand in both public and pri- out to spending on teachers—who are often Between vate sectors will respond to a change in the vocal, organized, connected, and contractu- 68% and public sector. A review of the impact of price ally obligated to be paid. It happens where 86% increases in public health clinics in seven spending is fairly high—Kenya spends more Malawi Ghana countries found that a substantial percentage than 6 percent of GDP on education—and Angola of visits to public providers deterred by price fairly low—Tanzania spends less than 2 per- Madagascar 58% Ethiopia Zambia increases are redirected to private ones— cent of GDP on education. The purpose is Côte d‘Ivoire although the magnitude of the effect varies not to single out Sub-Saharan Africa—the Sudan Zimbabwe across settings.144 Second, private providers phenomenon is widespread—or to pick on Niger may respond to changes in public provision. teachers. It is to suggest that �xing the prob- Mozambique An experiment in increasing fees in public lem requires dealing not just with technical Uganda Burkina Faso facilities in Indonesia in the early 1990s or managerial questions of how much to Mali found that the number of private dispen- spend on one input relative to others, but Cameroon saries and hospitals increased substantially, with the institutional and political contexts and that this resulted in only small changes in that generate these decisions in the �rst place. health outcomes.145 Identifying what contributes to an effec- Unless resources support services that tive classroom or appropriate medical treat- Note: Data refer to primary school–level work for poor people, they will be ineffective. ment is important for decisionmaking. In spending and primary school teachers. Source: Bruns, Mingat, and Rakatomalala (2003). The ef�cacy of spending is so varied that it is well-functioning systems service provision (c) The International Bank for Reconstruction and Development / The World Bank Governments should make services work 41 BOX 2.3 Why it’s so hard to “cost� the Millennium Development Goals How much will it cost to reach the Millennium or failure of turning that spending into outcomes. Development Goals? That question, crucial for gov- This does not mean “if this amount of money were ernments and donors who have committed to the spent, the Millennium Development Goals will be goals, is extremely dif�cult to answer. met.� It means “if the goals are met, here is what it will have cost.� What cost? For universal primary education completion, does “costing� mean putting a price tag Financing transitional costs. If institutional on enrolling all primary-age children in public reforms are necessary for sustainable improvements schools? With more than 100 million children of pri- in outcomes, the costs of those reforms should be mary school age not in school, multiplying the num- counted: for example, the cost of repurposing physi- ber in each country by average public spending per cal infrastructure or compensating redundant staff. primary student yields a total “cost� of about $10 bil- These costs are determined by country conditions. lion. But this number overlooks a simple point: chil- For example, the cost of a severance package will dren not in school might be harder to induce to depend on a country’s labor market, civil service reg- come to school, so the marginal cost of enrolling a ulations and norms, and other local factors. Given child could be higher than the average cost.These the uncertainty surrounding costs, it makes little children might have higher opportunity costs, so it sense to estimate transitional costs on a global basis. might require a larger subsidy to get them into school. Or they might live in remote areas, where it Interdependence and double counting. would cost more to build schools or to compensate Progress on each Millennium Development Goal them for traveling to more central locations. In addi- feeds back into the others. Safe water and good sani- tion, this approach implicitly assumes that spending tation contribute to better health. Good health on a particular target can be earmarked separately enhances the productivity of schooling. Education from other spending in the sector.Though that is promotes better health. Interventions that promote possible, it is not easy. one goal promote all of them. If the cost of reaching each goal is assessed independently, and the results Ef�ciency gains. The average cost calculation are totaled across goals, there is double counting.149 also ignores the weak overall association between spending and outcomes. Additional spending will be Multiple determinants. But the goals do not associated with only small increases in outcomes if just depend on each other—their determinants are the additional funds are spent with the average multiple, cutting across many sectors. Little is known observed ef�ciency (�gure 2.4).That means it will about the relative contribution of each factor to out- take astronomically high amounts to achieve the comes or about the magnitude of potential interac- goals. But what if the money is “well spent�? tion effects (see crate 1.1 in chapter 1). For example, A country-by-country simulation of spending in the impact of sanitation on mortality depends on 47 low-income countries adjusted the proximate access to safe water.The effectiveness of vaccines, determinants of primary completion success—pub- and thus their contribution to lowering mortality, lic spending as a share of GDP, the share of spending depend on preserving the “cold chain,� which that goes to teachers, the level of teacher salaries, depends on roads, other transport, and energy infra- pupil-teacher ratio, average repetition rate. It found structure. Precise estimates of these independent that average external resources of about $2.8 billion and interactive effects are not easy to come by. a year would be needed. (Since the simulation included domestically mobilized resources as a pol- In sum, costing the goals requires an estimate icy lever, the model yields the amount of external that distinguishes between marginal and average resources required).147 The average-cost approach to cost, incorporates the policy and institutional enrolling out-of-school children in these 47 changes required to make this additional expendi- countries yields a total incremental cost of $3.1 bil- ture effective, does not double- or triple-count given lion a year.148 the interdependence of the goals, and takes into “Costing� a change in proximate determinants is account the multiple determinants of each goal. No useful for identifying the �scal implications of a wonder that coming up with costs of reaching the change in policy, but it says little about the success goals is so dif�cult. can be improved through better teaching incompetent doctors or corrupt police. But materials, more reliable availability of drugs, trying to make services work in weak systems better training of health workers—that is, by an effective and least-cost solution— through technical improvements in the prox- applying a codi�ed set of actions or simply imate determinants of successful service pro- training providers—will end in frustration if vision. And management reforms may politicians do not listen to citizens or if reduce the frequency of shirking teachers or providers have no incentive to perform well. (c) The International Bank for Reconstruction and Development / The World Bank 42 WORLD DEVELOPMENT REPORT 2004 To understand why, it is important to dis- address these problems, and to motivate tinguish between institutional and manage- change, a civil society group introduced rial reforms. Reducing teacher absenteeism report cards in 1994 rating user experiences from 9 percent to 7 percent is a management with public services. The results—revealing issue; reducing teacher absenteeism from 50 poor quality, petty corruption, lack of percent to 9 percent is an institutional issue. access for slum dwellers, and the hidden Improving diagnostic recognition of speci�c costs of outwardly cheap services—were diseases is a management issue. Reducing widely publicized by an active press. widespread mistreatment of routine condi- The report cards gradually opened a dia- tions is an institutional issue. logue between providers and user groups— Institutional reforms seek to strengthen and eventually got a positive response from the relationships of accountability among the managers of public agencies. The state’s various actors so that good service provision chief minister set up a task force to improve outputs emerge—with all their proximate city governance. Follow-up activities—such determinants, including active management. as an in-depth report card for hospitals— Institutional arrangements need to take delved deeper into problems with individ- advantage of the strengths of the market— ual services. In 1999 a report card rated with its strong customer responsiveness, some services substantially higher, though organizational autonomy, and systemic pres- scores on corruption and access to griev- sures for ef�ciency and innovation. And the ance systems remained low. The initiative strengths of the public sector—with its was so successful that the Public Affairs ability to address equity and market failures Centre (which conducted the survey) col- and its power to enforce standards. This is laborated with local partners to prepare not about reducing or avoiding key public similar studies in other Indian cities.150 And responsibilities. It is about creating new other countries (the Philippines, Ukraine, ways to meet public responsibilities more and Vietnam) are adopting the approach. effectively. This might include alternatives to public production, but it could just as Participatory budget formulation in Porto easily include institutional changes to make Alegre, Brazil. The city of Porto Alegre, public agencies perform better. with a population of more than a million, developed an innovative model of budget Understanding what works and formulation. Citizen associations propose why—to improve services projects, which are then publicly debated. To produce better health, better skills, and The proposals are combined with technical better standards of living, service bene�cia- assessments, and the procedure is repeated ries, providers, and the state must work to determine �nal budget allocations. The together. How? Understanding what works, city made substantial strides. Access to why, in what context—and how to spread water went from 80 percent in 1989 to near successful approaches—is the subject of universal in 1996, and access to sewerage, this Report. Many successful institutional from less than 50 percent to 85 percent. innovations worldwide show clearly that School enrollments doubled. And with services need not fail. They offer lessons to greater citizen willingness to pay for better guide replication and to scale up solutions. services, city revenue increased by 50 per- A variety of stories illustrate the potential— cent. To make the process pro-poor, the and the challenge. poorest people had more voting power than others. The approach has proved a resound- Citizen report cards in Bangalore, India. ing success for the inhabitants (and for the In the early 1990s public services in Kar- political party, which repeatedly won elec- nataka’s capital city were in bad shape. A tions). Several other cities have since technology boom unleashed rapid growth. adopted similar procedures.151 Services were of low quality and corruption was rampant, affecting all income groups. Different stories point to other innova- To monitor the government’s failure to tions: greater transparency of school fund- (c) The International Bank for Reconstruction and Development / The World Bank Governments should make services work 43 ing in Uganda, citywide reform in Johan- and what makes some other innovations nesburg, South Africa, cash transfers to fail? Can they be replicated? To examine households in Mexico, statewide reform of these questions systematically—that is, to health services in Ceará, Brazil (see spot- learn from such examples—the Report lights). Only one of these innovations was develops a framework that incorporates the evaluated using an experimental design main actors—service bene�ciaries, the (Mexico’s Progresa). And not all have clear state, and providers—and describes how measures of change in outputs or out- each is linked by relationships of account- comes. But all hint at ways forward. ability (chapters 3 to 6). It then applies The various stories raise questions. Why these principles to speci�c reform agendas, were the innovations implemented? Whom exploring how those relationships play out did they affect? What made them work— in different sectors (chapters 7 to 11). (c) The International Bank for Reconstruction and Development / The World Bank spotlight on Kerala and Uttar Pradesh One nation, worlds apart States in one federal nation—following the same constitution, laws, and intergovernmental �nance system, and subject to the same election cycles—Kerala and Uttar Pradesh remain worlds apart in human development. Their different worlds mean dramatic differences in the quality of life for millions—Uttar Pradesh, with 175 million people, is larger than all but six countries in the world (Kerala has 32 million people).152 This is a story of achievement and failure, the power of public action, and the burden of of�cial inertia. to be well supplied, adequately maintained, in India, Kerala already led in human devel- W omen born in Kerala can expect to live 20 years longer on average than women born in Uttar Pradesh. Uttar Pradesh’s infant mortality rate is �ve times higher than Kerala’s (table 1). At and regularly staffed by teachers or physi- cians. Not so in Uttar Pradesh.153 A primary health center in Kerala left unstaffed for a few days may lead to public protests at the opment in 1956 when it was reconstituted as a new Indian state. Longstanding social movements against caste divisions, its cul- ture (including matrilineal inheritance in nearest district of�ce.154 But a rural school certain communities), and openness to for- the start of this century, one in three girls in in Uttar Pradesh can be nonfunctional for eign influences (including missionary-led Uttar Pradesh had never been to school: Ker- years and produce no civic protest.155 education) all helped. ala has universal enrollment. Kerala’s total Women’s participation differs widely in the But history is not all. Much of Kerala’s fertility rate is 1.96 births per woman (lower two states: more than 70 percent of primary spectacular achievements came after the than 2.1 in the United States and just above school teachers in Kerala are women, only mid-1950s. Adult literacy has risen from 1.7 in high-income European countries); 25 percent in Uttar Pradesh.156 around 50 percent in 1950 to more than 90 Uttar Pradesh’s fertility rate is 3.99 (substan- percent now and life expectancy at birth tially higher than the average of 2.85 for India Why did Kerala succeed where from 44 years to 74. The birth rate has fallen and 3.1 for low-income countries). Uttar Pradesh failed? from 32 to 18. In 1956 the Malabar region of Education and health services in the two Kerala lagged substantially behind the two states echo these differences. Studies sug- History helped. Even though its consump- “native� states (Travancore and Cochin) gest that public facilities in Kerala are likely tion-related poverty was among the highest with which it was combined to form the Table 1 The great divide: human development and basic services in Kerala and Uttar Pradesh new Kerala state. Today, the differences have Latest available data, in percent unless otherwise stated disappeared. Kerala Uttar Pradesh India Public action—and neglect Infant mortality rate (per 1,000 live births) 16.3 86.7 67.6 Dreze and Sen (2002) suggest that Kerala’s Total fertility rate (per woman) 1.96 3.99 2.85 success is the result of public action that Sex ratio (women per 1,000 men) 1,058 902 933 promoted extensive social opportunities Female school enrollment rate (age 6–17 years) 90.8 61.4 66.2 and the widespread, equitable provision of Male school enrollment rate (age 6–17 years) 91.0 77.3 77.6 schooling, health, and other basic services. Rural girls never in school (age 10–12 years) 0.0 31.7 26.6 They argue that Uttar Pradesh’s failures can be attributed to the public neglect of the Rural women never in school (age 15–19 years) 1.6 49.3 38.7 same opportunities. Immunization coverage rate (age 12–23 months) 79.7 21.2 42.0 Skilled delivery care (% of births) 94.0 22.4 42.3 • The early promotion of primary educa- tion and female literacy in Kerala was Rural population in villages with: very important for social achievements A primary school 90.1 75.1 79.7 later on. In Uttar Pradesh educational A middle school 87.1 31.9 44.6 backwardness has imposed high penal- A primary health center 74.2 4.4 12.9 ties, including delayed demographic An all-weather road 79.1 46.0 49.2 transition and burgeoning population Medical expenditure per hospitalization in public facility (Rs.) 1,417 4,261 1,902 growth. Women reporting: • Gender equity and the agency of women Health care provider respected need for privacy 93.0 64.0 68.2 appear to play a major role in Kerala’s success. Uttar Pradesh has a long, well- Health facility was clean 77.2 31.0 52.1 documented tradition of oppressive Skilled attendance at delivery is unnecessary 1.4 42.5 61.3 gender relations and extraordinarily Poorest 20% of households that prefer a public health facility 55.7 9.5 32.8 sharp gender inequalities in literacy and Rs = Rupees in women’s participation.157 Sources: National Family and Health Survey-2, 1998–99; IIPS, 2002; Census of India, 2001; National Sample Survey 1998–99. • Basic universal services in schooling, health care, child immunization, public (c) The International Bank for Reconstruction and Development / The World Bank Spotlight on Kerala and Uttar Pradesh 45 food distribution, and social security By contrast, in Uttar Pradesh caste and action, especially by the politically weak. differ sharply in scope, access, quality, class-based divisions, and the absence of And private action loops back to influence and equitable incidence. In Uttar compelling political alternatives to tran- public action. One set of reforms can lead Pradesh these services appear to have scend these divisions, led to poor political to further institutional evolution. A society been widely neglected and there has incentives for effective provision of univer- can be caught in a vicious cycle, as in Uttar been no particular effort to ensure sal, basic services. Political competition Pradesh, or be propelled by a virtuous one, results, particularly in schools. revolved around access to instruments of as in Kerala. As Dreze and Gazdar note: “In the state to deliver patronage and public Uttar Pradesh, the social failures of the state • A more literate and better informed employment to speci�c clients. Public are quite daunting, but the potential public in Kerala was active in politics and public affairs in a way that did not expenditures in the early years were accord- rewards of action are correspondingly high, appear to have happened in Uttar ingly concentrated in state administration and the costs of continued inertia even Pradesh. and remained well above expenditures on higher.�160 health and education. More recently, politi- As for Kerala, despite the many eco- • Informed citizen action and political cal parties have tended to underplay the nomic problems that linger and the new activism in Kerala—building partly on program or policy content of their plat- ones that have appeared, the remarkable mass literacy and the emphasis placed forms, and instead have publicized the eth- rescript issued in 1817 by Gowri Parvathi on universal services by early commu- nic pro�le of their candidate lists to Bai, the 15-year-old queen of the erstwhile nist and subsequent coalition govern- demonstrate commitment to proportional state of Tranvancore, certainly seems to ments—seem to have been crucial in representation of ethnic groups in the have come true in its bold aspirations for organizing poor people. In Uttar bureaucratic institutions of the state.159 the human development of her subjects. Pradesh traditional caste and power The rescript read: divisions, particularly in rural areas, Breaking out of vicious cycles “The state should defray the entire cost have persisted through more than 50 Public action can build on history, break of the education of its people in order that years of electoral politics—and such from it, or perpetuate it. Individuals’ abili- there might be no backwardness in the divisions have come to form the core ties to press their demands depend on their spread of enlightenment among them, that of political discourse and clientelist information, perceived rights, and literacy. by diffusion of education they might politics. Public action—by influencing information become better subjects and public servants Political incentives matter for service that citizens have, their legal protections, and that the reputation of the state might delivery and actual development out- and their schooling—influences private be enhanced thereby.�161 comes.158 Delivering broad, universal basic services has remained a credible political Figure 1 Kerala spent more on education and health, Uttar Pradesh on state administration platform in Kerala in contrast to the clien- telist, caste, and class-driven politics of Uttar Percent of total public expenditures 50 Pradesh. In Kerala, early governments with eco- nomic platforms emphasizing the provi- 40 State administration (Uttar Pradesh) sion of universal basic services established Education and health (Kerala) a political agenda that remained important 30 Education and health (Uttar Pradesh) in the coalition politics that followed. Political competition conditioned on State administration (Kerala) 20 promises to deliver better basic services showed up in early budget allocations (�g- ure 1): education and health services 10 accounted for a much higher share of pub- 1955 1959 1963 1967 1971 1975 1979 1983 1987 1991 1995 1999 lic expenditures as compared to what was Note: Public spending on state administration does not include interest payments. spent on state administration. Source: Reserve Bank of India Bulletins, 1955–1998. (c) The International Bank for Reconstruction and Development / The World Bank The framework for service provision 3 The public sector has generally taken on arrangements for the provision of services responsibility for the delivery of services requires an encompassing framework—to and frequently used civil service bureaucra- analyze which of the many items on the chapter cies as the instrument. This approach has menu of service reform is right for the had dramatic successes and—as chapter 1 time, place, and circumstance. documented—far too many failures. Much This Report’s framework starts from the remains to be done. Particularly for poor speci�c and works to the general. Start with people, there are widespread challenges in a child in a classroom, a pregnant woman providing affordable access, �xing dysfunc- at a clinic, someone turning a tap for water. tional facilities, improving technical quality, Each is seeking a service, and the proximate increasing client responsiveness, and raising determinants of success are clear. For any productivity. As chapter 2 noted, neither individual service transaction to be suc- economic growth, nor simply increasing cessful, there needs to be a frontline public spending, nor coming up with tech- provider who is capable, who has access to nocratic solutions is enough to meet this adequate resources and inputs, and who is challenge. motivated to pursue an achievable goal. The failures in service provision have The general question: what institutional not gone unnoticed. Indeed, there is a conditions support the emergence of capa- cacophony of proposed institutional solu- ble, motivated frontline providers with tions: civil service reform, privatization, clear objectives and adequate resources? democratization, decentralization, con- The answer: successful services for poor tracting out, provision through NGOs, people emerge from institutional relation- empowerment, participatory methods, ships in which the actors are accountable to social funds, community-driven develop- each other. (Please be patient, the rest of ment, user associations. With each of the Report works out exactly what that sen- these solutions comes a bewildering vari- tence means.) ety of techniques and instruments: This chapter does �ve things. It intro- demand-side transfers, participatory rural duces the analytical framework of actors appraisals, facility surveys, service score (individuals, organizations, governments, cards, participatory budgets. None is a businesses) and relationships of account- panacea. ability that will be used throughout the “One size does not �t all� is a truism but Report. It describes the characteristics of not very helpful. Everyone wanting to services that make creating those relation- improve services for poor people—from ships so crucial—and so dif�cult. It uses the the poor themselves to reform-minded framework and the characteristics of the professionals, advocates, political leaders, services to analyze why pure public sector and external agencies—asks: What size �ts production often fails—and why pure pri- me? Given the capabilities, resources, poli- vatization is not the answer. It lays out how tics, and incentives that I face, what can be the various items on the agenda for service done? What are the actions that would reform are related and how the Report will improve services for poor people in my address them. And it addresses the dynam- circumstances? To evaluate alternative ics of reform. 46 (c) The International Bank for Reconstruction and Development / The World Bank The framework for service provision 47 An analytical framework: cusses) and consistency (the same names Figure 3.1 The relationships of accountability have �ve features actors and accountabilities are used throughout). See box 3.1 for a glos- sary of terms used in this Report. For Language evolves through common usage, Delegating instance, recent work on the empowerment so no one is accountable when the word of poor people, extending the work of the accountability acquires so many different 2000/2001 World Development Report on Financing uses and meanings. What this Report means Actors Accountable poverty, suggests four elements that overlap by accountability is a relationship among (principals) actors in important ways with the analysis here: including (agents) actors that has �ve features: delegation, clients, Performing including access to information, inclusion and partic- �nance, performance, information about per- citizens, policy- ipation, accountability, and local organiza- policy- makers, formance, and enforceability (�gure 3.1). tional capacity.162 Others use the term makers providers Relationships of accountability can be as Informing accountability to refer only to the dimension simple as buying a sandwich or taking a of “answerability� (getting information job—and as complex as running a munici- about performance) or to “enforceability.� Enforcing pal democracy. This Report uses the term broadly. There are two motivations for this • In buying a sandwich you ask for it (del- broader approach. First, weaknesses in any egation) and pay for it (�nance). The sandwich is made for you (performance). aspect of accountability can cause failure. You eat the sandwich (which generates One cannot strengthen enforceability— relevant information about its quality). holding providers responsible for outputs And you then choose to buy or not buy a and outcomes—in isolation. If providers sandwich another day (enforceability), do not receive clear delegation, precisely affecting the pro�ts of the seller. specifying the desired objectives, increas- ing enforceability is unfair and ineffective. • In a typical employment relationship a If providers are not given adequate person is given a set of tasks (delegation) resources, holding them accountable for and paid a wage (�nance). The employee poor outcomes is again unfair and ineffec- works (performance). The contribution of tive. Second, putting �nance as the �rst the employee is assessed (information). step in creating a relationship of account- And based on that information, the ability stresses that simply caring about an employer acts to reinforce good or dis- outcome controlled by another does not courage bad performance (enforceability). create a relationship of accountability. To • In a city the citizens choose an executive be a “stakeholder� you need to put up a to manage the tasks of the municipality stake. (delegation), including tax and budget In the chain of service delivery the Report decisions (�nance). The executive acts, distinguishes four broad roles: often in ways that involve the executive in relationships of accountability with others (performance). Voters then assess • Citizens/clients. Patients, students, par- ents, voters. the executive’s performance based on their experience and information. And • Politicians/policymakers. Prime ministers, presidents, parliamentarians, mayors, they act to control the executive—either ministers of �nance, health, education. politically or legally (enforceability). • Organizational providers. Health depart- There are many other vocabularies for ments, education departments, water referring to these pervasive and critical and sanitation departments. issues from a variety of disciplines (eco- • Frontline professionals. Doctors, nurses, nomics, political science, sociology) and teachers, engineers. practices (public administration, manage- ment). This Report makes no claims of In the ideal situation these actors are coming up with a superior set of words. The linked in relationships of power and terms here have the virtues of completeness accountability. Citizens exercise voice over (a name for everything the Report dis- politicians. Policymakers have compacts (c) The International Bank for Reconstruction and Development / The World Bank 48 WORLD DEVELOPMENT REPORT 2004 BOX 3.1 A glossary for this Report Language is elastic—an asset reflecting the Compacts: The broad, long-term relationship tions for public and private service providers to diversity of human experience, but a liability of accountability connecting policymakers to operate. Usually accountability subrelationships when such overused terms as accountability organizational providers.This is usually not as spe- between politicians and policymakers (parodied lose their meaning.This Report, in developing its ci�c or legally enforceable as a contract. But an in the TV serial “Yes Minister�) are derived from service delivery framework, gives some explicit, veri�able contract can be one form of a the constitution, administrative law, or rules of commonly used terms (such as accountability) compact. public administration. speci�c meaning and we coin a few new terms. Discretionary services: Locally produced Service delivery framework (or chain): The We do not claim we have superior or better services, such as classroom instruction or cura- four service-related actors—citizens/clients, meanings, but we do try for internal consistency. tive care, where the teacher or doctor must exer- politicians/policymakers, organizational Accountability is a set of relationships cise signi�cant judgment on what to deliver and providers, frontline professionals—and the four among service delivery actors with �ve features: how, and where clients typically have a large relationships of accountability that connect information de�cit relative to the provider. Dis- them: • Delegating: Explicit or implicit understand- cretionary services that are transaction-intensive • Voice and politics: connecting citizens and ing that a service (or goods embodying the ser- are hard to monitor, both for the client and for vice) will be supplied. politicians. the policymaker, whether publicly or privately • Financing. Providing the resources to enable provided.They pose particular challenges for all • Compacts: connecting politicians/policy- the service to be provided or paying for it. the relationships of accountability. makers and providers. • Performing. Supplying the actual service. Frontline professionals: The teachers, nurses, • Management: connecting provider organiza- tions with frontline professionals. • Having information about performance. doctors, engineers, clerks, or other providers who come in direct contact with the client. • Client power: connecting clients with Obtaining relevant information and evaluating performance against expectations and formal or Long and short routes of accountability: providers. informal norms. Clients may seek to hold service providers • Short route of accountability: See long and short routes of accountability. • Enforcing. Being able to impose sanctions for accountable for performance in two ways. Client power connecting clients and providers is the inappropriate performance or provide rewards Strategic incrementalism: Pragmatic incre- when performance is appropriate. direct,“short� route of accountability. When such mental reforms in weak institutional client power is weak or not possible to use, environments that are not likely to fully address This Report de�nes four relationships of clients must use voice and politics in their role service delivery problems but can alleviate acute accountability: client power (over providers), com- as citizens to hold politicians accountable—and service problems while at the same time creat- pacts, management (by provider organizations of politician/policymakers must in turn use the ing the conditions for deeper and more frontline professionals), and voice and politics compact to do the same with providers.The favorable change—say, building capacity that (between citizens and politicians/policymakers). combination of the two is the roundabout, can respond to service delivery challenges.This Actors: Individuals, households, communi- “long� route of accountability. can be contrasted with, for lack of a better term, ties, �rms, governments, and other public, non- Management: The relationship of account- “incremental incrementalism� that merely solves governmental, and private organizations that ability connecting organizational providers one set of immediate problems but creates oth- �nance, produce, regulate, deliver, or consume and frontline professionals, comprising internal ers. For example, working around existing gov- services. In economic theory the actors who processes for public and private organizations ernment and governance structures with no hold others accountable are sometimes called to select, train, motivate, administer, and evalu- strategy for how these temporary measures will principals, and the actors who are held account- ate frontline professionals. These processes affect the long term. able are called agents. may be rule-bound in large bureaucracies, or Transaction-intensive services: Services that Client power: The relationship of account- idiosyncratic and ad hoc in small, private require repeated, frequent client-provider contact. ability connecting clients to the frontline service providers. Transaction-intensive services may be providers, usually at the point of service deliv- Organizational providers: Public, private discretionary and require constant, minute deci- ery, based on transactions through which clients nonpro�t, and private for-pro�t entities that sions (classroom teaching), making them very hard express their demand for services and can mon- actually provide services.These may range from to monitor.Or the technology may not require itor supply and providers. government line ministries with hundreds of much discretion (�re and forget) once there is Clients/citizens: Service users who as citi- thousands of employees to a private hospital client contact (immunization). zens participate individually or in groups (e.g., chain or from a vast urban water utility to a sin- Voice and politics: The most complex rela- labor unions) in political processes to shape and gle, community-run, village school. tionship of accountability. It connects citizens attain collective goals. As clients, individuals Politicians/policymakers: The service deliv- and politicians and comprises many formal and receive services to satisfy their household ery actors authorized by the state to discharge informal processes, including voting and elec- demand. All clients are citizens (in most settings) its legislative, regulatory, and rule-making toral politics, lobbying and propaganda, patron- but, depending on the service, not all citizens responsibilities. Politicians may be elected or age and clientelism, media activities, access to are clients. achieve their positions through nondemocratic information, and so on. Citizens delegate to Clientelism: The tendency of politicians as means.They can also be policymakers (the gen- politicians the functions of serving their inter- patrons to respond to political competition by eral who is president but also runs the military, ests and �nancing governments through their excessively favoring one group of clients over the telecom minister who administers the sale taxes. Politicians perform by providing services, another in return for political advantage (vote of frequencies). But more commonly policymak- such as law and order or communities relatively banks). Providing narrow supporter groups with ers are the highest nonelected of�cials—either free of pathogens. Citizens enforce accountabil- free public services or public employment, par- from a civil service or appointed. Politicians set ity through elections and other less de�nitive ticularly where shirking is not sanctioned, is general directions. Policymakers implement means, such as advocacy, legal actions, and often the way politicians practice clientelism. these directions and set and enforce the condi- naming and shaming campaigns. (c) The International Bank for Reconstruction and Development / The World Bank The framework for service provision 49 with organizational providers. Organiza- Politicians and policymakers. What dis- tions manage frontline providers. And tinguishes the sovereign state from all other clients exercise client power through inter- institutions is its monopoly on the legiti- actions with frontline providers (�gure mate use of physical force within its bound- 3.2). In low-income countries a �fth role, aries. From this monopoly, politicians played by external �nance agencies, affects derive the power to regulate, to legislate, to each of these relationships (chapter 11). tax—to set and enforce the “rules of the Weaknesses in any of the relationships— game.� Politicians are de�ned here as those or in the capacity of the actors—can result who control this power and discharge the in service failures. Providers can be made fundamental responsibilities of the state. directly accountable to clients (as in market This does not mean that electoral politics transactions) by passing decisions and pow- are always in play: some politicians are ers directly to citizens or communities—a heads of one-party states, some have “short route� of accountability. But, more imposed their control through military typically, the public sector is involved, so force, some arrive by election. In some sys- two key relationships—voice and com- tems executive politicians are dominant— pacts—make up the main control mecha- in others, legislative politicians. nism of the citizen in a “long route� of The other actors who exercise the power accountability. In either case, organizations of the state are policymakers. In some coun- (such as health, education, and water tries politicians are also policymakers. But departments) need to be able to manage in others there is a clear distinction between frontline providers. the highest nonelected of�cials of govern- ment—civil servants or appointees—and The four actors political actors. Politicians set general direc- Citizens and clients. Individuals and tions, but policymakers set the fundamental households have dual roles, as citizens and rules of the game for service providers to as direct clients. As citizens they participate operate—by regulating entry, enforcing both as individuals and through coalitions standards, and determining the conditions (communities, political parties, labor under which providers receive public funds. unions, business associations) in political processes that de�ne collective objectives; Organizational providers. A provider they also strive to control and direct public organization can be a public line organiza- action in accomplishing those objectives. As tion, whatever the name—ministry, depart- direct clients of service providers, individu- ment, agency, bureau (table 3.1). It can be a als and households hope to get clean water, ministry of education that provides educa- have their children educated, and protect tion services, an autonomous public enter- the health of their family. prise (autonomous public hospitals), a The role of citizens and clients as service bene�ciaries does not imply that all citizens Figure 3.2 Key relationships of power are alike or have the same views. Terms such as civil society and community are some- The state times used too casually. People differ in Politicians Policymakers beliefs, hopes, values, identities, and capa- of accounta bilities. Civil society is often not civil at all; ro u te bili Com ng ty pa Lo c ce many “communities� have little in com- t Voi mon. Individuals and households may dis- S h o rt r o u t e agree about collective objectives and work Citizens/clients Providers to promote their own views, both individu- Coalitions/inclusion Client power Management ally and through associations, sometimes at the direct expense of others. The capability Nonpoor Poor Frontline Organizations for collective action of citizens, a key ele- ment of service delivery, varies widely across societies. Services (c) The International Bank for Reconstruction and Development / The World Bank 50 WORLD DEVELOPMENT REPORT 2004 nonpro�t (religious schools), or a for-pro�t Frontline providers. In the end, nearly all (private hospital). It can be large (public services require a provider who comes in sector ministries with tens of thousands of direct contact with clients—teachers, doc- teachers) or small (a single community-run tors, nurses, midwives, pharmacists, engi- primary school). There can be several types neers, and so on. of providers (public, nonpro�t, and for- pro�t hospitals) and several providers of The four relationships each type delivering the same service in the of accountability same area (many independently operated Of politicians to citizens: voice and politics. nonpro�t and for-pro�t private hospitals). This Report uses the term voice to express When the organizational provider is in the complex relationships of accountability the public sector, one needs to be clear between citizens and politicians. Voice is about the analytical distinction between the about politics, but it covers much more. policymaker and the head of the provider The voice relationship includes formal organization. The policymaker sets and political mechanisms (political parties and enforces the rules of the game for all elections) and informal ones (advocacy providers—including the organizational groups and public information campaigns). provider. The head of the provider organi- Delegation and �nance between citizen and zation makes internal “policies� speci�c to state are the decisions about pursuing col- the organization. Clear conceptually, the lective objectives and mobilizing of public distinction is not always clear in practice, resources to meet those objectives. Citizens especially when the same individual plays need information about how actions of the both roles. For example, a minister of pub- state have promoted their well-being. They lic works may be the policymaker responsi- also need some mechanism for enforceabil- ble for making and enforcing the rules for ity, to make sure that politicians and policy- all providers—but also the head of the makers are rewarded for good actions and largest organizational provider of water ser- penalized for bad ones. If politicians have vices, directly responsible for management. abused their position, or even just not pur- Unbundling these roles to create a clear sued objectives aggressively and effectively, delineation of policymaking and direct pro- citizens need a variety of mechanisms—not duction responsibilities is one element in just periodic elections—to make politicians having clear lines of accountability. and policymakers accountable. Table 3.1 Organizational providers take a variety of ownership and organizational structures Ownership Type of provider Education services Health services Water services Energy organization (ambulatory curative care) Public sector Ministry/department/ Ministry of education Ministry of health Ministry of public Ministry of energy agency/bureau schools (national, state/ outpatient clinics works province, municipal) Public sector Autonomous Autonomous hospitals State water State electricity autonomous corporation universities corporations companies Not-for-pro�t sector Community owned Informal schools, Rural water Educo associations Not-for-pro�t Religious schools NGO-run clinics organization (Catholic, Islamic), NGO-run schools (such as BRAC) Private, for-pro�t sector Small for-pro�t �rms Private, nonreligious Private clinics Informal water vendors schools Large Hospital chains Private utilities Private utilities Note: Educo = El Salvador’s Community-Managed Schools Program; NGO = Nongovernmental organization; BRAC = Bangladesh Rural Advancement Committee. (c) The International Bank for Reconstruction and Development / The World Bank The framework for service provision 51 Of the organizational provider to the state: compacts. The relationships between pol- BOX 3.2 The many meanings of accountability icymakers and service providers can be Accountability is more a rubric than a single trains run on time�), but not offer any verti- thought of as compacts. The compact is not item, but it is a fruitful rubric for making cal accountability. But where the relation- always as speci�c and legally enforceable as useful distinctions. ship between clients and providers is very a contract, though a contract can be one Political accountability is the willingness strong (in some instances perhaps because of politicians and policymakers to justify their of the omnipresence of the ruling party, as form of a compact. Instead, it is a broad actions and to accept electoral, legal, or in Cuba), service delivery may work very agreement about a long-term relationship. administrative penalties if their justi�cation is well without much vertical accountability. The policymaker provides resources and found lacking. Even within “political�account- Even for a given type of accountability delegates powers and responsibility for col- ability one can have distinctions.With vertical there are distinctions. Formal horizontal political accountability, citizens individually or political accountability is the formal lective objectives to the service providers. collectively hold the state to account—say, description of institutions, and authority The policymaker generates information through voting or advocacy. Democracies among agents of the state. It may differ about the performance of organizations. must have some vertical accountability.With sharply from informal horizontal political Enforceability comes into play when the horizontal political accountability, agents of accountability, from the actual working of the state formally hold another agent of the institutions and effective control over deci- compact also speci�es the rewards (and state accountable—say, through the sions in state organizations. possibly the penalties) that depend on the “compact�relationship between policymak- service provider’s actions and outputs. The ers and providers. line between “the state� and “public sector Authoritarian states may manifest con- Sources: Goetz and Jenkins (2002) and Aghion siderable horizontal accountability (“the and Tirole (1997). organizational provider� is not always easy to draw. Of the frontline professionals to the organi- Accountability is not zational provider: management. In every the only relationship organization, formal and informal tools of The foregoing description is not reality, management provide frontline workers because it portrays only one direction in the with assignments and delineated areas of relationships between actors. The reality is responsibility, equipping them with the that actors are embedded in a complex set resources to act. In public agencies this of relationships, and accountability is not management function is at times blurred always the most important. Through vari- because providers are employees of “the ous forms of coercion, both subtle and bla- government.� But all the standard manage- tant, many states’ ability to impose obliga- ment issues of selecting, training, and moti- tions on citizens has proved much stronger vating workers in an organization apply to than the ability of citizens to discipline all organizations—private, NGO, govern- politicians and policymakers (box 3.2). And ment, whatever. All service provision orga- in many cases citizens approach the state nizations—whether a government min- and its agents as supplicants. istry, a religious body, a nonpro�t NGO, or Politicians often use the control over a for-pro�t �rm—have to create a relation- publicly provided services as a mechanism ship of accountability with their frontline of clientelism—for both citizens and providers. providers. In systems that lack accountabil- ity relationships, public service jobs (teach- Of the provider to the citizen-client: client ers, policemen) are given as political favors, power. Because the policymaker cannot which creates a relationship not of account- specify all actions of providers in the com- ability but of political obligation. A recent pact, citizens must reveal to providers their report on education in Nepal, for instance, demand for services and monitor the �nds that “teachers’ performance standards providers’ provision of services. Clients and are nonexistent. Most teachers are aligned organizational providers interact through with one of the many associations formed the individuals who provide services— on political party lines and appointment teachers, doctors, engineers, repairmen— and deployment practices are often deter- the frontline professionals and frontline mined as a result of individual’s contribu- workers. tions to political activities.�163 (c) The International Bank for Reconstruction and Development / The World Bank 52 WORLD DEVELOPMENT REPORT 2004 Services are allocated in ways that Individual interests reward (or punish) communities for their and collective objectives political support. Sometimes the ministry is A competitive market automatically creates the agent of the providers, not the other accountability of sellers to buyers. The key way around, and providers capture the poli- information is customer satisfaction, and cymaking. Providers also use their ability to the key enforceability is the customer’s control services and their superior social choice of supplier. Competitive markets status to intimidate poor people. Rather have proved a remarkably robust institu- than client power, there is provider power. tional arrangement for meeting individual The political scientist James Scott has interests. But they are not enough for ser- argued that the pressures of “authoritarian vices—for three reasons. high modernism� can mean that the state and its bureaucratic apparatus de�ne a • First, the market responds only to those “thin simpli�cation� in order to carry out with purchasing power, doing nothing to services—but that the domination of this ensure universal access or an equitable reality over citizens and their complex real- distribution, which societies often have ity can lead to unintended consequences. as a collective objective. • Second, the sum of the individual inter- Why establishing relationships ests may not produce the best outcome of accountability is so complex because markets may have failures of This Report moves beyond what the public various kinds. sector should do and emphasizes how pub- • Third, other collective objectives require lic action can be made most effective. public action. For instance, the state and Frontline workers have to have clear objec- society have a strong concern about the tives, adequate resources, technical capabili- role of schooling in the socialization of ties, and the motivation to create valued youth and may not want parents to services. This cannot be mandated. It is the choose for themselves. result of interactions between strong actors in each of the key service provision roles. The problem of monitoring The ideal: a state that is strong, not weak. Locally produced services—basic educa- Provider organizations that have a clear tion, health care, urban water supply and vision and mission of service provision, not sanitation—have three characteristics that ones that are internally incoherent and make it particularly dif�cult to structure merely process oriented. Frontline relationships of accountability. They are providers acting with professional auton- discretionary and transaction-intensive. omy and initiative, not tightly controlled There are multiple tasks and multiple prin- automatons. And empowered citizens who cipals. And it is dif�cult to attribute out- demand services, not passive “recipients� comes. who are acted on. Strong, capable actors need to be embed- Discretionary and transaction-intensive. ded in strong relationships of accountability. Services are transaction-intensive, and the But it is dif�cult to establish such relation- transactions require discretion. Teachers ships for these services. Why? must continuously decide about the pace and structure of classroom activity. Have • Because there are both collective objec- the ideas been grasped? Will another exam- tives and private objectives, a system that ple reinforce the idea or bore the class? A created only client power through choice doctor has to make decisions about diagno- (say) would meet only individual objec- sis and treatment based on the speci�c case tives, not the many public ones. of the patient. The examples differ from • Because of the multiple, complex objec- other public sector activities that are discre- tives of public production and co- tionary but not transaction-intensive, such production, it is dif�cult to create out- as setting monetary policy or regulating a come-based enforceability for providers. monopoly—or those that are transaction- (c) The International Bank for Reconstruction and Development / The World Bank The framework for service provision 53 intensive but not discretionary, such as tak- Attributability ing in bank deposits or controlling traf�c. The third problem in monitoring service Services may be transaction-intensive provision is that it is often very dif�cult to and discretionary, but some stages in ser- attribute outcomes to the actions of the vice provision may be less transaction- service providers because there are impor- intensive or discretionary (table 3.2). Even tant “co-producers.� As chapter 1 empha- in the health sector, services span the range. sized, health and education outcomes are For immunization, the appropriate action is mainly produced in households and com- nearly the same for each individual of a munities. The health of individuals given age (easily observed). The problems depends on their decisions about nutrition in implementation, while formidable, are (constrained by income), activity levels, primarily logistical. But for curative ser- personal hygiene practices (often con- vices, providers have to respond to com- strained by the availability of water)—and plaints from individuals and exercise discre- on community factors that determine tion in choosing treatment. exposure to pathogens. Even if people seek Services that are both discretionary and treatment when they are sick, the effective- transaction-intensive present challenges for ness of treatment depends in part on any relationship of accountability—because provider quality and individual compliance it is dif�cult to know whether the provider with the recommended therapies. has performed well. Administrative and The dif�culty in monitoring discre- bureaucratic controls that work well for tionary, transaction-intensive services is not logistical tasks are overwhelmed when they unique to the public sector—it is inherent in attempt to monitor the millions of daily services. Patients generally know how they interactions of teachers with students, feel. Studies of private practitioners in India policemen with citizens, case workers with commonly �nd practices that lead to short- clients, medical practitioners with patients. run improvement in symptoms (such as Rigid, scripted rules would not give enough steroid shots) but are not medically effec- latitude. tive—or are even counterindicated.165 Patients feel better, and this attracts repeat Multiple principals, multiple tasks. Public customers. But it does not create real servants serve many masters. Power and accountability, because simply being pleased water providers are under pressure from with the service is not suf�cient information. different segments of the market to cross- subsidize them—from producers to buy Table 3.2 Examples of discretionary and transaction-intensive services speci�c types of equipment, from people Sector Discretionary, not Discretionary and Transaction-intensive, who want more extensive connections, and transaction-intensive transaction-intensive not discretionary from others who want more reliable, con- Commercial banking Setting deposit rates Approving loans to Taking in deposits tinuous operation. The day-to-day pressure small businesses of local demand for health care can com- promise efforts in disease prevention and Social protection Setting eligibility “Case worker� Issuing checks to the criteria determinations eligible other public health activities that are not demand-driven.164 Policing Lawmaking de�ning Handling individual criminal behavior conflict situations Directing traf�c Personnel in health clinics are supposed to provide immunizations, curative care to Education Curriculum Classroom teaching Providing school lunches people who come to them, health education and other preventive measures to everyone Health Public information Curative care Vaccinations (whether they come in on their own or campaigns not), keep statistics, attend training sessions Irrigation Location of main Allocation of water Providing standpipes and meetings, and do inspections of water canals flows “in every village� and food. Police of�cers have to deal with Central banks Monetary policy Banking regulation Clearing house everyone from lost children to dangerous Agricultural extension Research priorities Communication with criminals. This diffusion blunts the preci- farmers sion of incentives (box 3.3). (c) The International Bank for Reconstruction and Development / The World Bank 54 WORLD DEVELOPMENT REPORT 2004 BOX 3.3 Creating conditions of accountability: the police Police are delegated substantial powers—to vague objectives, lots of discretion, little perfor- to allocate police time and visibility. Crime rates compel and, if necessary, to use violence. What mance information, few mechanisms of enforce- fell signi�cantly.This approach can back�re, objectives should they pursue, and how could ability (either internal or external), and the pub- though, if the desired outcomes are not well they be held accountable? lic authority to compel (and often too little speci�ed. Studies of police behavior in London budget). A frequent complaint of poor people is and Los Angeles showed that the monitored • “Client satisfaction� is not what should drive the abuse they suffer from the police. As one and numerically measured activities (crime rates police, for who is the “client�? Certainly not Kenyan put it recently “You cannot carry much and citizen complaints) improved markedly. But the criminals, and certainly not just the vic- money with you these days.There are too many other measures—community activities and tims: there are many objectives—creating a policemen.� crime rates, particularly for homicide—got safe environment, apprehending criminals, There are no easy answers.“Privatizing� worse. respecting individual rights and dignity. policing functions would face the same prob- So there is no general “optimal� solution. But • Police cannot simply follow a script—they lems: what would be the measure of output to there are solutions to particular cases, better or have to exercise discretion. If they went “by determine what the �rm should be paid? Crime worse in their adaptation to local circumstance. the book� and enforced every infraction, more rates? They are not under police control (and Creating more functional police services important activities would grind to a halt. they would deter reporting). Arrests? That would requires creating multiple institutional channels • They rely on many co-producers. Without the encourage false arrests to meet production quo- tas. Surveys of citizen perceptions of safety? of accountability—political (police are not sim- ply an instrument of oppression), compacts cooperation of citizens in abiding by the law, reporting violations, helping in investigations, These risk overzealous police violating the (policymakers can hold police in check), man- the job of the police would be impossible. rights of the socially disadvantaged to please agement (organizational strategies can And many determinants of crime are not the minority. Penalties for abuse of authority? inculcate dedication, loyalty, restraint), and client under the control of the police, such as eco- Police might then do too little. power (citizens have mechanisms to influence nomic trends, social changes, and Recent experience in several cities, notably police behavior directly, a free press). demographic shifts. New York, shows that better measurement of several important outputs is possible. Crime Sources: Moore and others (2002); The Economist The recipe for inef�ciency, abuse, and cor- rates were measured by neighborhood, (2002); Burguess, Propper, and Wilson (2002); Pren- ruption: simply turn individuals loose with reported regularly as a management tool, used dergast (2001). Many outcomes, even when observable to polar positions, using the �ve failures of the patient and the doctor, are not “con- services detailed in chapter 1—inaccessibil- tractable� in the sense of being able to prove ity, dysfunctionality, low technical quality, compliance to a judge or other mediator if a lack of client responsiveness, and stagnant dispute arises. productivity. Successes and failures of the Public production public sector and the market Two of the most powerful innovations of the Discussions of public action often juxta- long 20th century (1870–1989) are the pose two polar extremes for the institu- mutually reinforcing ideas of the nation- tional arrangements for services: traditional state, with extensive powers and responsibili- public production, in which all public ties, and the civil service bureaucracy. action and resources are channeled through Together they produced the consensus that a public sector organization with civil ser- governments have responsibilities for the vants; and market production, in which the welfare of their citizens, and that the most public sector takes a minimal role (but at effective way to ful�ll these responsibilities is least establishes the basic conditions for a through the direct production of services market, such as enforcing contracts). through a public sector organization with This Report seeks to help the public sec- civil service employees. The contested ide- tor meet its responsibility for health and ologies of the 20th century—communism, education outcomes. The public sector can capitalism, democracy—pale before the discharge its responsibility by engaging in a power of the twin ideas of a nation-state and variety of institutional arrangements for a public sector bureaucracy. These ideologies service provision, including direct produc- were merely notions of the uses for the tion, contracting out, demand-side trans- nation-state and its bureaucracy. fers, and so on. Before getting to them, it As just one example, schooling in the helps to illustrate the weaknesses of the two middle of the 19th century was almost (c) The International Bank for Reconstruction and Development / The World Bank The framework for service provision 55 exclusively in private hands (largely reli- effective organizations in the world are gious). Today the direct production of public agencies. schooling by the public sector—with the When the long route is not working, the nation-state the dominant service provider, framework provides a way to understand involved in every facet of schooling from the failures by identifying which relation- building schools, to determining the cur- ship of accountability was the weak link— riculum and texts, to training, hiring, and within the relationship of accountabil- assigning, and controlling teachers as civil ity, which was the missing dimension. servants—has completely triumphed as an idea, so completely that people forget it was • There are voice failures, when the state (controlled by politicians and policy- ever contested (box 3.4). makers) simply does not care about pro- Public bureaucracies are truly a blessing viding services—or does so in a strictly of modern life. All countries with high liv- venal or clientelist manner. The clearest ing standards have teachers who teach, sign of this: when too little budget is police of�cers who police, judges who devoted to services for poor people, and judge, public works that work, armies that when that budget is allocated to meet respond to external threats. Yes, bureaucra- political interests. cies might be frustrating, slow, inef�cient, and resistant to innovation. But the fantasy of “getting rid of the bureaucracy� would turn into a nightmare. No country has BOX 3.4 The “Progressive Era�: creation of modern developed without state reliance on an effective public bureaucracy to discharge bureaucracy the key functions of the state—though not Samuel Hays, in his study of the evolution of networks of human interaction.This upward always through direct production. So why conservation policy in the United States in shift can be seen in many speci�c types. . . : do some bureaucracies perform badly and the early 20th century, expresses eloquently the growth of city-wide systems of execu- the political and social tensions in the shift tive action and representation in both others well? And how do countries get from to modern bureaucracies: school and general government to super- badly to well? “The dynamics of conservation, with its sede the previous focus on ward represen- The analytical framework of the rela- tension between the centralizing tenden- tation and action; the similar upward shift in tionships of accountability provides a way cies of system and expertise on the one the management of schools and roads from hand and decentralization and localism on the township to the county and state. of diagnosing not just the symptoms of the other, is typical of a whole series of simi- “Examination of the evolution of con- poor performance (inef�ciency, corruption, lar tensions between centralization and servation political struggles, therefore, poor performance) and not just the proxi- decentralization within modern . . . society. brings into sharp focus the two competing mate determinants of these symptoms (lack The poles of the continuum along which political systems. . . . On the one hand the these forces were arrayed can be described spirit of science and technology, of rational of resources, low motivation, poor training, briefly. On the one hand many facets of system and organization, shifted the loca- and little capability). It also provides a way human life were bound up with relatively tion of decision-making continually upward of analyzing the deep institutional causes of small scale activities focusing on the daily so as to narrow the range of influences routines of job, home, religion, school and impinging upon it and to guide that deci- poor performance. recreation in which a pattern of inter-per- sion-making process with large, cosmopoli- In public sector production the direct sonal relationships developed within rela- tan considerations, technical expertness, link of client power is frequently missing, so tively small geographical areas. . . . On the and the objectives of those involved in the successful public production relies on other hand, however, modern forms of wider networks of modern society. On the social organization gave rise to larger pat- other however, were a host of political “long-route� accountability. What does that terns of human interaction, to ties of occu- impulses, often separate and conflicting, take? The policymaker must care about out- pation and profession over wide areas, to diffuse, and struggling against each other comes, including those for poor people. corporate systems which extended into a with in the larger political order.Their politi- That concern needs to be transmitted effec- far flung network, to impersonal—statisti- cal activities sustained a more open politi- cal—forms of understanding, to reliance on cal system . . . in which complex and esoteric tively to the public agencies that receive expertise and to centralized manipulation facts possessed by only a few were not per- public resources to provide the services. and control. . . .To many people the external mitted to dominate the process of decision- And the public agencies must hire techni- characteristics of this process—ef�ciency, making, and the satisfaction of grassroots cally quali�ed providers motivated to pro- expertise, order—constituted the spirit of impulses remained a constantly viable ele- “progressivism.�These new forms of organi- ment of the political order.� vide the services. When all this happens, as zation tended to shift the location of deci- it often does in developed countries, public sion-making away from the grass-roots, the service production is reliable and effective. smaller contexts of life, to the larger Source: Hays (1959). Indeed, some of the most admired and (c) The International Bank for Reconstruction and Development / The World Bank 56 WORLD DEVELOPMENT REPORT 2004 • There are compact failures, in which the determine who bene�ts; about the alloca- state fails to communicate clear responsi- tion of expenditures across inputs (how bilities for outputs or outcomes to the much for wages and other things). If public organization and fails to enforce resources are inadequate, if they are ineffec- any responsibility. Compact failures are tively applied to service provision for poor also associated with management failures, people, it is often because poor people’s in which the public sector organization voices are not being heard. fails to motivate its frontline workers. Nor is much information generated that would allow citizens to judge how effec- All this is embedded in a system in which tively their government is providing ser- the feedback loop from client satisfaction to vices. Since information is power, it is often both frontline and organizational providers closely guarded—or never created in the is cut. �rst place. Politicians seldom create infor- mation about outputs and outcomes. Indi- Voice. A common cause of the failure of viduals know about the quality of the ser- public service production is the apathy of the vices they confront, but they have a dif�cult state. Governments may care about some time translating that knowledge into public services for ideological reasons. But when power. Indeed, politicians may use the voice is weak (or divided or conflicted) and selective provision of services as a clientelis- the state is freed from the constraint of satis- tic tool to “buy� political support—or, fying its citizens, there are many possibilities worse, to enforce state control of citizens for failure. The state delivers little or nothing while weakening their voice (box 3.5). to its poor and socially disadvantaged citi- zens, reserving its few services for the elite, Compacts. The complex compact rela- including favored members of the govern- tionship fails in many ways. In failed or fail- ment. In these circumstances alternative ing states (such as those the World Bank strategies of public sector management will calls Low-Income Countries Under Stress, be powerless to create better services. or LICUS) there is no compact because the Many analysts and advocates point out state’s control is very shallow. This happens that resources devoted to services are inade- when countries are embroiled in long civil quate. But those budget allocations are the wars (Afghanistan, El Salvador, Somalia, result of political decisions: about the level Sudan) or large parts of the country are of taxation and mobilization of resources; beyond the reach of government (Democ- about the allocations of budgets across ratic Republic of Congo). activities; about the design of programs that Even in working states the compact rela- tionship between the state and public provider agencies is often extremely weak. BOX 3.5 Seeking services in the Arab Republic of Egypt The delegation and speci�cation of goals An anthropological study of urban Cairo connections, personal relationships, and are often vague or nonexistent, and there detailing the “Avenues of Participation�— outright bribes to of�cials.There also were rarely are clear responsibilities for outputs the ways residents coped with the demands explicit “patron-client� relationships: or links to outcomes. Budget allocations of the state, and sought its favors—revealed a pattern common in many countries. First, In a particular relationship I was able to and staf�ng for agencies are determined observe closely that the ties between without any direct relationship to past per- there is a huge gap between the formal and “patron�and the supposed “client�were very informal realities. As a manager of a family- close and reciprocal.The “client�received formance or clearly speci�ed objectives. owned shoe factory explained: This means that providers are often under- loans from the politician, gifts of food and We are caught in the middle of two totally sep- clothing for her family, publicly subsidized funded relative to announced rhetorical, arate systems that do not communicate with apartments, employment for her and her and unrealistic, targets. each other.One of them is the legal [formal] family, assistance with bureaucratic problems, system.The other one is what we call the tradi- and a great deal of information. . . . At election Without clear delegation of responsibili- tional system,which is much stronger than the time the client returned this service by orga- ties and identi�ed objectives, there is no way law.That is what really controls us.(p.205). nizing the election campaign and marshaling of generating the relevant performance infor- local political support in the district. mation for managing or assessing the organi- That gap means that, in approaching the state, individuals must rely on informal Source: Singerman (1995). zation. Without clear information on organi- zational objectives and progress, it becomes (c) The International Bank for Reconstruction and Development / The World Bank The framework for service provision 57 impossible to create enforceability. This also lowed strictly. That approach can succeed for discourages innovation and responsiveness. truly logistical tasks, but it can also be coun- There may be many isolated successes in ser- terproductive. By constraining the profes- vice provision and striking examples of pub- sional autonomy of frontline providers, it lic servants succeeding even against the odds. may frustrate self-motivated frontline work- But nothing in the system encourages the ers, driving them away and undermining the replication of successful innovations. development of strong providers. This is not to deny the enormous bene�ts The goal is to have providers with more that public provision has attained. But those capability, more autonomy, and more discre- bene�ts have often been in areas in which the tion in providing quality services. But more compact relationship is relatively easy autonomy requires more performance-based because the targets are numerical and provi- accountability. That is intrinsically dif�cult to sion is logistical. Strong states, even the polit- create because of the multiple (often unob- ically repressive, have been successful in pro- servable) objectives of public action, the viding services. Socialist states, such as China, demands of monitoring discretionary and have had great success in the social sectors. transaction-intensive services, and the dif�- But moving beyond the impressive logistical culty of attributing outputs or outcomes to accomplishment and improving quality has actions by providers. proved much more dif�cult. Even weak states Take schooling (chapter 7): good teaching can launch and sustain vertical programs of is a complex endeavor. The quality of a logistical delivery—expanding childhood teacher cannot be assessed strictly on the vaccinations in very troubled situations is a basis of student scores on a standardized classic example. But going from services pro- examination. Why not? Schooling has many vided in “campaign� mode to more discre- other objectives. It is dif�cult to isolate the tionary and quality-sensitive services has value added. And simply paying and promot- proved much more dif�cult. ing all teachers the same does not motivate good teaching—it can even lower morale Management. Failures of management are among motivated teachers. also common in public production of ser- Perhaps good teaching can be assessed vices. Frontline workers rarely receive subjectively by another trained educator—a (explicit or implicit) incentives for successful head teacher or school principal. But this cre- service delivery. There are no stipulations for ates the temptation to play favorites or, worse, service quality and quantity, no measurement to extract payments from teachers for good of effectiveness or productivity, few rewards assessments. So the autonomy of school heads or penalties. The provider organization mon- must be limited by accountability, to motivate itors only inputs and compliance with them to reward good teachers. There must be processes and procedures. Even so, some an assessment standard for school heads. But states have provided some services under all the problems of assessing good teaching these conditions, but the services remain lim- also apply to good school heads. Indeed, that ited, low in quality, high in cost. is how dysfunctional bureaucracies cascade The problems are deep. Quick �xes that into a morass of corruption, as upward pay- seem too good to be true probably are. One ments from those at lower levels buy good response to the corruption, absenteeism, and assignments or ratings from superiors. underperformance of providers is stricter monitoring. But if the objectives are not well The market known and if it is dif�cult to monitor behav- The “market,� as an idealized set of relation- ior, it is dif�cult to assess performance on the ships of accountability, relies more or less basis of real, relevant output measures. So exclusively on client power—and only on “accountability� is instead created by strict that part of client power that is based on rules, intended to prevent abuse, and choice, backed by purchasing power. Cus- attempts to monitor compliance with some tomer power is the main relationship of crudely measured proxies (attendance) or to accountability. The market has several reduce the activity to scripts that must be fol- strengths in the provision of services—but (c) The International Bank for Reconstruction and Development / The World Bank 58 WORLD DEVELOPMENT REPORT 2004 also many weaknesses. One strength is that influence—the short route. In some extreme customers will buy where they perceive the cases where the long route breaks down sud- greatest satisfaction—so organizations have denly, as in the aftermath of the breakup of incentives to be responsive to clients. Another the Soviet Union, reliance on the short route strength is that since the organizations are arises by default (box 3.6). But increased autonomous, they can manage their frontline reliance can be deliberate, forming the basis providers as they wish. Yet another is that of a wide variety of institutional reforms, with a variety of organizations providing ser- each with strong advocates. vices, each can be flexible with innovation In education people believe that schools and each has the incentive to adopt successful will improve with more use of choice innovations (or else lose resources). Markets through vouchers, greater community con- produce innovations and scale them up by trol, greater school autonomy, having more trial and error followed by replication and information about budget flows and more imitation—for organizational innovation as aggressive testing and school-based account- well as product innovation. ability. In health people believe that care will But for the services in this Report, the improve through greater demand-side market has three weaknesses. �nancing (and less public production), more use of vertical programs for speci�c diseases, • It responds exclusively to customer power, and community control of health centers. so there are no pressures for equity (much Others emphasize solutions that cut across less equality) in the allocation of services sectors: community-driven development, (though it is not obvious that political sys- participatory budgeting, power to local gov- tems lacking strong citizen voice have any ernments, “new public management,� and greater pressures for equity). civil service reform. • It will not, in general, satisfy collective All these proposals aim to improve ser- objectives (simply adding up individual vices by changing the relationships of objectives). For instance, if one person’s accountability. All recognize that, though use of adequate sanitation affects those there are many proximate causes of failure, who live nearby, individuals may under- the deep causes lie with inadequate institu- invest in sanitation. tional arrangements. If frontline workers • It can be effective in having customer (civil servants) in the existing organizations power discipline providers only when of public production are frequently absent, the customer has the relevant informa- have little regard for clients who are poor, and tion about provider performance. In lack the technical knowledge to perform their ambulatory curative care it is easy for services well, this inadequate organizational customers to know their waiting time capacity is the proximate cause of poor ser- and to know how they were treated. But vices. Too frequently those seeking improve- it is very dif�cult for them to know ment have focused only on internal organiza- whether the medical treatment they tional reforms—focusing on management of received was effective and appropriate the frontline workers. If organizational fail- for their condition. ures are the result of deeper weaknesses in institutional arrangements (weak political From principles to instruments commitment, unclear objectives, no enforce- This Report uses the framework of actors and ability), direct attacks on the proximate their relationships of accountability and determinants (more money, better training, power to understand the successes and fail- more internal information) will fail. ures of centralized public service produc- Different systems can underpin success. tion—and to evaluate reforms and new pro- For example, countries have very different posed institutional arrangements for service institutional arrangements in corporate provision. Given the failures and limitations �nance. Crudely put, in Japan �rms own of the traditional model of service provi- banks, in Germany banks own �rms, and in sion—the long route—greater reliance will the United States banks and �rms are sepa- inevitably be placed on more direct client rated. All three countries have very high levels (c) The International Bank for Reconstruction and Development / The World Bank The framework for service provision 59 BOX 3.6 Health care in Central Asia and the Caucasus: the long and short of it The upheavals that accompanied the breakup of ing to the degree of public responsibility each recognized private provider—growth in the pri- the Soviet Union had serious consequences for activity warranted, so the budgets for even vate sector was simply a matter of the market the health sectors of the resulting states. In the high-priority public goods were not protected. (the short route) taking over when the state Soviet era, the health system was run, virtually in The resurgence of some vaccine-preventable became incapable of ensuring services. its entirety, by the central government. It had its diseases as well as the growth of infectious dis- Similarly, the oversupply of hospital beds has problems. It was rigidly �nanced on the basis of eases (HIV/AIDS and tuberculosis especially) fallen by an average of 40 percent from its 1990 inputs rather than outputs. It was extremely attest to this.The “compact� leg of the long level. Again, whether this was due to a deliber- biased toward hospital and specialist care. It route was gone; political structures for the ate policy (a 55 percent fall in Uzbekistan— inef�ciently relied upon high-cost procedures “voice� leg were (and in some cases still are) yet almost all in public hands) or a simple necessity and long hospital stays relative to other indus- to develop, leaving a vacuum. because of austerity and closures after privatiza- trial countries. And it was not at all oriented All of the Commonwealth of Independent tion (the same 55 percent fall in Georgia) is an toward clients. But it worked. Services were free States (CIS) countries are struggling to replace open question. to all, and particularly in the resource-poor the former system while suffering from the twin Reforms under consideration in the CIS republics, many in Central Asia, it contributed to liabilities of declines in income and the legacy countries generally involve such client-centered levels of health status—low mortality rates and of an unsustainable degree of hospital—and mechanisms as insurance (a conditional high life expectancy—much higher than in staf�ng—intensity inherited from the former voucher) and capitation schemes, both of which other countries at similar levels of income. regime. allow payments to follow patients. Progress is It worked because the two legs of the “long The pace and deliberateness of reform slow, however. Institutions take time to develop route�functioned well enough. A commitment to strategies have varied substantially among the and the information collection systems neces- universal coverage of health and other social ser- CIS countries. In most there has been a marked sary for getting good results from insurance vices deriving from socialist principles substituted increase in the private sector and in fees—both programs are still lacking. for “voice�in the form of free political expression. informal and institutionalized (particularly in The starting point for the CIS countries is “Compacts,�or more speci�cally direct manage- Georgia and the Kyrgyz Republic)—in public very different from that of developing countries ment, were enforced through means of the sub- facilities. Both tendencies have meant that pri- in general—too much infrastructure and stantial control government had over state- vate �nancing has become a large part of the resources rather than too little. However, in employed providers.There may have been some health market—averaging around 40 percent many ways the solutions will be similar. Substan- support for this arrangement from the “short but ranging from under 20 percent in tially more reliance on the “short route� of route�due to monitoring by local party leaders, Uzbekistan to over 90 percent in Georgia. accountability is likely, with government being a but this was distinctly secondary given the strong Uzbekistan’s retention of a large public sector monitor and enforcer of the rules of the game hierarchic management capacity of government. reflects a more robust economy. Having natural regardless of who ultimately becomes the direct Then the compact collapsed. Accountability resources to sell led to a fall in income of only 5 provider. to policymakers could no longer be enforced— percent between 1990 and 2000 in contrast to there was no longer funding or control from the more typical declines of 30 percent in Armenia, center. Within the republics, dramatic declines in 45 percent in Azerbaijan, or the more extreme Source: Maria E. Bonilla-Chacin, Edmundo income led to similarly dramatic declines in cases of 65 percent in Moldova and 70 percent Murrugarra, and Moukim Temourov,“Health Care public funding for the sector. Also, since almost in Georgia. During Transition and Health Systems Reform: Evi- everything had been produced by the state, Even when there was no deliberate policy of dence from the Poorest CIS Countries,� Lucerne there was no history of setting priorities accord- privatization—the sale of public facilities to a Conference of the CIS-7 Initiative, January 2003. of income—so it cannot be that any of these ernments were imposed on them by the U.S. institutional arrangements is incompatible federal government and varied little across with economic development. At the same tribes. The imposed constitutions were rea- time, countries with �nancial arrangements sonably well-adapted to Apache social and very similar to one of these three have failed cultural norms, providing a reasonable �t to develop. With many proposed solutions, between formal and informal structures of does anything go? No. Solutions need to con- power. In contrast, the formal constitution form to certain principles, but the principles was at odds with Sioux norms and led to con- need to be implemented in ways that are tinuing discord between formal and informal appropriate to the time, place, and service. modes of exercising power, precluding the American Indians as a group are the poor- emergence of effective institutions.166 est minority in America. Some tribes, such as Many African scholars argue that the roots the Pine Ridge Oglala Sioux, have severe eco- of the problems in Africa today lie in the nomic problems (unemployment in 1989 legacy of colonialism. Nation-state bound- was 61 percent). But others, such as the White aries followed colonial power rather than Mountain Apache, do badly, but much better African realities. The struggle over how to (unemployment was only 11 percent). As adapt or transform the transplanted institu- part of attempts to control them, their gov- tions continues to influence debates today. (c) The International Bank for Reconstruction and Development / The World Bank 60 WORLD DEVELOPMENT REPORT 2004 Here is how Mamdani opens his study, Citi- icant elements of the non-poor. There is zen and Subject: unlikely to be progress without substantial “middle-class buy-in� to proposed reforms. In Discussions on Africa’s present predicament the words of Wilbur Cohen, U.S. Secretary of revolve around two clear tendencies: mod- ernist and communitarian. For modernists, Health, Education, and Welfare under Presi- the problem is that civil society is an embry- dent Lyndon Johnson in the 1960s:“Programs onic and marginal construct; for communi- for poor people are poor programs.�168 tarians, it is that the real flesh-and-blood De Soto’s study of rights to real estate in communities that constitute Africa are mar- urban areas emphasizes that not only are ginalized from public life as so many “tribes.� The liberal solution is to locate pol- poor people outside the bene�ts of having itics in civil society, and the Africanist solu- secure title and claim to their property, but so tion is to put Africa’s age-old communities are nearly all of the middle class. His study of at the center of African politics. One side the historical evolution of property rights in calls for a regime that will champion rights, the United States strongly suggested that the and the other stands in defense of culture. The impasse in Africa is not only at the level response to popular political pressure—not of practical politics. It is also a paralysis of top-down technocratic design—was the key perspective.167 to a broad-based system of property rights.169 Since poor people are excluded from many services, such as primary education or safe Reforming institutions to water, improvements in the system are likely improve services for poor to disproportionately bene�t poor people. people will be dif�cult But broad coalitions are not always suf�- Because institutional reforms change power cient because some services need to be tai- relationships among actors, they are politi- lored to destitute and disadvantaged groups cal reforms. But politics generally does not (as in situations of ethnic or gender exclu- favor reforms that improve services for sion). A common obstacle in the access to poor people. Such reforms require upset- services is that the socially disadvantaged are ting entrenched interests, which have the excluded—as a matter of policy, or because advantage of inertia, history, organizational they feel excluded due to their treatment by capability, and knowing exactly what is at providers, or due to actions of more powerful stake. Policymakers and providers are gen- social groups within the community itself. erally more organized, informed, and influ- The politics of services for disadvantaged ential than citizens, particularly poor citi- groups are even more dif�cult, because coali- zens. But reform is possible, even against tions made up exclusively of the powerless are these odds. often powerless. • Pro-poor coalitions for better services Change agents—reform champions increase the odds for success. Episodes of reform depend on reform • Change agents and reform champions champions, the entrepreneurs of public sec- can shape the agenda and follow through tor reform. They emerge from various on implementation. sources. Politicians can often pursue service • When the prospects for successful institu- improvements even when the conditions tional reform are not propitious, strategic are not propitious. They must act to create incrementalism may be all that is possible. and sustain pressures for reform. Profes- But pursuing it has the danger of being sional associations are often both the source merely incremental incrementalism. of pressure for, and resistance to, major innovations. Dissatis�ed with the progress Pro-poor coalitions in their �eld—education, policing, public In most instances making services work for health, sanitation—professionals emerge as poor people means making services work for champions for reform, putting pressures on everybody—while ensuring that poor people politicians and policymakers for reform. have access to those services. Required is a For instance, the campaign of Public Ser- coalition that includes poor people and signif- vices International for “Quality Public Ser- (c) The International Bank for Reconstruction and Development / The World Bank The framework for service provision 61 Table 3.3 Modern institutions took a long time to develop Institution/reform First Majority (of now Last United Kingdom United States developed countries) Universal male suffrage 1848 (France) 1907 1907 (Japan) 1918 1870–1965 Universal suffrage 1907 (New Zealand) 1946 1971 (Switzerland) 1928 1928–1965 Health insurance (the basis for what is now universal) 1883 1911 1911 Still no universal coverage State pensions 1889 (Germany) 1909 1946 (Switzerland) 1908 1946 Source: Chang (2002). vices� balances the unions’ role in protect- not emerge until the 1970s. Institutional ing the rights of workers with support for reform that changes the landscape usually innovation in public service delivery.170 moves at a glacial pace—but glaciers do move Linking the efforts of these “insiders� and and carve out new landscapes when they do. “technocrats� to broader coalitions of citizens The improvement of services, always is often a key element of success. pressing, cannot wait for the right conditions. Some arrangements, such as enclave Strategic incrementalism approaches to delivering services to poor Sweeping or fundamental reform of institu- people, may not be sustainable in the long tions is rare. It requires the right conditions. A run, even if they improve outcomes in the recent study emphasizes how long the devel- short run. Often driven by donors, these opment of political institutions in the now- actions can undermine national relationships developed countries took (table 3.3). Most of accountability (chapter 11). Sometimes “modern� institutions of “modern� political the desirable arrangement is to strengthen and economic governance that are recom- the weakest link. If the policymaker-provider mended today emerged late in the now- link is weak, contracting out services—such developed countries (at much higher levels as Cambodia’s use of nongovernmental orga- of income than developing countries today). nizations for primary health services—may And they spread slowly across countries. In be the preferred arrangement. But incremen- the United States universal white male suf- tal activities—pragmatic improvisation to frage was not achieved until 1870, female suf- make services work even in a weak institu- frage did not come until 1925, and true uni- tional environment—should be used to cre- versal suffrage did not come until (at least) ate more favorable conditions for reform in 1965. Switzerland did not adopt female suf- the longer run. Temporary work-arounds frage until 1971. Canada’s widely discussed cannot and should not substitute for creating “single-payer� style of health insurance did the conditions for reform. (c) The International Bank for Reconstruction and Development / The World Bank spotlight on Uganda Universal primary education—what does it take? Primary school enrollment in Uganda rose from 3.6 million students to 6.9 million between 1996 and 2001. What accounts for such a drastic increase in such a short time: Political will? Abolishing user fees? A good macroeconomic envi- ronment? Information to empower bene�ciaries? All of the above. Did emerging democracy make In the elections in Malawi in 1994 the P resident Yoweri Museveni’s decision in 1996 to make universal primary education an issue in the presidential election campaign broke with his earlier position. He had previously emphasized that the difference? The argument that free elections con- tributed to the success in Uganda is but- tressed by three observations. Education winning presidential candidate also made universal primary education part of his manifesto, but voting was much more polarized along regional lines than it was in was a salient issue for Ugandan voters. The Uganda. The winner would thus have had building roads and infrastructure would electorate had access to information about an incentive to continue to cultivate a provide access to markets to enable people to government performance in this area. And regional base of support, rather than to generate income that would pay for school- the success of universal primary education deliver on a national issue like education. ing. But in a radio address in March, he contributed to President Museveni’s con- Ugandan voters had access to various promised to give access to free primary edu- tinued popularity.171 sources of information about the education cation to four children per family, up to two For candidates to believe that they will initiative that enabled them to evaluate how boys and two girls, and all orphans. be judged on whether education promises the government had made good on its 1996 Although free primary education was are ful�lled, education must be important election promises. Major national dailies, only one part of Museveni’s manifesto, it for voters. Museveni’s 1996 commitment to such as The Monitor and The New Vision, soon became clear that the promise had universal primary education would have continued to give prominent coverage to struck a chord with the electorate. Finance been expected to draw national support, universal primary education issues. ministry of�cials recall that several of Musev- while Paul Ssemogerere’s positions on the Data collected by the Afrobarometer eni’s advisors repeatedly sent them messages northern rebellion, Buganda autonomy, project show that Ugandans believe that after campaign meetings, emphasizing how and reintroduction of political parties President Museveni has performed very the promise had resonated with the public. would have been expected to generate effectively, and in rating government per- The May 1996 election was Uganda’s regional support. formance they are most satis�ed with edu- �rst presidential election since the military Museveni’s victory did not depend on a cation policy. During the June 2000 survey, takeover in 1986. Though the elections did single regional base of support. He received 93 percent of respondents reported that not involve of�cial multiparty competition, more than 90 percent of the vote in they were either somewhat satis�ed or very President Museveni faced a credible chal- Uganda’s western region, 74 percent of the satis�ed with Museveni’s overall perfor- lenger, Paul Ssemogerere, the leader of the votes in Paul Ssemogerere’s home region mance. Eighty-seven percent of Ugandans Democratic party. Ssemogerere promised (central), and 72 percent of the vote in the reported that their government was han- to restore multiparty politics, negotiate eastern region. dling education issues well, while the aver- with the rebel movement in northern age across the 12 African countries was 59 Uganda, and grant greater autonomy to the percent. Buganda region, once an independent king- dom. These stances positioned him for sub- Figure 1 Enrollments increased dramatically stantial regional support in northern and after 1996 Macroeconomic stability Primary school enrollments in Uganda, and budgetary institutions central Uganda. He also declared that he 1996–2001 would match Museveni’s promise to pro- Democratic politics may have given the vide free primary education. Millions Ugandan government an incentive to deliver 8 on its education promise of 1996. But suc- In December 1996, soon after a landslide victory, President Museveni announced the cess in reorienting public expenditures abolition of school fees. Since then there has 6 toward primary education has also been a sustained shift in Ugandan public depended on stabilizing the macroeconomic expenditures in favor of education, espe- 4 environment and developing budgetary cially for primary schools. Spending on edu- institutions.172 Stable macroeconomic con- cation has risen as a share of government ditions have undoubtedly made it easier to 2 expenditures from an average of 20 percent forecast revenues and expenditures. Under in the three �scal years preceding the elec- more unstable macroeconomic conditions, tion to an average of 26 percent in the three 0 African governments like Malawi have found years following. Total enrollment in primary 1996 1998 2000 2001 it dif�cult to maintain a sustained commit- schools skyrocketed (�gure 1). Source: Murphy, Bertoncino, and Wang (2002). ment to increasing education expenditures. (c) The International Bank for Reconstruction and Development / The World Bank Spotlight on Uganda 63 Uganda’s macroeconomic stability since audited or monitored, and most schools 2001 both groups experienced a large drop 1992 has depended on budget reforms— and parents had little or no information in leakage. But the reduction in capture was from a cash budget system to the medium- about their entitlements to the grants. Most signi�cantly higher for the schools with term expenditure framework to the Poverty funds went to purposes unrelated to educa- access to newspapers, which increased their Eradication Action Plan. The framework tion or for private gain, as indicated by funding by 12 percentage points over aligns resources with budgetary priorities, numerous newspaper articles about indict- schools that lacked newspapers. while the cash budget system ensured that ments of district education of�cers after the With an inexpensive policy action—the overall �scal discipline is maintained if there survey �ndings went public. provision of mass information—Uganda are revenue shortfalls. The Poverty Action To respond to the problem, the central dramatically reduced the capture of a pub- Fund has been particularly effective in ensur- government began publishing data on lic program aimed at increasing access to ing that government spending priorities, monthly transfers of grants to districts in textbooks and other instructional materi- such as primary education, receive needed newspapers and to broadcast them on the als. Because poor people were less able than funds. It seems unlikely that universal radio. It required primary schools and dis- others to claim their entitlement from the primary education would have been sus- trict administrations to post notices on all district of�cials before the campaign, they tainable without these innovations in bud- inflows of funds. This promoted account- bene�ted most from it. getary institutions. ability by giving schools and parents access to information needed to understand and The power of information monitor the grant program. Figure 2 Amount of capitation grant due in delivering funds for education An evaluation of the information cam- schools actually received by schools, 1991–2001 In 1996 a public expenditure tracking sur- paign reveals a large improvement. Schools vey of local governments and primary are still not receiving the entire grant (and Share of funds Uganda shillings reaching the school schools revealed that only 13 percent of the there are delays). But the capture by inter- 6000 100 per-student capitation grants made it to the ests along the way has been reduced from Average capitation schools in 1991–95.173 In 1995 for every 80 percent in 1995 to 20 percent in 2001 5000 grant P1-P7 80 dollar spent on nonwage education items (�gure 2). A before-and-after assessment Average share of funds by the central government, only about 20 comparing outcomes for the same schools 4000 reaching the schools 60 cents reached the schools, with local gov- in 1995 and 2001—and taking into account 3000 ernments capturing most of the funding. school-speci�c factors, household income, 40 Poor students suffered disproportion- teachers’ education, school size, and super- 2000 Median share of funds ately, because schools catering to them vision—suggests that the information cam- reaching the schools 1000 20 received even less than others. Indeed, most paign explains two-thirds of the massive poor schools received nothing. Case study improvement. 0 0 evidence and other data showed that the In 1995 schools with access to newspa- 1991 1992 1993 1994 1995 2001 school funds were not going to other sec- pers and those without suffered just as Source: Reinikka and Svensson (2001), Reinikka and Svens- tors either. The disbursements were rarely much from the leakages. And from 1995 to son (2003a). (c) The International Bank for Reconstruction and Development / The World Bank Clients and providers 4 The well-being of poor people is the point of Report and this chapter try to give the term making services work. The value of public “empowerment� a precise and concrete policy and expenditure is largely determined interpretation. Speci�cally, the chapter dis- chapter by the value the poor attach to it. When cusses the potential for poor people to publicly provided and funded housing is left influence services by: vacant,174 when food supplies are not eaten, when free but empty public health clinics • Increasing their individual purchasing power. are bypassed in favor of expensive private care,175 this money is wasted. • Increasing their collective power over Improving services means making the providers by organizing in groups. interests of poor people matter more to • Increasing their “capacity to aspire�177: providers. Engaging poor clients in an allowing them to take advantage of the active role—as purchasers, as monitors, and �rst two by increasing the information as co-producers (the “short route�)—can needed to develop their personal sense of improve performance tremendously. capability and entitlement.178 How can public policy help poor people acquire better services through this route? By expanding the influence of their own When will strengthening choices. By having the income of providers the client-producer link depend more on the demands of poor matter most? clients. By increasing the purchasing power In the framework of chapter 3, improving of poor people. And by providing better client power—the short route of service deliv- information and a more competitive envi- ery—can overcome various weaknesses of the ronment to improve the functioning of ser- long route (�gure 4.1), even when services vices. Where such choice is not feasible, remain the responsibility of government. The governments can expand consumer power clearest case is monitoring providers. Clients by establishing procedures to make sure are usually in a better position to see what is complaints are acted on. going on than most supervisors in govern- Sad to say, governments and donors fre- ment hierarchies—who provide the compact quently neglect the possible role of poor and management. When the policymaker- clients in sustaining better services—or provider link is weak because of scarce or dif�- treat that role merely as an instrument for cult-to-manage supervisory staff, clients may achieving a technically determined out- be the only ones who regularly interact with come. Neither governments nor donors are providers. As discussed several times in this accustomed to asking the poor for advice. Report, improvements in basic education have Recent initiatives have begun to redress this often depended on participation by parents. through a variety of ways to increase partic- Although parents cannot monitor all aspects ipation by communities and civil society. of education, they can monitor attendance by But the potential for improvement has not teachers and even illiterate parents can tell if yet been adequately tapped.176 their children are learning to read and write. In short, the key is to enhance the power Citizens as clients can also make up for of poor clients in service provision. This shortcomings in the voice or politics relation- 64 (c) The International Bank for Reconstruction and Development / The World Bank Clients and providers 65 ship. If governments cannot or will not try to Figure 4.1 Client power in the service delivery framework determine and act on the desires of the pub- The state lic, or if the desires of poor people are system- atically ignored, there may be few options for poor communities but to develop mecha- of accounta ro u te bili nisms for getting services some other way. ng ty Lo The greater the differences among clients—their heterogeneity—the more that S h o rt r o u t e direct client power is likely to have an advan- Citizens/clients Providers tage relative to the “long route.� The greater Nonpoor Poor Frontline Organizations Client power the individual differences in preferences for the type and quality of services provided, the greater the importance of discretion on the Services part of providers and the more dif�cult it is to monitor the use of this discretion centrally. extend these services in the �rst place. There is Sometimes preferences differ geographically, no reason to believe they are all self-corrective so different levels of government may reflect through replicating aspects of the free market. this variation. But for many services, the het- Similarly, some settlements constitute erogeneity of preferences applies all the way communities with suf�ciently congruent down to the individual. Take courtesy and interests among members, egalitarian norms comfort (caring) relative to technical skill to protect the poor, mutual trust, and the (curing) in health delivery—or farmers with ability to mobilize information and to act constraints on their time and other workers collectively—that is, they have social capi- in the same community with different con- tal.179 But some clearly do not. How many vil- straints. Certainly people differ in the lages and urban neighborhoods are there in amounts of water and electricity they want, the developing world? Hundreds of thou- given their other needs. Government struc- sands? Millions? And how many kinds of tures may not be flexible enough to accom- social structures are represented? Ensuring modate this variety. And where local prefer- that poor people have a say in this variety of ences vary systematically between the poor circumstances demands that policies be and others, honoring poor people’s prefer- examined and designed with a great deal of ences over those of the better-off can be a local knowledge and an understanding of challenge. local conflicts and inequalities. Pretending For some collective action problems, gov- otherwise will almost certainly do real harm. ernments may not be located at the correct And some services, particularly for health level to solve them, no matter how willing and modern water and sanitation, need tech- they are to pursue the interests of the poor. nical inputs to be successful. Patients—as The boundaries of the political jurisdiction individuals or health boards—are good may not correspond to the boundaries of the judges of courtesy and attendance. But they problem. So schools are often the most are much less able to judge clinical quality or appropriate unit for management and opera- the appropriate mix of curative and preven- tion. Sanitation services need community tive services. And some health problems have pressure to ensure that everyone uses �xed- effects that spill over community boundaries. point defecation, but they are often organized Large pest-control initiatives and other forms around communities that are larger or of infectious disease control may seem a low smaller than villages, depending on the den- priority for any one group of citizens, yet will sity of population. A more active role for be effective only when all participate. Ulti- communities is needed in such cases. mately, some wide-scale government inter- It is important to avoid romanticizing vention is necessary. Still, emphasizing the either form of increasing client power—nei- power of clients is a welcome tonic for the ther choice nor participation is suf�cient for top-down, technocratic orientation that has all services. Market failures and concerns for characterized much development thinking equity lead societies to want to improve or until now. (c) The International Bank for Reconstruction and Development / The World Bank 66 WORLD DEVELOPMENT REPORT 2004 Increasing client power • Make better choices about which ser- through choice vices to demand. The most direct way to get service providers The �rst two work through providers, the to be accountable to the client is to make second two through clients. whatever they get out of the transaction depend on their meeting client needs and desires. That is, money (usually) or other ben- Provider behavior e�ts from providing the service should follow Discourtesy, social distance, abruptness of the client—the enforceability of a relationship care, discrimination against women and eth- of accountability, discussed in chapter 3. nic minorities, service characteristics mis- In market transactions, this is done by a matched to individual tastes—all are associ- buyer paying money to a seller. But that is ated with provider behavior. And all can not the only way. Payments by government improve with the purchasing power of to schools (and the pay of teachers) can clients. Indeed, that is why the private sector depend on the number of students enrolled is often seen as preferable to a public sector and continuing. The vast majority of pri- with staff paid by salaries (box 4.1). These mary education in the Netherlands is paid differences are echoed in studies from coun- for by government but delivered by private tries as diverse as Bangladesh, China, India, schools compensated in this way. Capita- Lao PDR, Thailand, and Vietnam. tion lists are the dominant method of pay For courtesy, caring, and convenience the for general practice medical providers in private sector usually has a distinct advan- several European systems, particularly the tage. Private practitioners usually provide United Kingdom. Overall consumer satis- services more convenient to the client. Lim- faction can be expressed through the possi- ited hours in public facilities (only in the bility of changing general practitioners, morning in farm communities) is often the determining their income. reason people go to a private practitioner.180 Vouchers issued to consumers are another What accounts for the difference? Not the method of linking service provider compen- training but the motivation: “. . . the same sation to consumer choices, even though the government doctor who was not easily or consumer is not the original source of funds. conveniently accessible, whose medication All health insurance with some choice of was not satisfactory and whose manner was provider is a form of voucher—one condi- brusque and indifferent transformed into a tional on being sick. And intrinsically moti- perfectly nice and capable doctor when he vated providers, whose sense of self-worth was seeing a patient in his private practice.�181 depends on having a large demand for their Why? Because the doctor wants the client to services, try for more patients under any pay- return. If the staff is paid through salaries, ment system. The essence of each of these there is no strong incentive to be accommo- methods is that client well-being translates dating. This is not lost on clients: “Anyhow, directly into provider well-being—the incen- they will get their money, so they don’t pay tives are aligned. much attention.�182 Discrimination, particu- Many service problems can be improved larly against ethnic minorities and women, by making sure that payment follows clients. and social distance are barriers to services Most of the evidence for this comes from even when the services are free, barriers that studies examining the effect of fees on the frequently yield to market forces.183 The arti- behavior of private providers (who must, of �cial scarcity of free services—ensuring course, operate this way) but it applies to all excess demand—induces rationing by some such methods. Payment can have four kinds other means (social status, personal connec- of bene�cial effects: tions, ethnicity), and poor people rarely have these other means. Groups coping with social • Improve provider behavior. stigma—such as prostitutes, who need to be • Increase supply and sustainability. part of the battle against HIV/AIDS—often • Increase vigilance and a stake in receiv- prefer the con�dentiality and more consider- ing better service from each transaction. ate behavior in private clinics. (c) The International Bank for Reconstruction and Development / The World Bank Clients and providers 67 The scarcity of commodities due to low there remain problems, such as ensuring the pricing may lead to other commonly quality of the newly established schools, but reported problems—illicit sale of materials these are secondary to getting girls to school. and the demand for under-the-counter pay- The revenues that providers raise from ments. Indeed, “free� public services are often charges at the point of collection are often the very expensive. Many countries have serious reason some services can continue at all. diversions of pharmaceuticals from the pub- Much of the success of the Bamako Initiative lic stock into private markets, where they in West Africa (see spotlight) stems from the instantly become expensive. In general, ser- supply of pharmaceuticals made possible by vices that most directly resemble market charging users for them. Bamako Initiative goods have a greater problem of diversion villages usually have drugs, other villages usu- and implicit privatization. ally don’t. Sustainability in piped water sys- In Eastern Europe the health systems are tems is almost always equivalent to �nancial often ranked among the most corrupt of public services (box 4.2). Under-the-table payments and pharmaceutical sales to open BOX 4.1 The private sector is preferred in Andhra markets are the main elements in this assess- Pradesh, India ment. If directives against such practices can- not be enforced, countervailing pressure is A study of consumer and producer attitudes was conducted in six districts in the southern needed (see box 3.6). Formalizing fees and Indian state of Andhra Pradesh.The study included 72 in-depth interviews and 24 focus groups. putting purchasing power in the hands of poorer clients is one possible source of such Private Public pressure.184 ATTITUDES OF DOCTORS Exemptions from fees can have perverse “They speak well, inquire about our health.� “Does not talk to me, does not bother (about effects by reducing this purchasing power. In my feelings or the details of my problems).� Benin a measure to raise female school “Ask about everything from A to Z.� “Don’t tell us what the problem is, �rst check, enrollment—waiving fees for girls—led “Look after everyone equally.� give us medicines and ask us to go.� teachers to favor the enrollment of boys and “They take money . . . so give powerful “They are supposed to give us Rs. 1000 and 15 medicine . . . treat better.� to raise informal fees for girls.185 Of course, kg of rice for family planning operations; they give us Rs. 500 and 10 kg rice and make us run the problem could have been solved by abol- around for the rest.� ishing fees for everyone (if the teachers could “Anyhow they will get their money so they continue to be paid) or by closer monitoring don’t pay much attention.� and enforcement by education of�cials. But CONVENIENCE in a system that has problems paying teachers “Treat us quickly. . . .� “Do not attend to us immediately.� and weak administrative capacity, bolstering the ability of girls to pay with vouchers seems “We spend money but get cured faster.� “Have to stand in line for everything.� more likely to succeed. “I know Mr. Reddy. He is a government doctor “Doctor is there from 9 a.m. to 4 p.m.—when but I go to him in the evening.� we need to go to work.� “Can delay payment by 5–10 days. He is OK “I have not been there, but seeing the Increase and sustain supply with that, he stays in the village itself.� surroundings . . . I don’t feel like going.� Greater purchasing power may simply COST increase supply and overcome bottlenecks “Recent expenses came to Rs. 500 for 3 days . . . “While coming out, compounders ask us for due to supply problems. In Bangladesh the had to shell out money immediately.� 10–20 Rs.� Female Secondary School program awards “We have to be prepared to pay, you never “Anyhow, we have to buy medicines from scholarships to girls if they attend school reg- know how much it is going to cost you.� outside.� ularly and gives secondary schools a grant ADVANTAGES based on the number of girls they enroll. Sec- “Even if I have to take a loan I will go to “Malaria treatment—they come, examine ondary school enrollment in Bangladesh is private place, they treat well.� blood, give tablets.� increasing, and faster for girls than for boys. It “For family planning operations.� also led to the establishment, at private “Polio drops.� expense, of new schools. Desires for single- “In case I do not get cured in private hospital, sex schools and separate toilet facilities for but it is very rare.� girls were mysteriously accommodated when Source: Probe Qualitative Research Team (2002). girls’ attendance meant more money. True, (c) The International Bank for Reconstruction and Development / The World Bank 68 WORLD DEVELOPMENT REPORT 2004 services.186 Farmers in southern India BOX 4.2 Bribery in Eastern expect the same from irrigation services Europe (box 4.3). Surveys in nine transition countries of Eastern and Central Europe* asked: “In your opinion, in Making better choices what area is bribery most common, widespread?� For some services consumer discretion is Health systems rank highest overall, but with important for allocating resources ef�ciently. answers ranging from 11 percent in Bulgaria to 48 percent in Slovakia. Since there has been an Households determine water and electricity overall contraction in public services with that in use, scarce goods that have costs associated economic activity, the most likely reason is that with them. And facing marginal costs is the these marketable services are naturally easy to charge for and dif�cult to maintain without infu- only way to ensure ef�cient use. The alterna- sions of funds from patients. tives: wasted water leading to shortages, unreliable service with serious consequences Other/ for the safety of the water supply, and peri- I don’t know 17% Health odic cuts in electricity familiar to most peo- 27% ple in developing countries. Protecting the Education poor in network services can be achieved 6% (assuming that meters work) with “lifeline� subsidies, in which the �rst few essential Customs units are free but full marginal costs are paid 11% beyond this level. In health care, as in water and electricity, Legal system more is not always better. Restricting Ministries/offices and police 16% 23% demand for curative services by pricing frees up providers, particularly public *Bulgaria, Croatia, Czech Republic, Hungary, Poland, providers, to do preventive health, for Romania, Slovakia, Slovenia, and Ukraine.The which there is little private demand.187 As diagram summarizes results averaged over these countries (weighted by population). the director of a prominent nongovern- Source: GfK Praha—Institute for Market Research mental organization providing health care (2001). to the very poorest in Bangladesh puts it: “Of course you must charge at least a token amount for services, otherwise you keep seeing people with paper cuts and other sustainability. There might be some subsidy minor things.�188 Similarly, crowding at element included in pricing, but systems to outpatient clinics at public hospitals can be get water to a private home depend on curtailed by charging enough so that people charges for that water. use a cheaper level of service. All these advantages can be obtained in Increased client stake—and vigilance ways other than charging fees at the point of The third argument for having money fol- service. As long as clients consider the low clients: when people buy things they resources used as belonging to them, the dis- make sure they get them, and they pay more cipline of market-like mechanisms can be attention to the quality of what they get. enforced. The Singapore Medical Savings Money is a profound source of power for Accounts do this by allowing people to apply poor people. When Zambian truck drivers funds not used for primary medical care to were expected to pay into a road fund, they other purposes, such as pensions.189 Coun- took turns policing a bridge crossing to tries with scarce administrative personnel make sure overloaded trucks did not pass. and supervisory capacity may certainly want Their money would have to be used to �x to enlist clients as monitors, and market the bridge. Women living in slum areas of mechanisms are one way of doing it. Rio de Janeiro proudly display bills they For any of these mechanisms to work, paid for water and sanitation—it proves however, there must be a real choice with their inclusion in society and their right to real options. Otherwise, giving schools pay- (c) The International Bank for Reconstruction and Development / The World Bank Clients and providers 69 ment. In some cases the market would be BOX 4.3 Payment and expected to wither away as the state increases accountability its capabilities. In the meantime, three cate- gories of policies can make the most out of A conversation with farmers in Haryana state in clients acting on their own behalf: India, who had been to see what had happened in reforming Andhra Pradesh (AP): • Increasing the power of the poor over Q: “What did you learn when you visited AP?� providers by providing them with A: “That the farmers are much poorer than us, �nance directly. but that they pay four times as much for water� • Increasing competition. Q: “The farmers in AP cannot be happy about • Increasing information about services that. . . .� and providers. A: “They are happy, because now the irrigation department is much more accountable to them . . . they know where the money goes Increasing the purchasing power of the poor. and they have a say in how it is spent. . . .� The big problem with services that can, in Q: “So then, you much richer farmers would be principle, be provided in markets is that willing to pay more?� poor people don’t have enough money to A: “Only if the irrigation department makes the same changes, otherwise we will refuse to pay for them. For market mechanisms to pay.� help the poor, their purchasing power must Q: “Ah, but this is just because there is a particu- be increased. The voucher mechanisms dis- lar Chief Minister who is pushing that now . . . cussed are a direct way of handling this for once he goes it will all go back to the same old way.� speci�c services. But additional mileage can A: “We also wondered about that, and so we come from more flexible transfers that can asked the farmers in AP about that.They told be used for purposes that the family chooses. us that ‘no matter who is elected as CM, we Flexible transfers can help to overcome will never allow the government to again the weakness of the citizen-policymaker link give us free water.’� by giving poor people more direct say in what Source: World Bank staff. gets delivered than even the political process would give them—the transfers become their money. Substantial work in South Africa has shown the bene�cial effects of cash pension ments on the basis of enrollments is not far payments on the health and well-being of all from what happens now in centrally owned members of a family.190 For services with and managed school systems, with all the large externalities, demand for the service problems we are trying to �x. Conversely, may not be great enough, even when the ser- market mechanisms with a natural monop- vice is free, so the Bolsa Escola program in oly don’t improve matters either. There is Brazil paid families to send their children to no denying that sparsely populated rural school, as did the secondary school program areas—where many of the world’s poor for girls in Bangladesh, while the Education, people live—are much more constrained by Health, and Nutrition Program (Progresa) in competitive supply than urban areas. But Mexico paid families to use preventive health even these markets may be “contestable� in care (see spotlight). the sense that other providers would be able Cash payments have problems though. to enter the market if the current provider First, giving unconstrained cash transfers to abused monopoly power or if monopolies poor people is often not politically palatable. were periodically granted on the basis of Second, cash payments always have to be competitive bids. administratively targeted, which requires determinations of eligibility. Everyone likes Policies to improve choice money, so self-targeting of cash transfers is Choice-based improvements alone cannot be not possible. If a government has a hard time a solution to the problem of bad services for getting goods and services into the hands of the poor, though some may remain as instru- the poor, it may well have an even harder time ments in a longer-run strategy by govern- getting cash, or cash equivalents, to them. (c) The International Bank for Reconstruction and Development / The World Bank 70 WORLD DEVELOPMENT REPORT 2004 Increasing the scope of competition. tioning competitive markets is the consumer’s Sometimes increasing competition merely awareness. The private sector is a mixed bag. means allowing a private sector to emerge Private “medical� providers vary from quite where laws previously restricted entry. Jor- good doctors (including senior specialists dan, after years of prohibition, allowed pri- from government hospitals in their off-hours) vate universities in 1990. Ten years later, to totally unquali�ed, untrained people, some enrollments in these institutions accounted of whom are downright dangerous.191 Private for one-third of all university students. or NGO schools may cater to speci�c skills not Bangladesh has had a similar experience in provided in public schools (foreign language, the past decade. This increase in competi- religious studies, arts and music) or they may tion allows governments to increase enroll- just be pro�teers. An essential part of improv- ments without extremely regressive subsi- ing peoples’ choices is to provide information dies to public universities. about these providers. Many times, people Competition can also be encouraged by simply don’t know enough to choose better or allowing subsidies to the poor to be worse services. And sometimes they identify portable between public and private good medical care with powerful medi- providers. Private providers may not exist cines—which is quite wrong and potentially simply because the public sector is free. dangerous. Governments can increase competition by Information can be advice to families on changing the form of subsidy from zero how to choose schools or medical care- price to competitive prices, with cash or givers192 or on how to take care of them- voucher payments to compensate. Univer- selves. This might be supplemented with var- sity education government loans, usable at ious certi�cation programs, standard setting, any eligible institution, can increase compe- and laboratory checks (say, for water purity). tition, improve quality in public facilities, Scorecards of public services should also be and reduce subsidies for all but students extended to private or NGO providers. On from poor families. the other side of the market, government In some cases competition is not possible, may want to directly improve the quality of at least not without substantial regulation. private services. Training, “partnership� Health insurance markets are notoriously arrangements, contracting, and other means prone to failure, and competition within of engagement can all be tried. But attempts them can lead to both inef�cient and to increase information should be subject to inequitable outcomes, since �rms can com- rigorous evaluation (chapter 6). pete by excluding the sick, not by being more ef�cient. Network services are also hard to Increasing consumer power run without a monopoly. But in each case through participation these markets can be contestable, capturing The accountability of providers to clients can much of the bene�t of competition. also be achieved when people voice their con- Some readers may think that the forego- cerns. In this case, enforceability is not ing arguments are just an attempted justi�ca- through clients’ money but through their tion for user fees. This is wrong, for all the direct interaction—encouragement and com- reasons put forth here. So, to make things as plaints. The scope for poor people to voice clear as possible, the pros and cons of user complaints individually is very small. In rich fees in general are laid out in box 4.4. There countries individuals get help from systems of are times when user fees are appropriate— tort law that can handle individual litigation and some when they are not. Based on the and from government-sponsored of�ces of primary goal of making services work for consumer protection or ombudsmen. But poor people, this Report argues against any these are rare to nonexistent for the poor in blanket policy on user fees that encompasses developing countries (they don’t always work all services in all country circumstances. so well for the poor in rich countries either). Some problems for which voice might be Increasing information to improve choices by expected to work are intractable. One consumers. One critical limit to well-func- example is corruption: the public might (c) The International Bank for Reconstruction and Development / The World Bank Clients and providers 71 resent under-the-table payments, but there bers than from individuals going it alone. may be no incentive to complain if, say, a So client power expressed outside market doctor is using public facilities and materi- transactions will almost always be expressed als at the same time. Clients know that the through collective action. service is still cheaper than if they had to go Strengthening participation along the to the market, and so do not complain.193 client-provider link can �x problems in the There is a deeper constraint: even when long route of government provision. So com- there is an opportunity to redress com- munity groups that take on complaining, plaints, monitoring and follow-through are monitoring, and other means of making sure public goods—the bene�ts accrue to the things work properly would be expected at entire group while the costs are borne by a some point to become institutionalized few. This is true for communities as well as within government (most likely local govern- individuals, but groups of people generally ment), or possibly to be supplanted by gov- �nd it easier to elicit support from mem- ernment as it improves. After all, collective BOX 4.4 No blanket policy on user fees The wide range of services and country circumstances discussed in this Report Is the service excludable? Do not charge for service (because you cannot). Pest control for Possible to keep people who No public health, surface (non-toll) roads, many police services. makes it impossible to claim that a par- do not pay from bene�ting? ticular level of user fees or none at all is appropriate in every case. User fees, as Yes with other public policy decisions, must balance protection of the poor, Transfer money to poor people and Can poor people be Can poor people Yes ef�ciency in allocation, and the ability charge user fees. distinguished from non-poor? be given money? to guarantee that services can be imple- Administratively and Yes Cash transfers or mented and sustained.The following politically? No Charge for service with exemptions vouchers or food? for poor people. Targeting can be flowchart summarizes the arguments administrative, geographical, or via and references in the text and raises self-selection. most of the issues necessary in deter- No mining whether user fees of any sort “Lifeline� price schedule. For water and electricity, charge full are appropriate in a given case.Three Can charges vary with Yes marginal costs of services for use above speci�ed maximum. points: amounts used? Make first few visits for medical care per year free for everyone. • First,“ef�ciency� is shorthand for No standard principles of public economics (see any textbook) that Charge for service. Empirically, this may apply to many services. Is service disproportionately often but not always require prices No Example: for higher education institute loan programs without used by poor people? subsidy. that equal social marginal cost and may include subsidies, taxes, or other Yes Charges are a necessary evil. Requires honest appraisal of ability interventions independent of their to deliver services along “long route.� If teachers or medical distributional effects. For example, providers cannot be supervised and medical stores not infectious disease control measures Will service be adequately maintained by government, then clients, by default, must bring delivered without user fees? No will have a subsidy element because purchasing power to bear. Revolving drug funds through the of their external effects regardless of Bamako Initiative, irrigation charges (see box 4.3), possibly many their impact on poor people. others including primary education if government is not reliable. Yes • Second, it is assumed that all subsi- Charge fees at a level that balances distributional effects with dies are paid for by taxes.The net effect on poor people depends on Will service be overused ef�ciency. Water (taps left running), electricity (interrupted without user fees? Is waste service from overuse). Also applies to curative care if staff time their contribution to tax revenue Yes likely to be large if prices are available for higher-priority public health activities is crowded out (possibly substantial when taxes are too low? or to outpatient clinics at hospitals when less expensive to treat based on agricultural exports) and on the same problems at lower-level facilities. their share of the deadweight loss that taxes impose on the economy. Do not charge for service. Best example: primary education. • Third, even when prices are not No Attendance is limited to one (school year) per child. Social value charged at point of service, commu- is considered high. Poor people use this more than non-poor nities may want to make (�gure 2.5). contributions to capital costs by, say, helping to construct or maintain schools. (c) The International Bank for Reconstruction and Development / The World Bank 72 WORLD DEVELOPMENT REPORT 2004 action is expensive—people, especially poor thus be compromised by too aggressive a people, have more pressing things to do with stance. This is true for other professionals as their time. They will want to transfer this well. In Kerala, maintaining staff at a health responsibility to permanent structures as fast center became dif�cult when local residents as they can.194 made too many demands on providers’ But local inputs and knowledge from time.197 direct participation may be needed for some Beyond monitoring, communities can time, possibly permanently, and govern- be the appropriate locus for more direct ment can help make those inputs more inputs, in effect becoming co-producers of effective. Education provides many of the services. Some services cannot be delivered better illustrations. Parents are in the best by state agencies very well because the envi- position to see what is happening in schools, ronment is too complex and variable—and and schools are usually the unit in which the cost of interacting with very large num- decisions are most effectively made. So giv- bers of poor people is too great.198 Sanita- ing parents power to influence school poli- tion programs often bene�t from local par- cies often has bene�cial results. In the exam- ticipation and inputs, since social relations ple of El Salvador’s Community-Managed in communities are often the best guaran- Schools Program (Educo—see spotlight), it tors of compliance with sanitation policies was the right to hire and �re teachers and and compliance must be universal if the the regularity of visits from the local educa- community is to reap the health bene�ts. tion committee, staffed in part by parents, Local perspective and knowledge are criti- that led to the increases in teacher and stu- cal in transmitting needed information. The dent attendance and in test scores. acceptability of messages on health-related Madhya Pradesh, India, has seen sub- habits, preventive health measures, hygiene, stantial improvement in test scores, com- sexual conduct, and other sensitive issues is pletion rates, and literacy.195 Community much greater when those messages are con- involvement is strong in recruiting teachers, veyed through informal face-to-face contact getting new schools built, and encouraging in discussions among small groups of indi- neighbors to enroll their children. Parents viduals with similar backgrounds. For have been helped by the ability to hire local, instance, organized discussions among infor- less-than-fully-trained teachers at a fraction mal women’s groups can enhance the credi- of standard pay scales for government bility and impact of behavior change efforts. teachers—with better results. It is possible, but unlikely, that outsiders may This last aspect of the program compli- learn enough of local mores to influence local cates scaling up. The ability to avoid con- conversations on these subjects.199 frontation with public sector unions has been a great advantage. Will teachers’ Tapping local social capital unions allow such recruitment to become Many communities have evolved means of standard?196 Do teachers hired at low wages solving longstanding problems requiring col- expect to be converted into full public ser- lective action. When the bene�ts of coopera- vants? For now, however, the involvement tion are great enough, there is a way to enforce of communities in Madhya Pradesh, which rules, and where there are no private alterna- is much greater than in other states, has tives, organizations often emerge on their made a big difference in performance. own.200 Communities have solved irrigation, Other policy initiatives that can also forestry management, nutrition, and other make client voice more effective include problems. Recently, governments (sometimes offering more convenient venues to air with help from donors) have started to learn complaints. Several studies have shown that from this experience, and have funded pro- the relationship between parents and teach- jects and programs that rely on, and require, ers is important: it should be supportive, the formation of local user groups and com- respectful, and cooperative, not punitive mittees to choose and implement develop- and confrontational. The success of local ment projects. Rather than give transfers of communities in improving education can income to individuals, which can be both (c) The International Bank for Reconstruction and Development / The World Bank Clients and providers 73 politically and administratively dif�cult, gov- of participation.204 More recent programs in ernments have channeled money through Indonesia have bene�ted from this experi- community groups. The various approaches ence and have been designed to elicit more that have been tried address two possible widespread participation (see spotlight on weaknesses in the “long route� of accountabil- the Kecamatan Development Program). ity through governments: implementation, or A real risk comes from the speed with the “compact� by single-purpose user groups, which groups are constituted and funds dis- and “voice,� which allows communities to bursed. Elites can mobilize more quickly, decide on projects to undertake. master the rules of submitting applications A recent evaluation of six early social (if they can read and the majority of the funds, most initiated in response to crises, community cannot), and present them- found that the programs were progressive, selves to the community as an effective con- though more between than within duit for receiving such funds. In one Sahe- regions.201 Special-purpose user groups have lian country a large fraction of project been more common. In water supply and funds was diverted for personal gain.205 sanitation particularly, there are numerous Much of the blame lies with the speed at cases of better implementation through which donors want to disburse funds such groups. In Côte d’Ivoire, when respon- (chapter 11) and with the limits this puts on sibility for rural water supply shifted from incentives and abilities to monitor the central government to user groups, break- behavior of leaders. Rushing to create social downs and costs were reduced.202 Some capital where it does not exist can do more local communities have used local contrac- harm than good. If there were ever a case tors, improving accountability and increas- for patience, this is it. It is not merely the ing ef�ciency through explicit contracts.203 creation of participatory formats but the When governments, especially local govern- encouragement of the abilities of poor peo- ments, are severely hampered in delivering ple themselves that will have longer-lasting services, these methods have the potential to effects. The policies to look for, then, may bring about marked improvements. be those of education, freedom of expres- These programs are new—and changing sion, transparency, and time. as lessons emerge. Because of their potential, The problem of capture is not limited to rigorous evaluation is a high priority. Which groups created for investment purposes. It aspects are replicable? How can pitfalls be also affects existing community groups and avoided? Some of the emerging lessons stem local governments. Both elitism and, in from the difference between groups that many cases, gender (men as opposed to emerge spontaneously and those that are cre- women) can determine who dominates tra- ated from above for the purpose of channel- ditional communities and local govern- ing money. ments.206 It is not clear that elite capture is always a problem. Wade (1988) proposes Capture. Groups constituted as a part of pro- that mobilizing community action may jects funded by outsiders may be particularly require the leadership of the more educated, prone to capture by elites. Local groups that connected elite. The lessons, though, are to evolve as a result of long-felt needs may or make sure that either the types of services may not be representative of the poorest funded by such methods have substantial people. But when those groups are used by public good characteristics (putting health higher levels of government or by donors to and education in a sort of “gray� area) or channel formerly unheard of sums of that the right to leadership is contestable. money, even representative groups tend to change. In Indonesia, when participation Developing government capacity. Some was mandated by national government to go special-purpose user groups, better funded through village councils, the increased par- than local governments, have drawn off ticipation of some members of communi- more capable of�cials to administer their ties was found to have a “crowding-out� funds (the same effect is seen at the national effect on others, leading to a net reduction level in other donor initiatives—see chapter (c) The International Bank for Reconstruction and Development / The World Bank 74 WORLD DEVELOPMENT REPORT 2004 11). One hypothesis is that this slows the delivered to all people is one of the many development of local government capacity. open questions on the agenda. But the opposite argument has also been made—that such groups are a catalyst for Client power in eight sizes developing local government capacity. In To sum up, increasing client power through northeast Brazil, social investment funds improved choice or direct participation will led villagers to organize and petition higher be important when people differ—are het- levels of government to, for example, guar- erogeneous in their preferences—or when antee a teacher to staff a school built by the either of the two legs of the long route to community.207 accountability is problematic. In terms of the decision tree (�gure 4.2) that deter- Sustainability. Participatory water pro- mines which of the eight types of solutions jects, underway since the 1930s and 1940s, is appropriate, client power matters at all have often improved water supply—at least three decision points. for awhile. But at some point water pumps and other pieces of expensive equipment Decision 1: Are politics pro-poor? Reliance break down. Covering the capital cost on client power should vary with the capac- (which is expensive) and obtaining the ity and orientation of government. Also technical help (also expensive) have always with the question of which level of govern- been the bottleneck for water projects in ment is problematic. When governments poor areas. When a new infusion of capital (central, local, or both) are pro-poor, they is necessary at short notice, the community may choose to enlist client groups as moni- must look either to donors or to regular tors or solicit their opinions regularly in sources of funds, such as taxes or other gen- sizes 3 and 4. Sizes 5 through 8, however, eral revenue. Eight or nine years after the require ways to avoid the problems of gov- original investment, are the donors still ernment. All four will involve getting infor- around? Do they have the same priorities mation to clients on their entitlements to they originally had? Can they respond and the performance of services. quickly to small individual requests? Often When levels of government differ in not. These demands will have to be met by their commitment to poor people, the role local government,208 and projects have been and sponsorship of user groups differ as evolving to work through them. well. If central government is a better cham- Such projects may have been a great deal pion of poor people, they may fund com- better than relying on inadequate govern- munities (if preferences vary between ment structures. The argument for them is them) or cash transfers or vouchers (if pref- strongest where the current government erences vary within them) in cases 7 and 8. system, especially the local government, is If local government is better, they can pro- weak, with few prospects for changing any vide or contract for these services. When no time soon. level of government is pro-poor, then This should, however, be a tactic that donors, if they are inclined to be involved at supports a longer-term strategy of develop- all, might choose to fund community ing governmental capacity—strategic incre- groups or organizations within civil society, mentalism, discussed in chapter 3. Caution being careful not to undermine the devel- is required when there appears to be a opment of government capacities. tradeoff between improving services in the short run and undermining delivery capac- Decision 2: Does heterogeneity matter? ity in the longer run. And the political con- Sizes 3, 4, 7, and 8 directly involve clients. sequences of participatory projects should When preferences differ by location then be the subject of careful evaluations. All this decentralization to local government or to complicates bringing these interventions to community groups (depending on the scale. It may be possible to replicate com- capacity and pro-poor orientation of the munity efforts in many places, but whether former) makes sense. If they differ by indi- this is the best way to make sure services are vidual, then purchasing power and compe- (c) The International Bank for Reconstruction and Development / The World Bank Clients and providers 75 Figure 4.2 Eight sizes �t all 1 Central government financing with contracting. Direct client input not Easy to monitor essential—citizen input through political process. Hard 2 Central government provision. Government may choose to enlist com- Homogeneous to monitor clients munities or users as monitors but is optional. 3 Deconcentrated central or local government provision with con- tracting. If individual, rather than community, variation in preferences Heterogeneous is important, transfers or vouchers targeted to poor can be used but Easy to clients are optional. If preferences vary by community, local government monitor can work and direct client input is not essential. Pro-poor Hard 4 Local or deconcentrated central government provision. Government politics to monitor may choose to enlist communities or service users as monitors. If preferences vary by individual, vouchers or cash transfers targeted to poor can be used. 5 Client power—experiment with contracts. If all levels of government are problematic, community user groups are essential and can be a source of contracts to private sector or NGOs. Funding may have to come from donors. If only local is problematic, center can fund com- Easy to munities or poor people directly with transfers. If only the center is monitor the problem, local government might provide adequately without direct client input. Clientelist 6 Client power—experiment with providers. Similar to 5 but relation to politics Hard provider cannot be with explicit contracts—more active monitoring Homogeneous to monitor of provider by the community is needed. If one level of government is clients pro-poor, it may enlist community input as in size 2. Evaluation and publicity of efforts of one community help others. Transfers or vouchers subject to strict rules possible even though service is uniform. 7 Client power—experiment with community control. Similar to 5 but requires more discretion, “choice� on part of communities (if funded Heterogeneous by donors or central government in case only the local government is clients the problem), by local government (if funded by donors and it is only Easy to the central government that is the problem), or by individuals. The monitor added discretion is needed due to heterogeneity of preferences. Explicit contracts with providers are possible. Hard to monitor 8 Client power—imitate market. Similar to 7 but explicit contracts are not feasible. Evaluation and publicity of efforts in one community help others. tition for individual business are preferable. work service users improve services either Providing information to clients is critical by choice in purchasing or by active partic- for translating their choices into better ser- ipation. vices. It is only in size 1, where government is perfectly capable of providing services Decision 3: Is monitoring easy or hard? directly, that client participation is optional. When monitoring is easier for clients than Possibly size 3 as well if government can for governments (at any level) then client accommodate varying needs of clients. For input may be required for sizes 2, 4, 6, and all other cases, the client needs to be placed 8. Parents of children, patients, and net- more �rmly at the center of service delivery. (c) The International Bank for Reconstruction and Development / The World Bank spotlight on the Bamako Initiative Putting communities in charge of health services in Benin, Guinea, and Mali In some of the world’s poorest countries, putting communities in charge of health services, and allowing them to charge fees and manage the proceeds, increased the accountability of local health staff and improved health services for the poor. T he Bamako Initiative in Benin, Guinea, and Mali reconciled tra- ditional community solidarity and provider payments with the objec- tives of the modern state.209 How? By than 10 percent of families used modern curative services. The approach focused on establishing community-managed health centers serving populations of 5,000 to 15,000 people. An • Standardizing diagnosis and treatment and establishing regular supervision. Scaling up incrementally strengthening the power of communities analysis of the main constraints in the three The Bamako approach was implemented over service providers. Policymakers bal- countries led to emphasis on service delivery gradually, with the support of UNICEF, anced this power with sustained central strategies focusing on the poor.211 Priorities WHO, and the World Bank, building on a involvement in subsidizing and regulat- included:212 variety of pilot projects.213 Since the early ing services—and in guiding community 1980s, it was progressively scaled up in the management. • Implementing community-owned revolv- three countries—from 44 health facilities in ing funds for drugs with local retention Benin to 400 in 2002, from 18 in Guinea to The initiative improved the access, avail- and management of all �nancial proceeds. 367, and from 1 in Mali to 559. This raised ability, affordability, and use of health services. Over the more than 10 years of • Revitalizing existing health centers, the population with access to services implementation in these three countries, expanding the network, and providing within 5 kilometers to 86 percent in Benin, community-owned services restored access monthly outreach services to villages 60 percent in Guinea, and 40 percent in to primary and secondary health services within 15 kilometers of facilities. Mali, covering more than 20 million peo- for more than 20 million people. They • Stepping up social mobilization and ple. Importantly, a legal framework was raised and sustained immunization cover- community-based communication. developed to support the contractual rela- age. They increased the use of services tionship with communities, the cost-shar- among children and women in the poorest • Pricing the most effective interventions ing arrangements, the availability of essen- below private sector prices, through sub- �fth of the populace. And they led to a tial drugs, and community participation sidies from the government and donors sharper decline in mortality in rural areas policies. Community associations and and through internal cross-subsidies than in urban areas. management committees were registered as within the system. Local criteria were Despite the various targeting mecha- legal entities with ability to receive public established for exemptions (table 1). nisms, affordability remains a problem for funds. many of the poorest families. But even with • Having communities participate in a biannual analysis of progress and prob- limited inclusion of the poorest people, lems in coverage with health services— Better health outcomes improvements were signi�cant.210 and in the planning and budgeting of for poor people services. Over the 12 or so years of implementation Revitalizing health networks • Tracing and tracking defaulters—and in Benin and Guinea, and more than 7 years In these three countries, serious disrup- using community representatives to in Mali, health outcomes and health service tions to the situation of health services had increase demand. use improved signi�cantly. Under-�ve mor- occurred during the 1980s as a result of a severe economic recession and �nancial Table 1 Reaching out to bene�t the poorest groups indebtedness. The health budget in Benin went from $3.31 per capita in 1983 to Disease Geographical Cross-subsidies Exempting the poor targeting targeting $2.69 in 1986. In Mali, rural infrastructure was almost nonexistent, and in Guinea, Focus on the burden of Focus on rural areas. • Higher markup and • Exemptions left to the health services had almost totally disap- diseases of the poor: Larger subsidies to co-payments on discretion of malaria, diarrhea, poorer regions diseases with lower communities peared—except in the capital city, respiratory infections, levels of priority • Exempted categories Conakry—during the last years of the malnutrition, • High subsidies for include widows, Sekou Toure regime. The vast majority of reproductive health child health services orphans poor families in the three countries did • Free immunization and oral rehydration not have access to drugs and professional therapy as well as health services. National immunization promotion activities coverage was under 15 percent, and less (c) The International Bank for Reconstruction and Development / The World Bank Spotlight on the Bamako Initiative 77 Figure 1 Under-�ve mortality has been reduced in Mali, Benin, and Figure 2 Improvements in under-�ve mortality Guinea, 1980–2002 among the poor in Mali Deaths per thousand Deaths per thousand births 450 400 Mali poor 1987 400 350 1996 Mali average Guinea average 350 300 2000 Guinea poor 300 Benin poor 250 250 200 Benin average 200 150 150 100 100 50 1972 1976 1980 1984 1988 1992 1996 2000 0 Source: Krippenberg and others 2003. Calculated from Demographic and Health Survey Poorest Second Middle Fourth Richest data for Benin 1996 and 2001; Guinea 1992 and 1999; and Mali 1987, 1996, and 2001. fifth fifth Source: Calculated from Demographic and Health Survey data 1987 and 1996 (based on births in the last �ve years before the survey). tality declined signi�cantly, even among the �ed with the quality of care, although 48 Community �nancing—a seat poorest. The poor-rich gap narrowed in the percent were not “fully� satis�ed. Health at the table three countries (�gures 1 and 2). In Guinea, care users found the availability of drugs to The community �nancing of key opera- the decline was steepest for the rural popu- be high (over 80 percent said drugs were tional costs bought communities a seat at lation and poorer groups. available) and the overall quality of care to the table. Donors and governments had to Immunization levels increased in all be good (91 percent). systematically negotiate new activities with three countries.214 They are very high in Greater access reduced travel costs, and community organizations. Governments in Benin, close to 80 percent—one of the the availability of drugs reduced the need to all three countries, with the support of highest rates in Sub-Saharan Africa. Immu- visit distant sources of care. Prices have donors, continued to subsidize health cen- nization rates are lower in Guinea and Mali, been kept below those of alternative ters, particularly to support revolving drug largely because of problems of access (�g- sources. In Benin the median household funds in the poorest regions. In Benin and ure 3). Coverage of other health interven- spending on curative care in a health center Mali today the public subsidy to health ser- tions also increased. The use of health ser- was $2 in 1989, less than half that at private vices is about the same per capita for rich vices by children under �ve in Benin providers ($5) or traditional healers ($7).217 and poor regions. In Guinea, however, pub- increased from less than 0.1 visit per year to Poor people still saw price as a barrier.218 lic spending has bene�ted richer groups more than 1.0. In Mali exclusive breastfeed- And a large proportion of the poor still do most. But all three countries face the chal- ing and the use of professional services for not use key health services in all three coun- lenge of emphasizing household behavior antenatal care,215 deliveries, and treatment tries. In Benin and Guinea the health system change and protecting the poorest and of diarrhea and acute respiratory infections allowed for exemptions, and most health most vulnerable. Establishing mechanisms increased for all groups, including the centers had revenue that they could have to subsidize and protect the poor remains a poorest (�gure 4).216 used to subsidize the poorest, but almost priority of the current reform process. In an independent evaluation in 1996 in none did. Management committees typi- Benin, 75 percent of informants were satis- cally valued investment over redistribution. Figure 3 Evolution of national immunization Figure 4 Antenatal care by medically trained coverage (DPT3), 1988–1999 persons in Mali by wealth group Percent Percent 100 100 Benin 75 Guinea 80 1996 Mali 60 2000 50 1987 40 25 20 0 1988 1990 1992 1994 1996 1998 0 Poorest Second Middle Fourth Richest Sources: World Health Organization, UNICEF, and Demographic and Health Survey data. fifth fifth Source: Analysis of Demographic and Health Survey data. (c) The International Bank for Reconstruction and Development / The World Bank Citizens and politicians 5 The most important political of�ce is that of pri- produce better service outcomes. Non- vate citizen. governmental and civil society organiza- chapter Louis D. Brandeis tions can help to amplify the voices of the poor, coordinate coalitions to overcome their collective action problems, mediate on Citizens’ voice in society and participation their behalf through redress mechanisms, in politics connect them to the people who and demand greater service accountability. represent the state—politicians and policy- Even when these measures have limited makers. Unlike the short route of account- scope, better information—through public ability between clients and providers dis- disclosure, citizen-based budget analysis, cussed in chapter 4, the long route of service benchmarking, and program impact accountability involves politics. That assessments—and an active, independent accountability has two parts: the relation- media can strengthen voice. ship of voice between citizens and politi- cians and policymakers (discussed here) and the relationship between policymakers and Citizen voice and political service providers (discussed in chapter 6). accountability This chapter asks several questions: Faced with classrooms without teachers, clin- Why don’t politicians in well-functioning ics without medicines, dry taps, unlit homes, democracies deliver education, health, and and corrupt police, poor citizens often feel infrastructure services more effectively to powerless.219 Elected representatives seem poor people even though they depend on answerable only to the more powerful inter- poor people’s votes? Why are public expen- ests in society if at all. When politicians are ditures systematically allocated to con- unaccountable to poor people as citizens, the struction projects and the salaries of bulky long route of accountability—connecting cit- state administrations, often at the expense izens with providers through politicians— of making services like schooling work? breaks down, voice is weak, and providers can And why, when the government does get away with delivering inadequate services spend money on services that the poor rely to poor clients.220 When poor citizens are on, such as primary health care, is service empowered, whether on their own or in quality so poor? Finally, what can citizens, alliance with others, their demand for particularly poor citizens, do when politi- accountability can make politicians respond cians fail to make services work for them? in ways that compensate for weaknesses else- Empowering poor citizens by increasing where in the service delivery chain. their influence in policymaking and align- ing their interests with those of the non- Services are politically powerful poor can hold politicians more accountable For poor people the only routine interaction for universal service delivery. Elections, with the state may be at the delivery point of informed voting, and other traditional services. Election platforms show that politi- voice mechanisms should be strengthened, cians are very aware that poor people’s per- because these processes—and the informa- ceptions of the state are shaped by the quality tion they generate—can make political of services. The 30-Baht Gold Card scheme in commitments more credible, helping to Thailand promised inexpensive universal 78 (c) The International Bank for Reconstruction and Development / The World Bank Citizens and politicians 79 healthcare and helped the Thai Rak Thai public agencies in 1993 do both, giving party win a landslide victory in the 2001 par- clients the right to redress through the Public liamentary elections.221 Service delivery was Complaints Bureau if corrective action for important for the Labor party’s successful noncompliance is not taken.226 2001 election campaign in the United King- Another complication is that the voice dom. A Prime Minister’s Delivery Unit has relationship links many citizens with many been set up to monitor progress.222 politicians—all with potentially very differ- Even when services do not �gure explicitly ent interests. When services fail everyone, in elections, politicians often seek to enlarge the voice of all citizens (or even that of the their political base by providing free public non-poor alone) can put pressure on politi- services or lucrative service-related jobs to cians to improve services for all citizens, their supporters. And people are increasingly including the poor. But when services fail concerned about accountability for services primarily poor people, voice mechanisms outside the voting process. In Brazil, India, operate in much more dif�cult political and and South Africa civil society organizations social terrain. Elites can be indifferent about are analyzing the allocation and use of public the plight of poor people.227 The political resources in the budget to understand their environment can swamp even well-orga- impact on the poor.223 With so much political nized voice. Protest imposes large costs on attention paid to services, why is the voice the poor when their interests clash with relationship often so weak? those of the elite or those in authority.228 It then matters whether society is homoge- Voice is the most complex neous or heterogeneous and whether there accountability relationship is a strong sense of inclusion, trusteeship, in service delivery and intrinsic motivation in the social and Voice is the relationship of accountability political leadership of the country. To between citizens and politicians, the range of expect poor people to carry the primary measures through which citizens express burden of exerting influence would be their preferences and influence politicians unfair—and unrealistic. (�gure 5.1).224 Accountability in this context Finally, voice is only the �rst part of the is the willingness of politicians to justify their long route of accountability. That compli- actions and to accept electoral, legal, or cates its impact on services, since the impact administrative penalties if the justi�cation is depends also on the compact relationship found lacking. As de�ned in chapter 3, between policymakers and providers. Even accountability must have the quality of strong voice may fail to make basic services answerability (the right to receive relevant work for poor people because the compact is information and explanation for actions), weak. But the reverse can also be true, as was and enforceability (the right to impose sanc- the case in the former Soviet Union. tions if the information or rationale is deemed inappropriate).225 One complication Figure 5.1 Voice in the service delivery framework is that voice is not suf�cient for accountabil- ity; it may lead to answerability but it does The state not necessarily lead to enforceability. Politicians Policymakers In principle, elections provide citizens of accounta with both answerability (the right to assess a ro u te bili ng ty Lo ce candidate’s record) and enforceability (vote Voi the candidate in or out). In practice, democ- S h o rt r o u t e racies vary greatly on both dimensions, as do Citizens/clients most attempts to exercise accountability. Cit- Coalitions/inclusion izen charters may spell out the service stan- dards and obligations of public agencies Nonpoor Poor toward their clients, but without redress the obligations may not be enforceable. In Malaysia the client charters introduced for Services (c) The International Bank for Reconstruction and Development / The World Bank 80 WORLD DEVELOPMENT REPORT 2004 Pro-poor and clientelist service ously, if delivery mechanisms do not delivery environments account for these speci�c country and ser- That voice is complex still begs the question vice differences, they are likely to fail, and of why, in societies where the average citizen the poor suffer. is poor, services fail poor people. The answer has to do with whether service The interaction of voice delivery settings are “pro-poor� or “clien- and accountability telist.� The distinction reflects the incentives When populations are heterogeneous, it facing politicians, whether services are matters whose voices politicians and policy- designed to be universal and available to the makers hear and respond to. Where popula- average citizen or vulnerable to targeting to tions are polarized around nonservice “clients� by political patrons, and, if for- issues—religious, ethnic, caste, or tribal mally targeted to the poor, whether they are background, for example—voters care in practice captured by elites (table 5.1). more about what politicians promise on Pro-poor settings are those in which politi- these polarizing issues than on services, giv- cians face strong incentives to address the ing politicians incentives to pursue other general interest. Clientelist political envi- goals at the cost of effective services. Where ronments are those in which, even though politics is based on identities and patron- the average citizen is poor, politicians have age, the poor are unlikely to bene�t from strong incentives to shift public spending to public services unless they have the right cater to special interests, to core supporters, “identity� or are the clients of those with or to “swing� voters.229 When the average political power. In failed or captured states citizen is poor, catering to special interests voice can become meaningless. Politicians at the cost of the general interest is clien- have neither the incentives nor the capacity telism. to listen. The distinction between pro-poor and Under what circumstances, then, is voice clientelist is clearly an oversimpli�cation, likely to lead to greater accountability? Elec- but it provides a useful way of thinking tions can lead to improved services if the about service delivery mechanisms. High- promises politicians make before elections are quality services for all are less likely if politi- credible. The framework of citizen rights, the cians cater to special interests rather than to right to information, service design, the influ- the interests of the average citizen. Making ence of the media, and administrative proce- services work for poor people is obviously dures for redress and appeal are all important more dif�cult in a clientelist environment for voice.230 So too is the effectiveness of the than in a pro-poor environment. Less obvi- institutions of accountability, such as parlia- Table 5.1 Pro-poor and clientelist service environments when the average citizen is poor Politicians’ Service delivery Inclusion Systemic incentives expenditure and exclusion service capture design Pro-poor No strong incentives to Promote universal Most poor people enjoy None cater to special interests, provision of broad the same access and preferring instead to basic services that service quality as address general bene�t large non-poor due to interests segments of society, network, political, including poor social, or altruistic people and the reasons non-poor Clientelist Strong incentives to Permit targeting to Poor people do not Systemic service capture cater to special interests, narrow groups of enjoy the same access by local or national elites; to core supporters, or to non-poor “clients� and service quality as ultrapopulist governments “swing� voters and sometimes to the non-poor, though (such as “Curley poor people but with speci�c groups of poor effects,� see box 5.2) features making “clients� may do so services vulnerable to capture by non-poor (c) The International Bank for Reconstruction and Development / The World Bank Citizens and politicians 81 ments, courts, ombudsman, anticorruption Political incentives for basic services commissions. And so too are higher stocks of If delivered effectively, basic services such as social capital, because they help overcome the primary health care and primary education collective action problem underlying voice, bene�t the poor disproportionately. But particularly for poor people.231 democratically elected politicians in coun- What can be done to strengthen voice, tries where the median voter is likely to be a particularly for poor citizens, in demanding poor person, or where poor people consti- better services? The answer depends greatly tute the majority of voters, often seem to on the political setting, but in functioning have little incentive to provide such basic democracies with elections and voting, at services. And voters seem unable to least three things should be done. strengthen incentives for politicians to • Deepen understanding and awareness of ensure better public services. Why? why the politics of service delivery is so How politicians and voters make deci- often clientelist and not pro-poor. sions and how politicians compete hold some • If the politics is clientelist, consider what answers.235 When politicians have incentives changes in the service delivery environ- to divert resources (including outright cor- ment might alter political incentives and ruption) and to make transfers to a few improve outcomes. clients at the expense of many, efforts to pro- vide broad public services are undermined. • When choosing how to deliver services, How easy it is for voters to learn about the factor in, to the extent possible, the pro- contributions of politicians to a particular poor or clientelist influence of political service—and therefore for politicians to competition on the incentives for service claim credit for the service—differs consider- delivery. Recognize and account for gov- ably by service (box 5.1). The degree of polit- ernment failure arising from clientelism. ical competition is important. For example, analysis suggests that an increase in the com- The politics of providing public petitiveness of elections seems to have a big- services to poor people ger effect on primary school enrollment than In 1974 only 39 countries—one in four— increases in education spending.236 were electoral democracies. By the end of Three factors therefore appear to be Figure 5.2 Democracy’s century World population by polity 2002, this had grown dramatically to 121 especially important for influencing politi- governments—three in �ve.232 Over the last cal incentives for service delivery: Percent 100 century, the percentage of people living in 90 democracies with competitive multiparty • How well voters are informed about the contribution speci�c politicians or polit- 80 elections and universal suffrage has increased ical parties make to their welfare. 70 dramatically (�gure 5.2). 60 Rapid democratization has brought rep- • Whether ideological or social polariza- 50 resentation and liberties, but not rapid tion reduces the weight voters place on 40 improvements in services for poor peo- public services in evaluating politicians. 30 ple.233 Most, if not all, new democracies are • Whether political competitors can make 20 low-income countries with substantial credible promises about public service 10 poverty. Services available to poor people in provision before elections. 0 these young democracies seem to be not 1900 1950 2000 much different from those available in non- Informed voters Protectorate/colonial democracies. In some cases services are The incentives for transfers targeted to dependence worse than those provided by ideologically informed voters are greater when voters in Monarchy committed but nonelected governments in general lack information about the quality (constitutional/ traditional/absolute) single-party, socialist countries. Whether of public services and the role their elected Restricted countries have elections or not seems not to representatives play in affecting quality. The democracy matter for public perceptions of corrup- same is true if uninformed voters are easily Authoritarian/ tion, and since corruption worsens service swayed by political propaganda, or if they totalitarian regimes delivery for poor people, by implication for vote on the basis of a candidate’s charisma Democracy public perceptions of effective services.234 or ethnic identity rather than record. Source: Freedom House (1999). (c) The International Bank for Reconstruction and Development / The World Bank 82 WORLD DEVELOPMENT REPORT 2004 Informed voting can be costly. Detailed Information de�ciencies thus lead voters to behavioral studies show that voters tend to give more credit to politicians for initiating adopt simple rules of thumb based on very public works projects (including school limited information about politics and construction), providing direct subsidies public policies.237 Most of the information for essential commodities, and increasing voters use is likely to be essentially “free,� in employment in the public sector (including that it comes incidentally with the perfor- hiring teachers and doctors) than for mance of social and economic roles.238 This ensuring that teachers show up for class kind of information tends to vary widely and can teach—or that doctors come to over the electorate, depending on occupa- clinics and heal. tion, social setting, and cultural norms. Vot- If voters vote with limited information ers also behave myopically, giving much or if they are uncoordinated but can be greater weight to events around election swayed by propaganda or bribes, special- times or to service outputs that are immedi- interest groups can capture policies by pro- ately visible.239 viding campaign �nance or mobilizing In principle, citizens could employ vot- votes.241 These interest groups need not be ing rules requiring very little information de�ned along rich-poor lines. They could and still motivate politicians to pursue be organized coalitions of voters (such as policies in their interest—if they could farmers or public sector employees) that coordinate their efforts.240 It is harder for lobby politicians to protect their interests, voters to coordinate rewards or penalties pushing for targeted policies at the expense for basic health and education because of of policies that would bene�t the many. the dif�culties in evaluating these services and attributing outcomes to politicians Social polarization (see box 5.1). Transaction-intensive public Social polarization can lead to voting based services such as education and health on social, ethnic, or religious identity rather depend on day-to-day provider behavior. than policy or service delivery performance. Quality is hard to measure and attribute. This too limits political incentives to pursue BOX 5.1 Why are public health and education services so dif�cult to get to poor people? When even the poorest of parents care deeply coordinated, clear preferences in health and Because of political problems of information about educating their children, why is it so dif�- education services.Furthermore, successful out- and credibility, public antipoverty programs are cult for them to do something to ensure that comes require supportive household behavior, more likely to take the form of private transfers, the village teacher actually shows up for work and very heterogeneous social and cultural such as food subsidies, electricity subsidies, agri- regularly? A political economy perspective on household norms may make households cultural price protection, construction projects, public service delivery suggests that basic respond differently to public interventions. and public sector employment. Programs of this health and primary education are very dif�cult to get right because they are transaction-inten- • Because of the dif�culty of regularly monitor- kind are easier to capture and more amenable to targeting than basic health and education, ing these services and of measuring and sive services with outcomes that depend attributing their long-term impact, it is harder which are more suited to universal provision.* crucially on the judgment and behavior of for politicians to claim credit for these That is why programs narrowly “targeted� to the providers, both dif�cult to monitor continually, services than for a road or a well. And politi- poor may not be optimal in the sense of having and on household behavior. cians who promise to improve these services the most impact on the economic well-being of poor people. • Learning takes place over long periods and may lack credibility and lose elections. For these reasons, politicians are likely to prefer the bene�ts of preventive health care are not always obvious. Compared with other, more infrastructure to human development, and visible public services—electricity or water are prone to using basic health and primary connections, rural roads, law and order— education services as patronage for clients, *van de Walle (1998) concludes, from a synthesis of monitoring basic education and health ser- rather than as universal services to be research on public spending and the poor, that there is vices makes large information demands on provided for the general good. a well-substantiated case for “broad targeting� of the both voters and politicians. poor by allocating greater resources to universal public So when poor people are uninformed, soci- services such as basic health and education. In contrast, • Poor voters may be uninformed because they ety is polarized along social or religious lines, �nely targeted food subsidies or other redistribution schemes may sometimes be detrimental to the inter- are illiterate. and politicians lack credibility or are prone to ests of the poor due to the burden of administrative • Where populations are socially polarized or het- clientelism, basic public services for poor people costs and unintended behavioral responses. See also erogeneous, households are less likely to have are the most likely to suffer. van de Walle and Nead (1995). (c) The International Bank for Reconstruction and Development / The World Bank Citizens and politicians 83 public policies in the general public inter- est. James Curley, an Irish Roman Catholic BOX 5.2 The “Curley effect� mayor of Boston in the �rst half of the 20th Described as “The Rascal King,� James the lowest population growth rate of any century, fanned class and religious divisions Michael Curley dominated politics in Boston U.S. city of comparable size. for electoral advantage and was repeatedly for half the 20th century, holding elected Glaeser and Shleifer (2002) call this the reelected despite the damage his policies did of�ce, among others, as four-time city mayor “Curley effect�—increasing the size of one’s between 1913 and 1950, besides serving political base, or maintaining it, through dis- to the city’s growth (box 5.2). two prison terms on corruption charges. tortionary, wealth-reducing policies.They use It is this effect of social polarization on Admired by working-class Irish families, Cur- it to shed light on the ethnic and class politics political incentives that partly accounts for ley was noted for his railing against the of service delivery when the net effect is to the empirical evidence on the negative cor- Protestant Yankee establishment and for his impoverish the overall community.They show rough-and-tumble ways. how the Curley effect may apply to Detroit, relation between ethnic heterogeneity and Curley used patronage, cash, and rhetoric USA, to contemporary Zimbabwe, and to the the availability of public goods.242 More to shape his electorate, driving the richer Labor party in the United Kingdom before its generally, basic public services, particularly Protestant citizens out of the city to ensure current reincarnation as New Labor. those that are not easily excludable such as his political longevity. Curley’s tools of The Curley effect demonstrates that patronage were public services, large con- clientelism need not bene�t only rich clients. primary education, can also deteriorate struction projects, and public employment. It can bene�t poor clients as well, and still where there is social fragmentation—some In his �rst year as mayor, Curley raised the imply substantial losses in ef�ciency through social groups do not want to pay for public salaries of police patrolmen and school cus- the misallocation of public resources. So, goods that bene�t other groups.243 todians but cut the salaries of higher-ranking clientelism results in inef�cient, targeted allo- police of�cers and school doctors (Beatty cations that bene�t only a few, as opposed to 1992). Miles of sidewalks were laid in Irish allocations to universal public services that Credible politicians neighborhoods, but the cobblestones of bene�t larger segments of the same poor swank Yankee neighborhoods crumbled and not-so-poor populations. Even when voters are informed, public pol- (O’Connor 1995). Boston did not flourish Sources: Glaeser and Shleifer (2002), O’Connor icy can fall short when the promises of under Curley: between 1910 and 1950, it had (1995), and Beatty (1992). politicians are not credible. When candi- dates cannot or do not make credible promises before elections (because aban- doning promises costs election winners lit- politician for building a school or assigning tle), incumbents are more insulated from teachers, but they can less easily verify that the disciplining effects of political competi- the politician is responsible when the build- tion. Challengers cannot mount effective ing is maintained or supplied, or when the campaigns because they cannot convince teacher is present and competent. If politi- voters that they will do a better job. Fur- cians cannot take credit for their efforts to thermore, if politicians are credible only to improve teacher quality, teacher quality is their “clients,� more public resources will be likely to be low—and voters are unlikely to allocated to these clients. This can have expect anything else. In Pakistan nonprofes- large implications for universal health and sional considerations have been common in education services. Incumbents enjoy the placement of teachers.245 The incentives greater discretion to pursue goals other facing local politicians have been important than those preferred by the majority of citi- factors in the low quality of rural schools zens who may be poor, goals such as provid- (box 5.3). ing narrowly targeted services to their sup- In many countries, politicians do not porters at the cost of more universal public campaign on their policy record, probity, or services that bene�t all.244 history of program involvement or on the policy record of their party. Voters then are Credibility and credit go hand in hand. likely to believe politicians who have shown Credibility problems also arise when politi- themselves to be reliable sources of personal cal competitors make credible promises but assistance. They might be locally influential their term in of�ce is too short to claim people who have helped families by provid- credit for policies with long maturing out- ing loans or jobs or by resolving bureau- comes. Promises of jobs or public works cratic dif�culties. Without well-developed projects can be delivered soon after an elec- political parties or national leaders who are tion. But promises to improve education credible, promises of targeted favors are all quality and outcomes are much less credi- that voters can rely on in making electoral ble. Similarly, voters can easily credit a choices. (c) The International Bank for Reconstruction and Development / The World Bank 84 WORLD DEVELOPMENT REPORT 2004 (local public goods, projects limited to a BOX 5.3 Better to build rural schools than to run them jurisdiction) or individuals and speci�c well in Pakistan groups (clientelism).246 What distinguishes clientelism? Clien- Elected of�cials in Pakistan have demon- roads, and water pipes close to supporters telism implies a credible threat of exclusion strated an extraordinary interest in target- and far from nonsupporters. ing services to their supporters, but much Voting by blocs of supporters makes from a stream of bene�ts if the voter less interest in services such as primary edu- patronage a more effective political strategy chooses to vote for the opposition.247 Thus cation that all voters can enjoy. Contribut- than the provision of well-functioning an incumbent politician can use clientelism ing to this outcome are three aspects of services that must be provided to all. Costly rural Pakistani politics: identity-driven poli- elections drive politicians to provide public to deter core supporters from switching tics, voting blocs that make it easy to iden- services to supporters who can be depended support. Clientelism is hard to pursue for tify core supporters, and costly elections. on to vote for them at low out-of-pocket cost. local or extensive public goods—bene�cia- Voter ignorance, poor information on Under these conditions, schools may ries are not reliable clients because they can political competitors, and the absence of get built for the corruption, employment, party credibility on broad policy issues and pro�t opportunities that construction support the opposition and still bene�t. encourage politicians and voters to build provides.Teachers are hired less on merit The Programa Nacional de Solidaridad personal relationships that make pre-elec- and more on how best to apportion patron- (PRONASOL) poverty alleviation program tion promises more credible. Because these age, particularly when absenteeism is not in Mexico spent an average of 1.2 percent of relationships are personal, they tend to be penalized.There are, in contrast, few based on narrow, excludable services systemic political incentives to make sure GDP annually on water, electricity, nutri- promised and delivered to core supporters. schools run well, teachers remain account- tion, and education in poor communities The distance between rural communities able, and children learn. between 1989 and 1994.248 Municipalities boosts the political ef�ciency of targeting dominated by the Institutional Revolution- political bene�ts—it is easier to site schools, Source: World Bank (2002l ). ary Party (PRI), the party in power, received signi�cantly higher per capita transfers than municipalities that voted in another party Credibility can make change dif�cult. (�gure 5.3). An assessment of PRONASOL Problems of political credibility can cause spending suggests that it reduced poverty by bad policies to become entrenched. Coun- only about 3 percent. Had the budget been tries often adopt poverty strategies based on distributed for impact on poverty rather subsidies for consumption and agricultural than party loyalty, the expected decline production, sometimes at the expense of would have been 64 percent with perfect tar- broad public services such as education and geting, and it would have been 13 percent health that might have resulted in lower even with an untargeted, universal propor- poverty and more economic growth. India tional transfer to the whole population. 249 subsidizes electricity, ostensibly for poor Even if voters want to vote for an opposi- farmers. Once political credibility is strongly tion party or candidate, they might be linked to a particular policy such as deliver- deterred by the fear of being penalized by ing subsidized electricity, these policies con- the withholding of funds by a central tinue to receive greater public resources than authority. So voters may end up keeping a they would if all political promises were party they may dislike in power in order to equally credible. Vested interests develop ensure funding for local public services. around suboptimal policies—rich farmers This is compounded by a coordination capture the power subsidy—which makes problem. Even if the majority of localities change even more dif�cult. wanted to vote against the incumbent party, without certainty about what other locali- Clientelism ties planned to do, the majority would end Clientelism is characterized by an excessive up supporting the ruling party to avoid tendency for political patrons to provide strategic miscoordination and the penalty private rewards to clients. Politicians allo- of loss of funds. cate public spending to win elections. To do Clientelism can also be the outcome of so, they can provide public goods that can political competition when the credibility improve everyone’s welfare (public goods of political competitors is limited—politi- that are extensive, such as law and order, cal promises are credible only to “clients.� 250 universal education, with no rivalry or Politicians with clientelist ties can ful�ll excludability). Or they can target localities campaign promises better than politicians (c) The International Bank for Reconstruction and Development / The World Bank Citizens and politicians 85 without them. When only clientelist promises mobilizing potential has been accompanied Figure 5.3 It paid to vote for PRI Mexico: PRONASOL expenditures according are credible, promises of construction and by an information revolution that has dra- to party in municipal government government jobs become the currency of matically simpli�ed information exchange Average expenditures per capita political competition at the expense of uni- and citizen access to of�cial information. (real 1995 pesos) versal access to high-quality education and Enthusiasm for direct citizen involve- 400 health care (as seen in box 5.3). Public works ment also comes from mounting frustration PRI 300 or jobs can be targeted to individuals and with the dominant mode of a national civil PRD groups of voters—clear evidence of political service delivering services that meet some 200 Other patrons ful�lling their promises to clients. It technically predetermined “needs� of the PAN is much more dif�cult to target the services of population.253 This frustration has led to 100 a well-run village primary school or clinic. greater interest in directly empowering citi- Cross-country evidence on public invest- zens and overcoming collective action prob- 0 1989 1990 1991 1992 1993 1994 ment supports the contention that credibil- lems, driven also by the �nding that civic ity and clientelism signi�cantly influence relationships and social capital are impor- Note: PRI = Institutional Revolutionary Party; PRD = Party of the Democratic Revolution; the provision of public services. There are tant determinants of government ef�cacy.254 PAN = National Action Party. Source: Estévez, Magaloni, and Diaz-Cayeros no variables that directly capture the credi- (2002). bility of pre-electoral promises or the extent Broad range of issues and tools of clientelism. But it is possible to argue that The rapid growth of citizen initiatives has in young democracies political competitors been described as a new accountability are less likely to be able to make credible agenda. It involves “a more direct role for promises to all voters and are more likely to ordinary people and their associations in rely on clientelist promises, and as these demanding accountability across a more democracies age, politicians are more likely diverse set of jurisdictions, using an to increase the number of clients since they expanded repertoire of methods, on the can count on client loyalty. A study sum- basis of a more exacting standard of social marizing the evidence shows that targeted justice.�255 Citizens are combining electoral spending—public investment—is higher in accountability and participation with what young democracies than in old and as would traditionally have been considered young democracies age, targeted spending the of�cial accountability activities of the increases.251 Corruption falls as democracies state. These initiatives address accountabil- age. These results are relevant for universal ity at various levels. Some are aimed at basic services since they are likely to be of strengthening voice in service delivery by lower quantity when public investment is enabling answerability and some at pushing high and of lower quality when corruption further for enforceability. These initiatives, is high. Similar cross-country evidence on and the state’s response, employ a number secondary and primary school enrollment of old and new tools, including tools based supports the view that credibility is a signi�- on information technology. cant influence on the provision of public These citizen initiatives cover a far-reach- services.252 ing array of issues, from improving law and order in Karachi256 to preparing citizen report cards. They vary tremendously in Beyond the ballot box: scale, ranging from global knowledge-shar- citizen initiatives to increase ing coalitions, such as Shack/Slum Dwellers’ accountability International,257 to community efforts in When elections are not enough to make ser- Mumbai to monitor arrivals of subsidized vices work for poor people, political pres- goods at local “fair price� shops in order to sure builds for new approaches that enable expose fraud in India’s public distribution citizens to hold politicians and policymak- system targeted to the poor.258 They also vary ers more directly accountable for services. in depth and reach. On election reform, they These activities do not replace the electoral range from generating background informa- process, but complement it to strengthen tion on election candidates and their perfor- the long route of accountability. The emer- mance in Argentina (Poder Ciudadano259) to gence of such citizen initiatives and their civil society efforts to implement and sustain (c) The International Bank for Reconstruction and Development / The World Bank 86 WORLD DEVELOPMENT REPORT 2004 an Indian Supreme Court judgment making ture into more dif�cult areas. The impact of it mandatory for all election candidates to these initiatives varies according to how disclose their assets and any criminal record. they are perceived by politicians and policy- On budget analysis, initiatives at one end makers and the government’s receptivity to seek to make national budgets accessible to change. Several studies link this receptivity citizens and at the other to promote village- to the stock of social capital.260 level participatory audits of local public One concern with some citizen initiatives expenditures (box 5.4). is that they can lead to conflicts of interest These citizen initiatives also use a broad and reduced accountability to poor people. range of tools, from door-to-door signature Facing funding uncertainties, many non- campaigns to cyber-activism. The rapid governmental organizations seek to diversify, growth of the Internet and communication starting from voice activities but moving on technologies has dramatically altered citizen to actual service delivery. When they become voice nationally and internationally, though advocates and providers at the same time, access is still limited by income and connec- there can be intrinsic conflicts of interest.261 tivity. Some innovative e-government appli- NGOs may suffer from their own lack of cations are reducing corruption and delivery accountability, internal democratic de�cits, times and increasing service predictability and gaps in their mandates.262 The award of and convenience. Karnataka, among India’s large service delivery contracts to a few big leading states in information technology, civil society organizations can exclude and has pioneered a computerized land records spell �nancial dif�culties for smaller organi- system to serve rural households (box 5.5). zations.263 And if community and civic groups are captured by unscrupulous leaders, Controversy and conflict of interest they can manipulate funding agencies and Two separate trends are discernible in citi- bene�ciaries for their own gain.264 zen voice initiatives: activities based on con- sultation, dialogue, and information shar- Information strategies ing, and activities more direct and to strengthen voice controversial, related to monitoring, com- Policies that increase information and coor- pliance, and auditing. Some activities start dination in voting, enhance the credibility of with indirect objectives, build internal political promises, and increase the ability of capacity and external trust, and then ven- civil society organizations to hold politicians BOX 5.4 Follow the public’s money The budget, a primary statement of government citizen voice in budget allocations and hearing), an open-air forum at which of�cial priorities, is for many citizens a black box, moni- implementation. records are presented alongside the testimony tored and assessed only by the traditional inter- derived from interviews with local people. nal accountability relationships within govern- Budget monitoring The Institute for a “Many people discovered that they had been ment. But it can be a crucial tool for citizens to Democratic South Africa makes information listed as bene�ciaries of anti-poverty schemes, influence and monitor public policy and ser- about provincial and national budget though they had never received payment. vices. Accordingly, participatory budgeting ini- allocations accessible to citizens. Its technical Others were astonished to learn of large tiatives are increasing rapidly in several coun- experts break public budgets down to facilitate payments to local building contractors for work tries.The challenge is to build the capacity of public comment. Special reports show how that was never performed� (Jenkins and Goetz citizen groups, to give politicians and policy- much money is allocated, say, to gender-related 1999). Until a state right-to-information law was makers the incentives to listen and act on citi- and children’s issues.The most direct influence passed in 2000—largely a result of the protest zen feedback, and to put out budgets that are of its work is in strengthening the ability of and lobbying efforts of the MKSS—its activists understandable and interesting to citizens. parliamentarians to participate more effectively had to obtain this information by appealing to in budget discussions. sympathetic bureaucrats. A similar national law Budget planning The most well-known was passed in 2003. budget planning initiatives come from city Budget auditing The Mazdoor Kisan Shakti municipalities in Brazil, such as Porto Alegre and Sanghathan (MKSS), a grassroots organization in Belo Horizonte. Neighborhoods indicate their the north Indian state of Rajasthan, has turned Sources: Andrews and Shah (2003), Singh and Shah spending needs at budget forums, and delegates ordinary citizens into �nancial auditors. Its key (2003), Goetz and Jenkins (2002), and Jenkins and then bring these needs to assemblies, ensuring innovation has been the jan sunwai (or public Goetz (1999). (c) The International Bank for Reconstruction and Development / The World Bank Citizens and politicians 87 BOX 5.5 Down to earth: information technology improves rural service delivery Karnataka state in India has pioneered Bhoomi the village accountant controlled the process, days in waiting time and (net of the higher user (meaning land), a computerized land record sys- with little of�cial or client monitoring. Even fee) Rs. 806 million in bribes tem serving 6.7 million rural clients. Its main where there was no fraud, the record system The resistance of village accountants had to function is to maintain records of rights, could not easily handle the division of land into be overcome in implementing Bhoomi.The tenancy, and cultivation—crucial for transferring very small lots over generations. chief minister, revenue minister, and members of or inheriting land and obtaining loans. Started Farmers can now get these records in 5–30 the legislature championed Bhoomi, which in 1991 as a pilot, the Bhoomi system now has minutes and �le for changes at a Bhoomi kiosk. helped.There are now plans to expand beyond kiosks in each of the state’s 177 subdistricts, ser- The entire process takes place in the vernacular, land transactions.The Indian government has vicing some 30,000 villages. Kannada. Clients can watch a second computer suggested that other states consider similar sys- Under the old system, applicants faced long screen facing them as their request is processed. tems to improve accountability and ef�ciency in delays (3–30 days), and nearly two out of three Users pay a fee of Rs. 15. In a recent evaluation, services that are vital to rural households. clients paid a bribe—70 percent paid more than only 3 percent of users reported paying a bribe. Rs. 100 (the of�cial service fee was Rs. 2).There The evaluation estimates that on average Sources: World Bank staff and Lobo and Balakrish- was little transparency in record maintenance— Bhoomi annually saves clients 1.32 million work nan (2002). and policymakers accountable are likely to group. That made benchmarking—system- improve services for poor people. Conversely, atic comparisons across time or space—easy. the lack of transparency in information dis- The Uganda information was not about the closure can come at a high price. Cases dur- general quality of education or general budget ing the earliest phase of the outbreak of support, or even about leakage from national Severe Acute Respiratory Syndrome in mid- education budgets. It was about one type of November 2002 in China were not openly transfer—capitation grants—disaggregated reported, which allowed a new and severe to the school level, responsibility for which disease to become silently established in ways was easy for parents and voters to assign to the that made further spread almost inevitable.265 school principal. In addition, the information Information campaigns have dramatically made clear to voters what the school should altered the behavior of politicians and policy- have received. The information in Buenos makers, but many have also failed to induce Aires was also speci�c to individual hospitals. change. Understanding when information Government responses to food crises campaigns can succeed is thus critical. show how the provision of high-level infor- mation is politically enforceable by voters. What makes for a successful First, a food crisis is a single, speci�c issue. information strategy? Second, responsibility for it is known to rest Tracer studies of spending on Ugandan edu- ultimately in the state chief minister’s of�ce. cation revealed leakages as high as 90 per- Third, there are no complicated issues of cent. Once the information was publicized, quality measurement—voters know imme- the budgeted resources reaching schools diately that they are bene�ting if they receive rose dramatically.266 Studies suggest that assistance. Benchmarking is a bit more com- newspaper readership and availability in plicated but still doable. Voters know if oth- India spur state governments to respond to ers less deserving receive assistance. But they food crises.267 In Buenos Aires, publishing do not know what effort governments the wildly different procurement prices paid should make in responding to food crises by city hospitals for similar products led to (which is different from the benchmarks in rapid convergence of prices.268 What did the Uganda tracer studies, where voters these information strategies have in com- knew exactly how much money should have mon? The information was speci�c. Political reached individual schools).269 interest in addressing the problem was high. Strong political or bureaucratic interest in And the information was electorally salient. correcting the problem. In some cases Speci�c information. The information (Uganda and possibly Buenos Aires), identi�ed speci�c government decisions, spe- national politicians did not bene�t, and ci�c decisionmakers, and the effect of the potentially lost, from leakages or inef�cien- decision on the voters individually or as a cies. That is, corruption was the product of (c) The International Bank for Reconstruction and Development / The World Bank 88 WORLD DEVELOPMENT REPORT 2004 bureaucratic shirking rather than political Citizen report cards: rent-seeking. Education had become a information as political action major issue for the president of Uganda, Other information strategies look directly at and his reputation was on the line. He had public service outputs (quality and quantity made public promises, followed by the of services provided by government) rather highly visible action of transferring more than inputs (prices paid, budgets committed funds to local schools. His ability to ful�ll and delivered). The best known are the citi- those promises was being undermined by zen report cards developed by the Public bureaucratic malfeasance. Once the malfea- Affairs Centre in Bangalore, India.271 Citi- sance was revealed, the fear of sanctions was zens are asked to rate service access and qual- enough to hold individual bureaucrats ity and to report on corruption and general accountable and produce rapid change. grievances about public services. Citizen In other cases, such as assistance in a report cards have spread to cities in the food crisis, there is considerably more room Philippines, Ukraine, and, on a pilot basis, for shirking. Citizens �nd it dif�cult to Vietnam. They have recently been scaled up know how large the crisis is, what resources in India to cover urban and rural services in are available, and how ef�ciently and equi- 24 states.272 The results have stimulated con- tably the resources are distributed. This siderable media, bureaucratic, and political uncertainty leaves room for political inac- attention and acknowledgment of their con- tion. But the consequences of government tribution to service improvements. inaction—starvation deaths and their Because citizen report cards focus on ser- reports in the media—are grave enough to vice outcomes, they do not provide voters with tarnish the chief executive’s reputation, information about speci�c decisions that spe- which gives the state administration a ci�c policymakers have made—or not made. major reason to avoid them. Famines do Nor do they give voters information (at least in not occur in democratic countries, even their �rst round) about service benchmarks, very poor ones, because the survival of the except to the extent that the agencies them- government would be threatened by the selves have established service standards opposition and by newspapers and other (repeat report cards do provide implicit bench- media.270 And the more citizens are marks from the previous report card). So it can informed about the crisis and the needed be hard for voters to assess, on the basis of one response, the more likely they are to hold report card, whether the results justify voting politicians electorally accountable. against the incumbents at the next election. Report cards seem to have had a more The issue is important electorally. Politi- direct influence on the heads and senior man- cians are not interested in improving perfor- agers of the municipal and utility agencies mance if voters do not care. Voters can be responsible for services, as in Bangalore. The well informed and know who the responsi- high visibility of report cards in the press and ble politician is, but still not hold the politi- civic forums turns them into league tables of cian accountable because other issues loom the ef�cacy of municipal agencies. The repu- larger. Where conflict is rife, or society is tational competition arising from the report polarized, the politician’s stance on conflict cards is enhanced by joint agency meetings on or polarization may dominate voter atten- the report cards attended by prominent social tion, allowing the politician to get away with and political leaders and citizens. poor performance on other issues. In But report cards clearly also perform a Uganda, the president made education a political function. Politicians can ignore poor central part of his election manifesto. In public services if they believe that voters can- Buenos Aires, municipal politicians may not penalize them for poor performance. Or if have been concerned that voters would view they believe that a political challenger cannot the corruption in hospitals as indicative of credibly promise voters better performance. deeper problems of malfeasance, in the city An NGO conducting broad surveys and issu- government and, because Buenos Aires is ing report cards on public services changes the capital, in other cities and the country. the equation. Now incumbent politicians are (c) The International Bank for Reconstruction and Development / The World Bank Citizens and politicians 89 confronted by an organized effort to improve • Supporting civil society organizations that public service delivery, which creates a latent show how to mobilize citizens and be a political force that is credible because there is credible voice for public service provision. no obvious personal gain to members of the • Supporting mass media development. NGO. The information that the NGO gener- First, improving the media’s ability to ask ates and disseminates is a political challenge, the right questions (reporting on whether both because of the demonstrated underlying government policy succeeds or fails, ability to mobilize citizens to answer a survey including how to identify the correct and the power of that information in the benchmarks). Second, improving the hands of informed voters. media’s credibility (independence from private interests that bene�t from govern- ment largesse and from government influ- Some implications ence, advertising revenue, or ownership). for information strategies Reducing barriers to entry is key here. These examples show that pure information Third, improving interaction with civil strategies work in fairly speci�c circum- society to generate information that reveals stances. But many information strategies are public malfeasance or nonfeasance.274 not designed around the speci�cs of a partic- ular country or service. Information about Decentralization broad aggregates of public sector perfor- to strengthen voice mance—whether based on surveys, budget Decentralizing delivery responsibilities for studies, or other methods—is less likely to be public services is prominent on the reform as politically relevant. Why? Because it does agenda in many developing countries (see not provide voters with a sense of how their chapter 10). Bolivia, India, Indonesia, Nigeria, representatives in government have hurt or Pakistan, and South Africa—to name a few— helped them. are all part of a worldwide movement to Like report cards, such information can decentralize. A key objective, usually linked to still be useful if voters can benchmark the the political motivation for decentralization, is information or if the very collection of the to strengthen citizen voice by bringing services information implies some latent political and elected politicians closer to the client. organization that could challenge incum- bents. But in many cases, the information Decentralization of service delivery collected is one-off—collected by donors and to local governments other foreign entities (posing no political Experience with decentralization varies. In threat by de�nition), by local survey �rms Bolivia the creation of rural local govern- (with no speci�c interest in social services), ments has been associated with dramatic or by civil society organizations (which care shifts in public allocations away from infra- deeply about public service performance but structure and into the social sectors—and a play no electoral role). sharp fall in the geographic concentration At the end of the day, these efforts tell citi- of public investments as they get more zens what they already know—that services evenly dispersed across regions.275 But oth- are bad. They might tell them exactly how bad ers have been less lucky, with increased and which services are worse than others— regional inequalities and the capture of roughly the information citizens already had, public resources by local elites. Since several but more quanti�ed. What citizens do not major decentralization reforms are just have, and what they need help in getting, is beginning (Indonesia, Pakistan), there is a information about how bad their neighbor- tremendous opportunity to rigorously eval- hood’s services are relative to others’ and who uate the impact of different institutional is responsible for the difference.273 In these designs on the quality of public goods. cases outsiders can help in several ways: What does it take for political decentral- • Supporting civil society organizations ization to improve universal, basic social that generate and use speci�c informa- services? Two conditions. First, voters must tion about service delivery. be more likely to use information about the (c) The International Bank for Reconstruction and Development / The World Bank 90 WORLD DEVELOPMENT REPORT 2004 quality of local public goods in making Citizen voice in eight their voting decisions. Second, local politi- sizes cal promises to voters must be more credi- Whether a political system is pro-poor or ble than regional or national promises. clientelist is dif�cult to assess and address. In principle, the impact of decentralization This is obviously the case for outside actors on informed voting and political credibility such as donors, but also for those within a could go either way. On the one hand, voters country, who are naturally influenced by the may make more use of information about history and traditions of their particular polit- local public goods in their voting decisions ical system, such as parliamentary democracy. because such information is easier to come by But the payoffs in service delivery for assessing and outcomes are more directly affected by whether the environment is pro-poor or local government actions. And political agents clientelist can be high. Even if the politics are may have greater credibility because of prox- clientelist, policy choices can be made that are imity to the community and reputations likely to yield better results than the mis- developed through social interaction over an guided application of policies that work well extended period. On the other hand, local only in pro-poor environments. voters may be apathetic about local elections Such choices can be combined with con- and have little or no information about the siderations of whether preferences are resource availability and capabilities of local homogeneous or heterogeneous (a feature governments. Social polarization may be of the relationship between clients and more intense because of age-old differences providers, discussed in chapter 4), and across settled communities. With closer social whether services are easy or hard to monitor relations between elected representatives and by policymakers (and therefore whether con- their clients, clientelist promises to a few vot- tracts between policymakers and providers ers may be easier to make and ful�ll. can be written, as noted in chapter 3 and dis- cussed in more detail in chapter 6). Simply Managerial decentralization put, the more people differ in their prefer- and political credibility ences, the more the decisions about service There has also been a push for institutionaliz- delivery should be decentralized. The harder ing greater autonomy of decisionmaking in it is for policymakers to monitor, the more schools, hospitals, and clinics—and encour- clients need to be involved and the stronger aging greater participation of citizens client power must be. through parent-teacher associations and Different combinations of these character- health committees. These institutional inter- istics lead to different choices, some a better �t ventions are also likely to address the credibil- than others, so that while no one size �ts all, ity of elected politicians. Politicians located at for illustrative purposes perhaps eight sizes the center far from the communities where might (�gure 5.4). None of these characteris- services are delivered cannot credibly tics or choices can be precisely rendered promise to improve service quality in such because countries lie on a continuum. But transaction-intensive services as basic health understanding them can help in thinking and education. At most they can commit only about the arrangements that are the most to providing such veri�able elements as infra- likely to make services work for poor people. structure, equipment, and salaries. Figure 5.4 also illustrates the broad service When responsibility for delivering and delivery arrangements and the implied policy monitoring primary education is completely choices that are appropriate under different centralized, the political incentives for settings. The biggest problem? The appropri- improving the quality of schooling are weak. ate choice is often not made. But if monitoring of providers is decentral- In many countries, policymakers assume ized (to clients), voters need verify only that that for transaction-intensive and hard-to- politicians have made resources available for monitor services (for example, primary edu- schools and clinics to decide whether to cation), their country or region has pro-poor reward or punish them at election time, and politics and little heterogeneity of preferences. politicians then can be more credible. So they chose central government provision (c) The International Bank for Reconstruction and Development / The World Bank Citizens and politicians 91 Figure 5.4 Eight sizes �t all 1 Centralized public financing with contracting out Easy to monitor Homogeneous Hard clients to monitor 2 Centralized government provision with regulatory oversight 3 Deconcentrated government or local provision or contracting out. Deconcentrated central government Heterogeneous Easy to may have an advantage over local government—at least in terms of financing—since service is easy clients monitor to monitor and central government is in a better position to achieve interregional transfers of income. Pro-poor politics Hard 4 Local or deconcentrated government provision. Lower-tier and local governments may have advantage to monitor over central government because they are better able to monitor local service quality and can be more easily held accountable. 5 Contracted provision and public financing that provides lots of information to strengthen voice and client power. Clientelist relationships dominate politics. Services used as currency of political Easy to patronage. Politicians have strong incentives for providing narrowly targeted transfers rather than monitor universal public goods and prefer credibility that clientelism yields. Information about politicians, their specific contributions, and service inputs and outputs can strengthen voice and have high payoff Clientelist and be cost-effective. Scorecards, tracking surveys, and client satisfaction surveys potentially most politics useful since nature of service and preferences are uniform and easy to monitor and compare across jurisdictions. Public disclosure and a free press essential. Strengthen voice and client power in Homogeneous Hard general to counter clientelism (applies to 5–8). clients to monitor 6 Encourage altruistic providers (NGOs). Copayments and fees to reduce capture. Public information campaigns. Heterogeneous 7 Decentralized, rule-based allocations, vouchers, private provision (because service is easy to monitor). clients Easy to monitor Disseminate local information on provider performance to help monitoring. 8 Possibly the hardest situation to deal with. Need to boost client power to increase monitoring. Hard Rule-based allocations, copayments, user groups, altruistic NGOs, information for mobilizing clients. to monitor Free press essential to monitor compliance. (option 2 in �gure 5.4). But if the service and a pro-poor service delivery environ- delivery environment is actually based on ment emerges, it should be possible to move clientelism, and preferences vary widely, then to the service arrangements described in conditions have been misread and services options 1 through 4. But to the extent they fail poor people. Decentralized provision do not change, then trying to scale up with with lots of client involvement at all levels options 1 through 4 and make services work may be called for in ways that create choice for poor people may be wishful thinking and mimic the market if services are to work and a waste of resources. (option 8). Under either clientelist or pro-poor envi- In general, services can be made to work ronments, having more and better informa- in clientelist settings by choosing arrange- tion pays off in strengthening voice. Informa- ments that reduce the rents from service tion about services that is speci�c, directly delivery that would otherwise be captured related to voters’ concerns, and framed in a through patronage and clientelism. These way that ensures political interest in address- are the situations depicted in options 5 ing service delivery concerns is likely to be the through 8. The appropriate service arrange- most effective. Information from impact ment for hard-to-monitor services such as assessments can show what works and why. curative care or primary education might Information about politicians can boost their then be option 6 or 8 depending on whether political credibility, strengthen incentives to preferences are homogenous or heteroge- provide universal public services, and avoid neous. If institutional arrangements change politically targeted goods and rent-seeking. (c) The International Bank for Reconstruction and Development / The World Bank spotlight on the Kecamatan Development Program Choice, participation, and transparency in Indonesian villages A new generation of community development projects in Indonesia illustrates many of the key elements of effective services. The projects transfer resources directly to local control, allowing a local decisionmaking body to choose among proposals from community groups. The three principles are: choice, participation, and transparency. Pastoral scene—or chaotic mess who reported participating in the govern- Scaling up with simplicity and trust Indonesia in the New Order era of Soeharto ment-organized village groups reported The �nancial and political crises that (1967–98) has been compared to a French being more likely to have spoken out about began in 1997 opened a window for Impressionist painting: viewed from a dis- village problems and to have done so effec- action. Projects were desperately needed tance, a beautiful pastoral scene, but viewed tively. But this impact, by crowding out the to help rural areas quickly. An improved closely, a chaotic mess. voice of others in the village, appears to design for community projects—based on The government launched top-down have been negative overall. the lessons of block grants under the IDT “blueprint� development programs in fertil- The problems with local governance and infrastructure investments under vil- ity, health, schooling, and poverty reduc- were obvious in projects. The �rst-genera- lage infrastructure projects—was being tion—implemented by a reasonably func- tion poverty alleviation programs—block piloted. It included open menus and more tional and capable bureaucracy. Viewed in grants to poor communities, under the IDT emphasis on community participation the aggregate, the results were spectacular. (Impres Desa Tertinggal)—used existing and decisionmaking. The crisis also cre- Gross domestic product per capita grew at village structures and were judged to have ated an opportunity to act on issues of more than 5 percent a year. Poverty fell from had very little impact, in quantity or qual- transparency, local accountability, and nearly half the population in the 1970s to 11 ity.279 A study of all projects in villages— corruption. percent in 1997. Infant mortality fell, fertil- including those initiated by villages on their The new Kecamatan Development Pro- ity fell, and schooling rose dramatically. own(found that village-initiated projects gram (KDP) provides block grants to eligi- The 1979 law establishing village govern- were much more likely to have sustained ble subdistricts—or kecamatan, an admin- ments was state of the art—on paper. With bene�ts than government-initiated projects istrative unit that includes roughly 10 to the goals of “decentralization� and “bottom- (�gure 1). 20 villages and roughly 30,000 people, up� planning, the law established locally The next generation of more participa- though its size varies enormously. Each chosen village heads accountable to a village tory projects—two rounds of village infra- subdistrict uses the funds to �nance com- council. The budget planning process incor- structure projects and water supply and munity proposals from the villages for porated village-level meetings to elicit bot- sanitation projects—showed that greater small-scale public goods (roads, wells, tom-up inputs into budget priorities. community engagement could have real bridges) or economic activities. Making But the reality of village leadership was payoffs. Water projects designed to incor- the decisions about which proposals to different. Creating multifunctional village porate participation had much lower failure fund is a subdistrict forum, including vil- administrative structures imposed order rates than conventional projects. And the lage delegations. and uniformity at odds with existing social costs of the village infrastructure projects The KDP’s design was based on simplic- structures, ignoring organizations with spe- were 30 to 50 percent lower than costs in ity, participation, self-reliance, transparency, ci�c functions (water) and traditional lead- projects using government construction. and trust. These principles may sound plati- ership (adat).276 Many villages had dynamic tudinous, but they pushed the design enve- leaders, but many others had leaders chosen lope in several directions. essentially by the regional (province or dis- Figure 1 Community-initiated projects: more trict) government, which had veto power likely to be maintained and in full use • Simplicity meant that funds were released directly to communities, elimi- over candidates.277 The village head was Percent projects in full use nating one role for regional (district and accountable to a council, but he also headed 100 provincial) governments. the council and chose many of its members. Most village heads were accountable 80 • Participation was encouraged, and upward—to regional governments—and locally chosen village and subdistrict 60 not to the villagers. The bottom-up plan- facilitators helped groups to prepare ning never really functioned: one analysis of 40 proposals and encouraged the dissemi- 770 village proposals found that, at most, 3 nation of information. 20 percent were included in district budgets.278 • Self-reliance reversed the usual depen- Empirical results from a recent survey in 0 dence of villagers on technical staff from 48 rural villages suggest that the govern- Community Private Government NGO ministries and government, permitting ment-driven organizations did not make sector villagers to hire the engineers and other village governance more responsive. Those Source: World Bank (2001g). technical help. (c) The International Bank for Reconstruction and Development / The World Bank Spotlight on the Kecamatan Development Program 93 • Transparency meant that all �nancial (and perhaps elsewhere) suggests the latter. community-driven project design because information was publicly available, and But whether it will “work� in those places is it can undermine technical quality. Given detailed information about the use of still an open question. Perhaps the principles the choice of a participatory bridge or an the funds was available in each village in can be implemented with the design adapted engineered bridge, most people would cross simple and easy-to-understand formats. to local circumstances. the engineered bridge. The question is how Second, are large external agencies (like to create a well-engineered bridge that • Trust made it possible to move from the World Bank) really capable of support- responds to community needs. Other ser- complicated formal accounting systems for releasing funds to disbursement sys- ing “big-time small development� projects? vices try to balance community control and tems that rely on minimal documenta- Perhaps yes, perhaps no. Some argue that technical quality: participation in health tion—but with built-in checks and external actors have no mandate or exper- care does not mean that medical science oversight. tise for engaging in local governance. And can be replaced. Should KDP-like mecha- preserving the traditional exclusive link of nisms be expanded with improved links to external agencies to formal public provider technical providers? Or should technical The project has so far been an imple- organizations may make them incapable of providers be strengthened and the “partici- mentation success, scaled up from 40 vil- contributing to the creation of needed local patory� role be channeled not into direct lages in 1998 to more than 15,000 in 2002. accountabilities. control and decisionmaking but into elect- It has moved into another cycle and been ing local of�cials? replicated in urban areas. Evaluation efforts, including an innovative attempt to This or that? Three strategic choices Local or regional governments. Regional directly measure the impact on corruption, Community-driven development projects governments often complain that moving are examining whether the KDP has such as the KDP raise three strategic resources directly to communities under- improved project performance. choices relevant to the design of service mines their authority, slowing the capacity This is not to suggest that the KDP is delivery. building needed for formal governance and free from flaws—it is a transitional project democracy. Proponents of community Narrow or sharp targeting. Community in a transitional situation, embedded in development respond that deep democracy funds, it is often alleged, are captured by existing institutions. There have been prob- depends on the kinds of transparency, deci- “elites� and will not be well targeted to the lems of corruption and poor technical sionmaking, open debate, and accountabil- poor. The KDP shows that the poor do ben- quality, and problems of local leaders “guid- ity that community projects build. Decen- e�t, but it is dif�cult to reach the poorest of ing� the participatory decisionmaking. But tralizing decisions about budgets and the poor or to change deeply held social the KDP does give villagers the structured programs to the provincial level—when prejudices simply by project design mechanisms of decisionmaking and trans- people have not developed traditions and (despite, for example, KDP mechanisms to parency. It also gives them recourse to force institutions of civic decisionmaking at the enhance women’s voice). Few large pro- the issue of better governance. Corrupt of�- local level (or have had them suppressed)— grams have shown greater ability to target cials have been sued. Money has been is risky business. In a transitional environ- the poor, and the very narrowly targeted recovered. Decisions have been reversed. ment, periodic elections alone are unlikely programs would not elicit engagement and Two lingering issues: First, is the KDP an to be suf�cient for public accountability. support. idiosyncratic product of its particular time The development of nonelectoral mecha- and place—or a model that can be replicated Technical or participatory projects and ser- nisms of public accountability (trans- elsewhere? That versions of it are being vice. Technical staff of the government parency, legal recourses, direct participa- launched in Afghanistan and the Philippines (and of many donor agencies) are leery of tion) is key. (c) The International Bank for Reconstruction and Development / The World Bank spotlight on Norway and Estonia Developing social services and building a nation One of the richest countries in the world, Norway today is the quintessential welfare state, with universal access to basic health and education. But this welfare state evolved over two centuries, with private systems only gradually giving way to state-run institutions. Making social services available to all was seen as part of building the Norwegian nation. Though geographically close, Estonia regained its independence in 1991. It is seeking to develop its social services and build a nation in a much shorter time, and under budgetary constraints. Norway: gradual change with top- larly monitor the health of infants and Estonia: starting over, with few down pressure schoolchildren. From 1890 onwards, the resources In 1860 the Norwegian national assembly health sector evolved through public-private At re-independence in 1991, Estonia wanted passed two laws—the Health Act and the partnerships, spurred on by pressure from to move away from its inherited systems to School Act—the �rst time the state took grassroots and philanthropic organizations. modern Western European approaches that responsibility for the health and education of As the state took on more responsibility for rested on progressive governmental, eco- its people. The Health Act, which established delivering universal services, a process that nomic, and social reforms—partly for accep- health commissions in every municipality, picked up in the 1930s, it did not have to build tance into the European Union. The new was promoted by the country’s social elite to the institutions from scratch: it could build on state had to quickly establish the mechanisms improve the welfare of Norway’s farming and institutions already built, organized, and of a modern welfare system. But there was lit- peasant communities so that the country �nanced by private actors and civil society. tle time to establish the system’s legitimacy. could compete with the more advanced The �rst priority in 1991 was services nations of Europe. The elite saw educating Reforming schools based on Estonian language and culture, poor rural households in personal and envi- From 1739, children were required to critical for national identity. Then came the ronmental hygiene as a key to this project, attend school from the age of seven until urgent need to improve ef�ciency and and the health commissions were charged they could read and undergo Lutheran con- equity. But economic dif�culties limited the with this task. Interestingly, members of the �rmation. The incentives to learn to read resources for reform. medical profession, which up to that point were strong. No reading meant no con�r- The health care system had to be com- had a somewhat lower status than other pro- mation, and no con�rmation meant no pletely reorganized. Unlike the situation in fessions (lawyers, priests, and the military), marriage license, no land holding, no per- Norway, the administrative, legislative, and saw themselves as the natural leaders of the manent job, and no chance of enlisting in regulatory powers in Estonia were all in one campaign. According to one doctor, appoint- the armed forces. Nevertheless, the rural place: the Ministry of Health. With little ing a lawyer to head the campaign (some- population resisted sending their children transparency and control, corruption flour- thing that was being contemplated) “would to school, mainly because they found the ished. do nothing to further the cause.�280 curriculum irrelevant for farming. To address the problem, the old state- But the health commissions faced sig- As with health, formally trained teachers funded system was replaced by health ni�cant dif�culties in getting their job became the driving force behind using edu- insurance, which facilitated transparency done. In addition to the facts that doctors cation to build the Norwegian nation. and a steady stream of �nance. A major were not trained in public health and their Teachers organized themselves in 1848 and challenge has been to convey the logic and work was poorly paid, the cultural divide advocated inclusion of education profes- long-term advantages of the new system. between the urban elites and rural farmers sionals in policymaking bodies. The School People suddenly had to pay for health care was an obstacle. For instance, although fer- Act of 1860 shifted responsibility for run- that used to be free. Drugs were sold at tilizer was a scarce commodity, the doctors ning schools from the clergy to an elected European prices. And, although the system were trying to get rid of the compost heaps school board (whose head would still be a has equity as a goal, the health status of a near people’s houses because of the “rotten� priest). As a result of populist and agrarian growing number of Estonians is declining, air that people were obliged to breathe. pressure, local school councils were able to especially that of the elderly, ethnic minori- Meanwhile, many of the services were appoint teachers, determine their own ties, and the unemployed. Around 6 percent being delivered by grassroots organizations. “education plan,� and introduce New Norse of the population is not yet covered by the Founded in 1896, the Norwegian Women’s as the language of instruction. But a grow- new national health insurance system. Public Health Association was running 14 ing labor movement was demanding more On many accounts, Estonia has suc- sanatoria for patients with tuberculosis by universal education, so that by 1889 a com- ceeded more than many other newly inde- 1920. The Association also advocated for mon school law was passed and education pendent countries. But in seeking to �nd its greater public intervention in health, getting �nally moved from religious training to own way of making services work, it has not the authorities to open public baths and regu- general learning and nation building. had the luxury of time. (c) The International Bank for Reconstruction and Development / The World Bank Policymakers and providers 6 Educated children. Good health. Clean, reli- be clear and backed with suf�cient resources able, and convenient water. Safe neighbor- for adequate and regular compensation. chapter hoods. Lighted homes. That is what citi- zens, poor and rich, want from services. If Good information on the actions of providers and the outcomes of those actions policymakers take responsibility for deliver- must get to the policymaker. And remunera- ing services, they must also care about these tion must be tied as closely to these outcomes outcomes and be sure that services as possible. Accountability is improved by: providers care about them, too. Chapter 5 discussed the �rst challenge: inducing poli- • Clarifying responsibilities—by separat- cymakers to reflect the interests of poor ing the role of policymaker, accountable people. This chapter takes up the second: to poor citizens, from that of provider inducing providers to achieve the outcomes organizations, accountable to policy- of interest to poor people. How? By choos- makers. ing appropriate providers. By aligning • Choosing the appropriate provider— incentives with those outcomes. And by civil servants, autonomous public agen- ensuring that policymakers do at least as cies, NGOs, or private contractors. Com- well as the clients themselves in creating petition can often help in this choice. those incentives. • Providing good information—an essen- tial step. Just monitoring the performance Compacts, management, and the of contracts requires more and better “long route� of accountability measures. Keeping an eye on the prize of The “compact� introduced in chapter 3 is better outcomes also requires more regu- composed of relationships of accountability lar measurement. It also requires �nding necessary for increasing the power of incen- out what works by rigorously evaluating tives for good performance (�gure 6.1). programs and their effects. Instructions to provider organizations must These steps are neither easy nor straight- Figure 6.1 Compact and management in the service delivery framework forward. Political pressures often make it impossible for policymakers to claim inde- The state pendence from the performance of service Politicians Policymakers providers. Compacts for the kind of services discussed here cannot be complete or have of accounta ro u te bili Com perfectly measured outcomes. Finding ng ty pa Lo c enough staff, regardless of their precise t employment agreements, is a real challenge S h o rt r o u t e Providers for many developing countries because of international migration and, for Sub-Saharan Management Africa, HIV/AIDS. And �nding out what Frontline Organizations works—determining the link between poli- cies and inputs and outcomes—is dif�cult, not just for technical reasons. Governments, Services donors, and provider organizations fre- 95 (c) The International Bank for Reconstruction and Development / The World Bank 96 WORLD DEVELOPMENT REPORT 2004 quently don’t want to know—or don’t want private goods. Private providers may be con- to take the risk of �nding out—what does- tracted by the public sector for services to n’t work.281 poor people or for true public goods (where a It may be dif�cult to measure the out- private sector is impossible, even in princi- comes of health, education, and infrastruc- ple). But they cannot be relied on to provide ture services, but it is possible. Improve- them on their own. That is why the public ments cannot be measured as precisely as sector should assume responsibility for basic tons of steel. But the outcomes of these ser- services, especially for poor people. vices are far more amenable to measurement The public sector has its problems, too. than many core functions of government. Chapter 5 asked whether, in ful�lling this Mortality rates, literacy rates, and the purity responsibility, policymakers have the incen- of water are observable in ways that “advanc- tive to “do the right thing.� The answer is ing the international interests of the nation,� often “no.� But even if policies are properly the goal of a foreign ministry for example, designed, it is dif�cult to get personnel to are not.282 And technical knowledge for staff facilities in poor or remote areas. rigorous evaluation of programs to reach Vacancy rates for doctors in Indonesia range the poor is certainly available. from near zero in Bali to as high as 60 percent A word on “compacts� versus “manage- in West Papua (formerly Irian Jaya), the ment.� The focus of this chapter is the com- province farthest from Java (box 6.1). pact between the policymaker and the The dif�culty in staf�ng such places varies provider organizations, not the details of the by job. It is greatest for the most highly edu- management of frontline providers by a cated people with the best alternative provider organization. Appropriate manage- employment prospects. Educated people in ment needs to be tailored to local circum- countries with few such people are almost stances. Focusing on the details of manage- always urban born and bred. In Niger 43 per- ment detracts from the more crucial cent of the parents of nurses and midwives relationship of the compact and indulges the were civil servants, and 70 percent of them tendency to micromanage. Here the empha- sis is on the principles for designing incen- tives. But management cannot be ignored BOX 6.1 A good doctor is hard entirely. Much of the (very thin) literature on to �nd what works and what doesn’t—on provider responses to changes in incentives—deals Public health centers in desirable locations with management reforms, so these experi- have modest vacancy rates, as low as 1.2 per- cent in Bali and near 5 percent in most of the ences must form the limited empirical base. provinces in the population centers of Java and Sumatra. For such remote areas as West Papua Misaligned incentives the vacancy rate reaches 60 percent, and for and service failures central Kalimantan more than 40 percent. Failures to reach poor people with effective Percentage of health centers without services can usually be attributed to a mis- doctors, by province, Indonesia 1992 alignment between the incentives facing 60 providers and outcomes. A private market Remote provinces left to itself cannot provide appropriate ser- vices to poor people. It will tend to serve 40 Poorer clients who possess the purchasing power for provinces a fairly narrow set of services. Modern, pri- 20 vate medical care by skilled professionals and private education are largely used by the bet- ter off. Water is an exception, with substantial 0 middle-class capture of public water, leaving Bali West Papua poor people to buy from expensive private sellers. The range of services the private sec- Source: World Bank (1994b). tor will provide on its own is limited to, well, (c) The International Bank for Reconstruction and Development / The World Bank Policymakers and providers 97 had been raised in the city.283 It is only nat- for poor people. The lack of conscientious- ural for them to want the same for their chil- ness, the mistreatment of students and dren. And it is naïve to simply say “pay them patients, and the loss of skills with time more.� Doctors in Indonesia would require (chapter 1)—all can be attributed to a com- multiples of current pay levels to live in West bination of the failure of incentives and a Papua.284 And giving providers too much dis- service ethos. Salaried workers with no cretion over where they serve may hurt the opportunity to advance and no fear of pun- poor, as in rural schools in Zambia (box 6.2). ishment have little incentive to perform well. Even when people accept jobs in poor Chapter 4 argued that discourtesy depended areas, their absenteeism is often astonishing on incentives, not training. If income does (see tables 1.2 and 1.3 in chapter 1). The rea- not come from clients, the policymaker must sons vary, but alternative earning opportu- hold providers accountable, particularly in nities are a major one for professions with monitoring and rewarding good behavior. easily marketable skills.285 This applies to Corruption—unauthorized private gain doctors and other medical personnel and to from public resources—is common in many teachers offering independent tutoring. services and also attributable to competing Again, the day-to-day imperatives for people incentives. In Eastern Europe under-the-table to make a living run counter to increasing payments to public servants and general cor- services to poor people. This is particularly ruption undermine the legitimacy of all gov- true where civil service pay is much less ernment services. They are particularly costly than private sector pay for the same skills. to poor people (box 6.3). Pharmaceutical Even when people are on the job, their mismanagement is everywhere: thefts from performance can compromise the outcomes public stores supply much of the private mar- ket in Côte d’Ivoire, India, Jordan, Thailand, and Zambia. Corruption responds to mone- tary incentives, but it also requires a lack of BOX 6.2 Provider discretion information on hidden activities and an can hurt the poor inability to impose sanctions. As Captain Shotover in George Bernard Shaw’s Heart- Funding of rural primary schools in Zambia break House put it,“Give me deeper darkness. from different sources Money is not made in the light.� Open infor- Funding per student (kwacha, thousands) mation can reduce both the incidence of cor- 30 Discretionary ruption and its corrosiveness.286 25 Community pressure can also subvert 20 the incentives to ful�ll the primary respon- 15 Staff sibilities of public providers. In many places 10 remuneration the public servant is a permanent member According of the community, facing substantial social 5 to strict pressures to bend rules to the bene�t of 0 rules local preferences. Sometimes this is good— Poorest 2 3 4 Richest it shows the flexibility to respond to local Wealth by fifths needs. But for some services, particularly those with punitive characteristics, it can Rural schools in Zambia obtain resources in cash and in kind (personnel). Cash transfers compromise the core duties of the provider. allocated by strict rules of per capita funding For example, forestry agents who are part of are distinctly progressive. Rural areas give sig- a community may be reluctant to report ni�cantly higher discretionary cash allocations illegal logging by their neighbors.287 A form to rich schools. Per-pupil teacher compensa- tion increases with the wealth of children of community pressure particularly harm- attending the school, reflecting higher staf�ng ful to the poor is the capture of services by ratios and the gravitation of senior staff to local elites. In Northern Ghana young, inex- richer areas. perienced, and poorly paid facilitators for Source: Das and others (2003). participatory projects found such pressure a major impediment.288 (c) The International Bank for Reconstruction and Development / The World Bank 98 WORLD DEVELOPMENT REPORT 2004 BOX 6.3 Bribery hurts the poor To gain access to health, education, and the justice system in Kazakhstan, poor people pay bribes simply to receive services they are entitled to (and to avoid “problems� getting them) while richer people pay to speed up service. In Romania, the poor pay substantially higher fractions of income in bribes. Kazakhstan: reasons for paying bribes Romania: Percent of income paid in bribes to health, education, and justice systems (of those paying bribes) Percent responses per income group Percent 140 12 Other 120 10 To avoid 100 problems 8 To receive 80 benefits 6 60 Speed 4 40 20 2 0 0 Poorest Middle Richest Poorest Middle Richest third third third third third third Note: Numbers add to more than 100 percent due to multiple responses. Source: Anderson, Kaufmann, and Recanatini (2003). Many, and usually most, providers in the for delivering services. In many cases, the pol- public sector are dedicated people whose icymaker is the legislature or a central min- interests are largely compatible with the pub- istry, the provider organization a line min- lic good. But their own needs of looking after istry. So many of the activities of the head of a family, ensuring their well being, having the “provider organization� will look like pol- friendly relations with neighbors—all pre- icymaking. But these are “internal policies� of vent them from providing suf�cient services the organization to achieve the overall goals to bene�t poor people. If the scale of opera- focused on here. (The literature on public tions needs to be increased to reach the poor, management explicitly cautions against sepa- even more incentives need to be changed at rating290 policymaking from implementa- the margin, whether monetary or not.289 tion, but that literature is concerned with management within the “provider organiza- tion� and not the separation proposed here.) Increasing accountability: Clear separation lends itself to much sim- separating the policymaker pler and less ambiguous accountability for from the provider the provider organization. When the policy- The many incentives that providers face blur maker is the provider organization, day-to- the focus on outcomes. Making a clear sepa- day pressures of management compromise ration between the role of the policymaker attention to outcomes on the ground. Take and the provider organization is essential for the desire to �nd and �x problems (see the aligning the incentives for the provider with spotlight on Johannesburg). When the poli- the �nal outcomes that policymakers want cymaker takes a separate role from the for citizens. Who is the policymaker, and who provider, it is easier to say “I don’t care what is the provider organization? The policy- your problem is, just tell me the vaccination maker is the person directly accountable to rates. Or the test scores. Or crime rates.� the citizenry, preferably the poorer citizenry. When roles are mixed, bureaucracies become And the provider organization is responsible insular and tend to hide mistakes. (c) The International Bank for Reconstruction and Development / The World Bank Policymakers and providers 99 Is this separation really necessary? Is it Multiple principals achieved in rich countries? Education, for The instructions of the policymaker to example, is frequently administered through providers are not the only ones that count. central ministries, which employ all teachers Public servants have to serve many masters. directly with little monitoring from central Education providers are under pressure staff agencies (such as �nance). Perhaps sepa- from parents of poor children (with the ration is not necessary for successful services. policymaker representing them), parents of But rich countries bene�t from a long children other than the poor, teachers’ evolution of the relationships between the unions, potential employers, various groups state and frontline providers. Almost all ser- in society that want (or don’t want) particu- vices provided directly to individuals in the lar items on the curriculum, and others. now-rich countries were originally provided Power and water providers are under pres- privately. They were eventually absorbed or sure from different segments of the market consolidated by a state institution that had to cross-subsidize them, from producers to been separate from the existing provider buy speci�c types of equipment, from peo- organizations. The state began as an indepen- ple who want more extensive connections, dent outside monitor and regulator of private and from others who want more reliable, activities. It largely retained that indepen- continuous operation. The day-to-day pres- dence as a monitor after the same activities sure of local demand for health care can became public (box 6.4). compromise efforts in disease prevention For the developing world the desire for and other public health activities that are rapid expansion of public �nancing and pro- not demand-driven.292 Whom is a provider vision short-circuits this historical develop- to listen to? ment. Both the monitoring and the provision are taking place simultaneously. This is not necessarily a bad thing—the poor might oth- erwise have to wait much longer for services to reach them. But it does show that the cur- rent institutional features of rich countries BOX 6.4 Learning to regulate may not transfer directly to poor countries Although the state pays for health care in incorporating private schools into the pub- without the establishment of a complemen- most Organisation for Economic Co-opera- lic network after a system of managing indi- tary regulatory structure, a structure that may tion and Development (OECD) countries, vidual schools had already developed. private practitioners still provide it (excep- Even the core networked services of need to be established beforehand. Without tions on the �nance side include the United water, electricity, gas, and railways—services this structure progress may be slow—possi- States; exceptions on the provision side, the now thought of as natural monopolies— bly slower than if a not-for-pro�t or private United Kingdom).The state carries out the began as purely private activities. In the sector were allowed to develop and later insurance function in these countries but United Kingdom early water systems were not the services. It is the insurance market sometimes developed with duplicate pipes brought under the supervisory wing of the that is hampered by severe market failure. laid by competing �rms. Only after cover- government. In Germany, the current system is a consoli- age was substantial were these Separating the policymaker from the dation, begun under Bismarck in 1883, of a rationalized—and then for reasons of pub- system of guild-based insurance schemes. lic health rather than duplication. German provider organization also helps to increase For most OECD countries, the current railways also began with duplicate lines. In the accountability of providers. But if the poli- system of public ownership or control other cases, companies worked out agree- cymaker knows what services to deliver, why dates from a time after World War II when ments that divided markets without the can’t providers just be given instructions—in a they had incomes at least as high as the duplication experienced in the railway mar- upper-middle-income countries today. kets in the United States and England—for contract—to do them? That is, why can’t out- Before these programs were brought under example, bus systems in many large urban puts just be speci�ed and paid for accordingly? public management, the state already had areas and German natural gas transport. regulatory powers over the medical profes- These were then brought under a regula- sion. The timing was fortuitous since it was tory regime but only after reaching quite Limits to accountability only in the 20th century that technical high service penetration. All public services face three problems that changes in medicine made public In each of these cases the independent oversight essential. regulatory capacity of the policymaker pre- make this solution impossible: providers Universal public education is also fairly ceded the incorporation of private face multiple principals, undertake multiple recent. It came out of a conflict between providers into a public system. tasks, and produce outcomes that are hard church and state. In the period of expansion to observe and hard to attribute to their of public facilities, the main mechanism was Source: Klein and Roger (1994). actions.291 (c) The International Bank for Reconstruction and Development / The World Bank 100 WORLD DEVELOPMENT REPORT 2004 Multiple tasks individual frontline providers depend on out- Personnel in health clinics are supposed to comes. All contracts will necessarily be provide curative care to people who come to incomplete, requiring at least some payment them. They are also to provide immuniza- of wages independent of outputs. When the tions, health education, and other preventive actions of the provider are speci�ed in great measures to everyone, whether they come in detail, the results are often less than optimal on their own or not. And they are to keep because of inflexible response to local varia- statistics, attend training sessions and meet- tion. The impossibility of specifying such ings, and do inspections of water and food. rules ahead of time is illustrated by “work-to- Police of�cers have to deal with everyone rule� strikes, in which strikers bring an activity from lost children to dangerous criminals. to a “grinding halt� by following rules entirely They are, at various times, investigators of to the letter (box 6.5). The balance between crimes, social workers intervening in neigh- control and flexibility is not easily struck. borhood and family disputes, and dissemi- Further, since the provider does many nators of information. This diffusion natu- things, some or all of them hard to observe by rally blunts the precision of incentives. the policymaker, there is the ever-present risk that payments for measured outcomes will Measuring and attributing displace hard-to-measure tasks (box 6.6). outcomes This risk has been discussed in the education literature as “teaching to the test.� When The most dif�cult problems, particularly for teacher compensation (pay or promotion the social sectors, are the dual problems of prospect) is measured by students’ perfor- measuring outcomes and attributing these mance on a standard test, there will be a ten- outcomes to the actions of providers. Test dency to downplay those aspects of pedagogy scores may adequately reflect certain educa- not covered by the test and to concentrate on tional goals, but abstract thinking and social those that are. In Kenya teachers manipulated adaptability are not so easily measured. The test scores by offering tutoring sessions aimed alleviation of pain is a subjective judgment speci�cally at these tests. There was no of the patient. Many outcomes, even when improvement in other indicators of quality, observable to the patient and the doctor, are such as homework assignments, teacher not “contractible� in the sense that, if a dis- absences, or teaching methods.293 pute arises, compliance can be proved to a Several industrial countries, in reforming judge or other mediator. And attributing the civil service or other providers of public impacts to provider actions is dif�cult in services, have tried to use performance con- almost all social services. tracts. The evidence of success is mixed:294 These problems make it impossible to have many problems are tied to the dependence of performance contracts that make payments to the policymaker on information the agency provides—a problem closely related to the regulation of private �rms. Some informa- BOX 6.5 Be careful what you wish for—part 1 tion used for performance contracts can be “Anyone who has worked in a formal orga- press a point on the municipal authorities easily falsi�ed or, less pejoratively, presented nization—even a small one strictly about regulations or fees, they sometimes in too favorable a light. For example, when governed by detailed rules—knows that launch a work-to-rule strike. It consists education reforms were instituted in the handbooks and written guidelines fail merely in following meticulously all the reg- utterly in explaining how the institution ulations in the Code routier and thereby United Kingdom, truancies were rede�ned as goes about its work. Accounting for its bringing traf�c throughout central Paris to a excused absences.295 smooth operation are nearly endless and grinding halt.The drivers thus take tactical shifting sets of implicit understandings, advantage of the fact that the circulation of tacit coordinations, and practical mutuali- traf�c is possible only because drivers have Overcoming the limits ties that could never be successfully mastered a set of practices that have Separating policymakers from provider orga- captured in a written code.This ubiquitous evolved outside, and often in contravention, nizations can help sharpen incentives to help social fact is useful to employees and labor of the formal rules.� unions.The premise behind what are poor people. Assigning policymakers the role tellingly called work-to-rule strikes is a case of devising a compact for the provider orga- in point. When Parisian taxi drivers want to Source: Scott (1998). nization and assigning provider organiza- tions the responsibility of management can (c) The International Bank for Reconstruction and Development / The World Bank Policymakers and providers 101 enable the use of higher-powered incentives to align the interests of the frontline provider BOX 6.6 Be careful what you wish for—part 2 with those of the policymaker representing The Sears Corporation lost a $48,000,000 most repeat business.That the intention of the poor, for the following three reasons. class action suit in which its automobile the instruction was to encourage courteous First, policymakers, balancing political pres- repair department was accused of deliber- behavior did not impress the court. sures, can help insulate providers from the ately sabotaging customers’ vehicles.The corporation was held responsible, having problem of satisfying masters with conflict- Source: Sears Automotive Center Consumer instructed its employees that bonuses Litigation, Action No. C-92-2227, U.S. District ing aims and offer unambiguous instruc- would be paid to those branches with the Court, San Francisco. tions. Second, provider organizations can face performance-based payments when individuals cannot. Third, managers of the provider organizations, if they have flexibility Cuba). What can be considered measurable over operational decisions, can supervise staff varies by the size of the organization—larger and choose the appropriate form of remu- ones being easier to hold to account. neration that best reflects local conditions. The problem of multiple tasks is partly a problem of economies of scale as well. Some Insulating providers from politics tasks can be divided into groups of comple- That providers have to satisfy many masters mentary activities—all immunizations as a reflects the inability of government to insu- group, say, or all health education activities late them from political pressures. While pol- based on home visits (chapter 8). Then a icymakers for education need to address fairly homogeneous organization can be concerns of potential employers, teachers’ charged with the responsibility to carry out a unions, or interest groups who want to influ- simpler set of tasks, with clear standards of ence curricula, there is no reason why this accountability. should affect day-to-day activities in a Reform in Johannesburg, South Africa school, or indeed any organization of front- (see spotlight), was in large part a reevalua- line providers. If the policymaking function tion of the appropriate set of services to be can be separated from the provider organiza- grouped together to deliver speci�ed outputs. tion, the policymaker can handle the politics Departments were reorganized so that their of the overall objectives of education while outputs were clearly identi�able and veri�- the provider can be given more precise able, with the department’s CEO able to instructions and be held accountable to the retain any savings over contract expenditures. policymaker. Poor people might legitimately At one extreme, commercial enterprises— delegate to policymakers curriculum devel- such as the athletics stadium, the airport, and opment as well as the responsibility to bal- metro gas—were simply sold to the private ance their interests with those of unions. sector and directly faced the forces of the market, where payment is very much depen- Organizations and individuals dent on outcomes.296 Individual providers will not accept perfor- mance contracts that leave them exposed to Management flexibility excessive risk. But the variability of aggregate Each of these potential effects depends on performance over all providers in an organi- managers in provider organizations having the zation—say, those dealing with infant mor- flexibility and authority to design the incen- tality for a district—is very much smaller, tives for the frontline providers in their organi- which provides a way of sharing the risk. zations. This allows them to adapt to local (or While a single doctor may not be able to sectoral) variation to see whether performance absorb the risk to income of the bad luck of pay or salaries with supervision works better. any particular patient, a district health board Flexibility for the manager is essential, a major would. Teams—schools, school districts, part of “institutional capacity.� Managers must health boards, city police departments—can have control over the pay scheme or the sanc- be the recipients of performance-based tions for poor performance. incentives where teachers, nurses, and police- Salaried systems work as long as there is men cannot (see spotlight on Costa Rica and the ability either to �re or to grant raises on (c) The International Bank for Reconstruction and Development / The World Bank 102 WORLD DEVELOPMENT REPORT 2004 the basis of merit. The worst case is when ments—for how many people sign up with salaried workers face neither sanctions for the doctor. But it is supplemented by speci�c poor performance nor increased pay or pres- additional payments for the provision of tige for good performance. Civil servants in immunizations to counter any incentive to Singapore enjoy high salaries and a lot of skimp on this priority service. prestige, but also work under a credible threat of being �red. A problem with some of the New providers for expanding supply recent reforms in developed countries insti- Where will the providers of services come tuting contractual relations with providers is from? One possibility is that competition for that they undermine the public service ethos. compacts will attract more provider organi- (“If I am to be treated as a mercenary, I might zations. The bene�ts from competition are as well act like one.�) Increased accountability reduced costs, greater effort, and better through monetary incentives was partly off- information—even when public provision is set by reduced accountability through inter- the dominant form, as long as public and nal motivation.297 Developing countries that other provider organizations are treated have instilled this sense of duty should be even-handedly. Three types of competition wary of compromising it. But they should be are relevant for services: competition in the brutally honest with themselves before market, competition for the market, and declaring this a major consideration. benchmarking. Sometimes performance pay is appropri- ate and necessary but should be a matter for Competition. Competition in the market local experimentation. Several health inter- simply means allowing private providers. For ventions have bene�ted greatly by introduc- health and education, such providers are ing performance-based incentives for workers everywhere, and in many places larger play- (box 6.7). In other contexts, those incentives ers than the government (chapter 4). Recent are precisely what is needed to obtain particu- technological advances have made it possible lar desired results. In the British National to open services formerly believed to be nat- Health Service most general practitioner pay ural monopolies to competition. Indepen- is determined on the basis of capitation pay- dent power producers, for example, can be used to sell electricity to a larger grid. The cost of allowing free entry into natural BOX 6.7 Incentive pay works for speci�c health monopolies is the risk of inef�cient duplica- interventions tion of investments. Ef�cient regulation is necessary but complicated. If political and The Bangladesh Rural Advancement Com- In Haiti, NGOs were given performance- mittee (BRAC), one of the largest NGOs in based contracts, directly from the U.S. administrative limitations on the indepen- Bangladesh, paid workers to teach mothers Agency for International Development, to dence and effectiveness of regulators are how to use oral rehydration therapy for chil- provide preventive health care services severe, allowing the duplication may be the dren with diarrhea. Independent of the such as immunizations, health education, lesser of two evils.298 providers, bonuses were paid on the basis prenatal care, and family planning. Again, an of surveys of random samples of 5–10 per- independent monitor, l’Institut Haitien de The impact of competition can go both cent of the mothers.The greater the num- l’Enfance, a local survey research �rm, was ways: the presence of the public sector can ber of women who could explain how to used to verify performance. Immunization impose indirect discipline on the private sec- make and use the rehydration solution, the rates increased dramatically along with sev- higher the payment. More than half of total eral other outputs. Interestingly, some of the tor, both on prices and on quality. In Malaysia compensation was paid as a bonus. NGOs experimented with performance pay a credible public health system has kept price The mothers’ knowledge increased dra- themselves but found lower morale and rises modest in the private sector.299 The ben- matically—to 65 percent of those taught performance when workers (low paid them- e�ts of public provision extend beyond the two years after the training. Most important, selves) faced such risky incomes.The NGOs, the teaching techniques that the workers while satis�ed with the high-powered numbers of patients treated publicly. Simi- used changed from standard lectures to incentives by which they were paid, found larly, the presence of quali�ed medical per- more hands-on demonstrations. Rather better ways to pay frontline providers in sonnel can force quality improvements in than have the right teaching technique accordance with local circumstances. private markets.300 speci�ed for them from on high, workers developed the best way to achieve the If natural monopolies exist, there can be measured outcome—�nding out for them- Sources: Chowdhury (2001) and Eichler, Auxilia, competition for the market. Potential com- selves what worked in their context. and Pollock (2001). petitors bid for concessions—compacts—to provide the service. Much government pro- (c) The International Bank for Reconstruction and Development / The World Bank Policymakers and providers 103 curement in richer countries uses this model. is particularly important because there are It requires the ability to let, monitor, and fewer “perks� for working in poor areas—pri- enforce the explicit contracts for the winner. vate earnings after working hours (for med- Recent innovations in the state of Madhya ical personnel and teachers) are lower, living Pradesh in India allow NGOs to compete for conditions harsher. concessions to primary schools. Payments are In the long run, a public sector with a strong conditional on improved test scores based on ethos of public service will be needed. In many independent measurement. One advantage places it already exists. It does no good to pre- the developing world has over earlier experi- tend, however, that expanding the civil service ence in Europe is that it has �rms with good under current recruitment and incentive reputations and experience in the supply of regimes will attract those best suited to serving water, power, and transport—and interna- poor people. In Nepal an anthropological tional courts for dispute resolution.301 But study showed that health staff’s view of their the recent experience of Enron in India’s jobs often differed from the of�cial view.302 Maharashtra State provides a reality check on Many staff saw the health program solely as a over-enthusiasm for these bene�ts. source of employment. A broader set of poten- Benchmark or yardstick competition can tial providers is needed to accept the compacts. be used when different providers are given NGOs—so much a part of the African parts of a larger system to run. Even when the scene and active in several other countries, public sector is the main provider of services, such as Bangladesh—are possible candidates. information from varying experiences can be They are a varied group. Many are not directly valuable. Information on costs of production involved in service provision, and many com- may be much cheaper to obtain by simple bine service with advocacy. Those that provide observation of one’s own activities than from services often have a great deal of autonomy, detailed technology assessments. Informa- choosing where and how to deliver services. To tion on consumer preferences may be that extent, they might be treated the same as cheaper to obtain by counting customers the rest of the private sector in planning public than by conducting market research. services. The government should not be in the For road construction in Johannesburg, an business of displacing them. explicit contract was made between the city NGOs that have a tradition of altruistic manager and an autonomous public agency, service can frequently be lower-cost produc- the Johannesburg Road Agency, to build a ers. In a recent study, religious NGOs provid- given number of kilometers of road for a ing health care in Uganda were found to offer negotiated price. The basis of the negotiation higher-quality service than their public sector was the set of historical costs in the public counterparts. They also paid lower wages than agencies. The manager of the autonomous the private sector and very much lower than agency then used both the public works the public sector. Unlike the private sector, department and private sector �rms as con- they were more likely to provide public health tractors. Competition among the contractors services (as opposed to simply medical care) determined subsequent allocations of funds. and to charge less. And they used an extra cash Even though it was not possible to �re person- grant to lower fees and provide more services, nel from the public agency, competition for such as laboratory tests, whereas the public funds ensured that the public agency would sector used the grant to increase pay.303 match the ef�ciency of the private �rms NGOs are often, though not always, better (which it did for many contracts). There able to reach poor people. A substantially could be a gradual shift to private provision, higher fraction of the clientele of NGOs pro- but only on the basis of proven performance. viding health care in Zambia comes from poorer segments of society than does the clien- Limits to competition and the search for tele of government facilities or private suppliers. For contracts that cannot be providers.304 But even they have a hard time complete, aspects of delivery outside the con- reaching the very poorest. NGOs may also be tract will remain a matter of trust. For the in a better position, with their greater flexibility provision of services to poor people, this trust and their internal motivation, to bring services (c) The International Bank for Reconstruction and Development / The World Bank 104 WORLD DEVELOPMENT REPORT 2004 to otherwise excluded groups (box 6.8). And altruism.306 Indeed, many appear to be run by smaller organizations can reach niche popu- former civil servants who have lost their jobs as lations that a broad-based bureaucracy may a result of the downsizing of public sectors but �nd hard to serve. who know how to approach donors and gov- In combating AIDS, community outreach ernment contracting agencies. A rapid expan- often needs to deal with prostitutes, drug users, sion of contracts for NGOs will tend to attract and very sick, stigmatized people. The same the same people, and their motives may be dif�culties in assigning public personnel to exactly the same as those of a for-pro�t �rm— remote areas have been found in reaching these requiring the same monitoring and care in con- subgroups. In Brazil, however, NGOs compet- tract enforcement. NGOs with a track record of ing for government funds were able to reach good performance and dedication to poor peo- high-risk segments of society that usually avoid ple are potentially very important elements of a public programs (such as prostitutes), to dis- strategy to extend services to the neediest peo- tribute 2.6 million contraceptives, and to take ple. But establishing a track record, by its very 11,000 hotline calls. The relative independence nature, does not happen as fast as donors would of NGOs from the core of the public service like. The development of trust takes time. may make it easier for them to fund their activ- ities from public resources by granting policy- New challenges to supply. Although there makers an extra layer of deniability. may be ways to extend the supply of providers The altruistic motives of people working by promoting competition and ef�cient con- in NGOs can overcome the incompleteness of tracting with NGOs and the private sector, two contracts. NGO providers are generally less recent trends in developing countries are mak- likely than for-pro�t providers to exploit the ing skilled professionals scarcer, or more dif�culties of monitoring contract terms for expensive. First, professionals—doctors, teach- their own bene�t. Their altruism may partly ers, and engineers—are increasingly part of outweigh a reluctance to locate in dif�cult, integrated global markets and recruitment remote, rural areas that are hard to staff with needs to compete at world wage rates. And it is civil servants. This possibility has led one ana- not only to the rich countries that staff are emi- lyst, thinking of Africa, to conclude that ser- grating. Botswana, for example, has been vices to poor people may, for the time being, recruiting teachers from other, poorer English- have to be left to such groups, particularly the speaking countries. The global market for ser- church.305 vices is changing rapidly due to international Once again, patience is called for. Donor agreements and could lead to new sources of enthusiasm has led to a massive proliferation of supply. Whether this turns out to help or hin- NGOs, many of them not at all motivated by der services in developing countries remains to be seen (see box 6.9). BOX 6.8 NGOs can be more flexible than government Second, HIV/AIDS, particularly in Sub- Saharan Africa, has dealt a major blow to the One advantage that NGOs may have over start in the kitchen?� While the two women ranks of service providers. More teachers died the public sector is the freedom from �xed were cleaning, they had a terri�c conversa- civil service rules or standard operating pro- tion about what was going on in that fam- of AIDS in Malawi in 2000 than entered the cedures. In some ways this reduces account- ily. When I told the story at a meeting, I was profession (see box 1.2). Botswana’s search for ability, but it can avoid unnecessary interrupted by the head of a university clini- teachers, originally to meet a burgeoning constraints. cal psych department who said,“What that demand for education, was given greater therapist did was unprofessional.� A social worker in a family protection pro- Well, all I can say is if we want effective urgency by the country’s AIDS problem. And gram calls on a family threatened with hav- interventions that have transformative just as demands for health service workers are ing a child removed for neglect. She’s effects on people, then we had better rede- increasing, their supply is being cut. greeted by the mother, who says “If there is �ne what is professional, or allowable in the one thing I don’t need in my life right now, expenditure of public funds. When a factor of production becomes it’s one more social worker telling me what scarcer, its use must be conserved—in one of to do.You know what I really need? To get (from Common Purpose by Schorr, 1997) two ways. First, techniques that are less skill- my house cleaned up.� In many countries there is no way for The social worker, who happened to be intensive can be chosen. Distance learning, publicly employed social workers to violate a highly trained clinical psychologist, the opinions of the university professor, but while not ideal for pedagogical purposes, may responded by saying,“Would you like to more independent NGOs could do so. need to be explored to save scarce teaching time. Similarly, it may be appropriate to use (c) The International Bank for Reconstruction and Development / The World Bank Policymakers and providers 105 water systems that require less technical used. Recent trends may merely have made inputs for maintenance. Second, some kinds this misallocation more costly. of services that happen to be highly skill- intensive may be reduced. Curative medical Monitoring and performance services that require trained professionals may All contracts—both compacts and man- be cut back relative to public works or public agement relationships within provider health education, more intensive in capital organizations—need to be monitored with and unskilled labor. It is possible that these independence and objectivity. With the separa- interventions (low-maintenance water sys- tion of the policymaker and the provider orga- tems, use of village health workers or tradi- nization, the policymaker will want to know tional healers) may always have been under- whether compact provisions are satis�ed. BOX 6.9 Is the GATS a help or a hindrance? For a new agreement that has so far not had mitments.These promises to eliminate or limit alizing commitments and new rules in areas much impact on actual policy, the WTO’s Gen- barriers to foreign supply were mainly the out- such as domestic regulation. Informed debate eral Agreement on Trade in Services (GATS) is come of negotiations—but some were volun- would undoubtedly help ensure that future viewed with a surprising degree of both hope teered, particularly in telecommunications. Most GATS rules and commitments reflect broader and trepidation. In the current Doha agenda existing commitments entailed little liberaliza- development concerns and not just the dictates negotiations, some look to the GATS to deliver tion beyond existing market conditions. Many of domestic political economy or external nego- much-needed reform of services from which the countries committed on tourism, �nancial, busi- tiating pressure. poor will also bene�t, while others see it as a ness, and telecommunication services, but rela- At this stage, however, the main issue is not threat to regulatory sovereignty and pro-poor tively few in health, education, and environmen- so much what the GATS forces countries to do or policies. tal services. Of the 145 WTO member countries, what it prevents them from doing, but that it In principle, multilateral negotiations can only 43 (12 developing) have made does not—indeed cannot—ensure the comple- foster reform in services, as in goods, by elimi- commitments in primary education, 52 (24 mentary action that is needed to deliver pro-poor nating or reducing protective barriers through developing) in hospital services, and none on liberalization.This raises a legitimate concern: in a mutual agreement and by lending credibility to water distribution (which was not an explicit complex area like services, trade negotiations alone the results achieved through legally binding part of the original negotiating list of services could lead to partial or inappropriately sequenced commitments.The expectation is that more sectors). reform. One possibility—already visible in some open markets and greater predictability of pol- The most serious charge against the GATS cases—is that less emphasis will be placed on icy will lead to the more ef�cient provision of is not its meager harvest of liberalization— introducing competition than on allowing a services.That is the rationale for the GATS. We after all the process has only recently begun— transfer of ownership of monopolies from address three questions: How much market- but that it deprives governments of the free- national to foreign hands or protecting the posi- opening has happened so far under the GATS? dom to pursue pro-poor policies. It is argued tion of foreign incumbents. Another is that mar- Does the agreement prevent recourse to the that the rules of the agreement threaten pub- ket opening will be induced in countries that complementary policies needed to ensure that lic education, health, and environmental ser- have not developed regulatory frameworks and the poor have access to essential services in lib- vices; outlaw universal service obligations and mechanisms to achieve basic social policy objec- eralized markets? Could the GATS process lead subsidized supply; and undermine effective tives.These flaws could conceivably make the to liberalization before other necessary reforms, domestic regulation. These charges do not poor worse off.The problem is accentuated by and how can this be prevented? seem well founded for three reasons. First, ser- the dif�culty in reversing inappropriate policy The GATS is certainly wide in scope. It vices supplied in the exercise of governmental choices that have been translated into legally applies to virtually all government measures authority are excluded from the scope of the binding external commitments. affecting trade in almost all services, including GATS, although the de�nition—services that The danger of adverse outcomes would be educational, health, and environmental services. are not supplied on a commercial basis or substantially reduced if two types of activities Moreover, in recognition of the fact that many competitively—offers scope for clari�cation. receive greater international support.The �rst is services require proximity between consumers Second, even in sectors that have been increased policy research and advice within and suppliers, trade in services is de�ned to opened to full competition, the agreement developing countries and outside to identify the include not only cross-border supply but also does not prevent the pursuit of domestic pol- elements of successful reform—and to sift the foreign investment and the temporary migra- icy objectives, including through subsidies or areas where there is little reason to defer market tion of service consumers and providers.The the imposition of universal service obligations opening from those where there is signi�cant broad reach of the GATS contrasts with the flexi- as long as these do not discriminate against uncertainty and a consequent need for tempered bility of its rules.The generally applicable rules foreign suppliers. Finally, the agreement recog- negotiating demands. An even greater need is for merely require of each country that its trade- nizes the right of members, particularly devel- enhanced technical and �nancial assistance to affecting measures be transparent and not dis- oping countries, to regulate to meet national improve the regulatory environment and pro- criminate among its trading partners.Thus, if a policy objectives, and its current rules on poor policies in developing countries.The devel- country were to prohibit all foreign supply and domestic regulations are hardly intrusive. opment community is already providing such make this fact public, then it would have met its However, the concerns noted above are not support, but a stronger link could be established general obligations. so much about what the GATS is but what it may between any market opening negotiated inter- The extent of market openness guaranteed become after the current (and any future) round nationally and assistance for the complementary by a country depends on its sector-speci�c com- of negotiations—which will aim for more liber- reform needed to ensure successful liberalization. (c) The International Bank for Reconstruction and Development / The World Bank 106 WORLD DEVELOPMENT REPORT 2004 Competition among providers helps, since ery. They are also clear public goods and core the policymaker will not feel locked into a responsibilities of government. Accurate infor- particular provider, obliged to ignore bad mation can motivate the public, particularly news. If the separation between the two is not the poor, to demand better services—from achieved, an independent regulator or auditor providers and from policymakers—and arm should be assigned the monitoring activities. them with facts. Knowledge of the real impact Clear and observable provisions make of programs helps the policymaker set priori- monitoring easier. When the provisions are not ties and design better compacts. Knowledge of so easily observed, the policymaker may want the impact of different techniques of service to enlist the help of other kinds of monitors. delivery helps the provider organization better The health program in Ceará, Brazil (see spot- ful�ll its compact. If the means to better ser- light), used applicants to the program who had vice is the alignment of incentives with out- not been selected as informal monitors. comes, knowing what those outcomes are and When monitoring is dif�cult because of the how services contribute to them is central. technical nature of the service, self-monitoring Good evaluation is the research necessary to by professionals may be necessary. In assign causality between program inputs and Bangladesh attendance by staff is much higher real outcomes. It should be directed at the full in larger facilities due to informal self-moni- impact of programs—not just the direct out- toring, among other factors.307 Professional puts of speci�c projects. But few evaluations associations can also serve as self-monitors, have been done well, even though most major establishing professional, ethical, and technical donors (including the World Bank) have always standards for medical care providers, teachers, made provisions for them. Evaluation, though and engineers. But the risk in self-regulation is primarily a responsibility of governments, is an that professional groups become effective lob- area in which donors can help. It costs a small byists for their members. fraction of the programs examined and a small A third source of monitors is the public. fraction of the value of the information pro- Even if clients are not the active monitors duced, but it does require some expensive tech- described in chapter 4—that is, they are not nical inputs. And since other countries will use purchasers of services or direct participants in the results, the international community should service delivery—soliciting information (as defray some of the costs. private business often does) can be useful in There are impediments to collecting such public services. Publicizing the results of information. Provider organizations often do scorecards led to a substantial improvement not want to acknowledge their lack of impact of many services run by the Bangalore Munic- (even if it does not affect their pay directly), ipal Corporation. This practice was replicated but knowing when things are not working is in most states in India. essential for improvements. Further, it is nec- When day-to-day monitoring to assess essary to know not just what works but also performance is not possible, independent why—to replicate the program and increase monitoring of the performance of services on the scale of coverage. an occasional basis can still be valuable—by bringing public information to bear on Provider incentives provider behavior. The Public Expenditure in eight sizes Tracking Survey in Uganda (see spotlight on Returning to the decision tree of �gure 6.2 Uganda) is an example. More regular public- from the perspective of provider incentives, ity of service characteristics on several dimen- the decision concerning the dif�culty of sions—such as absentee rates, regular delivery monitoring is, of course, key. When moni- of pharmaceuticals, hours of operation for toring is easy—sizes 1, 3, 5, and 7—oppor- electricity or water—could all mobilize com- tunities for more explicit incentives and the munity concern and informal influence. use of contracts should be explored. How- ever, contracting with a private sector is Evaluation often a bad idea for sizes 5 and 7. Such con- Generating and disseminating information tracts are a common source of corruption are powerful ways of improving service deliv- that governments �nd harder to manage (c) The International Bank for Reconstruction and Development / The World Bank Policymakers and providers 107 and citizens �nd harder to detect than if The boxes suggest eight sizes appropriate services were provided by government. in different circumstances. They also indi- When monitoring is dif�cult—the even- cate the relative dif�culty of carrying them numbered sizes—one goal is to improve the out—the degree of government failure ability to monitor with the methods dis- associated with them. Generally speaking, cussed in this chapter. More competition, the severity of the government failure more careful measurement of outcomes, increases with the size number. The degree the evaluation of the effect of inputs on of market failure needed to justify relatively outcomes, and the provision of incentives easy policies to carry out is modest, or, to groups of providers such as schools or equivalently, the highest-priority policies districts can all help. are those with large market failures or Figure 6.2 Eight sizes �t all 1 Central government financing with contracting. Performance contracts let by government. Contract recipients can be public providers, NGOs or the private sector as long as rewards can be made contingent on outcomes. Network infrastructure can have national oversight. Contracts Easy to require information for monitoring, low risk of government failure or of monitor contracting. 2 Central government provision. Technical reforms to allow better Hard to monitoring of outcomes that allow a movement toward size 1 may be monitor considered. These are the “New Zealand reforms.� Otherwise, (a) if Homogeneous clients difficult for government to monitor but easier for clients, consider using latter as monitors as in chapter 4; (b) evaluations of programs provide essential information. Low risk of government failure depending on monitoring abilities. 3 De-concentrated central or local government provision with contracting. As in size 1 but: performance contracts possible by local or de- Heterogeneous concentrated central government. Local or regional regulatory boards for clients Easy to network infrastructure. Low risk of government failure if government can Pro-poor monitor accommodate relevant variation in preferences. politics 4 Local or de-concentrated central government provision. Hard to As in size 2, some reforms to improve monitoring may be possible. monitor Contracts need more flexibility to accommodate differing preferences. Incentives to “teams� of providers. Regular evaluation of program success essential for replication and scaling up. Risk of government failure moderate since both monitoring and variation of preferences are challenges. 5 Client power—experiment with contracts. Communities may rely on contracting with private sector or NGOs. Contracts let by problematic levels of government are prone to corruption. If only local government is a problem, center can contract as in size 1 with local government as Easy to potential provider subject to contract monitoring. Government failure is monitor moderate if communities can be given more authority or only one level of Clientelist government is problematic. politics Homogeneous Hard to 6 Client power—experiment with providers. Explicit contracts difficult. clients monitor Communities (and donors) may want to rely on altruistic providers since monitoring is difficult (subject to caveats in chapters 4 and 11). Public dissemination of information is essential. Government failure moderate to severe. 7 Client power—experiment with community control. Similar to size 5. Heterogeneous Public disclosure essential. Evaluation of program performance clients Easy to important. Risk of government failure moderate to severe due to the monitor added problem of accommodating varying preferences. Hard to 8 Client power—imitate market. Similar to size 6. Local input important monitor from varying preferences. Evaluation of program performance and public disclosure is essential. Risk of government failure is severe. (c) The International Bank for Reconstruction and Development / The World Bank 108 WORLD DEVELOPMENT REPORT 2004 strong redistributive effects. For the hard- Scaling up, scaling back, est cases such as case 8, market failures and wising up must be quite costly to justify intervention, There is no “right� way to make sure services given the many legitimate claims on gov- reach poor people. The appropriate technical ernment. interventions—and the institutional struc- Including government’s ability to tures that generate them—vary enormously. implement—that is, the degree of govern- Education was expanded dramatically in ment failure to be expected—can lead to a Chile by markets and vouchers, in Cuba by a substantial re-ranking of public policies central ministry, and in El Salvador by local relative to conventional analyses. For school committees. Beyond trial and error, social security systems, for example, there scaling up means watching what you’re is no particular reason on conventional doing, evaluating whether it works, deter- economic grounds for the public sector to mining why it works or doesn’t, replicating send out checks to pensioners. But many success, and evaluating the replications as governments with well-developed admin- well. Sometimes things work for idiosyn- istrative procedures do it quite well, and cratic reasons—a charismatic (and literally there is no compelling reason to change— irreplaceable) leader or a particular (and market failures are not terrible but neither unrepeatable) crisis that solidi�es support for is it hard for government to do. Much of a politically dif�cult innovation. So one-time the controversy about whether rich coun- successes may not be replicable. Experimen- tries should emulate New Zealand’s tation, with real learning from the experi- reforms surrounds this point. New inno- ments, is the only way to match appropriate vations in contracting with a private sec- policies with each country’s circumstances. tor or with a government agency might Scaling up also means scaling back— improve the functioning of government abandoning failures unless a good, remedia- somewhat. But if government is already ble reason for failure is found. Abandoning doing tasks acceptably, the gains may be failures is harder than it sounds. Simply small and possibly not worth the disrup- admitting failure is hard enough, particularly tion caused by the change itself. for politicians. But with the severe resource When applied to the health sector some constraints in developing countries—they standard prescriptions are reinforced by are poor after all—badly performing pro- these considerations while others are chal- grams are simply unaffordable. Where pro- lenged. The provision of traditional public grams are intensive in management (and health services, such as pest control to pre- auditors and managerial talent are scarce) or vent infectious disease, is relatively easy to intensive in trained personnel (and teachers carry out. But staf�ng and maintaining a and doctors are scarce), states need to let go large network of primary health centers in of programs that are not working and �nd remote areas is often hard to do, even alternative ways to achieve better outcomes. though the redistribution effects are If the political will exists, the key to scaling potentially bene�cial. It might be wiser, up is information. Beyond evaluating pro- until government capabilities improve, to grams and projects, a continuing focus on try to get poor people to government facil- making services work for poor people—edu- ities, even to much maligned hospitals, cated children, better health, reliable water, than get facilities to poor people. Not only lighted homes, safer streets—depends on the would this address a serious market fail- continuing measurement of progress toward ure, the absence of insurance for expensive these goals. “What gets measured is what care, but it will be easier to implement counts.� This focus on outcomes helps poli- since working in less remote areas is more cymakers choose the best options for serving consistent with providers’ interests and poor people. It helps the providers know easier to monitor, with a smaller number when they are doing a good job. And it helps of larger facilities.308 clients judge the performance of both. (c) The International Bank for Reconstruction and Development / The World Bank spotlight on Cambodia Contracts to improve health services—quickly Cambodia began experimenting with different forms of contracting to improve health services in 1998. The lesson— thanks to good evaluation—is that contracting can help increase the coverage of some key services in a short time. M ore than 25 years of conflict left Cambodia with little health infrastructure. In the late 1990s its health indicators were among the worst in Southeast Asia. Average life expectancy at more than 20 percent of its planned health facilities functioning. All had very poor health service coverage. And all were com- parable in their socioeconomic status. Annual per capita recurrent spending by in contracted-in and control districts. The reduction was especially marked among the poor ($35 a year, or 70 percent), indicating better targeting and more ef�cient transfers of subsidies. birth was less than 55 years. Infant mortal- donors and government was higher in the Even though the health ministry encour- ity was 95 per 1,000 live births. And mater- contracted-out districts: $2.80 in the con- aged all districts to implement of�cial user nal mortality was 437 per 100,000 live tracted-in districts, $4.50 in contracted-out fees, only one contracted-in district estab- births.309 The public health care system districts, compared with $2.90 in control lished a formal user fee system and used the remained rudimentary: average facility use districts.311 These differences are large and receipts from the system to reward health was 0.35 contacts per person per year, and represent slightly less than 20 percent of the care workers with monthly performance patients complained of very low quality. health expenditures (including private and and punctuality bonuses. That could Then in 1998 the government con- excluding capital investments from the gov- account for slightly higher spending for this tracted with nongovernmental entities to ernment) in all of the districts. type of district.313 provide health services in several districts. There are several possible reasons for The contracting increased access to health Contracting for better results these pro-poor outcomes in the contracted services—and not at the expense of equity. All districts improved service coverage in a districts. short time. After only 2.5 years of the four- • The regular availability of drugs and qual- Contracting primary health care year experiment, all districts had achieved i�ed staff strengthened service provision services (in and out) their contractual obligations for most of at health centers in the villages, where the evaluation indicators.312 The use of most poor people are concentrated. Intervention and control areas consisted of health services among the poorest half of randomly selected rural districts, each with the populace increased by nearly 30 per- • The contracted nongovernmental orga- 100,000 to 200,000 people.310 Contractors nizations used a market-based wage and centage points in the contracted-out dis- were chosen through a competitive process bene�ts package to attract and retain trict (�gure 1). One possible explanation is based on the quality of their technical pro- health care providers. that the contracted-out districts did not posal and their price. Three approaches were used. charge of�cial user fees; they also discour- • A reduction in the private out-of-pocket aged health care workers from taking cost of services and a more predictable and transparent fee structure increased • Contracting out. Contractors had full “unof�cial� user fees by paying signi�- the demand for health care services by responsibility for the delivery of speci- cantly higher salaries to providers than in �ed services in the district, directly the other types of districts. the poor. employed their staff, and had full man- The pattern of increases is similar agement control (two districts). across a variety of service and coverage Figure 1 Percentage of illnesses treated at a • Contracting in. Contractors provided indicators (�gure 2). The contracted-out health facility for people in the poorest half of only management support to civil ser- districts often outperformed contracted-in the populace vice health staff, and recurrent operat- districts, which outperformed control dis- Percent ing costs were provided by the govern- tricts. But not all indicators were as respon- 35 ment through normal government sive. The share of deliveries assisted 30 channels (three districts). changed by only a small amount in all three 25 districts. And there was no difference • Control areas. The usual government between contracted-in and contracted-out 20 2001 provision was retained (four districts). 15 districts in the increase in vitamin A cover- 1997 A budget supplement was provided to con- age. The level of immunization in con- 10 tracted-in and control districts. tracted districts also remained quite mod- 5 Performance indicators were measured est, peaking at only 40 percent. 0 for all the districts by household, and health Out-of-pocket expenditures on health Control Contracted Contracted facility surveys, which were conducted in care services fell dramatically in the con- in out 1997 before the experiment. No district had tracted-out districts but increased slightly Source: Bhushan, Keller, and Schwartz (2002). (c) The International Bank for Reconstruction and Development / The World Bank 110 Spotlight on Cambodia Figure 2 Coverage of selected health indicators between own funds and, in one district, allocated a 1997 and 2001 in control and contracted districts of Cambodia larger share of user-fee income. The con- Percent trol districts, left to their own devices, 60 allowed workers to pursue private income- Control 50 Contracted in maximizing behavior through unof�cial Contracted out fees and private practice, to the detriment 40 of the public health care services for the 30 poorest of the poor. 20 Transparent and predictable fee struc- tures are important in improving access to 10 health services. Of�cial user charges were 0 introduced in only one contracted-in dis- –10 trict, in consultation with communities, to Antenatal Tetanus Assisted Full Vitamin A provide incentives to health workers. To care toxoid deliveries immunization coverage remove ambiguity about charges, a sched- Source: Bhushan (2003). ule of user fees was prominently displayed in all health facilities. This discouraged pri- • The availability of health services in vil- hold surveys and spot checks by government vate practice and helped bring “under-the- lages reduced travel expenditures to seek staff. Payments were linked to achieving tar- table� payments formally into the system. health care, and NGOs enforced rules gets, with bonuses for better-than-agreed-on Out-of-pocket spending on health fell in against informal payments by patients. performance. that district. No user fees were introduced Improving health services for the poor in the other two contracted-in districts, or Agreements on deliverables— requires that health workers be adequately in the control districts, where out-of-pocket and enforceable contracts compensated and effectively supervised and spending did not come down. Contracting health services to NGOs can supported. The NGOs working in con- Contracting health services to NGOs expand the coverage for poor people. In tracted-out districts revised the salaries of can be dif�cult for policymakers to accept. Cambodia it took agreements on deliver- health care providers, bringing them in line But the Cambodian experience shows that ables and an enforceable contract, which in with average salaries in the private sector. In it can be effective and equitable. It helped turn required an independent performance return, the NGOs required the providers to convince policymakers that the model veri�cation system. Once targets for 13 key work full time in health facilities and to could be adopted on a larger scale. They are health indicators were agreed on—for poor have no private practice. extending contracting to 11 poor and people—progress toward achieving them In the contracted-in districts, the NGOs remote districts, where the public provision was measured through independent house- supplemented provider salaries with their of services is dismal. (c) The International Bank for Reconstruction and Development / The World Bank Basic education services 7 An institutional arrangement for basic edu- ter 1—unaffordable access, dysfunctional cation should be judged by its production schools, low technical quality, low client chapter of high-quality learning, equitably distrib- uted. This requires that children be in responsiveness, and stagnant productivity. But not all countries face the same prob- school and that they learn. This in turn rests lems. In many of the poorest countries on education systems that create relation- there are enormous de�cits in affordable ships of accountability between citizens, access. Poor people have less access, lower politicians, policymakers, and providers, attainment, and lower quality than those with clear objectives, adequate resources, better off. In many countries public sector capable and motivated providers, progress provision is close to dysfunctional and rife assessments, and performance-oriented with corruption. The technical quality of managements. instruction and learning outcomes are Successful education systems vary widely. shockingly low, especially among poor peo- Some systems are centralized, others decen- ple. And even the most advanced economies tralized. Some have almost exclusively pub- struggle to make education systems more lic schools, while others provide public sup- productive. port to private providers. But not just Shortfalls in universal primary comple- anything goes. tion—a combined result of children who never enroll, children who do not progress, • The politics of schooling—particularly and children who drop out—reflect the fail- the effectiveness of the voice of poor ures in the system. In Madagascar only 52 people—determines both the school sys- percent of 15- to 19-year-olds in the poorest tem’s objectives and the public resources 20 percent of the population had ever that go to education. enrolled in school, and only 4 percent com- • The compact between policymakers and pleted even grade 5 (�gure 7.1). In Brazil 89 providers of schooling needs to balance percent of poor adolescents enrolled in the autonomy of schools and teachers grade 1, but only 30 percent completed with performance assessment. grade 5 because of high dropout and repeti- • Schools (and school systems) must be tion rates. In Turkey high retention through enabled to manage for performance— primary school, followed by a sharp drop in and, particularly, to �nd effective ways to progress to the next level, suggests that sys- train and motivate teachers. temic and institutional solutions are • Direct parent and community participa- required to increase achievement. In Ban- tion in schools, demand-side inducements gladesh only 60 percent of poor adolescents to expand enrollments, and choice—if have completed grade 1, and only 36 per- correctly designed—can be valuable parts cent have completed grade 5. of an overall plan for school improvement. Unaffordable access Common problems Despite at least 55 years of acknowledgment of service provision that universal literacy is the heart of develop- Education systems face the common prob- ment, and despite repeated rhetorical com- lems of service provision outlined in chap- mitments to universal enrollment, even the 111 (c) The International Bank for Reconstruction and Development / The World Bank 112 WORLD DEVELOPMENT REPORT 2004 modest goal of universal primary school Bangladesh 30 percent of students who completion has not been realized. Some completed grade 5 were not minimally countries have made huge strides—average competent in reading; 70 percent were not Figure 7.1 Poor children: less likely completion rates in Brazil expanded from less minimally competent in writing.316 to start school, more likely to drop out 15- to 19-year-olds who have than 50 percent in 1990 to more than 70 per- Evidence on learning outcomes is disap- completed each grade cent in 2000. But if countries continue at only pointing even in middle-income countries. Madagascar 1997 their recent rate of progress, universal pri- For instance, in the recent Programme for Percent mary completion would come only after International Student Assessment of the 100 2020 in the Middle East and North Africa, achievement of 15-year-olds in school, only 80 after 2030 in South Asia, and not in the fore- 5 percent of Brazilian students reached the seeable future in Sub-Saharan Africa. Organisation for Economic Co-operation 60 Richest fifth In the very poorest countries the attain- and Development (OECD) median in 40 ment de�cit is spread across the population, mathematical literacy (�gure 7.2). Fifty-six but in most it is concentrated among chil- percent of Brazilian students were at level 1 20 Poorest dren from poor households. In countries (of 5) in reading literacy, compared with 18 fifth with very low attainment, like Mali, most of percent for students in OECD countries. 0 0 1 2 3 4 5 6 7 8 9 the population is rural, and there are sub- Only 4 percent reached pro�ciency levels of Grade stantial de�cits in primary completion even 4 or 5, compared with 31 percent for OECD Turkey 1996–97 among relatively wealthier and urban fami- students.317 This is not to single out Brazil Percent 100 lies. In India the rural poor (poorest 50 per- for poor performance: Brazil is widely rec- cent) accounted for 72 percent of the de�cit ognized for its advances, and its willingness 80 in completion of grade 5 among 15- to 19- to participate in the study and its courage in 60 year-olds, and completion is higher among releasing the results demonstrate a strong boys than girls. In the Philippines the de�cit commitment to education outcomes (other 40 is much lower, concentrated among the countries have participated in examinations 20 rural poor and higher among boys than and then refused to disclose the results). In girls. addition, in an earlier comparison of 11 0 0 1 2 3 4 5 6 7 8 9 Latin American countries Brazil was tied Grade Dysfunctional schools with Argentina for second place in the Brazil 1996 Schooling completions and learning out- mathematics performance of 4th graders. Percent 100 comes may fall short because providers are dysfunctional. While most teachers try con- Low client responsiveness 80 scientiously to do their jobs, one recent sur- When communities are not involved in 60 vey found a third of all teachers in Uttar establishing, supporting, or overseeing a Pradesh, India, absent. Cases of malfeasance school, the school is often seen as some- 40 by teachers are distressingly present in many thing alien. Villagers refer to “the govern- 20 settings: teachers show up drunk, are physi- ment’s� school, not “our� school. In Voices of cally abusive, or simply do nothing. This is the Poor people often complain of absent or 0 not “low-quality� teaching—this is not teach- abusive teachers and demands for illegal 0 1 2 34 5 6 7 8 9 Grade ing at all.314 fees to get their children into school or to Bangladesh 1996–97 influence examination results.318 A study of Percent Low technical quality schooling in rural Nigeria found that vil- 100 lagers often stopped expecting anything The quality of instruction can also be low 80 because of low capability, weak motivation, from government schools, shouldering the and a lack of complementary inputs. In burden themselves.319 60 very-low-income settings learning out- 40 comes can be dismal. The 1994 Tanzania Stagnant productivity Primary School Leavers Examination sug- Creating and maintaining an institutional 20 gested that the vast majority of students had environment that promotes higher produc- 0 learned almost nothing that was tested in tivity and more learning is not easy. A 0 1 2 3 4 5 6 7 8 9 their seven years of schooling—more than recent set of studies documented that Grade Source: Analysis of Demographic and Health four-�fths scored less than 13 percent cor- spending per pupil in real terms has Survey data. rect in language or mathematics.315 In increased by 50 percent or more, often two- (c) The International Bank for Reconstruction and Development / The World Bank Basic education services 113 or threefold, in nearly all OECD and East Figure 7.2 Fifteen-year-olds in Brazil and Mexico perform substantially worse on standardized tests than students in OECD countries Asian countries. Yet in none of these coun- tries have test scores improved commensu- Distribution of mathematics test scores Distribution of reading test scores rately.320 The obvious implication of these OECD average two facts is that measured learning achieve- 0.5 0.5 ment per dollar spent has fallen dramati- 50 percent above 500 50 percent cally in every country examined. above 500 0.0 0.0 367 500 625 366 500 623 For higher-quality systems, Mexico strengthen the relationships 0.5 0.5 of accountability 8.6 percent 18 percent Despite enormous differences in attain- above 500 above 500 ment, equity, and learning across countries, 0.0 0.0 the features of school systems are strikingly 281 387 496 311 422 535 Brazil similar. Public production is almost always 0.5 0.5 the dominant—if not exclusive—means of government support of education. Whether 4.4 percent 11 percent in Argentina, Egypt, India, Indonesia, above 500 above 500 Paraguay, or Tanzania, public systems dis- 0.0 0.0 play age-grade organization of classrooms, 212 334 464 288 396 507 Normalized test score Normalized test score replication of social structures and inequal- Note: Distributions are approximated on the basis of the mean and standard deviation reported in the ities, and similar ways of training, hiring, original source. Source: OECD (2001). compensating, and promoting teachers. Despite these surface similarities, there are widely different outcomes. Both Nigeria and Singapore retain many of the organiza- • Voice, or how well citizens can hold the tional elements of British education. Yet on state—politicians and policymakers— one international achievement test in the accountable for performance in dis- 1980s Nigeria was among the worst per- charging its responsibility for education. formers while Singapore is frequently • Compacts, or how well and how clearly among the best. the responsibilities and objectives of That public provision has often failed to public engagement are communicated to create universally available and effective the public and to private organizations schooling does not imply that the solution that provide services (Ministries of Edu- is a radically different approach (complete cation, school districts). decentralization, total control by parent • Management, or the actions that create groups, generalized choice) or a narrow effective frontline providers (teachers, focus on proximate determinants (more administrators) within organizations. textbooks, more teacher training). Univer- • Client power, or how well citizens, as sal and quality education can come from clients, can increase the accountability of very centralized systems (France, Japan) or schools and school systems. from very decentralized systems with con- siderable local accountability and flexibility Effective solutions are likely to be mixtures (United States). Many countries have little of voice, choice, direct participation, and orga- private schooling, and some a great deal nizational command and control, with func- (Holland). Classroom practice is what mat- tional responsibilities distributed among ters. If the underlying causes of failure are central, regional, local, and school administra- not addressed, all these approaches can fail. tions. The pieces have to �t together as a sys- Chapters 3 through 6 developed a frame- tem. More scope for parental choice without work for analyzing service provision, looking greater information about schooling outputs at four relationships of accountability. In will not necessarily lead to better results. Infor- education, these are: mation systems that produce data on inputs (c) The International Bank for Reconstruction and Development / The World Bank 114 WORLD DEVELOPMENT REPORT 2004 but do not change the capabilities or incentives mate determinants (box 7.1). When teachers of frontline providers cannot improve quality. are not consulted in training design—often Schools and teachers cannot be made more the case—poor implementation is the result. accountable for results without also receiving Training may not be integrated into the sys- suf�cient autonomy and resources and the tem, as when teachers are trained in methods opportunities to build capabilities. Conversely, inconsistent with public examinations and so schools cannot be given autonomy unless they are reluctant to adopt them. Teachers often are given clear objectives and regular assess- have little incentive besides professional pride ments of progress. to adopt new methods. What successful education systems share is If the underlying problems are not a working structure of accountability: clear solved neither bureaucracy nor market will objectives, adequate resources, and capable work well. Increasing client power, by creat- and motivated providers. This Report focuses ing mechanisms for communities and par- on institutional reforms to achieve that system ents to improve their local school, is impor- of accountability—not on the proximate tant. But this short-route accountability is determinants of success, such as curriculum not enough. Improving services also design, pedagogical methods, textbooks, requires stronger mechanisms of long-route teacher training, school construction, or new accountability—accountability of politi- information technologies. Institutional re- cians and policymakers for education and forms will achieve desired outcomes by affect- improved pro�ciency in public administra- ing proximate determinants—and proximate tion with accountability of the education determinants that produce good education are bureaucracy for outcomes. There is no the outcome of well-structured and well-func- quick �x in an area as complex and exten- tioning systems. But efforts to improve proxi- sive as schooling, only the hard slog of grad- mate determinants through internal manage- ual improvement through strategic incre- ment initiatives have usually failed. Why? Not mentalism, which links current operational because of a lack of knowledge of what to do. actions with long-run institutional strate- But because of lack of the sustained bureau- gies and goals. cratic, market, parental, and political pressure needed to make things work. Citizens and clients, politicians The disappointing experience with teacher and policymakers: voice training shows the limit of a focus on proxi- In administration of all schools, it must be kept in mind, what is to be done is not for the sake of the BOX 7.1 The dismal state of teacher training in Pakistan pupils, but for the sake of the country. circa 1990 —Mori Anori, Japanese Minister of Education 1886–89 “Teacher training in this province is a mock- A national survey of Pakistan’s primary ery. We should close down the teacher train- schools suggests that these anecdotal Politics plays a key role in establishing ing institutes and stop this nonsense. I have accounts are only too true. Survey data on been teaching in a B.A./B.Ed. program for teaching practices “provide no basis for objectives for the education system—con- many years and see no signs that I have any statements that . . . teacher training makes a cerning both distribution and quality—and impact on the students I teach.� substantial difference to how teachers in mobilizing resources. The reason is that —A university education instructor quoted teach.� A 1998 study of teacher training schooling, especially at the basic level, has in Warwick and Reimers (1995). suggests that “staff and faculty are profes- sionally untrained, political interference is become an important element in a child’s “Most inmates of this system [two teacher common, resources and facilities are poor socialization. training institutes] have no respect for and badly utilized, motivation and expecta- Those who control the state use schooling themselves, hence they have no respect for tions are low and there is no system of others.The teachers think the students are to promote beliefs they consider desirable. accreditation to enforce standards.� Embed- cheats, the students think the teachers have ded in an education system that was funda- Nearly everywhere this means that schools shattered their ideals. Most of them are dis- mentally unaccountable and lacked any promote a sense of national identity, a illusioned.They have no hopes, no aims, no outcome orientation, teacher training national language, and loyalty to the nation- ambitions.They are living from day to day, reflected worst practice. watching impersonally as the system crum- state—in competition with more local or bles around them.� Sources: Warwick and Reimers (1995); Kizilbash ethnic af�liations—and, in more extreme —Nauman (1990). (1998), p. 45. cases, a speci�c political indoctrination. Modern states—from Third Republic France (c) The International Bank for Reconstruction and Development / The World Bank Basic education services 115 to Ataturk’s Turkey—have also used public Because the educational system had no schools to supplant or suppress religious coherent, consensual focus: “For reforms to instruction.321 Authoritarian states have used stick, there �rst needs to be a vision for the schooling to disseminate a single acceptable future with agreed long-term objectives ideology—for example, Soeharto’s promo- derived from stakeholders: informed dia- tion of the �ve principles of pancasila322 in logue with parents, employers, religious lead- Indonesia. These examples are not the excep- ers, school leavers, and others. The absence of tion but the rule: countries around the world such a shared long-term quality-of-services explicitly use schooling to inculcate ideas strategy that focused scarce resources on about the proper organization of society. quality rather than quantity has left the edu- cation sector open to the imposition of ideas Voice and the objectives of schooling from outside: from donors, with agendas of Schooling has become a battleground for questionable value to the country’s situation, political conflicts. Different groups want dif- or from graduates returning with overseas ferent—often contradictory—things from degrees and ill-informed, though well-inten- schooling. Poor parents see education as an tioned, agendas of their own.�323 opportunity for their children to lead better Democracy is not necessary for excellent lives, but they may also want education to schools. The huge variation in commitment reinforce traditional values. Elites may want to schooling across the states of India is universal education but often promote pub- enough to suggest that electoral democracy is lic spending on higher education for the also not suf�cient for voice to lead to universal bene�t of their own children. Urban and education (see spotlight on Kerala and Uttar business coalitions may favor more educa- Pradesh). But the absence of democracy or tion because it increases the productivity of other means of effective citizen voice has a their workers, or industrialists may quietly huge downside. While one-party states occa- oppose “too much� education because it sionally produce good results (see spotlight makes workers restive. One recent study of on Costa Rica and Cuba), many authoritarian owners and managers of modern factories regimes have no interest in expanding educa- in Northeast Brazil that were moving to cut- tion or improving its quality. There are two ting-edge business practices revealed a dis- risks: the system is effective but its goals are turbing lack of support for expanding edu- completely set by politicians and policymak- cation. Many felt that a primary education ers, or the system is ineffective because politi- (eight years) was helpful, but more than that cians and policymakers have goals other than was “dangerous� because it created workers effective provision of services. The results: too who were less docile. Many commented that few resources are allocated to education, too “too much education is a bad thing.� few of those resources reach poor people, and (Tendler 2003). Politicians may want to resources are allocated ineffectively (because deliver on promises of universal schooling providers are more influential than citizens). while also using the education system to As more countries move to more democ- provide patronage jobs (the example of Pak- ratic modes of choosing leaders, citizen con- istan, in box 5.3, is not unique). Teachers trol over the structure and content of curric- and their unions want high-quality univer- ula gains prominence. Having a common sal education but also higher wages. negotiated vision of the objectives of public To get what you want, you need to know support for schooling makes it easier to move what you want. But what a society wants to the other stages of improving the quality of from its schools is not simple and cannot be schooling—mobilizing and allocating decided by experts alone. A recent study of resources, communicating objectives to attempts to improve the quantity and qual- providers, and delegating responsibility and ity of basic education in an Asian country in autonomy to schools. Without a clear vision the 1990s concluded that even many peda- of goals, reform is reduced to a focus on gogically and internally sound reforms did inputs and process alone. not have a sustained impact on teaching The greater the demand for education, the practice or student learning. Why? sharper the vision. In Malawi, Uganda, and (c) The International Bank for Reconstruction and Development / The World Bank 116 WORLD DEVELOPMENT REPORT 2004 most recently Kenya, a commitment to uni- higher levels of education. Or systems are inef- versal education was a popular stance— �cient in translating resources into outputs. A although a dif�cult commitment to match common problem is that teacher salaries, even with resources (see spotlight on Uganda). at very low wages, crowd out all other inputs. A recent study found that 44 of 55 countries Adequate resources, examined allocated more than 70 percent and adequately distributed half (23) allocated more than 80 percent of To achieve educational goals politicians and spending to salaries. Such levels of spending policymakers—either autonomously or often imply either inadequate supplies to other through the pressure of citizen voice—must inputs or formal or informal levies on parents. provide adequate resources. To learn effec- Empirical studies also show that increases in tively, children need affordable access to teacher salaries have little or no association infrastructure, inputs, and instruction—far with learning outcomes (discussed further from the case in many countries. A recent below). Many studies estimate the impact of study of �nancing the global Education for selected classroom instructional materials or All initiative compared successful and less school facilities to be some 10 times that of successful countries along three dimensions: teacher salaries (this is not to say that simple “equipment-based� approaches will succeed). • Revenue mobilization for primary edu- Another common problem is devoting cation (overall taxation rates, the frac- resources to reduce average class sizes, which tion of spending on schooling, the frac- often results in inef�ciently small classes— tion of that spent on primary schooling). boosting unit costs and limiting access. • Unit cost of a year of effective schooling Public resources are politically distrib- (teacher salaries and class size). uted, so the effective distribution of resources • Internal ef�ciency (years of schooling is an issue of voice. A review of the empirical provided per primary school completer). evidence suggests that the common pattern Even with adequate �scal effort, reason- of too few resources to high-productivity able costs, and internal ef�ciency, many inputs is so ubiquitous—�gure 7.3 gives just countries do not generate enough resources two of many possible examples—that it is to achieve universal completion. For these likely generated by a political economy that countries there is a compelling case for addi- fails to adequately incorporate the voice of tional international assistance (see box 2.3). poor people. Changing this distribution of But in many cases the resources are simply resources requires more than a technocratic not used effectively. They are allocated to the adjustment—as Brazil has shown by its wrong mix of inputs. Too great a share goes to reforms in the 1990s. Because poor people are almost always the Figure 7.3 Increases in test scores per dollar spent on different inputs last enrolled, additional spending that Northeast Brazil (1980s) India (1990s) expands access is more favorable to poorer households than existing spending. A study Teacher salary 1 Teacher salary 1 in India found that even though educational expenditures on average were not more pro- Ensuring all Facility poor than a uniform transfer would be, the teachers have 3 years secondary 1.9 improvement 1.2 poor bene�ted more than proportionately at intervention the margin when enrollments in primary school Teacher table, education expanded (since the better-off One additional pupil tables and chairs, and other 7.7 square foot 1.7 were already in school).324 So education per student “hardware� expenditures that expand access are better Packet of Full packet of targeted to poor people than resources that instructional 19.4 instructional 14 materials materials exclusively raise quality. But the quality-quantity tradeoff is not a 0 5 10 15 20 0 5 10 15 20 Increase in test score per dollar, Increase in test score per dollar, simple choice between creating additional relative to teacher salary relative to teacher salary school places or improving instruction. A Source: Pritchett and Filmer (1999). major problem for poor children in nearly (c) The International Bank for Reconstruction and Development / The World Bank Basic education services 117 Table 7.1 In Madagascar, at higher levels of education unit costs are much higher and participation of the poor much lower Ratio of cost of a year of Cumulative public spending Share of poorest 40 percent Share of poorest 40 percent in higher education to the cost on graduates of each level in those who complete each those who reach each level of a year of primary school (percent of GDP) level (percent) (percent) No schooling 0 0 37.6 57.8 Primary (grades 1–5) 1 0.4 7.3 33.7 Lower secondary (grades 6–9) 2.75 1.25 0.5 3.1 Upper secondary (3 years) 5.5 2.56 * * Higher (4 years) 19.6 8.84 * * * indistinguishable from zero. Sources: World Bank (2001c) and analysis of Madagascar Demographic and Health Survey. every environment is that they drop out of tute and disadvantaged has political dan- school with greater frequency, in part because gers as well. Systems that focus mainly on the quality of the schooling they receive is so poor citizens, leaving the middle classes no low. So quality improvements need to accom- stake, tend to be �nancially less sustainable pany quantity improvements. and to experience less pressure for account- Spending on primary schooling is mildly ability—and so tend to be inef�cient and progressive, but that on higher levels of edu- unconcerned with quality. cation is not. With children from the poorest households unlikely to reach higher levels of Policymakers and organizational schooling, and with greater per student providers: compacts spending at higher levels than at lower, chil- dren from richer households capture the bulk I do not care that teachers are offended by it. I am less interested in the teacher’s method of teaching of educational spending. In Madagascar a than in the result she achieves. . . . There should be single year of higher education costs 20 times a test at the end to see whether the results are being that of primary schooling—and only 3 per- achieved. . . . Let us who represent the community cent of children completing lower secondary say here and now there should be a [test] no mat- school are from the poorest 40 percent of ter who may oppose it. . . . If we want to see that a certain standard is reached and we are paying the households (table 7.1). Relative cost alone is money we have the right to see that something is not the issue. It is whether funding across lev- secured for that money. els is equitable and ef�cient—or driven Eamon de Valera, Irish Prime Minister, 1941326 exclusively by elite politics. The political conditions required for ade- The line separating the state as education pol- quate budget allocations for education are icymaker (setting the rules of the game) and not obvious. Simple answers like “democ- as major organizational provider (running racy� are attractive—but just not true. India, the school system) is typically blurred. The democratic since independence, has wealth minister of education frequently wears both gaps in education attainment larger than any hats. Often there is no interest in measuring other country with comparable data. At least results, so there is no way of making the pub- one empirical study suggests that nondemoc- lic provider accountable for results. ratic countries spend more on education.325 Clarifying objectives and the roles of pol- But there is a risk that these governments care icymakers and providers is a �rst step. With- not about the quality of education but about out specifying desired outputs and outcomes using schools for religious, secular, or there is no way to say whether resources are national indoctrination. In countries with suf�cient (suf�cient to do what?) or used democratic elections, schooling opportuni- effectively (relative to what goal?). Vague ties can be limited and education resources oversight and vague goals reduce manage- devoted to patronage and clientelism if ment to compliance with formal rules for voice is weak and control rests with a nar- inputs and processes. The resulting lack of row elite. Targeting resources to the desti- clarity often results in “mission drift� and (c) The International Bank for Reconstruction and Development / The World Bank 118 WORLD DEVELOPMENT REPORT 2004 distracting struggles within the ministry of tion-intensive. It has multiple outputs that education. Lacking a clear mission, the edu- differ in measurability and in the dif�culty cation ministry is often accused of being cap- of attribution. And it involves a complex— tured by a teachers’ union rather than repre- and not well understood—relationship senting the collective interest in schooling. between inputs and outputs. High-perfor- The Irish Prime Minister’s insistence on mance schooling conveys skills, attitudes, testing is a common reaction to the perceived and values. Some steps in this process can failure of schools: a temptation to de�ne the be reduced to a detailed script. And some output of the school system exclusively as test aspects of instruction can be replaced by scores and then to hold schools accountable technology. But face-to-face interaction and for those scores. But accountability too nar- flexibility are crucial to high-quality rowly measured distorts the education sys- instruction. Instructors need to be capable tem. Only what gets measured gets done. of exercising discretion—in assessing stu- The strict primary completion examination dent mastery, providing feedback, and tai- brought in so con�dently by the Irish govern- loring the instructional mode to the student ment in the 1940s was gone by the 1960s, in and subject matter. This classroom behavior large part because of these concerns. is extremely dif�cult to monitor. The compact between policymakers and Schooling has multiple outputs—some organizational providers should create an easily assessed, others not. Assessing mastery environment in which all schools have the of simple skills through standardized testing means and motivation to provide high-qual- is fairly straightforward. But it is dif�cult to ity learning. Whether there is public produc- assess how well schooling has conveyed a tion or government funding of a range of conceptual mastery that allows application providers, the compact should focus on out- to real-world problems. It is still more dif�- puts and outcomes. This requires a means of cult to assess how well schooling has assessing a school’s contribution to the col- encouraged creativity. And it is even more lective objectives of education, and creating dif�cult to assess how well schooling has an environment for organizations to inno- conveyed values. Assessing success is further vate and bring those innovations to scale— complicated because different actors assign school autonomy with accountability. different values to different objectives. Designing an accountability system is The use and abuse of accountability dif�cult because it is dif�cult to attribute Creating accountability in schooling is dif�- speci�c outcomes—or even outputs—to cult. Schooling is discretionary and transac- speci�c actors. If a 15-year-old has mastered Table 7.2 Schools account for only a small part of variance in student learning outcomes (percent) Share of total variance across students I II III IV Due to differences in student Due to differences across Fraction of total variation Share of total variation in performance within schools schools attributable to student student test performance background differences that is (a) school speci�c across schools and (b) not attributable to student background differences across schools (II minus III) Brazil 55 45 25 20 Russian Federation 63 37 17 20 Czech Republic 48 52 4 18 Korea, Rep. of 62 38 14 24 Mexico 46 54 32 22 Developed country average 66 34 20 14 Source: OECD (2001), Annex B1, table 2.4. (c) The International Bank for Reconstruction and Development / The World Bank Basic education services 119 algebraic concepts, who deserves credit and many goals societies have for their schools. in what proportion? The parents’ genes? Performance measurement is not an attempt The child’s nutrition? The parents’ motiva- to reduce the output of schooling to the ability tion and efforts? The child’s peers? The of students to answer questions on standard- child’s primary school math teachers? ized examinations. The dangers of test-based Another teacher who motivated the child to school accountability have been debated for at do well in all subjects? The child’s current least 140 years (box 7.2). You get what you pay algebra teacher? for. But there are also dangers in too little Nearly all empirical studies of measured attention to performance. It is important to learning achievement agree that home back- distinguish among the three types of assess- ground accounts for most of the explainable ment: sample-based assessments to track per- variation in learning outcomes, especially in formance over time, “gatekeeper� examina- primary grades. The same studies disagree tions that are high stakes for students, and widely about how much can be attributed to assessments of school performance. a child’s school. The recent Programme for International Student Assessment study Tracking progress. One way to strengthen found wide variation in differences in student the compact between policymakers and edu- performance within or between schools cation providers is to develop measurement (table 7.2). Half or more of the variation in and reporting systems that allow investiga- performance across schools was due to varia- tion of value for money. Standardized exami- tion in students’ socioeconomic status, not to nations are a relatively inexpensive device for factors under school control. In poorer coun- monitoring progress and effectiveness. But tries the effect of schools is larger—and that few education systems in the developing of parental background smaller. But, in gen- world have disaggregated the cost of running eral, identifying the school’s value added is a school, and even fewer know how that cost not simple. is associated with learning. So there is almost Even for outputs easier to specify and no reporting based on such measurements. measure, not much is known about how The lack of information leads to an inability inputs affect them. Economists summarize to act accordingly. this relationship under the metaphor of a When the data are revealed, they can be “production function.� Little is known about surprising. One study that generated data this function because instruction involves relating expenditures and learning at the human beings—teachers and students—in all their complexity. For instance, there is ongoing, vigorous debate about the relevance BOX 7.2 Test-based accountability—nothing new under of class size for student test scores. Some the sun assert that class size is irrelevant, or nearly so. Test-based school accountability might ments provided teachers (who at the time Some assert that reductions in class size have seem like the latest thing. It isn’t. British leg- had little training) with clear indications of such a salutary impact on performance that islation for school funding in 1862 included what was valued and tangible awards for they are a cost-effective means of improving a system of “payments for results.� In addi- achievement. performance.327 After more than a century of tion to a base grant (based on number of Opponents raised the same arguments children and attendance), schools received made today.Teachers will “teach to the test� widespread use of classroom instruction, a grant for each student who passed a and ignore subjects not covered by the test intelligent, well-meaning, and methodologi- series of tests given by school inspectors in (such as history and geography).Test-based cally sophisticated researchers are still debat- reading, writing, and arithmetic. accountability will lead to teaching meth- ing such a seemingly simple issue. That shows Proponents of the testing argued that ods that emphasize rote memorization and performance-based transfers were only cramming. One educator argued that “pay- how truly complex the research questions common sense since public money was ment for results� would “be remembered are—the results will vary across time, con- involved. As one parliamentary proponent with shame.� tent, and context. reasoned: paying for performance will This particular system of “payment for either be cheap (because few schools meet results� was abolished in 1890. But the the standard) or expensive (because many debate continues today. Assessment systems students have high performance)—but it National assessment systems are essential for will not be both expensive and ineffective. Sources: Based on Bowen (1981) and Good and monitoring educational achievement. But per- Educational historians claim that the pay- Teller (1969). formance measurement is as complex as the (c) The International Bank for Reconstruction and Development / The World Bank 120 WORLD DEVELOPMENT REPORT 2004 Figure 7.4 School success depends on more than chances, parents will exert pressure on the spending per student Primary school pass rate in Mauritania school system for better examination results. Where public examinations are lim- Pass rate (percent) ited and educationally inadequate, perverse 100 pressures can worsen true educational qual- 80 ity in the interests of better examination scores. 60 School-based accountability for examina- 40 tion results. School accountability is con- troversial—with good reason. There is 20 empirical evidence that accountability mech- anisms based on examination results lead to 0 “teaching to the test� and to attempts to 0 5,000 10,000 15,000 20,000 manipulate results. Evidence from locations Unit cost (Mauritanian currency) as diverse as rural Kenya (see chapter 11 and box 7.5 later in this chapter) and urban Source: Mingat (2003). Chicago shows that accountability raised examination scores—but also that teachers school level in Africa found little connec- manipulated the students taking the exam, tion—Mauritanian schools with similar and taught to the test. spending had pass rates of less than 5 per- But teaching to the test is a criticism only cent and more than 95 percent (�gure 7.4). if the test is not a reliable assessment of the Needed for active management are skills that are the objective of public sup- data—on school costs, on the characteris- port for schooling—or if the tests divert tics of students, and on school performance teachers from more productive activities, on cognitive achievement tests. Once such as teaching higher-order thinking. implemented, these sample-based systems There is a tradeoff between what the test can be gradually scaled up to provide more costs (in design, testing, and scoring) and census-like measurements. how well it captures desired schooling out- put. Tests in some circumstances could Gatekeeper examinations. In most coun- divert teachers from more productive to less tries examinations are seen as a fair way of productive activities, such as “drills.� But in allocating limited school places. One study many cases performance is so weak that suggests that the impact on student perfor- even “less productive� but learning-ori- mance of centralized curriculum-based ented activities would be an improvement. examinations is as large as that associated with differences in parental education or with substantially more formal education Figure 7.5 Centralized exams have a strong impact for teachers (�gure 7.5). Since centralized on student performance examinations make relevant information Incremental test score widely available, they can be useful for gen- 20 erating accountability. Math The impact of public examinations on 15 the incentives of various actors points to Science systemic considerations. For instance, 10 teacher training programs often attempt to instill pedagogical techniques that promote 5 higher-order thinking skills. But when gate- 0 keeper public examinations assess only rote Centralized Parent completed Teacher with memorization, teachers frequently revert to exams tertiary/secondary PhD/MA/ similar methods. And if public examina- education bachelor degree tions have a major impact on students’ life Source: Wößmann (2003). (c) The International Bank for Reconstruction and Development / The World Bank Basic education services 121 There are three important technical comes. The roles of the ministry of educa- Figure 7.6 In Chile, good schools service students from every level of design issues with school-based account- tion can be unbundled so as to separate edu- socioeconomic status ability. First, the characteristics of students, cation policy from the operation of schools. their peers, and their families are far and A more explicit compact relationship can be Fourth grade basic SIMCE score 100 away the largest determinants of variation in made with organizational providers, per- performance. Any attempt to judge schools haps even with multiple providers within 80 on their level of performance will therefore the same jurisdiction. This structure can be judging the socioeconomic composition give clearer guidance on desired outputs and of the school—a “good� school might sim- outcomes, freeing school heads and teachers 60 ply have wealthier students. This is true on to pursue de�ned goals. average. Some schools serving poorer popu- Nicaragua created autonomous public 40 lations perform well or even very well (�g- schools guided by a school directive council ure 7.6). And some schools with wealthier comprising the school head, elected teach- 20 0 20 40 60 80 100 students are mediocre. ers, parents, and students. The school High Low To focus on school value added rather retained revenue from students, and the Index of socioeconomic status than differences based on school popula- council could make decisions about person- Note: The SIMCE is a standardized test in Chile. tions, scores can be empirically adjusted for nel, �nance, and pedagogy. The average Source: Mizala and Romaguera (2000). the composition of the student body (box school autonomy reported by these schools 7.3). Or assessments can measure changes in was between that reported by traditional student performance (which assumes that public schools (very little) and private socioeconomic composition is roughly con- schools (almost complete). The degree of stant). Or a threshold can be set that all self-reported school autonomy was posi- schools—whatever their student composi- tively correlated with student performance tion—are expected to achieve. on test scores at the primary level (though A second design issue in school-based not at the secondary)—but autonomy on accountability is statistical sampling. In paper was not. In a study in Chile very little many schools the number of students is of the variation (less than 1 percent) in small enough to result in considerable vari- three measures of self-reported autonomy ability. That means that even schools with of teachers was between the four types of strong improvements over time will have schools—public, private voucher, private years when scores are lower than in previ- paid, and Catholic voucher. More of the ous years—simply because of the mix and variation was between schools of the same number of students. It also means that a program of rewards or punishments for performance would disproportionately BOX 7.3 School-based performance awards in Chile reward and punish small schools relative to large schools. The third design issue is Since 1996 Chile has had an award for “top- correlation between socioeconomic status whether to reward good performance or performing� schools in each region. Ninety and awards. percent of the award goes directly to teach- Next, an index of school performance is intervene in bad performance—or both. ers (in proportion to their hours of employ- calculated based on standardized tests in ment), and 10 percent is allocated to the Spanish and mathematics in grades 4, 8, School autonomy schools.The awards are given every two and 10.The index is weighted for average years. test level (37 percent) and improvement in Accountability and autonomy are twins. Schools are divided into comparison test scores (28 percent) and includes other Traditional public sector bureaucracies have groups within each region of the country criteria such as “equality of opportunity� (22 little autonomy because accountability is based on location (rural, urban), education percent)—based on student retention and linked to rules and procedures, which allows level (primary only, secondary with no “discriminatory practices�—and “initia- primary), and socioeconomic status of par- tive� (6 percent)—based on regular devel- for little discretion. The heads of individual ents (according to information collected as opment of group pedagogical activities. schools are often bound by process require- part of the examination and an of�cial The program has been through three ments and so have little autonomy to “index of vulnerability�). In 2000–01 this rounds of selection, with 2,520 schools hav- actively manage their schools—to de�ne a classi�cation produced 104 comparison ing received awards once, 1,084 schools groups. In this way the performance of poor twice, and 360 schools in all three rounds. mission, choose instructional staff, inno- rural schools is not compared head-to-head vate, or encourage performance. Granting with that of richer urban schools. Analysis greater autonomy requires new forms of suggests that this procedure diminishes the Source: Mizala and Romaguera (2002). accountability based on outputs and out- (c) The International Bank for Reconstruction and Development / The World Bank 122 WORLD DEVELOPMENT REPORT 2004 teaching methods, new instructional inputs, BOX 7.4 Two large-scale cross-national assessments new use of the latest technology. The problem of learning is that there is too little systematic learning from innovation and too little replication of The Third International Mathematics and formance data on 265,000 students in 32 proven innovations. Science Study (TIMSS) created a data set on (mostly OECD) countries along with infor- student performance and characteristics mation from students and principals about The contrasting use of rigorous evalua- and on institutional characteristics of the themselves and schools.The conclusions for tions in health and education is striking. In schooling system such as use of centralized what schools can do to make a difference most developed countries no drug can be examinations and central, local, and school were: decisionmaking responsibilities. Analysis of used until it is proven safe and effective, and the performance of more than 266,000 stu- • Students’ reported use of school the standard of proof is the randomized resources was more closely associated dents from some 6,000 schools in 39 with performance than principals’ reports double-blind clinical trial. But in schooling, (mostly OECD) countries yielded the follow- instructional practices for hundreds of mil- of resource de�ciencies. ing conclusions: • The ratio of students to teachers matters lions of children can be changed because a • Money cannot buy quality in present where it is high, while in the typical range new technology appears promising. Or schooling systems. there is a much weaker association with because a group of experts thinks so. Or • Incentives are the key to success. performance. because the practice has been tried in a pilot • Schools should be allowed to decide • Three factors of school policy are associ- program (and subject to “Hawthorne� autonomously on operational tasks. ated with better student performance: • Schools must be made accountable. school autonomy, teacher morale and effects, the nonreplicable impacts that commitment, and other teacher factors occur simply as a result of the increased • Teachers’ incentives have to focus on such as expectations. improving student performance. attention from any innovation). Or because • Competition between schools creates • Three classroom practices reported by it has been shown to be statistically corre- students show a positive association with incentives for improving performance. performance: the extent to which teach- lated with success, subject to all the dangers A second study, the OECD Programme ers emphasize performance, teacher-stu- of improperly inferring causation. There is for International Student Assessment, dent relations, and the disciplinary assessed “young people’s capacity to use strikingly little use of randomized con- climate of the classroom. their knowledge and skills in order to meet trolled experiments as a routine manage- real-life challenges, rather than merely look- • Successful students are more likely to do ment practice—despite its eminent feasibil- homework. ing at how well they mastered a speci�c ity for many classroom practices (box 7.5). school curriculum.�The study collected per- Sources: Wößmann (2003) and OECD (2001). A recent example of evaluating a school- ing innovation illustrates the power of flexi- bility in design—and the power of evalua- type (between 15 and 18 percent), and most tion. A remedial education program, was between teachers in the same school.328 established as a collaboration between the Teacher autonomy in classroom tasks government and a nongovernmental orga- consistently emerges as a determinant of nization (NGO) in two cities in India success (box 7.4). The principles developed (Mumbai and Vadodara), hired local in chapter 6 are apt: discretion and deci- women to teach catch-up classes for stu- sionmaking power need to be delegated to dents who were falling behind. The program those with the relevant information and was inexpensive—$5 a child a year. A rigor- professional skills. Centralized control of ous evaluation based on the randomized teacher assignment and assessment can design of the program found it very effective cause bureaucratic paralysis. But making at boosting learning, especially among schools autonomous in curriculum design, poorer children. The evaluation showed examinations, assessment, and �nance can that, at the margin, extending the program lead to excessive variability across schools. would be about �ve times more cost-effec- tive than hiring new teachers.329 The pro- Innovating, evaluating, gram is implemented now in 20 Indian and scaling up cities, reaching tens of thousand of children. The goal of school autonomy and account- But “there is a particular irony to educa- ability is to create a system in which organiza- tion reform . . . [as] pockets of good education tional providers have strong, sustained incen- practice . . . can be found almost anywhere, tives to improve outputs. The problem is not signifying that good education is not the a lack of innovation—there is a continual result of arcane knowledge. Yet the rate of stream of new modes of teacher training, new uptake of effective practices is depressingly (c) The International Bank for Reconstruction and Development / The World Bank Basic education services 123 BOX 7.5 Randomized experiments in Busia district, Kenya Since 1996 a group of researchers has been pre-test.This suggests that because the treatment for hookworm, roundworm, working with a Dutch nonpro�t (International textbooks in this instance were too dif�cult for whipworm, and schistosomiasis found that it Christelijk Steunfonds) supporting schools in the typical student, the books did not matter. reduced absenteeism by a quarter. Does this rural Kenya to estimate the impact of various mean that health is all that matters? No. While Teacher incentives. Everybody knows that interventions.Through random selection some attendance improved, test scores did not. teacher incentives are crucial since teachers are schools were chosen to implement the Three observations. First, things that every- undermotivated.Yet a study on incentives for interventions �rst, with the other schools to fol- body “knows� to be important did not work as teachers based on student test scores found low.This allowed the researchers to test a num- planned, whereas the intervention with lower that “teachers responded to the program ber of ideas. expectations had large impacts. Second, these primarily by seeking to manipulate short-run results from a hundred schools in an isolated test scores. . . . [T]eachers’ absence rates did not Textbooks. Everybody knows that textbooks area of Kenya have been getting enormous aca- decline, homework assignments did not are important and that their lack is a major con- demic attention because there are so few rigor- increase, teaching methods did not change.� straint on effective instruction.Yet the �rst study ous, randomized evaluations of schooling inter- Does this mean that teacher incentives don’t found “no evidence that the provision of text- ventions.Third, the �ndings from each matter? No.Teachers did change their books in Kenyan primary schools led to a large intervention do not reveal universal, behavior—they “conducted special coaching positive impact on test scores, nor is there any immediately generalizable results, but they sessions and encouraged students to take the evidence that it affected daily attendance, grade reveal that speci�cs matter and that learning test.�This suggests that you get what you pay repetition, or dropout rates.� Does this mean about what works needs to be local to be useful. for—whether you like it or not. that textbooks don’t matter? No. Although text- books did not increase the performance of the Deworming. Deworming does not feature Sources: Miguel and Kremer (2001); Glewwe, Ilias, typical (median) student, they did improve per- widely in the education effectiveness literature. and Kremer (2000); Glewwe, Kremer, and Moulin formance for students who did the best on the Yet a randomized trial of an inexpensive medical (1997). low and effective schools are often found just Accountability is, of course, dif�cult to a few blocks from dysfunctional ones.�330 de�ne. Is it accountability within the The U.S. Agency for International Develop- bureaucracy (so that policymakers choose ment (USAID) attempted to refocus its and replace principals based on perfor- efforts in education in Africa from “proxi- mance)? Is it the direct participation of par- mate determinants� to a more systemic ents or school councils in choosing school approach that focuses on internally driven management? Is it parental choice? identi�cation and scaling up of good prac- There are alternatives. One is to allow the tices. A recent review of USAID projects most competent actors (principals, teach- based on systemic reform found, not surpris- ers) to run more than one school. This ingly, that implementation was dif�cult would allow the more competent to affect because it went to the heart of the relation- greater numbers of children—and reduce ships of accountability among actors in edu- the sphere of influence of the less compe- cation—and that was intensely political. tent. A second way is to systematize a vari- Even so, recent work at USAID explores ety of standard-provision models that are solutions to the challenges of linking author- easy to replicate and franchise, whether the ity, accountability, and transparency to franchise is a bureaucracy or a private strengthen basic education through institu- provider. Franchise models should be based tional reform. There are several ways of on local research on what capable principals expanding and scaling up good practice.331 currently do in a variety of real settings as The most obvious way is to use greater well as on citizen dialogue around the school autonomy—leaving scope for school emerging models. Models could also be management to de�ne a school mission, based on statistical analysis of the maxi- mandate, and tactics—and greater account- mum “output� produced by schools, using ability to enable the monitoring of perfor- the average level of resources that schools mance. The autonomy and accountability can typically mobilize. create incentives to adopt proven successful None of these approaches to learning practices, to evaluate the effectiveness of about learning is possible without assess- homegrown initiatives, and to create a sense ments of outputs—not just standardized of pride and commitment in the school. exams but assessment of all the relevant (c) The International Bank for Reconstruction and Development / The World Bank 124 WORLD DEVELOPMENT REPORT 2004 outputs of schooling. Nor is any possible of income) or too high (where pay is several without enough organizational autonomy times higher than needed to attract a qual- for individual schools or groups of schools ity pool of teachers). But appropriate com- to decide how best to act. pensation involves more than the level of pay. It is the overall attractiveness of the Organizational and frontline profession and the structure of compensa- providers: management tion that motivate performance.332 Teacher Managing for effective services means get- pay is usually linked to factors that show lit- ting people with the right skills and training tle association with student performance— in place (capacity). It means giving them mainly seniority. Teacher earnings thus the right infrastructure and inputs to work exhibit much less variance than earnings of with (logistics). And it means ensuring the workers in other occupations. Compensa- motivation (both extrinsic and intrinsic) of tion should reward good teaching, not just frontline workers. The typical public school longevity. is often handicapped in these endeavors in Motivation and capabilities nearly every possible way. Individual school The schooling process is so complex—the managers often cannot choose their own dif�culties of attribution so severe—that teachers and cannot dismiss them—even simple proposals of “pay for performance� for good cause. Teacher training and capac- for individual teachers and principals have ity building are often ill-designed and rarely proved workable.333 But a total lack of poorly integrated, and so become irrele- connection between incentives and perfor- vant. Logistical issues are beyond an indi- mance allows excellent teachers working in vidual school’s control—with decisions centralized and bureaucratic. Compensa- adverse circumstances and those who never tion structures tend to be tied to seniority show up to be paid the same amount. This and level of education or training, not to undermines the morale of good teachers demonstrated mastery of skills. And and drives them out of the profession. although pay, or other extrinsic motivation, But motivation is affected by more than is not the only motivator for education pro- money, as a study of teachers in three types of fessionals, the typical structure of working schools in Merida, República Bolivariana de conditions and pay undermines even the Venezuela (nonpro�ts, state, and national), intrinsic motivation of providers. shows. Catholic Fe y Alegria schools—which cater to low-income families—emphasize Employment relationship school autonomy and teacher input in deci- and structure of compensation sionmaking. Even though pay is roughly the same as in state and national schools, teacher There is no single best approach to com- satisfaction—and student performance—are pensation, capacity building, and classroom much higher (table 7.3). autonomy. Indeed, one of the major bene- Enhancing teachers’ capabilities is clearly �ts of greater autonomy is that it allows fundamental to good-quality schooling, but more experimentation and more flexibility experience with teacher training is frequently in implementation and replication. With disappointing, mainly because of too little school autonomy, organizations can try dif- transfer from training to classroom practice. ferent compensation schemes, training Teachers need training that lets them do their methods, and modes of parent-teacher job better. But autonomy, motivation, and interaction and can evaluate them relative assessments of providers (based on outputs to output and outcome objectives. If the and outcomes) are needed for training to public sector can specify what it wants from improve outcomes. a school—a clear compact—it can leave teacher compensation to school manage- ment and let the best system win. Client power Teacher pay can be too low (where infla- Client power is a weak force in public tion has eroded real salaries to the point school systems. Channeled into narrow where teachers resort to alternative sources interests, it has little impact. In nearly all (c) The International Bank for Reconstruction and Development / The World Bank Basic education services 125 Table 7.3 Autonomy and outcome in Merida, República Bolivariana de Venezuela, in the mid-1990s School type Cost per student and Salaries as share of Teacher satisfactiona Student performance Retention rate, grades per student hour operating expenses (average of math, 1–6 (percent) (bolivar) (percent) reading, writing, percent) State (escuelas integrales) 190, 24 95 3.75 40 51 National 160, 32 99 3.57 39 42 Private (Fe y Alegria) 155, 31 88 4.02 53 100 a. Rating of 5 indicates complete agreement with the statement, “I’m satis�ed with my work.� Source: Navarro and de la Cruz (1998). countries parents can choose schools for • The emergence of community-run their children, within limits. But choice has schools in El Salvador showed that they little or no overall impact on school quality performed as well in test scores and in stu- because there is typically no effect either on dent dropout rates as schools operated by schools that lose children or on those that the ministry of education, which catered gain them. Even when parents abandon the to wealthier children (see the spotlight on public system and pay for private schools— Educo). as is happening in many countries—little • In Cambodia a donor-�nanced initiative systemic pressure for change is created sought to improve schools by stimulat- because government resources continue to ing greater community engagement in flow into public schools. Direct parental schools and using direct budget transfers participation in schools is also typically a to schools (box 7.6). weak force since there is little about public schools that parents can affect. Often the • Evidence from Argentina supports the idea that parental participation together school head and teachers themselves have with school autonomy raises student per- little or no autonomy to make changes. Par- formance.334 ent organizations are simply a means of mobilizing additional resources for the • NGOs can help both through direct engagement with communities and school. through creating and disseminating infor- There are ways to change this, to use client mation—as in the system of school infor- power to improve outcomes. One is to mation for communities in Nepal assisted involve citizens directly in the assessment and by Save the Children-UK. operation of schools. Another is to use demand-side subsidies to increase access for poor people. A third is to make provider resources depend on client choice—to have money follow students. None is a panacea, BOX 7.6 School improvement in Cambodia but each can be part of a strategy for school To improve school quality, the Education advise the government on how to improve improvement. Quality Improvement Project in Cambodia its education policies.The animators are uses a participatory approach and perfor- supported by a network of technical assis- mance-based resource management. Oper- tants at the local level, who provide peda- Direct participation: community ating in three provinces, the project covers gogical and organizational support. involvement in schools 23 percent of the primary school The project has stimulated lively population. Local school communities iden- dialogue at the school, cluster, and adminis- Since students, and indirectly their parents, tify their needs and make proposals for trative levels on how to improve schools. It interact daily with the education system, they change and investment. Funds are delivered has also set in motion a process of change have valuable information about provider directly to school clusters by the Ministry of in the administration of schooling and in performance that tends to be ignored in Education. teaching and learning practices. As a result, Change management is supported by unprecedented responsibility has been purely bureaucratic systems. Several success- district-based “animators,� who draw gen- devolved to school and local administrators. ful experiences with giving parents a formal eral lessons from the experience with the role in school governance have heightened school’s quality improvement grants to Source: World Bank (2002c). interest in this model: (c) The International Bank for Reconstruction and Development / The World Bank 126 WORLD DEVELOPMENT REPORT 2004 Informal and community schools exist in teaching methods and will encourage teach- many settings, most often when parents take ing to the test when there are gatekeeper matters into their own hands and arrange examinations. Ensuring that the poorest are for teaching outside the formal system. This not excluded from this process is essen- is often supported by NGOs and religious tial—and dif�cult. Experience with school- organizations. A recent review of initiatives based control in South Africa suggests a key in Ethiopia points to the potential of this role for training parent groups: without the support for expanding access to schooling training the more advantaged populations (box 7.7). The big question is how to link bene�ted while poorer and less powerful these efforts to the formal system so that groups lost out. informal schools are not a dead end. Direct participation in schools raises the Greater parental involvement in school dif�cult issue of user fees and their relation- management has its risks. Parents need ship to community engagement. Some access to relevant information and the argue that as long as locally collected fees power to effect change. Their focus should are retained by the school, fees are a good be on performance, not on micromanaging thing, for two reasons. First, empirical the classroom, where teachers should have studies suggest that centrally controlled professional autonomy. It is fairly straight- resources are almost universally devoted forward for parents to assess whether the largely to payroll, while resources collected instructor is present and not abusive to stu- at the school level raise school quality by dents. But high-quality teaching cannot be much more than equivalent resources from reduced to scripted actions. Parents often higher levels.335 A study in Mali showed that have a very conservative perspective on paying fees left parents better off (on aver- age) because the value of increased school quality was much larger than the fee itself.336 Second, if communities are to feel pride in their school and empowered by BOX 7.7 Alternate routes to basic education in Ethiopia their participation, then parents should be expected to make some contribution. Pay- Ethiopia is a large country with a heteroge- neous population. Education levels are low: • Involving community members in moni- ment may come in-kind, such as labor for toring the attendance of teachers and construction of the school, rather than as only 24 percent of children complete primary students. school.There are very few schools in poor direct fees for use. and remote areas: only about 30 percent of • Targeting class sizes of about 35 But these potential bene�ts of greater 10-year-olds in rural areas have ever attended students. school. But recent innovations sponsored by • Recruiting teachers and teaching assis- community engagement have to be weighed NGOs show other ways of getting schools to tants from local areas and paying them against the apparently large negative effects these children. less than professional teachers. on enrollments of even very low user fees in Programs run by six NGOs reveal how expanding school places is possible even in • Spending more on textbooks, other poor countries and against the increases in instructional materials, training, and remote areas—at reasonable cost and with- supervision. inequality from relying on fees (see box out sacri�cing quality.The NGO ActionAid 4.4). Some might argue the ideal is a com- The results are promising. Children proposed adapting school models used by attending these schools continue on to promise of a fairly apportioned fee on com- the Bangladesh Rural Advancement Commit- higher grades. Moreover, learning does not munities to generate ownership but with tee in Ethiopia, and since then several other NGOs have sponsored similar programs.The appear to have suffered.Test scores in the signi�cant exemptions for poor households second grade were about 20 percent higher schools share several features: (subsidized from a central fund). Recent than in government schools, and scores in experience with such targeting (as in South • Compressing four years of the of�cial the fourth grade were only slightly lower, even though the schools catered to Africa) suggests that it is dif�cult to make curriculum into three years. • Streamlining the curriculum to reduce children from poorer families. All this at a this work.337 repetitiveness and remove elements lower cost per student. Issues remain, however—particularly deemed irrelevant to local needs. about scaling up these programs to reach Demand-side transfers • Using instructional routines that appeal more children, the more so since some ini- Many governments use scholarships or con- to children, such as songs or teaching in tial success was driven by a few energetic ditional transfers (households receive bene- groups. individuals. • Scheduling classes on days and times �ts if children are enrolled) to expand approved by the community. Source: Ministry of Education Ethiopia (2000). enrollments. The Education, Health, and Nutrition Program of Mexico (Progresa) (c) The International Bank for Reconstruction and Development / The World Bank Basic education services 127 has drawn considerable attention because— ment support for these schools—as for unusual for this type of effort—it was struc- Catholic schools in Argentina or Islamic tured to allow rigorous impact evaluations. schools in Indonesia—or there is support to The program has resulted in substantially parents who choose private schools—as in higher transitions to secondary school.338 In Chile, the Czech Republic, the Netherlands, Bangladesh a study found that conditional and New Zealand. transfers of rice raised enrollments. The What kind of relationship should gov- program was also cost-effective relative to ernments have with nonpublic providers? other interventions, though the government One decision is whether to allow demand- recently moved to monetize the bene�ts due side transfers or scholarships to be used in to concerns about leakage.339 Indonesia nonpublic schools. Colombia used scholar- introduced a large scholarship program in ship programs for private schools to expand response to the economic crisis in 1997. The enrollments for poor students. The fact that program helped maintain junior secondary participants were chosen randomly from a school enrollments.340 pool of applicants allowed for rigorous Conditional cash transfers have proved impact evaluation, which found signi�cant effective in expanding enrollments, but they positive impacts for scholarship recipi- have shortcomings. They focus on enroll- ents.341 But even though the program was ment without creating incentives for both targeted and apparently effective, it improving quality. To the extent that was discontinued—for bureaucratic and demand-side transfers use funds that would political reasons. A second decision involves otherwise have been devoted to school more generalized support for nonpublic improvement, there is the risk of expanding schools. In general, it is hard to say anything quantity at the expense of quality. There about “choice� without provoking contro- was widespread concern that school feeding versy, but here are four tries. programs in India were “too successful� in attracting students. Schools were flooded General subsidies to private schooling— with underage children not ready for learn- neither disaster nor panacea. Although ing, which put even more pressure on qual- there is a wide-ranging and still inconclusive ity at the critical lower grades. empirical debate about the impact of gener- alized choice, providing general subsidies to Resources and client choice private schooling has never been a disas- The Universal Declaration of Human Rights ter—or a panacea. The Netherlands has had (Article 26) asserts that parents have a “right full school choice among public and to choose the kind of education that shall be denominational providers since 1920, with- given to their children.� Despite this apparent out terrible repercussions. Chile has had endorsement of parental choice, there is little choice since 1980, and while there is some consensus about its role. controversy about whether it has produced In practice, there is a large amount of substantial gains in measured learning out- choice. A substantial fraction of schooling is comes (Hsiao and Urquoila 2002), no one carried out by a range of private providers: argues it has been a disaster. New Zealand has for-pro�t schools, religious and denomina- had school choice since 1991, and in a recent tional schools, NGO-operated schools, and assessment of 32 countries, came in third in community-owned and -operated schools. math and sixth in reading and science liter- In some countries the proportion of chil- acy. The Czech Republic and Sweden have dren in private schooling is rising rapidly— had public �nancing of private schools since even without public support. In Pakistan the 1990s. the proportion of urban students in public So choice is neither an ivory tower schools fell from 72 percent in 1991 to 60 notion that could never work in practice percent in 1996 to 56 percent in 1998—with nor an ideological Trojan horse that would most of the shift to private, nonreligious destroy public schooling. It is also not a schools (religious schools accounted for universal remedy. The successful expansion only 1 percent). Sometimes there is govern- of choice has nearly always been embedded (c) The International Bank for Reconstruction and Development / The World Bank 128 WORLD DEVELOPMENT REPORT 2004 in a more general program of school reform other elements as well. Choice as part of a and improvement. package of reforms can have three bene�ts: Parents who exercise choice perceive themselves to be better off. But schooling • The introduction of choice forces an unbundling of roles. To have effective transmits beliefs and values, which implies choice the government must be explicit a distinction between meeting the collective in its dual role of setting the rules for all goals of citizens for publicly �nanced providers and managing schools as the schooling and satisfying the clients of largest provider. schooling. Parents acting as citizens may want publicly supported schools to encour- • Deciding how to regulate private providers age all children to be tolerant and respectful can force a discussion of the output and of other people’s beliefs. Yet these same par- outcome goals of education that can ents acting as clients may want their chil- improve accountability in all schools. dren to receive instruction in the absolute • And choice often creates new acceptance of correctness of a particular set of beliefs. A assessments for monitoring providers— system that satis�es every individual par- which can be expanded to all schools. ent’s demands as a client might fail to meet the collective goals of citizens for publicly Designing choice around the politics. Pol- supported schools. Doubts about choice icy decisions about choice are intrinsically often arise from the impact of schooling on political. The United States prohibits public socialization.342 But this argument cuts support to schools run by religious organi- both ways: if socialization is chosen by an zations. Cordoba, Argentina, has actively authoritarian government to repress indi- supported Catholic schools.343 Holland vidual or group rights, choice is all the more explicitly supports both Catholic and important. Protestant schools. Rather than being based on the perceived relative effectiveness of the Using taxes for private schools requires different schools, these policy choices seem accountability. While parents should be to reflect differing public opinion at the allowed to choose their child’s education time the decisions were taken—for exam- and create their own accountability, using ple, historical concern about Catholic influ- taxes for private schools requires public ence among the Protestant majority in the accountability. For choice to be effective in United States, a predominantly Catholic creating greater accountability, parents population in Cordoba, and a more even need timely, relevant information. This will distribution of religions in Holland. Simi- not necessarily emerge spontaneously larly, the suppression of Islamic schools in because it depends on comparable assess- some countries and support for them in ments across schools. Policymakers could others, or the decision to ban private publicize that a speci�c school meets mini- schools in Pakistan and Nigeria in the mum standards through easily visible infor- 1970s, has little to do with school effective- mation tools, such as symbols prominently ness. The promotion of choice through displayed in the school. A more sophisti- vouchers in the Czech Republic has been cated approach could involve broadly dis- seen as a reaction to the use of schools for seminated census-like information on out- political indoctrination. puts and outcomes—perhaps normalized If school choice is a political given, an by socioeconomic status. effective school system can be designed around that constraint. If school choice is politically precluded, an effective school sys- Making choice part of a package of reforms. tem can be designed around that as well. The public sector always remains an impor- tant provider, and choice complements reforms to improve the public sector. Advo- Getting reform going cates of school choice emphasize the poten- This chapter is about changing the relation- tial bene�cial effects of competition—for ships of accountability to produce better which there is mixed evidence. But there are educational outcomes by creating the insti- (c) The International Bank for Reconstruction and Development / The World Bank Basic education services 129 tutional conditions for the technically right strongly linked to economic performance. things to happen. But how can institutions But it is much easier to mobilize technocratic be changed? How do openings for reform policymaker support for certain types of get created and exploited? Decentralization edcuation reform (narrow accountability) can create opportunities. Reform champi- than others (pedagogical improvement). ons can emerge from political, business, professional, or parental interests. And Teachers and teachers’ unions. Effective teacher groups can promote—or resist— teachers are the backbone of any educa- change. tional system, but how can the power of teachers be harnessed for educational Decentralization improvement? Some believe that teachers Decentralization can be driven by a desire have too little power, arguing that educa- to move services closer to people. But suc- tional reforms ignore teachers. Followed cess depends on how it affects relation- through, this view can lead to reforms that ships of accountability. If decentralization ignore classroom and school-level realities, just replaces the functions of the central further demoralizing teachers and under- ministry with a slightly lower tier of gov- mining reforms. Others believe that teach- ernment (a province or state), but every- ers, especially teachers’ unions, have too thing else about the environment remains much power and focus exclusively on the same—compact, management, and wages and working conditions (box 7.8). client power—there is little reason to Both sides can marshal empirical evidence. expect positive change. The assumption is Much of the debate stems from the joint that decentralization works by enhancing function of teachers’ unions as profes- citizens’ political voice in a way that results sional organizations, which exist to pro- in improved services. But this could go mote ef�cacy, advance professional knowl- either way on both theoretical and empiri- edge, and advocate views in public policy; cal grounds. Decentralization is not magic. and as agents of collective bargaining, It must reach the classroom. And it will which emphasize resources and working work only to the extent that it creates conditions. greater opportunities for school reform (chapter 10). Reform champions BOX 7.8 Education reform and teachers’ unions in Latin Getting education reform on the agenda is no mean feat, and getting reform politically America supported and implemented is even more Reforms to promote greater parental Education of its intention to transfer teacher dif�cult. While individual parents are pow- involvement, more school autonomy, more colleges to public universities set teachers erful advocates for their children, that does emphasis on results, and changes in the and students at those colleges “rioting in the training, selection, assignment, and streets, breaking windows, attacking police, not necessarily translate into system compensation of teachers are politically throwing rocks, and setting cars on �re� improvement. Educators and progressive explosive—particularly with teachers’ (images the government used to mobilize forces among teachers often emerge as unions. A study of �ve attempts at educa- public opinion against the unions). tion reforms that included many of these Teachers’unions wanted governments to champions of education reform because elements in Latin America in the 1990s address the issues of teachers’wages and they are most acquainted with the prob- found that teachers’ unions opposed nearly working conditions and were concerned that lems inside the classroom and school. But it all of them—emphatically and stridently. decentralization and school autonomy would is much easier to mobilize educator sup- “Teacher’s unions in Mexico, Minas intrude on more familiar relationships and Gerais, Brazil, Bolivia, Nicaragua, and Ecuador negotiations between a centralized school port for certain types of education reforms followed similar strategies in opposing edu- administration and a centralized union. (system expansion, increased resources, cation reform. All used strikes to assert their Even when governments pushed pedagogy improvement, technical curricu- power . . . against unwanted changes.The reforms through, conflicts with the unions lar reform) than others (increased power to disrupt public life, to close down made implementation problematic, since schools and ministries, to stop traf�c in capi- successful reform requires teacher partici- choice).344 tal cities, to appeal to public opinion—were pation. Local or national politicians or tech- familiar actions to them.� In April 1999 the nocrats can also be forces for education announcement by the Bolivian Ministry of Source: Grindle (forthcoming). reform, particularly if they perceive it is (c) The International Bank for Reconstruction and Development / The World Bank 130 WORLD DEVELOPMENT REPORT 2004 In too many countries discussions can reinforce professional ethics and mutual between the government and teachers’ accountability. They can be used to organize unions are no different from discussions teacher input on technical issues of educa- between a large company and its unions. The tional reform, such as assessment, classroom relationship between policymakers and autonomy, student discipline, and teacher teachers’ organizations needs to shift from a training. If unions refuse to take on that role, pure bargaining game to a positive-sum preferring to concentrate on wages and work- game. This is easier said than done. As profes- ing conditions, there are no �rm guidelines sional development bodies teachers’ unions for how reformers should cope with that. (c) The International Bank for Reconstruction and Development / The World Bank spotlight on Educo Educación con Participación de la Comunidad en El Salvador By contracting directly with communities, El Salvador dramatically increased the primary school enrollment of children in poor and remote areas—without reducing the quality of learning. E l Salvador was wracked by civil war throughout the 1980s. Some 80,000 people died—in a total population of roughly 5 million—and many more were wounded and disabled. Income per capita community schools, bearing the cost them- selves and paying teachers when they could. The government seized on this model of community-based schooling as the basis for a formal program that would be �nancially percent of the new students enrolled in Educo schools (�gure 1). By 2001 there were almost 260,000 students enrolled in Educo primary schools, 41 percent of all students enrolled in rural schools—and fell almost 40 percent between 1978 and and administratively supported by the min- more than 100,000 children enrolled in 1983.345 In 1989 the conservative Republi- istry: Educación con Participación de la Educo preschools, 57 percent of all children can Alliance Party won a majority in the Comunidad, or Educo, with the goal of in preschool. national assembly, with Alfredo Cristiani as encouraging the establishment of preschools Even as enrollments increased rapidly, president. Despite contentious negotia- and primary schools, or classrooms in exist- there is little evidence that learning quality tions, a peace accord was signed in January ing schools. suffered. A survey of 30 Educo primary 1992.346 Begun in 1991, Educo targeted 78 of the schools and 101 traditional schools in 1996 The war had severely damaged the edu- country’s poorest rural municipalities (of found no signi�cant differences in average cation system. Communication between 221 urban and rural municipalities). By 1993 math and language test scores among third the central ministry and schools broke the program was expanding to all rural graders in the two types of schools.352 A fol- down, supervision collapsed, and many areas, including many areas formerly under low-up study in 1998 found that grade pro- teachers, viewed by some as government opposition control. But not all of the “popu- motion and repetition were similar across “agents� and by others as agents of social lar schools� established during the war were the two types of schools as well.353 As the opposition, abandoned their posts. By 1988 incorporated into Educo. Some observers innovation matured, the institutional more than a third of the country’s primary claimed there was selective inclusion based arrangements that it introduced took hold schools had closed.347 And by the end of the on political favoritism; others saw not incor- and ensured rapid expansion of school war some 1 million children were not in porating popular schools into a government places and enrollments of poor children, school.348 program as a way of sustaining spontaneous seemingly without a substantial cost in community-based education.350 quality.354 Establishing Educo— Each Educo school (or section within a Education with the Participation traditional school) is operated by a Com- Parent visits to classrooms made much of Communities munity Education Association (ACE)—an of the difference The Ministry of Education quickly identi- elected committee made up primarily of That Educo schools served the poorest of El �ed expanding access to basic education students’ parents—that enters into a one- Salvador’s students, in the poorest areas, and raising its quality as central goals— year renewable agreement with the min- makes these results all the more astonish- both to rebuild national unity and to pro- istry. The agreement outlines rights, ing. How did they do it? Using retrospective mote long-term economic development. responsibilities, and �nancial transfers. The data that allow controls for child, house- Minister of Education Cecilia Gallardo de Ministry of Education oversees basic policy hold, teacher, and school characteristics— Cano, a reform proponent from the “mod- and technical design. Using the money ernizing� wing of the Republican Alliance directly transferred to them, ACEs select, Figure 1 Students enrolled in traditional rural hire, monitor, and retain or dismiss teach- and in Educo primary classes Party, was intent on lessening the distrust between former combatants. ers. Teachers at Educo schools are hired on Thousands of students one-year renewable contracts. Parents are 700 But skepticism was high. The Ministry of Education was not trusted in many parts taught about school management and how 600 of the country and by organized groups to assist their children at home.351 500 Traditional such as the National Association of Teach- 400 ers. Expansion of the traditional education Three-quarters of new enrollments 300 system was viewed suspiciously as a covert Educo succeeded in many respects. From a 200 means of reasserting national control and pilot phase of six ACEs in three depart- 100 building political support in opposition- ments, it scaled up nationally to all of the Educo 0 dominated areas.349 country’s departments by 1993. Rural pri- 1990 1992 1994 1996 1998 2000 2002 During the war many communities had mary enrollments increased from 476,000 Note: Figures for 2002 are estimates. recruited local teachers and established in 1992 to 555,000 in 1995—with over 75 Source: El Salvador Ministry of Education. (c) The International Bank for Reconstruction and Development / The World Bank 132 Spotlight on Educo Figure 2 Educo promoted parent involvement, that teachers were less likely to be absent in Is the Educo model applicable which boosted test scores Educo schools (averaging 1.2 days of elsewhere? Parent visits to Increase in test scores absence a month rather than 1.4 days). Stu- Educo’s achievements might appear idio- classrooms in previous associated with a visit dents in Educo schools were also absent less month syncratic. The end of a bloody civil war that (three fewer days a month) than students in had thrown the traditional education sys- Number of visits Percent increase 6 5.7 6 5.7 traditional schools.356 In addition, Educo’s tem into chaos opened up a unique oppor- more flexible compensation scheme tunity to change the way schools were man- 5 5 resulted in greater variability in teacher aged. Based in part on coping strategies 4 4 3.8 earnings, which suggests that parent associ- during the civil war, El Salvador had a his- 3 3 ations used compensation to motivate tory of community involvement in school greater effort among teachers.357 Offering management. Indeed, the community asso- 2 1.6 2 or withholding future employment itself ciations appear to have worked better in 1 1 was an incentive, and one that ACEs used. places that had prior experience in commu- 0 0 Turnover among Educo teachers was high, nity organization.358 In addition, in the Traditional Educo Math Language which suggests that job loss was not an idle aftermath of the war there was an unusually schools schools threat. large pool of educated people without jobs Source: Adapted from Jimenez and Sawada (1999). (coinciding with the rapid expansion of Converging with traditional university places fueled by opening higher and statistically adjusting for the fact that schools education to the private sector). unobserved abilities of children might sys- Educo’s administration has become These factors suggest that the Educo tematically differ between the two types of embedded in the Ministry of Education, model might not be directly replicable in a schools—researchers found that commu- and Educo has developed into a major different setting. But some lessons are gen- nity involvement explains much of Educo’s schooling model in the country. Aspects of eral. First, with political will it is possible to success. traditional and Educo schools have been change the relationships between the actors Parents are more active in Educo converging. Traditional schools now have in basic education. Second, schools can be schools. And their involvement affects more parent participation in school gover- transformed to work in ways that promote learning (�gure 2). Each classroom visit by nance and management, and are more enrollment, participation, and learning— parents was associated with signi�cantly autonomous with supporting block even for children from the poorest house- higher math and language test scores �nancing. Similarly, the pay packages of holds. Third, getting parents to participate regardless of the type of school. Parents teachers are more similar: Educo teachers effectively in managing schools can help were more active informally as well: they receive the same salaries and bene�ts as overcome some of the potential pitfalls in were more likely to meet with teachers or to teachers in traditional schools. Even so, a the provision of education services—espe- assist teachers in monitoring attendance or key distinction remains: Educo teachers cially monitoring schooling in remote maintaining school furniture.355 are hired (and potentially �red) by parent areas. Fourth, it is possible to scale up small How did Educo and parent involvement committees while those in traditional innovations to have a signi�cant impact on affect test scores? At least part of the story is schools are not. national outcomes. (c) The International Bank for Reconstruction and Development / The World Bank Health and nutrition services 8 Poor people in most countries have the worst • Getting highly transaction-intensive and health outcomes. They are pushed further discretionary individual-oriented clini- into poverty due to ill health. And they are cal services to poor people is most chal- chapter often excluded from support networks that lenging. To influence quality, poor clients enhance the social and economic bene�ts of should have greater power—through good health. Unlike education, health and third-party payments, information, and nutrition outcomes of poor people are pro- greater oversight of health workers and duced by households—with contributions facilities. Organized citizens can exert from many services. And health and nutri- this power by contributing �nancial tion services contribute to other aspects of resources and co-producing and moni- human welfare, such as protecting people toring the services. But insurance market from catastrophic health spending. They failures, asymmetries of knowledge, and should thus be judged by the way they con- conflicts of interest mean that govern- tribute to the poor’s health outcomes, to pro- ments need to invest in purchasing key tecting citizens from impoverishing health services to protect poor households and expenditures, and to helping the poor break foster a pro-poor professional ethos. out of their social exclusion. • Population-oriented outreach services— Throughout history, poor people have standardized services that can include vec- often paid health providers directly. But this tor control, immunization, or vitamin A short route from client to provider is supplementation—are easier for policy- blunted by asymmetries of information and makers to monitor. Even governments conflicts of interest. Another problem: poor with limited capacity can provide these people lack the money for market transac- services—or write contracts with public tions. A variety of market failures—disease- or private entities to provide them. Build- related externalities and fragmented insur- ing coalitions to strengthen poor people’s ance markets—and concerns for equity collective voice is essential to ensure ade- justify public intervention in �nancing quate public resources for those services. health and nutrition services. But govern- ments �nd it dif�cult to monitor the per- • For community and family-oriented services that support self-care—such as formance of health workers, especially information and social support for pro- those delivering highly discretionary ser- moting breastfeeding or safe sex—com- vices, such as clinical care. And since insur- munity and civil society organizations ance-market failures affect everybody, the and commercial networks are often well non-poor often capture public �nancing of placed to provide services close to poor health care. households. Governments can establish Health services are failing poor people partnerships and provide information not because of lack of knowledge for pre- and targeted subsidies. venting and treating illnesses but because health systems are trapped in a web of failed Policymakers need to be accountable for relationships of accountability. To break out health outcomes—which means greater invest- of this trap, service delivery arrangements ment in monitoring and evaluation mecha- can be tailored for three classes of services: nisms that capture disparities in health.359 133 (c) The International Bank for Reconstruction and Development / The World Bank 134 WORLD DEVELOPMENT REPORT 2004 The health of poor people and poor.363 The poor also often suffer from BOX 8.1 Ethnicity Health outcomes improved in the second higher rates of noncommunicable diseases and health half of the 20th century, a trend likely to such as depression and cardiovascular dis- continue in many countries. But hopes for eases in North America or alcohol-related In the United States,according to 2000 data,indigenous Americans an ever-improving trend are fading as ailments in the Russian Federation. Malnu- and Alaskan Natives have a life progress slowed down in the 1990s. At the trition is a double burden: poorest groups expectancy �ve years lower than current pace most regions of the developing have both high rates of malnutrition and that of the general population.Aus- world will not reach the Millennium Devel- diabetes and obesity.364 tralia’s Aboriginal Health Service notes that in 1996 life expectancy opment Goals for health by 2015 (�gure Improving health outcomes for the poor was 20–25 years lower for abori- 8.1). Infant mortality rates are increasing in is a complex task. In addition to income, gines than for their non-aboriginal Central Asia. Under-�ve mortality is on the other household factors influence health out- counterparts.In Chocó,Colombia, where 90 percent of the popula- rise in 22 countries in Sub-Saharan Africa. comes: age, social status, religion, residence tion is of African descent,the male Stunting is rising in many African countries (chapter 1), ethnic background (box 8.1), and infant mortality rate is more than and remains high in South Asia. In 1995, gender—particularly in South Asia. Girls in 90 per 1,000 live births,while the India are 30 to 50 percent more likely to die 500,000 women died worldwide as a result national average is 25 per 1,000.In Guatemala people born in the cap- of complications associated with pregnancy, between the age of one and �ve than boys.365 ital have a life expectancy 10 years mostly in developing countries. The AIDS Maternal mortality depends mainly on health higher than those born in the epidemic is expanding in Africa, India, services while nutrition and under-�ve mor- department of Totonicapán,where tality depend on many other services, such as more than 96 percent of the popu- China, and Russia, along with a resurgence lation is indigenous.The ratio of of tuberculosis.360 Adult mortality rates have education, water, food security, communica- stunting prevalence in indigenous worsened in the Russian Federation and tion, electri�cation, and transportation. The versus nonindigenous populations some of its neighbors. AIDS epidemic has particularly challenged is more than double in Colombia, The outcomes are consistently worse policymakers and providers to look at links Peru,and Ecuador.The 1996 South African Census reveals that despite among the disadvantaged. In low-and mid- with other sectors and focus more on behav- the government’s systematic dle-income countries, under-�ve mortality ior and societal values. efforts,the infant mortality rate was rates are 2.3 times higher among the poorest 5.5 times greater among the black population than among the white �fth of the population than among the rich- Health services can work population,a signi�cantly larger est �fth. Stunting rates are 3.4 times higher for poor people disparity than would be predicted (�gure 8.2).361 The rich fare well in absolute Experience from Brazil, Chile, Costa Rica based on income differences. terms. In Pelotas, Brazil, infant mortality for and Cuba (spotlight), Iran (box 8.2), Nepal, Source: Torres Parodi (2003). the richest 7 percent of the population in Matlab (Bangladesh), Tanzania, and several 1993 was comparable to the average for the West African countries (spotlight) shows Netherlands in 1998.362 that health services, if delivered well, can Communicable diseases, malnutrition, improve outcomes for even the poorest and reproductive ailments account for most groups. A health program in the Gadchiroli of the mortality gap between high- and district in India reduced neonatal mortality low-income countries and between the rich rates by 62 percent. Midwifery services and community hospitals are linked to dramatic Figure 8.1 Reaching the MDGs in health: accelerate progress reductions in neonatal and maternal mor- Trends in under-�ve mortality by region tality in Sri Lanka and Malaysia. In Uganda Deaths per thousand Deaths per thousand and Thailand government efforts changed 250 250 sexual behavior, reducing the prevalence of Sub-Saharan Africa HIV. In low- and middle-income countries 200 200 services promoting oral rehydration therapy led to a decrease in diarrhea-related child 150 South 150 Asia mortality.366 Latin America and the Caribbean Health services also help protect the income 100 100 East Asia of the poor. Locally managed �nancing 50 50 and the Pacific schemes in Niger, contracted-out services in Middle East and North Africa Eastern and Central Cambodia, and insurance schemes targeting Europe 0 0 poor people in Thailand and Indonesia helped 1970 1985 2000 2015 1970 1985 2000 2015 reduce out-of-pocket spending and extended Source: World Development Indicators database (2003). the reach of the safety net among the poor.367 (c) The International Bank for Reconstruction and Development / The World Bank Health and nutrition services 135 Health services, when they work, can also Figure 8.2 Reaching the MDGs in health: focus on poor households contribute to greater self-reliance and social Stunting prevalence among children 3–5 years old, by inclusion of poor people. They have been used wealth group as entry points to broader development activi- Richest fifth ties, as in the Democratic Republic of Congo, Peru where community health �nancing schemes Second richest triggered the emergence of cooperatives to Morocco commercialize agricultural products.368 Middle fifth Turkey But those who need the most often Ghana Second poorest get the least . . . Despite these successes the availability of good Kazakhstan Poorest fifth health services tends to vary inversely with need.369 Poor groups and regions have less 0 10 20 30 40 50 access to sanitation and vector control.370 An Percent analysis of 30 countries shows that the use of Source: Analysis of Demographic and Health Surveys, 1990–2002. health care interventions is consistently lower among people living on less than $1 a day than among richer groups (�gures 8.3 and 8.4). the poor are 39 percent less ef�cient than the rich at translating drinking water, sanitation, . . . pay the most . . . and health services into better health. In Mali, Illness pushes households into poverty, the poor are only 16 percent less ef�cient, but through lost wages, high spending for cata- the gap has been widening over time: while strophic illnesses, and repeated treatment for the availability of inputs is increasing, poor other illnesses. The share of household non- households’ ability to transform those inputs food expenditures spent on health is higher into health is lagging behind. among poorer than richer groups. Patients sell assets to �nance health care, as 45 per- cent of rural patients do in the Kyrgyz BOX 8.2 Making health services work for poor people in Republic.371 Health expenses are estimated to the Islamic Republic of Iran have pushed almost 3 million Vietnamese In 1974 the infant mortality rate in Iran was about 84 percent of rural communities.The into poverty in 1998. Out-of-pocket spend- 120 per 1,000 live births for rural areas and rest were covered by mobile teams. Health ing pushes poor households deeper into 62 in urban areas. By 2000 it stood at 28 for centers offering emergency obstetrical care poverty, but it also pushes households that urban areas and 30 for rural areas. Maternal 24 hours a day (three midwives) and trans- were not poor into extreme poverty. The mortality rates dropped in rural areas from port for referral were created covering 370 per 100,000 to 55 between 1974 and about 20,000 people each. poor seldom enroll in voluntary insurance 1996. Immunization rates, treatment of child Despite the large number of medical schemes and rarely bene�t from compulsory illnesses, and antenatal care increased dra- professionals in Iran, staf�ng these facilities schemes. Even when they do, they still incur matically and are now at comparably high was a challenge because personnel did not signi�cant direct health care costs in the levels in rural and urban areas, although readily accept posting to rural areas.The skilled attendance for deliveries remains health houses were therefore staffed by form of insurance premiums, copayments, lower (75 percent) in rural areas than in female and male health workers known as and payments for noncovered services.372 urban (95 percent). All this, despite the fact behvarz. Selected from the villages where that 76 percent of rural children had no they were to be stationed, with the partici- . . . and lack the power to produce health insurance in 1997 and 56 percent of pation of village authorities, the behvarz rural women were still illiterate. were required to have eight years of formal good health How did this happen? In 1980, after the schooling.Their training lasted two years, Poor nutrition practices, careless handling of Iranian revolution, a new constitution was mainly local, and consisted of on-the- water and waste, and inadequate care for ill- bound the government to provide basic job training with supervisors and peer train- health bene�ts to the “disadvantaged� ers. A simple health information system— ness are major contributors to poor health. (mostazafeen).The most immediate concern the Vital Horoscope—enabled behvarz to Illiteracy, women’s ignorance of health issues, was to increase access to care in rural areas. identify families with child and maternal and lack of decisionmaking power are often Allocations of resources for rural services health problems and link them with health the causes (�gure 8.5).373 Studies of the rich- increased and today are about a third of the services. health budget in the rural regions. By March poor gap in health outcomes show that the 2002 there were 16,340 rural health houses, Source: Mehryar, Aghajanian, and Ahmadnia poor and the non-poor may respond to the each covering about 1,500 people, serving (2003). same level of inputs differently.374 In Senegal (c) The International Bank for Reconstruction and Development / The World Bank 136 WORLD DEVELOPMENT REPORT 2004 Figure 8.3 Poor people use high-impact services less health status to a speci�c course of action. Below Above This makes them imperfect judges of health poverty line poverty line providers. Although responsiveness to Treatment for ARI patients’ needs is often better in the private sector, the technical quality of private ser- Skilled delivery care vices varies broadly from very good to very DPT3 coverage rates bad (chapter 4). The technical complexity of clinical services confers considerable power Antenatal care on providers to influence the nature and Complementary feeding quantity of services they provide—to their Use of oral own �nancial bene�t. This is well illustrated rehydration therapy by recent increases in caesarian sections in 0 20 40 60 80 both high- and low-income countries.376 Percent of population below and above Governments often aim at solving con- $1 poverty line using health services flicts of interest through direct service pro- Source: Authors’ calculations, based on Demographic and Health Survey data, weighted average for 30 low- and middle-income countries. vision with tight administrative control and enforcement of a public ethos. But monitor- Figure 8.4 Richer groups do well in absolute terms ing whether services are actually delivered Poorest fifth Richest fifth and of adequate quality may be dif�cult. India These “bureaucratic� failures are particu- larly high for highly discretionary and trans- Turkey action-intensive services such as diagnosing Indonesia and treating an illness. Absenteeism rates in health clinics is high377 (chapter 1) and Zambia although technical quality of services is Cameroon often slightly better in public than in private services, quality shortcomings are still ram- Egypt pant in the public sector. In Tunisia in 1996 Guatemala only 20 percent of pneumonia cases were managed correctly and 62 percent of cases Philippines received antibiotics inappropriately.378 0 20 40 60 80 100 Governments also address failed insur- Percent of births attended by skilled personnel ance markets by running “free� public hos- Source: Gwatkin and others (2000), Demographic and Health Survey data, pitals. The bene�ciaries of these hospital weighted average for 30 low- and middle-income countries. services are usually the non-poor in urban areas (chapter 2). They use their political Market failures clout to ensure that public spending for and government failures these (expensive) hospitals is maintained— Market failures and a concern for equity call often at the expense of services that could for some government �nancing of health have a real effect on poor people. and nutrition services. One type of market Many health services are private goods, failure is the underprovision of services to and all countries have a private health care prevent or treat individual illnesses that spill market. Most industrial countries started over to the general populace. Another is the with private health systems. In low- and breakdown in insurance and credit markets, middle-income countries out-of-pocket impoverishing people.375 The concern for spending represents a large share of health equity is either a social choice or based on spending even in countries with well- the notion that health is a human right (see functioning public systems (�gure 8.6). And box 2.2). in the last 20 years there has been tremendous Conflicts of interest and the capacity of growth in private provision (often uncon- services to do serious harm also justify gov- trolled) and private spending on health.379 ernment involvement in service provision. Worldwide, richer groups generally Patients �nd it dif�cult to attribute their resort more to the private sector, but the sit- (c) The International Bank for Reconstruction and Development / The World Bank Health and nutrition services 137 Figure 8.5a Poor women do not know much Figure 8.5b Husbands say no to contraception about HIV Richest fifth Poorest fifth Poorest fifth Richest fifth Bolivia Tanzania Turkey Senegal Tanzania Peru Peru Turkey Kenya Mali Bangladesh Madagascar Philippines Indonesia ˆ d’Ivoire Cote Guatemala Namibia Comoros Uganda Bolivia Central African Republic India 0 20 40 60 80 100 0 20 40 60 80 100 Percent of women who reported Percent of women who have knowledge their husbands do not approve of of sexual transmission of HIV/AIDS, contraceptive use, by wealth group by wealth group Source: Authors’ calculations from Demographic and Health Source: Authors’ calculations from Demographic and Health Survey data, 1995–2002. Survey data, 1995–2002. uation differs by country. Richer groups often provided by the public sector. Even in also use public facilities more—which indi- India, poor people, who turn mainly to pri- cates that subsidies are not well targeted. vate providers to treat illnesses, rely on the The public-private mix varies by type of public sector for vaccination (93 percent) service (�gure 8.7). The private sector is and antenatal care (74 percent).380 involved in many critical services, including The boundaries between public and pri- disease control and child and reproductive vate services have blurred. Many govern- health. But immunizations, family plan- ments subsidize privately provided ser- ning, and skilled delivery care are more vices—all high-income countries do and so, Figure 8.6 A public responsibility, but private spending matters By income Selected and regions countries Private (out-of-pocket) Low income Public Georgia Lower middle income Cambodia Upper middle income India Private High income Cameroon (pooled) Egypt Mexico South Asia Bangladesh East Asia Malaysia Middle East Guatemala Sub-Saharan Africa Mozambique Latin America and Caribbean Saudi Arabia Europe and Central Asia Turkmenistan 0 20 40 60 80 100 0 20 40 60 80 100 Percent of public and private expenditures on health as a proportion of total spending Source: WHO (World Health Organization) (1999), National Health Accounts, updated 2002. (c) The International Bank for Reconstruction and Development / The World Bank 138 WORLD DEVELOPMENT REPORT 2004 Figure 8.7 The public-private mix differs between poor and rich, and among interventions Public and private facility treatment rates Public and private facility delivery rates for acute respiratory infection Richest fifth Poorest fifth Richest fifth Poorest fifth Sub-Saharan Africa Zambia Uganda Mozambique Madagascar Ghana Public Private Ethiopia Cameroon Benin Private Public South Asia Pakistan Nepal India Bangladesh Middle East, North Africa Yemen Morocco Egypt Latin America, Caribbean Peru Nicaragua Haiti Guatemala Dominican Republic Brazil Bolivia East Asia, Pacific Vietnam Philippines Indonesia Cambodia Europe, Central Asia Turkey Uzbekistan Kyrgyzstan Kazakhstan 100 80 60 40 20 0 20 40 60 80 100 100 80 60 40 20 0 20 40 60 80 100 Proportion of births attended in a health facility Proportion of children with fever and cough who, within 2 weeks, are seen medically Source: Authors, based on Demographic and Health Survey data 1995–2002. increasingly, do middle- and low-income of public services. Informal payments boost countries, as in the Thai Social Security the cost of “free� maternity care in Bangladesh Scheme, Poland’s social insurance system, to $31 for a normal delivery—a quarter of a and Uganda’s subsidies to not-for-pro�t pro- household’s average monthly income—and viders. Many public facilities charge user fees, to $118 for a caesarian section.384 introducing a market-like transaction in the delivery of public services,381 and poor peo- Applying the framework: ple spend substantial sums to use them.382 classes of services There has also been widespread growth in Chapters 3 through 6 developed a frame- informal payments to public providers in work for analyzing service delivery that iden- Africa, East Asia, and Eastern Europe,383 ti�es two routes for poor people to obtain which represents an informal marketization services. One is the short route, where clients (c) The International Bank for Reconstruction and Development / The World Bank Health and nutrition services 139 exert power over providers by dint of their The task is slightly easier for services that money and ability to enforce discipline. The support self-care by families and communi- other is the long route of public accountabil- ties, such as information to support changes ity from poor clients to policymakers, and in nutritional or sexual behavior. This ser- from policymakers to providers. For health vice is discretionary because it has to be tai- services, the short route often fails because of lored to the family’s social environment, but conflicts of interest and asymmetries in it requires less intensive professional trans- information. Poor people also lack money action than services responding to individ- for market transactions. But the complexity ual illness. Short- and long-route failures of services and the heterogeneity of health are small. needs make it dif�cult to standardize service Some professional services, despite being provision and to monitor performance—a technical and transaction-intensive, are also major “bureaucratic� failure of the long well standardized and less discretionary. route of accountability. Lack of voice of the Such services include those that serve homo- poor in decisions over the use of the collec- geneous needs of a population—such as vec- tive purse is another failure of the long route. tor control, immunization, supplementation Clinical services are highly discretionary with vitamin A, or screening for diabetes. and transaction-intensive, requiring indi- These population-oriented services can be vidually tailored diagnostics and treatments. delivered through outreach to the poor. So one leg of the long route, which requires When health interventions have large exter- the policymaker to monitor the provider, is nalities such as communicable diseases con- dif�cult. Yet failures in the insurance market trol, the short-market-like route is unlikely call for government involvement in high- to work. If technology allows standardiza- cost services, such as inpatient services or tion, this delivery arrangement is then the catastrophic illnesses. But often these gov- prime choice as policymakers can monitor ernment-�nanced, high-cost services bene- performance and tightly control delivery. �t primarily the non-poor. So the other leg Technology is constantly evolving, and no of the long route (voice) also fails. This intervention belongs automatically to one means that poor people have to revert to the category (box 8.3). Service delivery ap- short route of direct client purchases of a proaches are continually being developed service. But the asymmetry of information that reduce the need and dif�culty of moni- between the client and the provider—and toring for both government and citizens. the client’s lack of money—cause this route Countries have standardized highly technical to fail too. A quandary! interventions into less discretionary services BOX 8.3 The changing mix of cure and care: who treats what, and where? Throughout the 20th century, service institutions required. Hospital tuberculosis treatment els of success. Independent practitioners have responded—albeit slowly—to rapid (the sanatorium) was replaced by outpatient developed �rst in Western countries and have changes in health technology. Countries choose clinical care—thanks to antibiotics . Screen- been the cornerstone of Western systems.The combinations of “delivery modes�based on costs ing followed by treatment—DOTS (Directly hegemony of hospitals in the Western world and international standards but also on country- Observed Treatment Therapy)—were later is no older than the 20th century. In contrast, speci�c characteristics such as geographic and standardized to allow delivery through com- hospitals have played a much larger role in density constraints, transport and infrastructure munity outreach. Similarly, new treatments the provision of outpatient care in Eastern capacity, existing health infrastructure inherited for HIV and cancer cut lengthy hospitaliza- Europe and Central Asia, and in Latin America, from previous technological innovations, labor tion requirements. The care and cure Vietnam, and Sri Lanka. In Africa health sys- market characteristics, training and orientation of functions of the hospital also evolve. Hospi- tems developed through hospitals and providers, and so forth.What is delivered as inpa- tals are being transformed into long-term- mobile clinics since the beginning of the 20th tient treatment, outpatient hospital, health center care centers for the elderly, while complex century, with primary health care emerging or home visits; and what by lay people, nurses, procedures are increasingly conducted in only in the 1980s. Different skill levels are also general practitioners, or specialists is far from ambulatory clinics. Home-based nursing care used for similar interventions. Health techni- being standard across countries. is being revived. cians and nurses have been successfully used in Mozambique to perform caesarian • Technological progress triggers modi�cations • At similar levels of technology countries have sections, while other countries use general in the nature, type, and quantity of services opted for different models with comparable lev- practioners or skilled obstetricians. (c) The International Bank for Reconstruction and Development / The World Bank 140 WORLD DEVELOPMENT REPORT 2004 Figure 8.8 Making health services easier to deliver, through standardization and empowerment Complex intervention: high Empowerment discretion, transaction intensive, Standardization high asymmetry of information, heterogeneous needs Clinical services Support to self-care individual-oriented Outreach family-oriented population-oriented Information campaigns, Medical treatment of a Vector control or presumptive Malaria social marketing of insecticide- malaria case intermittent treatment treated nets Community mobilization for Treatment of opportunistic Systematic HIV or TB screening HIV and TB HIV, community DOTS infections including TB Peer support for breastfeeding, Nutritional rehabilitation Micronutrient supplementation Nutrition food forti cation Retail contraceptives, Essential obstetrical care Antenatal screening, Reproductive community information for tetanus toxoid health birth spacing Peer support for home care for Medical treatment of Child immunization Child health child illnesses pneumonia Initiated by the client Initiated by the policymaker that can be initiated, controlled, and deliv- vices—witness the universal childhood ered by the government through reaching immunization, polio eradication efforts, out to the poor. Such interventions can be and the elimination of onchocerciasis. addressed to population groups with simi- Egypt and Mexico have successful itinerant lar needs. Another way is to reduce the clinics. But sustaining these efforts for the transaction intensity of services—and the poor is problematic. Low demand and poor need for government monitoring—by quality reduce effectiveness of family plan- empowering poor people to drive services ning services in India. Governments can that correspond to their needs and charac- control and monitor such services easily, teristics (�gure 8.8). but they often underinvest in them. In tran- Table 8.1 presents some of the key obsta- sition countries in Europe and Asia, moving cles limiting the coverage of the poor with from publicly provided services to social regard to health and nutrition services. insurance led to confusion about the gov- Poor access to information and to networks ernment’s role in health, which led to the distributing health-related commodities is a neglect of these services and the subsequent major impediment to the use of commu- re-emergence of communicable diseases. nity- and family-oriented support to self- Low-income countries have dif�culty care. Rural populations often have no access ensuring physical access to clinical services to retail condoms, soap, water containers or for a large part of their population. In Chad bed nets. If they do, price remains a barrier. only 20 percent of the poorest �fth live less Fewer than 30 percent of the poorest �fth of than one hour from a health facility. Middle- the population have access to the media in income countries have more trouble ensur- such countries as Bolivia, India, Morocco, ing affordable and quality care to excluded, and Mozambique. Few poor households vulnerable, and dif�cult-to-reach segments receive extension workers, with visits peak- of the population. Inequitable risk pooling ing at 20 percent of Indonesian and Peru- still leaves the poor exposed to the �nancial vian households . risk of illness in most of these countries. Most countries have had some success in Studies report large inef�ciencies in the way increasing the physical access of poor com- health facilities are run. In Turkey, a study munities to some population-oriented ser- found that only 54 of 573 general hospi- (c) The International Bank for Reconstruction and Development / The World Bank Health and nutrition services 141 Table 8.1 Selected examples of obstacles for the delivery of health and nutrition services to the poor Dimensions of performance Community- and family-oriented Individual-oriented clinical Population-oriented outreach of health services support to self-care services General Knowledge Demand Low literacy and low knowledge of High demand, but large asymmetry of Low demand for key services caring practices among the poor, information between people with low literacy (e.g., immunization in India) particularly women levels and providers; low access to technical information Empowerment Intrahousehold allocation unfavorable No control over socially powerful providers; Low decisionmaking power of women in to women or children, particularly in no inclusion of the poor in risk pooling public life, particularly in the Middle East South Asia schemes Affordability Low affordability of safe water, food, Low affordability of insurance premiums as High transport and opportunity costs for radio, magazines, bed nets, condoms, well as service fees and impoverishment due the poor even when services are free etc. to catastrophic illnesses Social and cultural Social and cultural factors affect Social distance from providers: social class Information on bene�ts of services not access nutritional and caring practices and ethnic minorities, castes in India, tailored to local values and social norms Western-educated providers in Africa Physical access Poor access to safe water, commercial Low geographical access to facilities in Africa Increased access and demand through networks, and media mainly, but also in poor areas of other regions mobile strategies but large dropout rates Availability of human Norm setters, opinion leaders, Major poor region–rich region and rural-urban Health workers’ training and resources community elders, peers may resist imbalance in quali�ed human resources remuneration modes do not provide change incentives to deliver those services Availability of De�cient markets: no bed nets, Fake, dangerous drugs mainly in Sub-Saharan Often not very sensitive to market consumables condoms, contraceptives, etc., in Africa and East Asia dynamics retail Organizational Inappropriate information channels: Poor facilities amenities, inconvenient opening Often mainly supply driven, even centrally quality e.g., information on HIV and condoms hours, poor attitude of staff planned, with little involvement of in clinics instead of schools or bars recipients Technical quality Noncomparable standardized Broad variations in quality of care in private Standardization allows quality assurance, information, diffusion of erroneous services, poor use of drugs, overprescription but supervision can be inadequate information leading to quality shortcomings Input/technical Technical ef�ciency of provision of Poor input mix: inappropriate investment/ Integration in clinical services is not ef�ciency information/community support often recurrent balance, lack of nonsalary recurrent effective if use of clinical services is low not known inputs and there are no incentives Resource Low level of multisectoral coordination Leakages of drugs, and funds; absenteeism of Vertical program approaches may have management personnel, moonlighting/informal practices opportunity costs by diverting resources of public servants away from other services and creating skewed incentives Governance Supply Capture of subsidies by richer groups Lack of transparency in �nancing of health Financing often dependent on donors facilities, large capture by richer groups in low-income countries Adapted from Claeson and others (2003). tals—public and private—could be consid- speci�c obstacles to service delivery (box ered to be operating ef�ciently.385 8.4). This clearly requires de�ning the With a broad variety of situations and accountabilities of those involved and problems, the key issue is to �nd a balance ensuring that there are suf�cient resources, in the public-private mix to minimize the information, and enforcement mechanisms consequences of both market and govern- to make the relationships work. The mix of ment failures in �nancing and providing client power and government action will services. Increased resources for health ser- need to differ according to the nature of ser- vices will translate into better results for the vices, the country, its institutions, and its poor only if used to address the country- government.386 (c) The International Bank for Reconstruction and Development / The World Bank 142 WORLD DEVELOPMENT REPORT 2004 BOX 8.4 Buying results to reach the Millennium Development Goals Where are the bottlenecks? What would be the cost and impact of removing those bottlenecks? Improving self-care in Madhya Pradesh Cost per capita Percent Scenario 3: Increasing Scenario 2: Increasing Scenario 1: U.S. dollars 100 access to support services use of healthy home Increasing 0.8 practices quality of home 0.6 3 practices 80 0.4 2 60 0.2 Low access 1 0.0 0 40 –5 Low Low –10 initiation quality –15 20 of home –20 practices –25 0 –30 % children % population % children % children Under-five mortality rates living with with access breastfeeding breastfeeding mother or to an assistant exclusively exclusively both parents nurse midwife 1–3 months 4–5 months Outreach services in Ethiopia Percent U.S. dollars 100 Scenario 3: Increasing access Scenario 2: Scenario 1: 0.5 3 Increasing utilization Reducing 0.4 drop-out 80 2 0.3 0.2 60 0.1 0.0 0 40 –5 Low 1 utilization –10 20 Drop out –15 0 –20 % fulfilled % population <15 km % children % children Under-five mortality rates need for of functional immunized with immunized (auxilliary) nurses health clinic BCG or DPT1 with DPT3 Clinical care in Madagascar Percent U.S. dollars 100 5 Scenario 3: Increasing Scenario 2: Scenario 1: equity of access Increasing Improving 4 80 demand for quality of 3 3 3 clinical services treatment 2 60 1 2 0 0 40 1 –5 Low –10 utilization –15 20 Low –20 quality –25 0 –30 % fulfilled % population % respiratory % respiratory cases Under-five mortality rates need for < 5 km of a cases seeking taken to a health clinical staff functional care outside facility or skilled Impact on health outcomes (1/10,000) health clinic the home health worker (reduction from baseline over 5 years) (c) The International Bank for Reconstruction and Development / The World Bank Health and nutrition services 143 BOX 8.4 Buying results to reach the Millennium Development Goals (continued) Budgeting approaches often don’t link the money determines a real-life “coverage frontier� (at 80 preparation of investment and recurrent bud- spent with the expected results.The impact of percent). In Ethiopia the coverage with at least gets that speci�cally support strategies most public funds invested in health services and their one dose of DPT (diphtheria-pertussis-tetanus likely to improve ef�ciency of demand (scenario contribution to the Millennium Development vaccine) in relation to the geographical access is 1), support increased demand (scenario 2), or Goals are thus dif�cult to assess. Budgeting the the main bottleneck, while the dropout between increase access to health services and availabil- contribution of health and nutrition services to the �rst and third dose of DPT constitutes ity of health staff (scenario 3).This enables Min- goals should address three questions: another obstacle to adequate coverage with istries of Health and Finance to determine the outreach services. Increasing the use of immu- spending needed to support the goals and to 1. What are the major bottlenecks hampering nization services through information and com- encourage policymakers to ensure funding of the delivery of health services, and what is the munication (scenario 2) as well as defaulter priority activities that address the major cover- potential for improvement? tracking (scenario 1) can increase the adequate age bottlenecks. The analysis of the determinants of the baseline coverage to over 80 percent. In Madhya Pradesh coverage of interventions in terms of availability 3. How much can be achieved in health in India, late initiation and non-exclusiveness of of critical resources (human and material), physi- outcomes by removing the bottlenecks? breastfeeding (BF) are the main bottlenecks cal accessibility, demand (utilization of the ser- identi�ed in self-care. Increasing the use (early The increase in coverage in speci�c vices), and continuity and quality of health ser- initiation of BF, scenario 2) and quality (exclusive- interventions achievable through well-targeted vices helps identify the main constraints to the ness of BF, scenario 1) of healthy home practices strategies and budgets such as those de�ned increase in coverage. through information and the creation of a sup- above can be converted into measures of In Madagascar the bottlenecks of clinical care portive environment can increase effective cov- improved health outcomes using various epi- are mainly low quality, utilization, and human erage to nearly 80 percent. demiological models. Preliminary analysis can resource availability. Increasing the demand for show the reduction in under-�ve mortality rate clinical services (scenario 2) especially the use of 2. How much money is needed for the expected that can be expected from removing the bottle- health facilities or skilled health workers for the results? necks of clinical care, outreach services, and self- treatment of severe respiratory infections (sce- Once coverage bottlenecks and the potential for care in Madagascar, Ethiopia, and Madhya nario 1) could raise the effective coverage of this improvement (coverage frontiers) have been Pradesh, India. intervention to almost 60 percent.The potential identi�ed for each mode of health services for improving the availability of health staff in delivery, the cost of the strategies can be calcu- poor rural areas (scenario 3) is limited, and this lated using country-speci�c data.This allows the Source: UNICEF and World Bank (2003). Strengthening client power Paying for services confers power These are our people. This is our money, you can- The poor often �rst use the commercial sec- not touch it. tor to purchase key commodities for President of Benin’s Health Committees, improving health. In Colombia and Mexico responding to the Ministry of Health’s community cooperatives distribute insecti- attempt to centralize community-owned cide-treated bed nets. In more than 40 coun- health funds and return them to the Treasury tries social marketing programs have relied on market incentives. They have increased Strengthening client power can improve the use of bed nets, condoms, contracep- services for the poor by substituting for or tives, soap, and locally produced disinfectant correcting the weaknesses of the long route for water treatment (which has reduced the of accountability. Throughout the world, risk of diarrhea by 44 to 85 percent).387 poor people are engaged as purchasers, co- Modest copayments can also provide an producers, and monitors of health services. entry ticket to clinical services for poor peo- For poor clients to have more control over ple by reducing capture of supposedly free health providers means: services by richer groups (box 4.4). Con- trolled studies in several countries388 �nd • Making the income of health service improvements in the use of services among providers depend more on demand from poor people after copayments increased the poor clients. transparency and accountability of pro- • Increasing the purchasing power of the viders to poor clients (�gure 8.9). But to be poor. pro-poor copayments need to be retained • Fostering the involvement of poor people locally and tied to performance, and they in co-producing and monitoring services. need to contribute to the income of pro- • Expanding consumer power to use com- viders rather than compensate for inade- plaint and redress mechanisms. quate public funds (chapter 4). (c) The International Bank for Reconstruction and Development / The World Bank 144 WORLD DEVELOPMENT REPORT 2004 Figure 8.9 Community-managed Yet for many services the purchasing A problem with subsidies is how well health services increase utilization and reduce spending power of poor people remains insuf�cient they reach the poor. Thailand’s low-income Randomized control study of three to overcome price barriers (table 8.2). Mar- insurance scheme has used demand-side districts in Niger, 1992–94 keting programs for condoms are unlikely subsidies—health cards—for clinical ser- Percent increase in number of to be pro-poor in the early stages, and a vices, but about a third of the bene�ciaries curative consultations focus on cost recovery excludes the were not poor and half the poor did not 120 poorest.389 Governments can then provide bene�t.398 Demand-side subsidies also have Poorest 25% subsidies, as is often done for food to limits if provided in isolation. Even well- 80 address malnutrition. The subsidies may targeted food-related transfer programs sel- Average population need to be very high to substantially dom have measurable impacts on malnutri- 60 increase use among poor people, as demon- tion unless accompanied by programs to strated in the programs to distribute free promote breastfeeding, complementary 40 condoms to sex workers390 and to offer bed feeding, hygiene, and care of childhood ill- nets as part of antenatal care.391 Market seg- nesses. In Mexico the success of the Educa- 20 mentation, tier pricing, and product differ- tion, Health, and Nutrition Program (Pro- entiation can be helpful. In Malawi, highly gresa—see spotlight) was made possible by 0 subsidized bed nets for pregnant women a parallel program of itinerant health differ in color and shape from regular mar- teams.399 –20 ket-priced nets.392 Cross-subsidizing pre- ventive and maternal and child health ser- Coproducing health services Control User fees Taxation vices using the margins on fees for adult Self-care is a particularly important type of with lower clinical care made the �rst more affordable service co-production, relied on by poor user fees in Bolivia.393 and rich alike, and more common in indus- Percent change in health expenditures Direct transfers to client households— trial countries than the use of professional per episode demand-side subsidies—can also boost services. Support to families and communi- 20 client power. Vouchers have a good record in ties can help poor communities reduce mal- promoting use of some well-de�ned ser- nutrition, as in East Asia where community- 0 vices. Food stamps can increase food con- based programs were linked to service sumption.394 In Honduras and Nicaragua delivery structures, often village outlets for – 20 families receive �nancial stipends under the primary health care. Government employ- condition that they use key preventive health ees were trained as facilitators of nutrition- –40 services.395 Financial support through relevant actions coordinated and managed vouchers for consultations reduced sexually by volunteers selected by local communities. –60 Control User fees Taxation transmitted diseases for sex workers (box In Honduras the strongly community- with lower 8.5). And poverty funds ensured third-party owned AIN-C program reduced severe user fees payments for the poor in China.396 How the malnutrition by 31 percent. Civil society Source: Diop, Yazbeck, and Bitran (1995). subsidies are managed matters, however. In organizations can serve as intermediaries Tanzania, the poor pregnant women targeted between clients and providers. Women’s by the bed net program almost never used support groups have helped spread the prac- the vouchers, probably because the subsidies tice of exclusive breastfeeding in Africa and were too low.397 Latin America. Other successes include greater use of oral rehydration solutions, better environmental health, and better Table 8.2 Affordability remains a problem for the poor AIDS interventions.400 Ratio (in percent) of average annualized price of health services to income Communities are often involved in build- Russia South Africa Brazil Tanzania ing health facilities. But they have also taken Poorest 11.9 34.6 20.6 4.4 over provision and management of profes- Second poorest 5.3 20.3 11.6 2.7 sional health services, as in the more than Middle 3.3 13.3 7.2 2.0 500 community-led health clinics in Peru, Second richest 2.1 9.2 4.1 1.3 which cover more than 2 million people.401 Richest 0.8 3.2 1.2 0.5 Cooperative pharmacies have sprung up in Haiti, Nigeria, and Singapore. The availabil- Sources: Authors’ calculation based on Living Standards and Development Survey from Brazil (1996); South Africa (1993); Tanzania (1993–94); Uganda (1999); and Russia Longitudinal Monitoring Survey (1998). ity of high-quality drugs has been improved (c) The International Bank for Reconstruction and Development / The World Bank Health and nutrition services 145 through client-controlled revolving funds in Mongolia, Vietnam, and West Africa and BOX 8.5 Vouchers for sex through Ethiopia’s special pharmacies. Par- workers in Nicaragua ticularly resilient to crisis, these efforts can A voucher program has been operating in voucher.The incidence of sexually transmit- be a source of stability and sustainability. In Managua, Nicaragua, since 1995, with the ted diseases among sex workers who used the Democratic Republic of Congo zones de objective of increasing the use of reproduc- the vouchers dropped by 65 percent in the santé continued to provide immunization tive health services by female sex workers. �rst three years of the program. Vouchers through community-managed services after The vouchers entitle the sex workers to free are now available to the clients of sex work- services at selected for-pro�t, NGO, and ers and their partners. more than two decades, without state or public clinics.The clinic turns the voucher in donor support.402 to the voucher agency, which reimburses Client �nancial and managerial control the provider at an agreed-upon fee per Source: Gorter and others (1999). over health staff—the power to hire and fire—can help ensure that services are actually delivered. In Mali user associations hire and pay professional health staff serv- provision. The transparency of direct com- ing more than 6 million people (spotlight munity control is lost. Scaling up would on the Bamako Initiative). There, as in then be the kiss of death. Peru, provider attendance is high, with strong local ownership of community cen- Information and monitoring ters. But the approach can have problems. One of the strongest levers for strengthen- The relationship between communities and ing client power is information, a critical health personnel can become antagonistic, instrument for changing self care behaviors. requiring state intervention to arbitrate Under government leadership in Thailand, disputes. And the right of local health messages on AIDS were broadcast every boards to hire and �re staff may mean little hour during the peak of the epidemic. But in countries facing severe human resource behavior change is possible only when shortages. In Zambia during the health grounded in an understanding of cultural reform most districts hired the staff already norms and the links between behavior and working in the health facilities. In Mali the disease—and when local actors are shortage of nurses made recruitment dif�- involved. Community-based nutrition pro- cult, loosening the control of clients over grams have often improved nutritional sta- frontline providers.403 tus through information exchanges, as in Solidarity mechanisms to protect Tamil Nadu, India. In addition to giving socially and economically disadvantaged women greater control over resources, the groups by pooling �nancial resources can microcredit programs of the Bangladesh strengthen communities in their negotia- Rural Action Committee also increased tions with health care providers. Micro- knowledge.407 insurance schemes for health have been Clients can exert more leverage on linked to such traditional modes of self- providers by participating in decisionmak- organization as rotating savings—as with ing and by monitoring some aspects of ser- abotas in Guinea-Bissau and traditional vice delivery. Community monitoring of community schemes in the Philippines.404 health service performance in Bolivia and In Germany and the Republic of Korea, Vietnam raises community awareness of small, not-for-pro�t insurance societies key demand-side barriers and helps ensure became independent purchasers of health that services meet community needs. Civil services, contracting with public and pri- society representatives are also part of vate providers.405 The proximity of these health boards in New Zealand. These schemes to their insured members allows approaches increase the transparency of effective monitoring. But the risk pool’s management. But these health boards also small size makes them very vulnerable. Re- remain vulnerable to capture by local elites insurance can help,406 but the added com- unless institutional mechanisms ensure the plexity can lead to the same managerial and representation of disadvantaged clients governance problems in large-scale public (chapter 5).408 (c) The International Bank for Reconstruction and Development / The World Bank 146 WORLD DEVELOPMENT REPORT 2004 Since ancient times, city councils have been BOX 8.6 Making health insurance work for poor people involved in curtailing the spread of plague, cholera, and leprosy. But the state as the main Copayments are not always regressive (see insurance, with egalitarian bene�ts. But this chapter 4), and prepayment and insurance is often dif�cult to implement in developing actor in health emerged only in the 20th cen- are not inherently pro-poor. In Tanzania’s and transition countries with weak taxation tury, starting in centrally planned systems— Community Financing Funds, poorer capacity (in China the proportion of gross the Soviet Union and others—and then groups’ copayments subsidized care for domestic product collected by the state fell spreading to Western Europe after World War wealthier insured groups.The poor enroll in from 30 percent in 1980 to 10 percent in voluntary health schemes in small numbers, 1999).Taxes in low-income countries are II, with the government as unchallenged as in Cameroon, Ghana, and Rwanda—and also often regressive. leader of the formal health system. Many in smaller proportions than richer groups, as An alternative is for governments to developing countries have followed, with in the Democratic Republic of Congo, subsidize insurance for the poor, using Guinea, and Senegal. Employment-based income-related sliding scales for premiums health services being part of the social con- social insurance systems have been and copayments. Income-related contribu- tract between the (often) newly independent observed to bene�t richer groups most in tions were successful in promoting equity in states and their citizens. The model was not Latin America and Africa. A review of Asian Israel and in the Gomoshasthaya Kendra always successful—witness the remaining community �nancing schemes concluded community �nancing scheme in that equity was low. Matching grants for Bangladesh. Governments can also develop health gaps between poor and rich in middle- insurance premiums are often captured by speci�c schemes for the poor, as was and low-income countries. But some coun- the local elites, as in Thailand during the successfully done in Indonesia and tries—Brazil, Malaysia, Sri Lanka, and Costa early days of the health card.The grants Thailand. Both approaches require some Rica (chapter 1)—have done better than oth- sometimes simply lead richer groups to means testing and the adequate and timely substitute private care for public services. compensation of providers. And leakages ers in making the long route of accountability One way for governments to provide are to be expected. work for poor people by: income protection to the poor is through progressive taxation or payroll-based social Source: Preker and others (2001). • Channeling collective resources to poor people through multiple allocation mechanisms that combine targeting dis- Enforcement and regulation eases of the poor, poor regions, service Legal and other dispute mechanisms— delivery providers close to poor people, ombudsman services, the judicial system, and and vulnerable groups and individuals. pressures on professional associations—can • Developing coalitions that bring poor enforce provider accountability. But there is people into the policymaking process little proof that such mechanisms work in through both elections and advocacy by favor of the poor. A study in the United States civil society. in 2000 found that 97 percent of people with • Establishing accountability for progress negligent injury did not sue. Better-off clients on outcomes, particularly among poor went to court, while poor clients did not. In people through increased information India women from lower castes were less and citizen monitoring of health services. likely to use consumers’ courts. Seventy-six percent of Indian doctors believed the Con- Which services and for whom? sumer Protection Act to be only moderately How do we know if voice works for poor effective because of weak enforcement and people? There is no conclusive evidence that resistance from professionals. Sanctions are either one of the main resource-generation often a weak threat, because health providers mechanisms for health services emanating tend to protect one another, and accountabil- from collective action—social insurance ity channels do not favor poor clients. Ram- (Bismark model) or general taxation (Bev- pant corruption in the legal system of many eridge model)—works better for the countries also works against poor people.409 poor.410 To be pro-poor they both require some level of cross-subsidy—through Strengthening poor either differential premiums or progressive citizens’ voice taxes. More than prepayment, third-party payment—whether through insurance or Simply don’t vote for your mayor . . . if he doesn’t provide you access to your health program. other solidarity funds—is what makes the Managers of the Health Agent Program difference for poor people (box 8.6). that reduced infant mortality There is also no clear answer on whether in Ceará, Brazil providing universal access to a limited set of (c) The International Bank for Reconstruction and Development / The World Bank Health and nutrition services 147 Table 8.3 How do we know whether poor people’s voices have been heard? Key questions Evidence Problems Is there a constitutional or legislative commitment • Most countries have constitutions that express • Lack of prioritization or targeting gives the non-poor to guarantee some level of health service to all? some commitment to universal access or rights to incentives to seek more services or shift from health care. Few have an expressed commitment to private to public providers. the poorest segments of society. Rights-based • The distribution of health resources is influenced arguments such as the Special Session For Children by more vocal and often urban populations and is of the United Nations General Assembly can provide concentrated in a highly visible urban hospital legal grounds for claiming access to services. infrastructure, as in Nepal. With the exception of a • With the exception of Mexico, Turkey, and the few countries (e.g., Costa Rica, Malaysia), public United States, all OECD countries today offer their spending on health care often disproportionately populations universal protection against the cost bene�ts the non-poor (chapter 2). of illness. Are the diseases that affect the poor priorities for • Cost-effective packages of services addressing the • Cost-effective health interventions have also been public action? overall burden of disease often overlap with the captured to a large extent by richer groups. diseases that affect the poor most. • Epidemiological pro�le often polarized between rich and poor. Are services that are close to the poor given • Bene�t incidence studies show primary health • Primary care also bene�ts richer groups more in priority? care to be signi�cantly more pro-poor than many countries. hospital-level care. Do disadvantaged areas bene�t as much or more • Population and needs-based funding increase the • The wealthiest areas often receive larger than richer areas? health funding for poor groups. Brazil led richer government subsidies than poorer regions, as states such as Parana to have their share reduced documented in Bangladesh, the Kyrgyz Republic, to the bene�t of poorer states in the Northeast. Mauritania, Mozambique, Pakistan, and Peru. • Political resistance to equalization. Do children, women, and the elderly bene�t from • Thailand’s exemption policy for children and the • Poor children and poor women may not be reached. public services? elderly has been largely successful. Mozambique’s And services leave out poor male adults, whose exemption for treatment of illnesses in children at welfare indirectly affects the welfare of children and the primary care level has been partially successful: women. 65 percent of children were exempted. Do individual households bene�t from speci�c • Exemptions can be effective when funds are • Exemptions have a poor track record of serving the protection measures? available to compensate providers, as in the insurance poor, often bene�ting civil servants and their families. fund in the Kyrgyz Republic or the Type B scheme in • Assessment of individual targeting left to individual Thailand. Ghana’s program of fee exemptions for the providers, which generates a conflict of interest. poor was initially successful, then faltered when providers were not compensated. Sources: Pearson and Belli (2003) and Soucat and Rani (2003a). services or targeting poor people is the way to information asymmetry between citizens and go. The debate is often not in these terms. Most politicians. But decentralization has had Organisation for Economic Co-operation and mixed results in health.411 It has not always Development (OECD) countries use both meant increased resources for poor areas. approaches. Many middle-income countries Transferring the provision function to local are moving gradually, developing multiple pro- governments has often overwhelmed them, grams to protect children, women, and the leaving them with little capacity and incen- poor, as in Colombia, Indonesia, Iran, and tives to develop the policy function and Turkey. These programs may eventually encourage citizen oversight. The transfer of merge into a national system of universal cov- ownership of assets—hospitals and clinics— erage, as happened in industrial countries and to local government has also created incen- recently in Thailand. Geographical, age, and tives for rent-seeking by local elites. In individual targeting—despite leakages—often Uganda allocations to health services reach the poor. But a combination of mecha- declined when districts received responsibil- nisms seems necessary, and each country has ity for service delivery, personnel manage- to assess whether mechanisms in place are ment, and allocation of health resources. successful (table 8.3). Spending on primary health care fell from 33 Decentralization has often been imple- percent to 16 percent during 1995–98, and mented with the hope that it would better the use of maternal and child health services align spending with local needs, reducing the declined signi�cantly412 (chapter 10). (c) The International Bank for Reconstruction and Development / The World Bank 148 WORLD DEVELOPMENT REPORT 2004 Pro-poor coalitions An altruistic vision can nonetheless motivate With widespread capture of health services coalitions between providers and citizens for by the non-poor, building pro-poor coali- better services. Associations of rural doctors tions to influence health spending is critical. in Thailand and midwives in Guatemala The democratic process conveys what poor have promoted alternative modes of health citizens value in health and health care. The service delivery that better reach the poor. extent of health bene�ts and the �nancing of health services are an electoral issue in Information and monitoring to industrial countries—and, increasingly, increase accountability for outcomes everywhere. In Thailand access to health ser- Creating policymaker accountability to citi- vices was a key plank in the political platform zens for delivering health services is a dif�cult of the Thai Rak Thai party, and its proposal task. A patient dying while waiting in line at a for universal health insurance, with a single public hospital makes front-page news. But 30-baht copayment, was eventually imple- the thousands of children who die of treat- mented.413 Eliminating hunger was a major able or vaccine-preventable diseases do not theme in the recent elections in Brazil. get equal time. Politicians �nd it easier to Beyond the ballot box, poor citizens can claim success for building a hospital and pro- communicate their preferences through viding employment to nurses and doctors national consultations, such as those for than for reducing malnutrition among a poverty reduction strategy papers (chapter nomadic group. Yet with a web of market and 11). A review of the consultation process for government failures affecting those services, 25 poverty reduction strategy papers shows policymakers are uniquely placed to create a that the poor care about health services, espe- vision conducive to better outcomes. They cially access, price, and social distance. In the need to be accountable for equitably distrib- state of Oregon in the United States, tradeoffs uted health outcomes, protecting citizens in bene�t coverage are discussed in consulta- against impoverishing health expenditures tions with the public. In the Netherlands the and helping the poor escape their social Dunning Report, proposing criteria for exclusion, and not for the provision of grow- rationing health services, relied heavily on ing quantities of services. citizen consultations. Citizen involvement in Better informed and educated citizens can budgeting is now an avenue for more voice in make politicians more accountable. Civil health service delivery, as in Porto Alegre, society organizations can bridge the asym- Brazil (chapter 5).414 metry of information between poor citizens Advocacy by civil society organizations and policymakers. They can bring commu- can put the interests of groups forgotten or nity participation into research, to ensure discriminated against on the agenda, as in the that the perspectives of poor people influence campaign and court action to ban the use of policy. Monitoring the performance of gov- quinacrine for chemical sterilization in India. ernment policies, with report cards as in Ban- Such organizations can build coalitions for galore and Ukraine (chapter 5), can work the interests of the disadvantaged. Pressure well, particularly when income-disaggregated through self-help groups and NGOs has trig- data are available. gered a productive dialogue on the public Civil society can also serve as a watchdog. response to AIDS in the Philippines and A Belgian health publication set in motion six South Africa. royal decrees of protective legislation by Yet despite years of endorsement of par- drawing attention to health hazards. In ticipation, most health systems still resist Bolivia a census-based, impact-oriented greater involvement by civil society in influ- approach combines pragmatism with quanti- encing health care allocations. Reforms in tative and qualitative approaches.415 But gen- the health sector often engender opposition erating information to support pro-poor by powerful unions and professional associa- coalitions is a challenge. Epidemiological sur- tions, which have a much stronger power veys may neglect the needs of peripheral base than poor citizens and can �ght to social groups, such as minority groups or maintain the status quo, as in New Zealand. those affected by less common diseases. (c) The International Bank for Reconstruction and Development / The World Bank Health and nutrition services 149 Compacts: provider incentives to serve the poor BOX 8.7 The government as active purchaser of health outcomes through strategic contracting As long as men are liable to die and are desirous to live, a physician will be well paid. Community nutrition projects in Madagas- borhoods selected by the communities and Jean de La Bruyère, 1645–1696 car and Senegal contracted out nutritional supervised by physicians hired by the pro- services.Transaction costs for introducing jects. Open tendering was used to select the and maintaining the contracts were 13 per- supervising NGOs in Madagascar. Contrac- Even when policymakers truly care about cent of total project costs in Madagascar tual agreements speci�ed the services to be health services for poor people, it is not easy and 17 percent in Senegal. Both projects provided and the number of bene�ciaries to translate policy into reality. Making the were well targeted to the poor communities. to be served, monitored monthly by the compact between policymakers and providers Activities included growth monitoring, user community and the project. health and nutrition education services for Childhood malnutrition declined signi�- work for the poor implies that governments: mothers, food supplementation for mal- cantly in both projects. An evaluation in nourished children, and referrals to health Senegal after 17 months of project imple- • Benchmark performance for services centers and home visits when necessary. mentation showed almost zero prevalence they can monitor easily; In Madagascar the project also had a of severe malnutrition among children aged • Foster autonomous providers for clinical social fund for income-generating activi- ties, and in Senegal an effort to improve 6–11 months and a reduction in moderate malnutrition from 28 percent to 24 percent services; access to water. among children aged 6–35 months. • Establish a strong monitoring function. The frontline providers in Madagascar included women from the targeted neigh- Source: Marek and others (1999). Buying results Widespread de�ciencies in the technical quality and ethics of frontline providers serv- population-oriented services that can be ing the poor—whether public or private— standardized. Explicit contracts have proven reveal an incentive problem.416 The solution effective in serving poor, hard-to-reach lies in some form of compact between the groups.418 Output-based fees for services can policymaker and provider to align the be paid to providers when quantity and qual- provider’s incentives—already acknowledged ity can be de�ned, thus contributing to by La Bruyère in the 17th century—with the improvements in productivity. In Bangladesh policymaker’s wishes. How countries can cre- NGO �eld health workers were paid on the ate incentives to make autonomous or dis- basis of their clients’ knowledge of oral rehy- persed health service providers accountable dration therapy. for outputs and outcomes depends on the Implicit contracts with focus on speci�c nature of services and the capacity to create outputs can also work, as demonstrated by accountability for public objectives through the universal childhood immunization cam- purchasing and regulation.417 paigns of the 1980s, and the Vitamin A sup- Outcome-based contracts are dif�cult to plementation and itinerant health teams in implement because health outcomes are Egypt, Indonesia, and Mexico.419 Malawi and often slow to change, dif�cult and expensive Uzbekistan have achieved immunization to measure, and affected by multiple factors coverage of more than 90 percent among the other than health services. Such contracts are poor with very low per capita spending. But especially dif�cult to write when outcomes Bolivia, Guatemala, and Turkey, with higher are linked to a variety of services, both profes- levels of spending, are much less successful sional and nonprofessional, as in efforts to in providing equitable coverage (�gure 8.10). reduce under-�ve mortality or HIV preva- Performance-based contracts, whether lence. But the experience of Madagascar and directly with health providers or with pur- Senegal with nutrition programs for the poor chasers or insurers, need to align money with shows that it is possible, at least for malnutri- intent, taking into account the variations in tion (box 8.7). effort required to produce a given output in Less dif�cult to implement are output- poor and disadvantaged regions and in based contracts that specify criteria for better-off communities.420 Costs of services disbursing public subsidies—on the basis of can be much higher in remote rural areas, increasing immunization rates, for example. and broad variations have been found in the Output-based contracting is particularly suc- cost of immunization between regions in the cessful for easily monitorable, single-product, same country.421 (c) The International Bank for Reconstruction and Development / The World Bank 150 WORLD DEVELOPMENT REPORT 2004 Figure 8.10 High spending does not ensure more equitable immunization providers the responsibility for rationing Ratio of poor to rich for DPT3 immunization services, opening the possibility for cream- 1.2 skimming. Providers can lower their stan- Uzbekistan dard of care, deny service, or insist on addi- tional informal fees (box 8.8). Egypt Equitable service distribution then Malawi Zimbabwe Kazakhstan Namibia requires providers to have both the techni- Kyrgyz Republic Nicaragua cal capacity to inform the rationing process Zambia Ghana Peru Brazil and a pro-poor ethos to make this process 0.8 Kenya Benin Philippines Colombia bene�t the poor—a combination often dif- Tanzania Bangladesh �cult to achieve. High- and middle-income Vietnam Dominican Republic Indonesia countries are therefore increasingly delegat- Burkina Faso Guatemala Paraguay Turkey Uganda Morocco ing the undertaking of clinical service con- Mali Bolivia Cambodia Cameroon tracts to an autonomous—often paras- Côte d’Ivoire Guinea Mauritania India Haiti tatal—social insurance organization that Togo Madagascar ˆ 0.4 negotiates agreements for services to be Niger Pakistan Central Mozambique provided. They rely on more sophisticated Ethiopia African Republic contracts and monitoring systems using Nigeria complex payment methods: for example, Chad case-based payments in Georgia or diag- nostically related groups in Hungary. 0 Identifying poor target groups and 1 10 100 1000 monitoring results are also more dif�cult Per capita public expenditure on health (US$)—log scale for clinical care, because needs are more Source: Authors’ calculations using Demographic and Health Survey data and World Bank data. dif�cult to de�ne. Cambodia’s contract, which included speci�c provisions to increase clinical services for the poor, had Performance benchmarks for more dis- mixed results (spotlight on Cambodia). cretionary or multitask services, such as Incentives for treating those most in need clinical care, are more dif�cult to establish. can be strengthened by calibrating pay- Which services for whom, when, where, ments to providers on the basis of impact how, and how much are dif�cult to specify on health outcomes, with higher payments in advance. Effort and quality are not read- for emergency obstetric care, as in Burkina ily veri�able. Providers have more gaming Faso, or with higher payments for treating space.422 Targeted payments to increase cer- the poor, as in the Kyrgyz Republic (box vical smears in the United Kingdom led to a 8.9). To exempt the poor from fees, how- short-term increase in the number of smears. But in the long term providers reor- ganized their activities so that they would continue to receive the payments without delivering the services.423 BOX 8.8 The risks of Clinical service contracts also take con- capitation payments siderable capacity to write and enforce.424 In 1999 in Poland, Mrs. K. contributed a health Purchasing based on output—per visit, per premium amounting to 7.5 percent of her salary, case, per hospital day—causes the quantity but was not able to see a public-private of clinical services provided to rise (some- insurance doctor “free of charge� even once. Sev- eral times she tried to get an appointment with times mainly among better-off groups) and the doctor in the outpatient clinic where she the costs to escalate. Health reforms often signed in. But she was told every time that “there try to contain costs by establishing �xed- are no tickets� for that day. In the end she went price contracts—capitated payments, or to the private internal medicine doctor and paid out of pocket.425 prospective global budgets—using caps to keep costs down and shifting risks to the Source: World Bank (2001d). provider. But this leaves to frontline (c) The International Bank for Reconstruction and Development / The World Bank Health and nutrition services 151 ever, providers need to receive timely and adequate compensation. BOX 8.9 Modulated Contractual arrangements often need to payments for combine soft capacity-building components providers according to meet unserved needs with rewards for to income criteria performance. Experience with perfor- mance-based contracts for immunization as The Kyrgyz Republic created an insurance fund part of the Global Alliance for Vaccines and to purchase services from health facilities, com- Immunizations shows that zero-based con- pensating them for increased use of services and providing greater �nancial access to the poor. tracts and the threat of resource withdrawal Facilities charge fees to patients and claim pay- were not very effective in raising coverage.426 ments from the insurance funds under a sliding Variations across communities can be taken scale based on �ve categories of patients: self- referred, uninsured, insured, partly exempt, and into account through a two-tier allocation fully exempt. Facilities get higher compensation system. The �rst tier can be allocated to a for clients from the partly exempt and exempt local government on a capitation basis—as categories, who pay lower fees. Sixty-eight per- in Argentina, Brazil, Ethiopia, and Poland— cent of poor patients preferred the copayment system over the previous theoretically free sys- and the second tier to an insurance fund or a tem, which often required “informal� payments. purchasing agency in charge of maximizing Under the new payment mechanism, many the ef�ciency of resources and purchasing informal payments were replaced by formal an appropriate mix and quantity of clinical copayments.The very poor have access to treat- ment through a reserve fund. services—as in Korea and New Zealand (�g- ure 8.11). Source: Kutzin (2003). The impact on the poor of the changes in provider payments from input-based to more complex output-based is not conclu- sive, and there has been little evaluation. But sector. An autonomous parastatal hospital the separation of policymaking from pur- in France that enjoys large �nancial and chasing and service provision creates clearer management flexibility and an NGO hospi- channels of accountability. Autonomous tal under government contract in Canada providers have more flexibility in ensuring are not very different. the appropriate input mix including hiring For population-oriented services that can and �ring, as in Kenya and Zambia.427 Pur- be standardized, governments can generally chasing bodies are more independent actors, write contracts for public or private pro- subject to a double line of accountability viders. Policymakers can specify the service because they are often �nanced by govern- ments and households through insurance Figure 8.11 Citizens exert power on both providers and purchasers systems. Governments need to be involved in these agencies to de�ne which services the Policymaker Com poor need most, price the subsidy support pac t Buying needed by the poor, and limit providers’ outcomes conflicts of interest. ce Purchaser/ Voi Social insurance Com pac t Buying Selecting providers Voice outputs There is no presumption that one type of provider—public, for-pro�t, or not-for- Citizens Service provider pro�t—is likely to be better than any other. C l i e nt p o w e r Public health facilities can be remarkably Su pp e s ef�cient, as in Malaysia, or largely ineffec- o rt rvic to s l se e lf - c a nica tive, as in middle-income Gabon, where r e • O u tr e a ch • Cli immunization levels have stalled at under 30 percent. Nor are NGOs necessarily pro- Social Health Income poor. When contracted by governments, inclusion outcomes protection (MDGs) NGOs also tend to get closer to the public (c) The International Bank for Reconstruction and Development / The World Bank 152 WORLD DEVELOPMENT REPORT 2004 characteristics and monitor and enforce the populations although management and contract. Government providers deliver quality problems have been observed.431 high immunization rates in Tajikistan or Tunisia—but in Cambodia and Haiti so do Regulation and enforcement contracted NGOs. As contractors NGOs Governments can also use market regulation can extend service outreach and test new to counter conflicts of interest. In Hungary approaches to service delivery.428 accreditation mechanisms have been quite To support self-care, private services— successful in establishing quality criteria for whether for-pro�t or not-for-pro�t—often providers. To reduce supplier-induced over- are most appropriate. Private for-pro�t supply and compensate for the lack of invest- providers can be very ef�cient in providing ment in poorer settings, most countries use information and distributing commodi- certi�cates of needs (in United States), plan- ties—witness the success of social market- ning boards (Australia), or health maps ing. For information and social support, (Spain and most of Africa). But most low- grassroots organizations, small private income countries do not regulate their phar- providers, and community-based organiza- maceutical market successfully, though tions often do the job better than rigid pub- Cambodia has had some success. When reg- lic organizations. Governments can con- ulation fails, a combination of user educa- tract some key services such as information. tion and provider training can yield the Yet public services can sometimes be more greatest bene�ts.432 Government involve- pro-poor than NGO services. In Ceará, ment through national tendering, price cap- Brazil, “many NGOs delivered services in ping, or tariff reduction has also influenced clinics while the public service sent its the prices of pharmaceuticals.433 Overall the workers into households�429 (see spotlight). enforcement of regulatory controls is often For clinical services the contrast is sharper. weak, focusing mainly on personnel licens- Public provision works well when there is a ing. The same political and institutional fail- strong public ethos, the politics are pro-poor, ures hampering health service delivery affect and rules are enforced. For-pro�t pro- the legislative, administrative, and judicial viders—quali�ed or not—are usually in tune services needed to make regulation work. with demand, but the inherent conflicts of Expanding access to professional health interest require external control. In Lebanon care providers—particularly midwifery, sur- and elsewhere, expansion of the private sec- gical skills for reducing maternal mortality, tor did not bene�t the poor much.430 In and clinical skills for reducing neonatal India, private providers serving poor groups deaths—is a priority to reach the Millennium are often less than quali�ed. Governments Development Goals. In many countries the can then exert control only if they have imbalance between rural and urban areas in sophisticated regulation and purchasing terms of skilled health workers is extreme. In functions. When these do not exist, partner- Turkey, there is one doctor per 266 people in ships between government and civil society the richest region and one per 2,609 in the organizations can compensate by strengthen- poorest. In Ghana and Senegal more than ing clients’ power over clinical providers. half the physicians are concentrated in the Because not-for-pro�t organizations capital city, where fewer than 20 percent of often bene�t from strong intrinsic motiva- people live. Health workers lack opportuni- tion and professional ethos, government ties in rural areas where turnover is high. can also write open-ended contracts with Rural workers are less likely to be female and them and still expect that providers will do educated. Lack of services (school, water) and the right thing. Service delivery by NGOs access to training/education are major incen- can help repair the link between policymak- tives to leave rural areas. Undifferentiated ers and providers for clinical services with salary structures are disincentives to work in signi�cant bene�ts for the poor. In areas where the poor live, because it is dif�- Guatemala, about a third of the population cult to supplement low salaries with alterna- is now served by NGO providers, signi�- tive income from activities such as private cantly increasing access for indigenous poor practice, teaching, and consulting.434 (c) The International Bank for Reconstruction and Development / The World Bank Health and nutrition services 153 BOX 8.10 The human resource crisis in health services When the international community set out in High rates of absenteeism reflect disenchant- ing too: in 2001 the United Kingdom approved 1955 to eradicate malaria, Africa was left out ment with working conditions. Studies of health 22,462 work permits for nurses from developing because it lacked adequately trained person- professionals in Ghana, India, Mozambique,Tanza- countries. nel. Today, Africa still lacks such personnel, and nia, and Uganda show that the health work To retain internationally marketable health yet it must deal with multiple disease-control force—nurses and physicians in particular—feel staff, poor countries will have to offer efforts. In Burkina Faso, the average number of overworked and underappreciated. In Guinea, internationally competitive wages and bene�ts. physicians per 100,000 people was 3.4 in the Mauritania, Poland, and Russia health staff wages That requires replacing inflexible civil service 1990s compared with 303 for nine industrial have declined in real terms. policies with more flexible approaches.Training countries. In Zambia the already low number of There has been considerable emigration of speci�cally oriented to national markets can also physicians, at 8.3 per 100,000 people in the health professionals from developing countries. help. Countries that emulate the training stan- 1960s, declined to 6.9 in the 1990s. To meet the More than 600 South African doctors are regis- dards of industrial countries tend to be more vul- requirements of the priority health interven- tered in New Zealand, at a cost to South African nerable to poaching (Ghana).There is evidence tions recommended by the World Health Orga- taxpayers of roughly $37 million. An estimated from Ethiopia and the Gambia that community nization, Chad would require a sevenfold 61 percent of Ghanaian doctors trained between nurses and health of�cers with curricula not increase in health personnel. 1985 and 1994 left the country. Nurses are leav- internationally certi�ed are less likely to migrate. Human resources constraints for clinical care Percent assisted deliveries Sri Lanka El Salvador Sudan Solomon Colombia Azerbaijan Cape Verde Turkey Iran South Africa 80 Belize Zimbabwe Namibia Tajikistan Paraguay China Comoros Vietnam Benin Nicaragua Lesotho Egypt Bolivia Indonesia Peru The Philippines Cameroon Honduras Iraq Myanmar P. N Guinea Togo Kenya Senegal Zambia Côte d’Ivoire Madagascar Ghana Guatemala 40 Sierra Leone Niger India Constrained Mauritania Morocco Tanzania countries Burkina Faso Cambodia Haiti Gambia Mali Lao Niger Yemen Pakistan Chad Eritrea Bhutan Bangladesh Equatorial 0 Guinea 0 10 100 1,000 10,000 Medical personnel (doctors and nurses) per 100,000 population (log scale) Sources: Liese and others (2003), Commission on Macroeconomics and Health (2001), Ferrinho and Van Lerberghe (2003), and Reinikka and Svensson (2003b). Chile, Mexico, and Thailand435 have used workers from underserved areas or social �nancial and non�nancial incentives to groups as done in Indonesia and Iran. In the encourage quali�ed staff to work in rural Bangladesh Rural Advancement Committee areas. In Indonesia doctors were also allowed (BRAC) community workers are trained to to supply private services during or after duty seek out the extremely poor in need of urgent hours. Other countries have tried to establish medical care. But success requires careful new credentials, as for health of�cers in design and evaluation. In El Salvador low- Ethiopia, and trained community workers in skilled health promoters posted in rural vil- India and Brazil (spotlight on Ceará). lages did little to improve health or health- Another approach includes progressively seeking behavior.436 The global crisis in the upgrading the skills of traditional providers, labor market for clinical services also requires such as community midwives in Malaysia, or innovative strategies to get professional ser- encouraging the hiring and training of health vices to rural areas and the poor (box 8.10). (c) The International Bank for Reconstruction and Development / The World Bank 154 WORLD DEVELOPMENT REPORT 2004 The complexity and dispersed nature of ity rates and contraceptive use in most Sub- clinical health services and the potential for Saharan countries in the 1990s have been conflicts of interest make self-monitoring by concentrated in the urban, richest popula- providers critical for effective service deliv- tion segments. In Tanzania the declining use ery. Historically, peer regulation has been the of skilled delivery care between 1993 and common response to a conflict of interest. 1999 can be attributed mainly to declining But because of state dominance in many use among the poorest groups. When col- countries, professional bodies are fairly lecting information on income is dif�cult, weak. Provider-driven changes in the organi- alternative indicators can be used, such as zation of service provision can yield substan- ethnicity, caste, region, gender, linguistic tial bene�ts for clients, as in initiatives such group, or religion. Countries as different as as Health Workers for Change.437 The Ger- Colombia, Indonesia, Iran, Mexico, and the man health system is largely self-regulated. Philippines use community maps to identify Professional associations in Zimbabwe high-risk individuals and households in maintain professional ethics and standards need of home visits and special attention.440 among public and private nurses.438 Associa- tions of midwives in Guatemala and New Zealand develop and promote a pro-poor Six sizes �t all? ethos (box 8.11). Which accountability mechanisms should be emphasized to ensure that health Information and monitoring resources go where they should? There is no Decentralization, devolution, and output- single path. The many things that influence based contracting of services increase the the short and long routes of accountability importance of timely and accurate informa- call for different responses. For health and tion for monitoring performance. National nutrition services, one size does not �t all. and international statistics do not yet capture What works varies by country and type of the range of practices or the performance of service. A strong command-and-control all health care providers. Most ministries of approach can achieve much if policymakers health know little about the private sector, have a solid mandate or the ideological which makes it hard to develop partnerships drive to make tough choices about which or contracts, although countries are attempt- health services to deliver and to whom—as ing to conduct provider surveys as in in Cuba,441 Malaysia, or Iran. Greater inclu- Poland.439 Information on access, quality, and sion of poor people in the political debate ef�ciency is scarce and often noncomparable. can influence policymakers, as in Brazil And because many factors outside the health where the pro-poor orientation of health sector affect health status, cross-sectoral policies has improved over the past 10 years monitoring and planning are also required, as (spotlight on Ceará). But when the mandate Thailand’s National Economic and Social is less clear and implementation levers are Development Board does regularly. weak, the short route of accountability Monitoring of average outcomes or ser- through client partnerships with private vice utilization patterns often does not reveal and community-based providers gives poor where change is occurring. Changes in fertil- people more control over services. BOX 8.11 Developing a professional ethos in midwifery In the words of a professor of midwifery, ones involved, the danger is that a ‘for midwife’ We need to be able to form a contract, . . . and fol- “Midwives should be able to take on a more culture develops, protecting midwives and per- low up on it, all the time respecting woman’ s enabled,‘for women’ role.This then has implica- petuating problems. . . . The formal process can individuality and the culture in which she lives. tions for regulation, which should be ‘self regu- also be backed up by a less formal process (i.e. This all implies enough education to do this well, lated’ to a point—but should also have input to peer review) to ensure lots of midwife to midwife and enough power to influence the system.This that process from women themselves, and from contact and learning.This ‘with-women/for- is what I would describe as ‘professional’.� fellow professionals. . . . Midwives should be very women’ stance can then form a foundation for involved in the process, . . . but if they are the only what ‘professionalism’ looks like for midwifery. . . . Sources: ICM (2003) and Davies (2001). (c) The International Bank for Reconstruction and Development / The World Bank Health and nutrition services 155 Figure 8.12 Six sizes �t all Provision Financing 1 Government contracting as services are Public financing: as collective action standardized and quality and quantity can be needed, funds to come from the public monitored. Output-based contracting is purse, most often taxes. Financing can be possible. Decentralization can pose problems integrated into social insurance—if the Yes as economies of scale can be important latter is well functioning and pro-poor— because of network externality and public with public subsidy. Demand-side Population-oriented Pro-poor goods. subsidies to stimulate demand. outreach services politics? 2 Voice: pro-poor coalitions putting pressure Public financing as those services are No on governments with limited capacity to unlikely to be driven by the market. focus action on services they can monitor, Alternatively collective action can build reaching the whole population. Focused into large civil society organizations (e.g., public provision or contracting out to BRAC Bangladesh). Yes intrinsically motivated providers (NGOs, CBOs) to compensate for potential problems due to deficient government. 3 Government contracting with enlisting of Public subsidies: “supply side� to users’ groups and individuals in monitoring commodities and information, e.g., social Easy to providers (fees/food stamps/vouchers). marketing; “demand-side� (vouchers) to monitor by Outcome (nutrition) and output (knowledge, individuals and communities. government? Yes Family- goods)–based contracting of local NGOs or oriented community-based organizations if capacity to Pro-poor monitor results exists. support politics? to self- care 4 Client power: imitate the market and foster Copayments: competition likely to drive No role of civil society organizations in prices down as asymmetry of information Yes disseminating information and commodities: is low. Use of local/community solidarity No commercial networks, cooperative, networks. peer/support groups, community based organizations, local NGOS. Easy to 5 Government provision or contracting of Social or government insurance made monitor intrinsically motivated providers. Public progressive through differential by users? provision implies standardization of contributions. Public subsidies to the benefits package and a well-accepted insurance premiums, exemptions for rationing approach combined with a strong copayments for the poor, and/or third-party public ethos. Contracting possible with payments (e.g., poverty funds). Equalization intrinsically motivated providers (e.g., of resource allocation for poor regions: NGOs). Contracting for-profit providers matching grants, poverty-sensitive No Yes only if sophisticated purchasing capitation. function and capacity to measure outputs exists. Individual- oriented Pro-poor 6 Client’s power, voice and providers’ ethos. Copayments, cross-subsidies, and clinical politics? Coproduction: community-owned/managed/ microinsurance. Copayments to avoid services monitored services, e.g., drug revolving elite capture and ensure presence and funds, community-based services. Litigation: responsiveness of providers. Local consumer courts/ombudsman. Voice: report retention of locally generated funds. No cards, pro-poor coalitions. Self-regulation: Cross-subsidies between services. Local intrinsically motivated providers/NGOs; solidarity networks, micro-insurance. foster professional ethos through Strengthen collective action outside of professional organizations. government arena: autonomous purchasing agency/social insurance under citizens’ oversight. So if one size does not �t all, can six sizes? When the long route of accountability Figure 8.12 attempts to capture some typical works well for poor people—their concerns situations that could provide guidance. Situ- are included in the political process—public ations vary according to the homogeneity of action bene�ts them. Governments can the health needs, the nature of services, and provide or contract out standardized popu- the characteristics of the political process. lation-oriented services (1), and provide (c) The International Bank for Reconstruction and Development / The World Bank 156 WORLD DEVELOPMENT REPORT 2004 demand-side subsidies to poor families for making poor citizens’ voice heard by fostering those and for appropriate self-care (3). civil society groups, and building pro-poor Needs for clinical care can be made homo- coalitions for services requiring collective geneous through a technocratic rationing of action. Governments with limited capacity services based on equal bene�ts. In this case can then be pressured to focus on contracting an integrated service delivery approach—or population-oriented services, in partnership a universal single payer system—can be with intrinsically motivated providers—com- appropriate, as in Cuba, Finland, and Viet- munity or civil society organizations—to nam (5). make sure the services are delivered (2). But these conditions are far from univer- Serving heterogeneous needs where the sal. Needs for self-care or clinical services long route of accountability is not working are rarely homogeneous. This heterogeneity is common in developing countries and can be accommodated in a pro-poor con- requires enlisting poor people as monitors, text through decentralization and flexible investing in client’s power. Commercial and output-based and outcome-based contract- media networks, cooperatives, and commu- ing combined with equalization of subsi- nity-driven development activities are then dies between rich and poor regions (3 and best used to support self-care (4). Micro- 5). Subsidies for clinical services can be pro- insurance schemes, and community co- vided to local medical schemes (as in Ger- managed health services and drug funds are many and Poland), to speci�c schemes for especially relevant for clinical care (6). But poor people (as in France, Indonesia, and this power is not enough to avoid conflicts Thailand), or directly to poor groups of interest. Litigation can be of limited help. (poverty funds in China) (5). More importantly, altruistically motivated The long route of accountability may also providers, such as not-for-pro�ts, can help not be working—either because richer groups foster a stronger pro-poor ethos supported capture the political process or because the by professional self regulation (6). bureaucratic process—compact—does not None of the solutions is fast or easy. But deliver. In these cases more investment in the success is clearly possible, as hundreds of long route through taxation and strengthened examples have shown. “It does not matter government actions is unlikely to do much for how slowly you go so long as you do not the poor. Instead, more needs to be invested in stop� (Confucius, 551–479 BC). (c) The International Bank for Reconstruction and Development / The World Bank spotlight on Costa Rica and Cuba Good health at (initially) low income Costa Rica and Cuba have both attained very low levels of infant mortality in the last 50 years. For Costa Rica, this is eas- ily explained by rapid income growth and attention to traditional public health and, in recent years, to innovative approaches to publicly funded health care. Cuba, on the other hand, has maintained lower levels of infant mortality than many industrial countries and has eliminated diseases common to developing countries while remaining very poor. The achievements came with a community-based health system with numerous health workers, highly motivated staff, and close monitoring and evaluation of outcomes. Can Cuba’s model survive the economic reversals of the 1990s and pressure for a more open and free society? C osta Rica and Cuba both have very similar, low infant mortality rates—almost as low as Canada’s though at much lower incomes (�gure 1).442 Their routes to this happy circum- mortality. After recessions in the 1980s, growth has resumed and progress on health status continues. One way to attain good health from initially low income is surely to stop having a low income. Second was the creation of a commu- nity health program, with specialists tend- ing patients in clinics as well as at home, school, or work. In the mid-1980s this community-based stance, however, have been quite different approach was intensi�ed with the Family In 1945, infant mortality—measured in The Cuban puzzle— Doctor Program. The goal: to place a doctor deaths of infants under one year per 1,000 good health without growth trained in primary health care and a nurse live births—was 100 in Costa Rica and 40 in The puzzle is Cuba. How has Cuba man- in every neighborhood (serving about 150 Cuba, respectively. Up to 1960 Costa Rica aged to maintain an infant mortality rate at families). By 2001 there were more than made progress largely due to economic least as low as that of any developing coun- 30,000 doctors—a ratio of one family doc- growth and aggressive public health pro- try in the Western Hemisphere and quite a tor for every 365 Cubans.444 Services are grams.443 Hookworm was eliminated with a few industrial counties as well? The sus- free, although nonhospitalized patients are program starting in 1942, and public health tained focus of the political leadership on required to co-pay for medicines. campaigns accelerated after the revolution of health for more than 40 years surely played While this approach clearly contributes 1948. As a result, malaria, tuberculosis, and a big part. After the revolution, universal to better health outcomes, it is also expen- most diseases that were preventable by vac- and equitable health care was one of the sive. Indeed, Cuba spends substantially more cines at that time were also eliminated by government’s top three goals. The govern- of its gross domestic product on health than 1960. In stark contrast, Cuba’s admittedly ment sees good health as a key performance other Latin American countries: 6.6 percent low level of infant mortality stagnated under indicator for itself. in 2002. (Average public spending on health a particularly corrupt political regime. Despite low infant mortality before the is 3.3 percent in Latin America and the Since 1960, progress in Costa Rica has revolution, rural areas lagged far behind Caribbean, but some other countries also been rapid but not too dif�cult to explain. urban areas. The new government, com- spend substantial amounts—Costa Rica 4.4 Costa Rica’s real income per capita mitted to changing this, concentrated on percent and Panama 5.2 percent).445 increased by 25 percent from 1960 to providing health care to rural areas. It 1970—the same rate, coincidentally, that required all new medical school graduates Specifying what you want—and infant mortality declined. Income growth to serve for one year in rural areas. It also keeping track of what’s going on of 40 percent by 1980 along with the uni- increased the number of rural health facili- The Cuban health model rests on three pil- versalization of coverage for health care saw ties. In 1961 the government nationalized lars: giving clear instructions to providers, a further decrease of 60 percent in infant mutual-aid cooperatives and private hospi- motivating staff, and monitoring and eval- Figure 1 Infant mortality in Cuba: low in the tals, which left the public sector as the sole uating the system.446 Clear guidelines are 1950s, even lower by 2000 provider of health services—a feature of the provided through national specialist advi- system that remains today. At that time sory groups—which draw up standards and Deaths per 1,000 live births 160 many of the country’s medical profession- technical procedures (and evaluate the per- Chile als left the country (as many as two-thirds formance of physicians and specialists)— 120 by one estimate). and regulations that standardize activities Costa Rica In the mid- to late 1960s there were two in the national hospital care system. Dominican Republic major innovations in the health system. Health staff in Cuba typically are highly 80 Jamaica First was the establishment of policlinics— motivated. Medical training emphasizes the the basic unit of health services—each altruism of medical service—often culmi- 40 staffed by several specialists and nurses and nating in service of one or two years Cuba Canada serving a population of 25,000–30,000. abroad. This is volunteer service, but there 0 1940 1950 1960 1970 1980 1990 2000 This was combined with campaigns to are strong social pressures for it. Serving in immunize many more people, control vec- poor rural areas in Cuba remains a right of Note: 1945 refers to years 1945–1949, 1950 to 1950–1954. Source: 1945, 1950 from United Nations Demographic Year- tors (such as mosquitoes), and promote passage for many newly trained doctors. book 1961. Data for 1960–2000 are from UNICEF. good health practices. Television programs lauding health workers (c) The International Bank for Reconstruction and Development / The World Bank 158 Spotlight on Costa Rica and Cuba engaged in international solidarity missions about providers. Their complaints can go tected, with public spending on health raise their pro�le and contribute to a sense through the health system—such as the exceeding 10 percent of GDP in 2000. But of pride in Cuba’s doctors. policlinic that coordinates the local health in real terms, spending had gone down. Cuba also keeps close track of what’s facilities, the municipal health council, or Health outcome indicators worsened in the going on in health facilities. Monitoring is hospital administrators. Or they can go early and mid-1990s, recovering only some- strong, with information flowing in many through political channels—say, to the local what by the end of the decade. directions. The main elements are: representative of the People’s Power Assem- As health infrastructure suffered, so did bly, which is required to respond. Despite transport services. Public transport had all • An integrated national health statistics this monitoring, there is limited direct citi- but disappeared by the early 1990s, and fuel system that collects data routinely from zen control: participation in administrative shortages limited the use of private cars. service providers. Indicators of particu- and health councils does not entail much Cubans resorted to walking miles to work, lar concern, such as infant mortality, are more than setting broad targets.447 Like- and the use of bicycles skyrocketed.449 collected with high frequency—some wise, citizens play only a small role in set- The economic reversal also appears to even daily. ting priorities within the health sector, and be weakening motivation among staff. • Regular inspection of, and supervision between health and other sectors. Physicians are paid relatively well, earning visits to, health facilities. almost 15 percent more than the average Can Cuba sustain the system? national wage.450 But their pay is in local • Annual evaluations of health techni- cians on the technical and scienti�c The 1990s were dif�cult for Cuba. The col- currency, with purchasing power declining results of their work. In addition, a ran- lapse of the socialist system in Europe and steadily over the past decade. The legaliza- domly selected sample undergoes exter- in the Soviet Union and the tightening of tion of a separate “dollar economy� has nal evaluation. the economic embargo by the United States made occupations that pay in dollars led to a severe economic contraction. Cuba highly prized. Stories of doctors shirking • Annual reports by the Ministry of Pub- lost the trading partners that had provided their formal duties to join this parallel lic Health and the provincial and municipal health directorates to the most of its imports of medicines, food, fuel, economy—driving taxicabs, for example— People’s Power Assembly. and equipment used in agriculture and are common.451 mining. Between 1988 and 1993 imports of Time will tell whether an approach that Monitoring and evaluation go beyond medicines fell by more than 60 percent. By relies on a publicly paid doctor for every statistical and expert assessments. Public 1994 agricultural production had fallen by 150 families can be sustained in times of dissemination of health indicators, at the almost half. Drug shortages persist today.448 economic hardship—and with competition end of each year, draws citizens into the Government spending on social ser- from an economy that relies more on the process. In addition, citizens can complain vices, particularly health care, was pro- dollar. (c) The International Bank for Reconstruction and Development / The World Bank Drinking water, sanitation, and electricity 9 Drinking water, sanitation, sewage disposal, In rural network and non-network set- electricity, rural roads, and urban transport tings, community and self-provision domi- influence human development outcomes nate. The policymaker as standard setter chapter (crate 1.1). As with education and health and capacity builder in support of the client services, the impact of infrastructure ser- is missing. To avoid ensuing problems, such vices on human development is direct (e.g., as arsenic in Bangladesh’s rural drinking reducing water-related diseases, which rank water, policymakers need to support clients among the top killers of children). The in ensuring service quality and access. impact is also indirect, through economic Externalities in sanitation in rural, non- growth.452 But like education and health, network settings are best contained within the these services are failing poor people. village or community. So supply-side support Focusing on water, sanitation, and elec- at the household level should be comple- tricity services, this chapter uses the Report’s mented with interventions at the community service delivery framework to �nd out why level—be it information about hygiene or and to show how things might be improved. subsidization of latrines—that are designed to The reform lessons from these services, rep- spur household demand and create commu- resenting both network and non-network nity peer pressure for behavior that internal- services, are also likely to apply to other izes the externalities. In urban settings, where infrastructure services. demand for sanitation services may be greater, For networked services, such as urban property rights and facilitating private water and electricity, regulating providers response can support collective efforts. and ensuring that poor people have access to affordable services are the main reasons for government intervention. This brings the The state of water long route of accountability into play. But and sanitation services poor citizens have a weak voice because water About 2 of every 10 people in the developing and electricity are particularly vulnerable to world were without access to safe water in patronage politics. Providers end up being 2000; 5 of 10 lived without adequate sanita- more accountable to policymakers than to tion; and 9 of 10 lived without their waste- clients, which breaks the long route of water treated in any way.453 There have been accountability. gains, but despite the many global commit- The solution is to separate the policy- ments, notably the U.N. Decade for Water makers from the providers—and to make and Sanitation, access to water and sanitation providers more responsive to clients. Dis- lags far behind the milestones set in the persing ownership through decentraliza- 1980s. Nor do aggregate trends in the 1990s tion and private participation, promoting give comfort (�gure 9.1). The share of people competition through benchmarking, ensur- with access to these services in Africa and ing alternative access by using independent Asia—where the world’s poor are concen- providers, and charging for services are trated—has fallen, remained constant, or ways of separating policymakers from increased only slowly. providers and strengthening compacts, Innumerable city and town studies con- client power, and voice. �rm the UN-Habitat Report’s key message 159 (c) The International Bank for Reconstruction and Development / The World Bank 160 WORLD DEVELOPMENT REPORT 2004 Figure 9.1 Little progress in access to improved water and sanitation, 1990 and 2000 Infrastructure and the Water supply coverage by region Sanitation coverage by region accountability framework Percent Percent for service delivery 100 100 Countries are trying different approaches to Urban Urban 75 75 address failing water, sanitation, and electric- ity services. These include decentralizing to 50 50 local governments, private sector participa- Rural Rural tion, regulatory reform, community-driven 25 25 development, and small independent providers. Some approaches try to make ser- 0 0 vices work for poor people through targeted 1990 2000 1990 2000 1990 2000 1990 2000 1990 2000 1990 2000 interventions. Others seek to improve ser- Africa Asia Latin America Africa Asia Latin America and the and the vices overall—on the premise that making Caribbean Caribbean services work for all is necessary for making Source: WHO, UNICEF, and Water Supply and Sanitation Collaborative Council (2000). them work for poor people. The same approach has worked in one setting and failed in another, and different approaches have worked in seemingly the same setting. that water and sanitation services are too What is needed is a way to think about the often failing communities.454 Full-pressure, institutional and political characteristics of “24-7� water supply remains a pipe dream infrastructure services to understand what in many cities. Because a quarter to half works where and why. (and more) of urban water supply remains unaccounted for, many cities are turned Accountability in infrastructure into leaking buckets (�gure 9.2). The lim- services ited number of network access points must Chapters 3–6 of this Report develop a frame- be widely shared, which dramatically work for analyzing how well the actors in ser- increases waiting times and often simply vice delivery—clients and citizens, politicians overwhelms the system. Rural infrastruc- and policymakers, and service providers— ture often goes to seed: more than a third of existing rural infrastructure in South Asia is estimated to be dysfunctional.455 Figure 9.2 24-hour water: a pipe dream Poor people bear a disproportionate Karachi share of the impact of inef�cient water and sanitation services. Fewer poor people are Delhi connected to a network. When they do have Chennai access, the installation has to be shared Nakuru among many more people (�gure 9.3). And Kathmandu the prices they pay are among the highest, Calcutta generally more than those paid by more Phnom Penh affluent households connected to the piped Dhaka system (�gure 9.4). The price differential is Manila partly a result of inef�ciencies—the Jakarta inequitable practice of subsidizing piped Nairobi water, lack of scale economies for indepen- Colombo dent providers, or worse, providers taking Bangkok advantage of poor people’s lack of choice. But some of the price differential can also Hong Kong reflect the flexibility and convenience of Beijing services offered by independent pro- 24 16 8 0 20 40 60 viders—no connection charges or access to Hours of water Percent of water quantities of water that are more affordable available in a day unaccounted for for poor people. Source: Human Settlements Program (2003). (c) The International Bank for Reconstruction and Development / The World Bank Drinking water, sanitation, and electricity 161 Figure 9.3 Water and sanitation by Short route of accountability. In a simple poorest and richest �fths market transaction, the buyer holds the Accra seller accountable for the product bought, Percent rewards the seller by repeating business, or 80 penalizes the seller by choosing another Poorest fifth provider. This accountability is “short� 60 because the client can hold the provider directly accountable, without any interme- 40 diaries. Small, independent providers in 20 water and sanitation and their clients are Richest fifth usually in such a market relationship. 0 In Dar-es-Salaam, Tanzania, a cholera outbreak in 1996 forced the sewerage and Jakarta sanitation department to loosen its 80 monopoly on cesspit cleaning and allow private providers in. There is now an 60 emerging competitive private market for 40 cesspit cleaning—households can choose a provider based on price and (easy-to-mon- 20 itor) performance. Besides allowing entry and implementing regulations on sewage 0 disposal, the city’s role has been small.456 ~ Paulo But service and market conditions that Sao automatically give clients power—through 80 choice, ease of monitoring, and market 60 enforceability—are not always present for infrastructure services. So the route of 40 accountability has to be long. 20 Long route of accountability. Govern- ments worldwide deem it their responsibil- 0 ity to provide, �nance, regulate, and in No water at Sharing toilets other ways influence infrastructure services. residence with 10 or more They do it for two good reasons: market households failures and equity concerns. First, net- Source: Human Settlements Program (2003). worked infrastructure services exhibit Figure 9.4 Alternative sources of water: poor people pay more Price of water per liter, U.S. dollars hold each other accountable within four rela- tionships (�gure 9.5): 0.010 Bicycle water • Client power connects service users with vendor providers. delivering to non- • Voice connects citizens with politicians service Handcarts and policymakers through the political area 0.005 delivering to process. homes Water • Compacts connect policymakers through Water vending Standpipes implicit or explicit contracts with providers trucker drawing responsible for services. Kiosks House water from Utility connection mains • And management connects provider organizations with frontline across-the- Lima Kampala Bandung Dar es Salaam counter providers. Source: Human Settlements Program (2003). (c) The International Bank for Reconstruction and Development / The World Bank 162 WORLD DEVELOPMENT REPORT 2004 Figure 9.5 Accountability in infrastructure services Board, a corporation owned and operated by government. The state The short and long routes of accountabil- Politicians Policymakers ity need to work together. Indeed, even for of accounta cesspit services in Dar-es-Salaam, govern- ro u te bili Com ng ty pa ment regulation was necessary to ensure that Lo c ce t Voi the small private operators complemented S h o rt r o u t e the public provider and complied with Citizens/clients Public and private utilities sewage disposal guidelines. Effective solu- and providers tions are likely to be a strategic mixture of the Coalitions/inclusion Client power Management short and long routes of accountability as a Nonpoor Poor Frontline Organizations system in which the clients, the policymaker, and the provider are linked in accountability relationships that make services work for Wate s r, sa nitation, other service poor people. economies of scale, or network externali- Why infrastructure services fail poor ties, that make it technically more ef�cient people: patronage to have a single distributor of the service. In Because the family has been without daytime sanitation the externalities come literally water for the past decade, the children have never from spillovers. Yes, households in Dar-es- seen water come out of their home faucets. . . . The Salaam were willing to pay for improved faucet flows only between midnight and 4 a.m. in sanitation with larger health bene�ts to the most of Baryo Kapitolyo. MWSS, you know that. Did you care? city. But free-rider problems, where one Dahli Aspillera, a citizen of Manila, on the person’s behavior hurts others with eve of the privatization of Manila’s public impunity—as in the case of runoff from water agency, Metro-Manila Waterworks open defecation in many parts of Asia and and Sewerage System (MWSS), in 1997 Africa—require community or government intervention. Second, societies care about Where water, sanitation, and electricity are equity, and governments often redistribute publicly managed, the accountability to cit- resources—such as a lifeline water sub- izens is achieved when the state ensures that sidy—to ensure the minimum equitable utilities, boards, and government depart- service access that markets cannot. ments provide ef�cient and equitable ser- Network externalities, collective action vices for all citizens, including the poor. problems, and distributional goals thus When the state is unsuccessful and the voice provide powerful reasons for the govern- relationship is not effective, the long route ment to be involved. The arrangements of accountability has failed. then are no longer primarily between the In 1997 the MWSS was typical of service client and the provider, and new account- utilities, boards, and government depart- ability relationships become important. ments that consider politicians and policy- The �rst of these arrangements is voice— makers as their real clients. Politicians— citizens delegating to politicians the respon- responding to equity concerns or, more sibility to ensure the infrastructure ser- likely, to short-term political gain—often vices they want. The second is through the keep prices for infrastructure services well compacts between policymakers and below those for cost recovery. This makes providers—to design the service delivery service providers dependent on politically framework, choose a provider, and ensure motivated budget transfers for survival—or that it meets citizen expectations. Voice and when transfers are not forthcoming, on ser- compacts together become the “long route� vice cutbacks that attract no penalties from of accountability. In Bangladesh the prime policymakers. minister and her power minister are, in State-owned water and electricity pro- principle, accountable to citizens for the viders then cease to function as auton- performance of the Power Development omous service providers.457 They become (c) The International Bank for Reconstruction and Development / The World Bank Drinking water, sanitation, and electricity 163 BOX 9.1 Clientelism in service delivery Patronage weakening accountability in the citizen-provider chain Politicians Operational subsidies/ Appointment Political of directors favors Untendered Employees Overstaffing Utility company contracts Contractors Poor quality of service Artificially depressed Unconnected population tariffs High prices Connected population With patronage, the compact between the politician vice quality, and precarious �nances.The scarcity of and provider—the utility or board—is neither trans- resources for investment leaves much of the popu- parent nor determined by universal client needs. lace without adequate services and forces them to Politicians exert their control by appointing (and dis- rely on expensive or inconvenient alternatives. missing) company directors and by providing public The clientelist model broadly describes the poli- subsidies to �nance investments and prop up ailing tics of urban and rural regional utilities in both water enterprises. In return for this patronage, water com- and electricity sectors. It also applies to local admin- panies are often obliged to supply political favors in istrations in charge of urban or village-based the form of excess employment, the depressing of services (funding of piped networks or community tariffs, political targeting of new investments, and toilets or even public investment in deep tubewells). the distribution of contracts on the basis of political criteria.The consequences: spiraling costs, low ser- Source: Foster (2002). extensions of policymakers. The policy- system, such as tubewells. Citizens or their maker and provider begin to fuse into one groups respond to rationed access by sup- role. When this happens, policymakers can porting politicians who favor them as their no longer hold providers accountable for clients over politicians who push for uni- delivering to all citizens, services deterio- versal access. This strengthens the ability of rate, and poor citizens as clients are left politicians to use patronage. The account- powerless. ability linking clients, politicians, policy- The dynamics of this relationship can be makers, and providers is displaced by even more debilitating for poor clients. patron-client relationships—clientelism— Over time providers become a strong polit- on both legs of the long route of account- ical force, influencing the policymaker. In ability (box 9.1).458 effect, providers capture the policymaking In such settings, the breakdown in voice process, exerting pressures through orga- for poor citizens is reinforced by their loss nized labor or their ability to control service of client power. Dahli Aspillera’s question— delivery for the politician. With deteriorat- did you care?—reflects both a sense that the ing service levels, policymakers and client cannot penalize the provider for poor providers ration access. This has an impor- service and a deeper reality that the long tant implication when lumpy investments route of accountability has failed the citizen. are needed to gain access to services— If failure of voice is at the root of weak ser- whether through an electricity grid, a vil- vice delivery in water, sanitation, and elec- lage water network, or even a stand-alone tricity, what are the options for reform? (c) The International Bank for Reconstruction and Development / The World Bank 164 WORLD DEVELOPMENT REPORT 2004 Urban water networks vide incentives to remove patronage and com- pensate for the weak voice of poor people. Who is the Water Board accountable to? Question asked of the Managing Director Strengthening the compact: of the Hyderabad Water Board by a consumer, Hyderabad, September 2002 decentralizing assets Devolving responsibilities to different tiers of In cities and towns, where scale economies policymakers and separating powers between prevail, water systems have major network them can create the right incentives to suppliers—generally a public sector provider, improve service delivery. First, by having ser- such as the Lagos water board in Nigeria, or a vice and political boundaries better coincide, small municipal water department, as in Cha- decentralization can strengthen voice and pai Nawabganj in Bangladesh. Some of these accountability. Second, when the center is in providers belong to local governments—as in charge of both regulatory and service delivery the case of the Johannesburg water utility; responsibilities, it has few incentives to hold some to a state government—as is common itself accountable. Devolving services to in India; and some—like MWSS in Manila— another tier of policymaker triggers incentives to central governments. For all, the relevant more compatible with having the center (or an questions are whether there is a clear delin- upper-tier government) oversee the regulatory eation of roles between the policymaker and framework. Finally, devolution creates an the provider—and whom the provider is opportunity to benchmark performance and accountable to, the policymaker or the client? use �scal resources and reputation as rewards When voice and politics fail, the distinction to support ef�cient service provision. The con- between the two is blurred, and the provider is testability for resources in this context requires accountable to the policymaker. a tier with �scal capacity and without service Four reform strategies can potentially sep- provision responsibilities—appropriate for arate policymakers and providers: decentral- the center (or a state in a federal system). izing assets, using private participation in Devolving responsibilities to local gov- operations, charging for services, and relying ernments has had mixed results in water on independent providers to give clients and sanitation, often leading to the loss of choice. The �rst two aim to influence com- scale economies, eroding commercial via- pacts, the second two to strengthen client bility by excessive fragmentation, and even power. All are politically dif�cult to imple- constitutional conflicts between municipal- ment. That is not surprising, since strong ities and upper-tier governments.459 The political forces—not technocratic failures— historical experience of industrial countries blur the roles of policymakers and providers. offers lessons for addressing these problems The issue is whether these strategies can pro- (box 9.2). BOX 9.2 Decentralization and the water industry—in history In France water assets have historically been Examples include Elizabeth and Hackensack, under company structures, mostly owned and devolved to the commune—the lowest tier of both in New Jersey. run by municipalities, but many were under pri- government. Clusters of communes have inte- Interestingly, for France and some areas in the vate operation if not ownership. In England the grated the industry by delegating water and United States, the limited capacity of the smaller national government consolidated the local sanitation services “upward� to private or semi- local governments provided the incentives for pri- water systems into regional bodies, moving public companies.The functional boundaries of vate companies to serve clusters of political juris- from 1,400 in World War II to 187 in 1974 and 10 the companies cut across several communes, dictions. In both France and the United States the in the 1980s, all eventually privatized. In the which continue to own the assets but contract multijurisdictional coverage prevents the water Netherlands, also under central government out the management of services. provider from being captured by any one local mandate, the municipal companies were con- In the United States water and sanitation body—thus maintaining the separation from verted to regional companies to support the assets are also devolved to local governments. local policymakers. expansion of services to rural areas. But the Where local governments have been carved up The approach was different in England and companies remained under the ownership of into small political jurisdictions and individual the Netherlands. At the outset of the 20th cen- municipalities and provinces. water works are impractical, privately owned tury in both countries, oversight and direct pro- companies have emerged to provide regional vision of water services were in the hands of Sources: Lorrain (1992); Seidenstat, Haarmeyer, and services covering several local governments. local authorities. In the Netherlands these were Hakim (2002); Jacobson and Tarr (1996). (c) The International Bank for Reconstruction and Development / The World Bank Drinking water, sanitation, and electricity 165 Fragmentation and the loss of scale The bottom line: upper-tier governments economies can be partly addressed by permit- can influence the design of compacts at the ting interjurisdictional agreements. In the local level through legislation and incentives. French syndicat model, municipal jurisdic- However, as demonstrated by experience in tions can cede the right to provide water and Latin America, decentralization processes sanitation services to a company jointly owned have not always been designed with suf�- by several local authorities. Bolivia’s water law cient care to allow these kinds of bene�ts to explicitly allows for multi-municipal compa- be reaped. The success of managing service nies. Colombia empowered its regulator to reforms during decentralization will depend enforce mergers of nonviable local water agen- on whether broader decentralization policies cies, but ironically exempted the smallest of can ensure that local politicians and policy- the municipalities that would have bene�ted makers bear the consequences of policy deci- most from this rule. Brazil’s state companies sions. Ensuring that decentralization can were created through voluntary agreements separate policymakers and providers at the with municipalities, �nanced by central funds. local level requires that it also separate roles These examples suggest an important and responsibilities of the different tiers of approach for aligning general decentraliza- government (chapter 10). Without that sepa- tion with sectoral priorities. When authority ration, decentralization may simply transfer is being decentralized, a window usually patronage to local levels. opens for central government to influence the restructuring of local services. Decentral- Strengthening the compact: ization gives the center the ability to negotiate using private participation the restructuring of devolved assets through in operations �scal incentives—say, by deciding to retain Over the past decade, private participation the liabilities while devolving only the assets. has grown signi�cantly in water, sanitation, Where devolution has already happened, and electricity in different forms and across the center can provide incentives such as many regions (box 9.3). In general, private �scal grants to subnational governments participation in infrastructure has been that are dependent on milestones of institu- advocated for many reasons, including tional reform. Australia’s federal govern- accessing management expertise and private ment provided grants to states to reform the investment and introducing incentives in the water sector. The South African govern- operations of infrastructure services. Private ment is also using central �scal incentives to participation is also a direct way of separat- support municipal restructuring and to ing policymakers and service providers influence reform of urban services, includ- through two aspects of the accountability ing water and sanitation. India’s federal chain—compacts and voice. government is exploring a similar policy In the design of compacts, private pro- instrument—the City Challenge Fund—to viders generally require explicit contracts create incentives for general urban reform, that de�ne up front the service responsibili- including municipal services. ties of the provider and the policymaker, the Such �scal incentives are more effective if regulatory and tariff parameters, and issues allocated competitively to local tiers of gov- of access by poor households. In addition, ernment. But this requires information so the the process of contracting private providers center can compare the performance of dif- can strengthen the voice channel, particu- ferent local governments, promoting compe- larly if advocacy groups and public informa- tition and accountability. It also requires that tion mechanisms are involved in the process. the policy and legal framework enable local Indeed, service delivery standards and ser- governments to have the flexibility to reform vices for poor people are often explicit in the service delivery—to form regional compa- policy debate on private participation in nies and use contracting, for example. Coun- water and sanitation. tries such as Pakistan and South Africa that In many industrial countries the involve- have recently embarked on decentralization ment of the private sector in service deliv- have adopted such legislation. ery enabled governments to develop the (c) The International Bank for Reconstruction and Development / The World Bank 166 WORLD DEVELOPMENT REPORT 2004 BOX 9.3 Trends in private participation: water, sewerage, and electricity Investment commitments in projects with private participation in developing countries, 1990–2001 Water and sewerage projects Electricity projects 2001 U.S. dollars (billions) 2001 U.S. dollars (billions) 10 Manila water system concessions 60 Brazil 8 Chile Aguas privatization All other Argentina 40 developing 6 concession countries 4 20 2 0 0 1990 1991 1992 1993 1994 1995 1996 1997 19981999 2000 2001 1990 1991 1992 1993 1994 1995 1996 1997 19981999 2000 2001 Cumulative investment, water and sewerage Cumulative investment, electricity projects projects (total $40 billion) (total $213 billion) Middle East and Middle East and North Africa North Africa South Asia 0% 0% 1% Sub-Saharan Sub-Saharan Africa South Asia Africa Europe and 1% 10% 2% Central Asia 8% Europe and Central Asia 9% Latin America Latin and the East Asia East Asia America Caribbean and Pacific and Pacific and the 43% 38% 32% Caribbean 52% Water and sewerage projects by type (total 202 projects) Electricity projects by type (total 832 projects) Divestitures 8% Divestitures Greenfield 39% projects Greenfield Concessions 56% projects 44% 28% Concessions Management and 4% Management and lease contracts lease contracts 20% 1% Private investment has been far higher in electricity than in water and vate investment. Finally, in electricity privatization dominates; in water, sanitation. Not surprisingly, the decline in private investment in the late management contract and concessions—public ownership—remains nineties was more pronounced in electricity. In both sectors, the impact the norm. of “large deals� and country speci�c changes are visible—reflected also in the geographic concentration—East Asia and Latin America—of pri- Source: World Bank, PPI Project Database. (c) The International Bank for Reconstruction and Development / The World Bank Drinking water, sanitation, and electricity 167 BOX 9.4 Private participation—in history England. In London private companies sup- companies, which were then progressively ing led to conflicts over �scal transfers from plied water for more than 400 years with little taken over by municipalities and operated as municipalities to companies. Not surprisingly, government restriction on entry. Companies public utilities.The amalgamation was public ownership increased, and with it the pub- competed against each other, invested in promoted by central regulation and facilitated lic system inherited the tradition of managing service and quality innovations, and increased by municipal politics. A major motivation was to and regulating water as an economic good. household connections. By the 19th century use the companies to deliver services more London’s extensive water system helped make regionally to rural areas. By the time the public France. Starting with private provision of that city “one of the best housed and healthiest sector took over, principles of economic man- water at the local government level and main- cities in Europe, with a death rate lower than agement of water services had become well taining it from the mid-1600s onward, France birth rate by about 1800, at a time when most embedded in the political system. Arms-length evolved toward public ownership and private European cities were devourers of men.� Ninety- management of public utilities by municipalities provision through different types of �ve percent of London residents received piped became the norm. management and lease contracts.The reasons: supply from the private companies, and a major- scale issues (small local authorities), the history ity had direct home connection. United States. Between 1800 and 1900, U.S. of the French legal system, and the role of voice Technological change led to signi�cant cities experienced a tremendous growth of in controlling policymakers.The issue of �re price competition, industry consolidation, and water works. Initially dominated by private own- �ghting did not come up in France, perhaps higher prices. And the improved water supply ers, half of them were public by 1900.The shift because cities were built with vastly different increased demand for flush toilets, which to public ownership emerged because of con- materials and densities. created problems of sewage removal. tracting problems between municipalities and Parliament responded with regulation, and by companies over water for �re �ghting.The dif�- Sources: Tynan (2002), Schwartz and Maarten 1908 the private system was nationalized. (In the culties of establishing contracts when cities (2002), Crocker and Mastens (2002), and Lorrain (1992). 1980s England shifted back to private were growing rapidly, and several urban confla- provision.) grations, offered opportunities for both private companies and governments to evade perfor- Holland. Between 1853 and 1920 the water mance targets or force renegotiations of sector was dominated by private water supply contracts. A lack of metering and direct charg- capacity and political setting to regulate, Ulitmately, like decentralization, private price, and manage water in public and pri- provision offers an opportunity to influence vate contexts (box 9.4). But in today’s devel- the relationships of accountability. And like oping countries private participation is being decentralization, its success depends on flung into a context of institutional rigidity, design and implementation (box 9.5). Experi- not necessarily conducive to the organic ence so far suggests that regulation and infor- growth of formal private participation. mation—two interlinked parts of overall sec- Using private provision to drive a wedge into tor reform—are important in successfully patronage makes managing private partici- implementing private sector participation in pation intensely political—but potentially water, sanitation, and electricity sectors and in powerful for increasing accountability. promoting greater voice in service delivery. The proof of this potential is already evi- dent. Formal private participation in water Regulation. A regulatory system in this and sanitation has led to greater demand for Report’s framework is best de�ned along the accountability—this, despite accounting for a dimensions of accountability between the pol- small part of total investment in water and icymaker and the provider—delegation of sanitation. During the 1990s private invest- responsibilities and �nance, information ment accounted for only 15 percent of total about the performance of the provider, and investment in water and sanitation, covering enforcement (chapter 3). The regulator could less than 10 percent of the world’s population. be responsible for speci�c elements of the Even in Latin America, where private provi- accountability chain—just providing informa- sion has made the greatest inroads in the tion on performance or also ensuring enforce- water sector, it only covers 15 percent of the ment. Sometimes the policymaker is the regu- continent’s urban population.460 In addition, lator, and sometimes a dedicated third party in contrast to electricity, for example, public has this responsibility. Sometimes even an ownership and not divestiture of assets association of providers can self-regulate. remains the norm in the sector. Whichever method is followed, the regulatory (c) The International Bank for Reconstruction and Development / The World Bank 168 WORLD DEVELOPMENT REPORT 2004 BOX 9.5 Private participation in water and sanitation can save poor people’s lives, and money In the 1990s Argentina embarked on one of the Distribution of new connections following Responding to the need for alternatives for largest privatization campaigns in the world as private sector participation in water and reaching poor people, one of the Manila conces- part of a structural reform plan.The program sanitation services sionaires has developed a system for water deliv- included local water companies covering ery in densely populated, hard-to-reach slum approximately 30 percent of the country’s New Connections areas. In the Bayan Tubig (“Water for the Commu- municipalities. Child mortality fell by 5–7 Percent nity�) program, the use of appropriate technologi- percent in areas that privatized their water ser- 35 cal standards, client participation in maintenance, vices.The largest gains were seen in the poorest 30 and community-based organizations in interme- municipalities, where child mortality fell by 24 diation and mapping of the network reduced percent. Overall, privatization of water services 25 water costs for poor families by up to 25 percent. prevented approximately 375 deaths of young 20 To increase affordability, the concessionaire has children each year. 15 introduced an interest-free repayment scheme Aggregate data from other sources on the Argentina over a period of 6 to 24 months. Between 1991 distribution of new water connection by 10 and 2001, the program provided water connec- Bolivia income quintile from three countries in Latin 5 tions to more than 50,000 households—this America con�rm the results of the pro-poor 0 Chile despite the fact that the contract of one of the impact of private sector services. As the data Poorest 2 3 4 Richest Manila concessionaires is under review. show, 25–30 percent of the network expansion Income quintile was targeted at the lowest 20 percent of the Source: Galiani, Gertler, and Schargrodsky (2002); income pro�le. Source: Foster (2002). Water and Sanitation Program (WSP-AF) (2003). process has to separate policymaker and national level, or at the state level if policymak- provider and preserve its own independence. ing and provision are done at the local level. Another option in a multi-tiered government Organizing regulation: one size does not �t is to use local regulation but have the appeals all. Where voice is strong and supported by process at a different level. In the United States an effective legal system, the policymakers the Constitution provides an overall frame- and the judiciary do the regulating. In France, work for property rights while state regulatory where the compact for water is between commissions oversee the operations of pri- municipal policymakers and a private com- vately owned local utilities. Local governments pany, regulation is done primarily through regulate public utilities directly. municipal monitoring of contracts, with some support from central authorities. Regulation and sector reform. The account- In countries without a tradition of sepa- ability framework clari�es the conditions rating policymakers and providers and with under which a regulator will be effective in discretionary policymaking, credible regula- supporting sector reforms. Just as account- tion requires a third party—an agency—to ability is blurred if any one of its relationships set or interpret regulatory rules. Several for- is broken (see chapter 3), the effectiveness of mal safeguards can support the indepen- a regulator is abridged if delegation of dence of a regulatory agency from political responsibilities and �nance between the state influence.461 Some examples: earmarking and the provider is incomplete. That is the funds for the regulatory agency, hiring staff case in the electricity sector in some states in from the market without being restricted by India. In other words, an independent regula- civil service rules (competence and capacity tor is needed to enforce the separation are important elements of gaining credibility between policymaker and provider, but if the and independence), ensuring that the hiring separation is not initiated through general and �ring of regulators are protected from reform to begin with, the regulator may well the political interference of the executive and be ineffective. A regulator cannot substitute legislative branches, and not linking the for broader sector reforms. terms of staff to electoral cycles. At the same time an effective regulator can A multi-tiered governmental structure help sustain sector reform. A recent study of offers additional scope for protecting the inde- about 1,000 concessions in Latin America pendence of a regulator by placing it at the showed that even a moderately well-func- (c) The International Bank for Reconstruction and Development / The World Bank Drinking water, sanitation, and electricity 169 tioning regulator can temper opportunistic communities, labor unions, and other interest renegotiations of contracts.462 The study con- groups. Neither process was flawless, but both cludes that where a regulatory body exists in a opened the door to greater accountability. An country, the probability of a renegotiation is open process is needed to broaden the partici- 17 percent; where none exists, the probability pation of communities in the policy debate is 60 percent. on private provision—otherwise narrow interest groups can capture the information Regulating the public sector. Sector regula- and representation. tion is often discussed in the context of pri- Community involvement is also essential vate sector participation. But issues of in the regulatory process—but it has not been monopoly behavior and service performance suf�ciently encouraged. A review of urban are also relevant for public sector provision— water utilities in Latin America and Africa perhaps even more, because the contracts concludes that giving consumers little infor- between the policymaker and the public mation about the process of reform and tariff provider are often not explicit. Independent setting—and limiting their opportunity for regulation of public providers is therefore comment before taking regulatory deci- equally important. But unless public sions—weaken the regulatory process and providers have operational flexibility and are the credibility of reform, and make tariff brought under explicit compacts—and changes—however justi�ed—dif�cult to unless all the relationships of accountability implement.465 are applied—it is not clear how regulation of Organizing consumers is, however, not an the public providers would have an impact easy task. There are major free-rider (and on service standards. In particular, because related �nancing) problems in developing most of the instruments of modern regula- countries that prevent consumers from orga- tion are based on �nancial incentives, in the nizing themselves to a degree where they can absence of user charges regulation of public be an articulate voice in the regulatory providers would be ineffective. In Chile pub- process. The problem is even more severe for lic sector regulation was introduced in the poor consumers. In industrialized countries, context of sector reforms, which included relatively well-developed consumer associa- greater provider autonomy in operations and tions perform this role reasonably effectively. economic pricing of water. This helped cat- Where competent and effective consumer alyze regulatory capacity in the public sec- associations are absent, the asymmetry tor—an important asset, now that Chile has between consumers and providers becomes privatized water services. more acute, and the regulator risks being cap- tured by the provider. The role of information. With private provi- Examples exist of regulatory bodies engag- sion more needs to be done to deliver on the ing communities—especially poor communi- demand for greater voice—informing com- ties—more actively. In Jamaica the regulator munities about the why and how of private reaches out to communities through local sector contracting. A public opinion poll in churches; in some cases in Brazil special con- Peru found support for privatization of elec- sultative or advisory bodies have been created; tricity among only 21 percent of the citizens. and in Peru regulators have made extensive But when informed that privatization was to use of the radio to engage and communicate be undertaken through a transparent process with communities.466 But these are few exam- and tariff increases would be regulated, sup- ples only—much more needs to be learned port increased to 60 percent.463 In Manila the about how to organize and access communi- concession process was preceded by a wide- ties in the regulation of services. spread public campaign by President Ramos, Managing private participation also re- who convened “Water Summits� to bring quires information on how private players together different stakeholders.464 In South are performing relative to their contract Africa Johannesburg’s water management and the performance of other public and contract was also undertaken after signi�- private providers. This information, which cant—and often dif�cult—consultation with is critical for regulators, also strengthens the (c) The International Bank for Reconstruction and Development / The World Bank 170 WORLD DEVELOPMENT REPORT 2004 relationship between citizens, politicians, and Strengthening client power: policymakers. For private provision to have a charging for services catalytic impact on the sector, information is User charges provide operational autonomy essential on the performance of both the for the provider, support client power, and public and private sectors. But too little infor- elicit greater accountability from the state (box mation has been available on the perfor- 9.6). Without access to enough revenues from mance of the public sector and through few the clients, service providers depend on the credible sources. Leveling the playing �eld policymaker for �scal resources to maintain between public and private providers—as service provision. In addition, if the seller is discussed later—and benchmarking their not dependent on the buyer for at least some performance are essential in getting the best part of revenues, the provider will have little out of private participation in the sector. incentive to respond to the client. At the same Overall, the impact of private sector partic- time, given the politics of water pricing, imple- ipation is best leveraged within a broader menting user charges can quickly elicit a con- reform context—greater separation of policy- sumer response—as in Johannesburg, Manila, makers and providers for all public providers; and very visibly in Cochabamba, Bolivia. greater participation of communities in the process of private participation and in the Implementing user charges. Drawing on the regulatory framework; and greater use of power of user charges to leverage accountabil- benchmarking of both public and private ity in service delivery requires, as discussed providers. The Australian approach is in- earlier, effective regulation to address mono- structive. An enabling framework and a poly provision. But more importantly, the crit- national competition law level the playing ical policy issue is how to increase tariffs. There �eld for all public and private providers. Sec- are two implementation issues: the �rst is syn- toral legislation provides guidelines for service chronizing tariffs with quality improvements, provision. The central government provides and the second is ensuring that there is a safety �scal incentives to support change at the state net to safeguard basic affordability. level. A variety of delivery approaches are sup- In many countries, bringing the tariffs to ported—corporatization (Melbourne), man- cost-recovery levels would require signi�cant agement contract (Adelaide), vertically inte- adjustment and rebalancing of tariffs among grated public utility (Sydney), multi-utility residential, business, and industrial cus- (Canberra). Regulation differs between states tomers. In Indian cities the charges on resi- and is backed by independent regulatory dential users are less than a tenth of the oper- agencies as well as benchmarking done ating and maintenance costs. Industrial users through an association of water providers. pay ten times more but are below the bench- mark for operating and maintenance costs in two-thirds of the metropolitan cities and 80 BOX 9.6 Charging for percent of smaller cities.467 Even if there is a water—in history willingness to charge, how can the transition Treating water as an economic good and charg- to prices be managed? ing for services enabled France and the Nether- Charging cannot be assessed independent lands to use private provision to jump-start the of the broader policy framework and the sector’s development. In France the private sec- credibility of service providers. Policymakers tor remains the major service provider of water and sanitation services. In the Netherlands the are obviously concerned that services will not system shifted from the private to the public improve enough to justify the price increases. sector. But in both countries charging users for Central to a price increase is what comes water remained the norm, which enabled �rst—the increases or service improvement? providers to sustain service delivery at arm’s length from local government and gave them Guinea entered a lease contract for water ser- greater incentives to be responsive to the needs vices in its major towns and cities in 1989. of the clients. During the �rst six years of the contract, the Sources: Lorrain (1992); Blokland, Braadbaart, and government subsidized a declining share of Schwartz (1999). the private operator’s costs while tariffs were adjusted gradually toward cost recovery, (c) The International Bank for Reconstruction and Development / The World Bank Drinking water, sanitation, and electricity 171 which avoided a major tariff shock. This of a competitive market, and in others they jump-started the move to cost recovery and are controlled by a few groups. better service delivery. It also gave the reform Enhancing the role of independent credibility in a region that had little experi- providers as part of the short route of ence with private provision.468 For various accountability is a key policy challenge. How reasons the lease contract expired in 1999 and can this be achieved?470 By recognizing inde- was not renewed, but the pricing strategy pendent providers and giving them legal sta- remains relevant for other countries. tus, by ensuring that network providers are Similarly, subsidies to poor people could not given exclusive supply, by enabling be better targeted and designed, which greater partnership between formal public would enable user charges to be imple- and private network providers and small mented overall. Chile has a nationally independents, by ensuring that the regulatory funded household water subsidy. Colombia framework for network providers gives the uses geographic targeting. South Africa has a flexibility to enable contracting with inde- national lifeline tariff system that guarantees pendent providers, by enabling small-scale each household 6 kiloliters of water a provider associations and working with these month.469 Given the substantial divergence umbrella bodies to introduce appropriate between piped water prices and the high cost levels of regulation, and by enabling poor of the inferior alternatives that many of the people to gain access to multiple independent poorest are forced to use, there is often a providers while keeping their regulation strong case for giving highest priority to con- more focused on health and issues related to nection subsidies rather than subsidizing the groundwater depletion. use of water by those who already enjoy Of particular concern is the effect of access to the piped network. Connection bringing in a formal private provider in an subsidies also have the advantage that they area dominated by independent providers. are easier to target (since lacking access to This issue was not addressed in the design of service is already a strong indicator of the Cochabamba contract—where the pri- poverty) and cheaper to administer (since vate provider was given exclusivity rights— relatively large one-time payments are and it contributed to the contract’s cancella- involved). Generally it is more ef�cient to tion.471 In reality, if coverage targets are subsidize the connection costs for low- de�ned in such a way that they can be met income households, but there are alternative with the services of small independent options for designing connection and con- providers, the operator will have an incentive sumption charges that bene�t poor people. to encourage their involvement. Ultimately, tariff adjustment and subsidy mechanisms are technocratic tools that can Rural areas: network be designed and applied in many ways. What and non-network systems is critical is to turn payments for services into Rural settings are complex in their settlement a political tool for reducing patronage and patterns, ranging from dense settlements in strengthening client power of poor people. South Asia to dispersed communities in many African countries. Suppliers include house- Strengthening client power: relying hold systems in Bangladesh, water vendors in on independent providers Laos, and community-managed local piped As the example of pit operators in Dar-es- water systems in Ghana. Across all situations, Salaam suggests, small independent pro- the client-provider link is the norm. Under- viders are a common feature in providing standing why the long route of accountability water and sanitation services across income is needed to support this client power, and groups. Their organization varies from how this can be done, are the main service household vendors of water, small network delivery challenges in rural areas. providers, and private entrepreneurs to coop- eratives. In some cases they are the primary Community-managed networks suppliers, and in others they supplement the In countries as diverse as India and Kenya, formal provider. In some cases they are part water boards or engineering departments (c) The International Bank for Reconstruction and Development / The World Bank 172 WORLD DEVELOPMENT REPORT 2004 have traditionally been responsible for provide examples of institutional mecha- delivering water services to rural communi- nisms for supporting community-based sys- ties. Top-down in their approach, with little tems. They are all “works in progress,� and skill in community mobilization, and learning from them will offer insights on backed by �scal support from central gov- how to advance rural community-based sys- ernment, the boards scaled up physical tems of delivery. investment. But they had little success in Local governments can form the institu- ensuring sustainable operations and main- tional and �nancial support for expanding tenance. Indeed, these boards face the same community-based systems. With access to a problems of state capture inherent in the tax base, local governments can provide patronage model of service delivery. resources to cover periodic capital expendi- Given the failures of top-down institu- ture, provide temporary �scal support to tions, some countries are shifting to com- communities to adjust to economic shocks, munity-managed systems—often supported and facilitate access to technical assistance. by donors, as in India and Ghana. Commu- Uganda and South Africa provide examples nities are involved in the design and man- of arrangements in which local govern- agement of their water systems, paying for ments are part of a larger �scal decentral- operations and maintenance costs. Govern- ization program with own resources and ments, generally central governments, pay a greater autonomy. Local governments thus signi�cant part of the capital costs. Donor- strengthened can support community- funded project management units, backed based programs. Even in India, where local by not-for-pro�t organizations, often form panchayats do not have as much autonomy, the technical and organizational backbone the relations between local governments of these systems. and user groups are evolving. Where neigh- The client-based model puts the client at boring small towns have effective providers, the center of the accountability relation- these can be contracted in by rural local ship, but many challenges remain in scaling governments to support their communities. it up.472 In Côte d’Ivoire a national utility run by a private partner has responsibilities for • Communities require technical support urban centers and smaller towns. The in the medium to long run to manage national utility uses cross subsidies—with water systems, and donor-funded pro- the capital city providing the �scal sur- ject management units are not well plus—to support the smaller urban centers. suited for this. Expansion of its responsibilities to rural • Communities pay for current operating areas is now being tried. The early lessons costs, but replenishing capital invest- have not been successful but the approach is ments and covering higher tariffs—to still evolving.473 pay for rising power costs, for example— Finally, communities can contract with a are not easily managed through group third party or an independent provider to contributions. manage local network systems. In China • Communities are not homogeneous— formal cooperatives (rural companies) run problems of exclusion and elite capture on commercial principles with very high can be the same as in government sys- cost recovery.474 In several African countries tems. And different communities may village entrepreneurs manage water systems have differing abilities to form cohesive under contract. In East Asia small indepen- groups. dent providers are being organized to take • Ef�cient technologies that require scale on operational responsibility on a conces- economies are not selected because of sion basis. In each case, the process is orga- the focus on village-level associations. nized through group consultation and endorsement. While small systems can be Supporting client provision contracted by community organizations, Three approaches—local governments, re- villagewide systems may again require the gional utilities, and independent providers— support of policymakers at the local level. (c) The International Bank for Reconstruction and Development / The World Bank Drinking water, sanitation, and electricity 173 Self-provision. Households managing wells and providers will not suf�ce; policymakers and hand pumps are common in large parts are needed to support communities of rural Asia and Africa. Nowhere is self- provision more dramatically showcased Sanitation than in Bangladesh, where shallow aquifers Policy issues in sanitation need to be dis- and the market provision of hand pumps cussed in the context of the private and pub- enabled households to directly manage lic goods dimensions of the sector. To the water services and replace pathogen-conta- extent it is primarily a client-provider rela- minated surface water with groundwater. tionship, households invest in sanitation Service delivery improved—less waiting systems and contract independent providers time, no quantity limits, and the conve- for the removal of excreta. To the extent the nience of household connection. And the public goods dimensions are dominant, pol- health impact, which included a decline in icymakers need to support collective action diarrhea-related deaths, was remarkable. to change behavior at the household and Missing was any attempt to monitor community levels, and organize common water quality. Finding arsenic in the infrastructure for excreta removal. groundwater caught everyone by surprise. Access to sanitation services has often The government had withdrawn from the been seen as an issue of subsidizing latrines rural water sector, assuming that access was and prescribing latrine technology. This now fully addressed by the private market supply-driven approach, emphasizing the and household efforts directly. In addition, �scal and engineering aspects of sanitation, in a unitary system of government, there has failed. In response, some countries have was no local government to respond to the been shifting toward “complete sanita- crisis. In rural Bangladesh today, a policy- tion�—focusing on community and house- maker is needed to support communities, hold behavior and sanitation practices.475 manage externalities, and understand the This involves breaking the fecal-oral chain technological choices for addressing the by encouraging households to change arsenic crisis (box 9.7). More broadly, for a behavior—shifting away from open defeca- collective good such as the monitoring of tion, washing hands, keeping food and water quality, a partnership between clients water covered, using safe water, focusing on BOX 9.7 Fighting arsenic by listening to rural communities The arsenic contamination of shallow aquifers aquifers are contaminated. Government, donors, wells, well-sharing, and other mechanism may may be undoing the success of rural drinking and NGOs are advocating several options: shift- not work. Indeed they have not yet been water provision in Bangladesh. While the num- ing to alternative water sources, including some successful as solutions. ber of individuals showing symptoms of arsenic surface sources; sharing of uncontaminated Communities strongly indicated a poisoning is still low—despite the high concen- tubewells in villages; sinking deep tubewells in preference and willingness to pay for tration of arsenic in the water—between 25 and public areas; and promoting household �ltering centralized, community-based �ltering systems, 30 million people may be at risk in the future. technologies.The latter, if successful, would pre- such as local piped-water systems with a central The �rst response to the crisis by serve the use of shallow tubewells—decentral- �ltering point for chemical and biological conta- government and many donors was denial.This ized, household means of water access—that minants.The piped water network systems was followed by an effort to test all water have de�ned the “water miracle� of Bangladesh. introduced in the Bogra area by the Rural Devel- sources and hand pumps.There were various In all of this, little effort was made to understand opment Academy suggest the potential of such technological and logistical problems—which is the preferences of rural households. systems in Bangladesh.This has been con�rmed not surprising in view of the fact that arsenic A WSP-BRAC (Water and Sanitation by preliminary data, which show the cost effec- contamination of this scale has not been faced Program–Bangladesh Rural Advancement Com- tiveness of piped water in settlements that have anywhere in the world.These problems were mittee) team undertook a comprehensive sur- 300 or more households. If implemented further complicated by a lack of coordination vey of household preferences for different broadly, this approach would dramatically and blurring of roles among government, approaches to arsenic mitigation in selected change the nature of water institutions in rural donors, and nongovernmental organizations areas of rural Bangladesh.The results reveal that Bangladesh—a change that communities are (NGOs). communities place a high premium on conve- willing to undertake. The efforts so far have revealed that surface nience. Unless the alternatives are as convenient water does not contain arsenic and that not all as the current hand pumps, the shift to dug Source: Ahmad and others (2002). (c) The International Bank for Reconstruction and Development / The World Bank 174 WORLD DEVELOPMENT REPORT 2004 children’s hygiene behavior and maintain- But even if a subsidy is required, the �s- ing a clean environment. The use of cal contribution could be delivered to the hygienic latrines is a result of this process of community, rewarding collective action, changing behavior. self-regulation, and the elimination of Because the health impact of a house- open defecation. Take one of India’s hold’s sanitation practices is affected not largest states—Maharashtra state, with 97 only by the household’s behavior but also by million people. It subsidized latrine con- the practices of the community, there is a struction by households below the poverty collective action problem. The provider’s line only to discover that close to 45 per- role in ensuring information and social sup- cent of the latrines were not being used. port to households through community So it shifted its subsidy to a competitive structures becomes critical. Success depends scheme (the Gadge Baba scheme) that on making people see themselves as a com- rewarded communities for good sanita- munity, where every member’s behavior tion practices, using an information cam- affects the other—a daunting challenge and paign to de�ne the principles of sanitation perhaps the reason why sanitation has and publicizing the names of winning vil- always lagged behind demand for water. lages. Reputation, recognition, and com- munity rewards became the catalyst. Over A participatory focus in rural areas a short period an estimated 100,000 Because communities need to manage sani- household latrines were built, and for tation collectively, innovative participatory every rupee of state resources, local approaches are required to generate spending on sanitation and related infra- demand for it, especially in dispersed settle- structure increased by 35 rupees. ments. The shift from open to �xed-point Local compacts. Making the shift to better defecation may be motivated by health, sanitation practices is the �rst objective— safety, and privacy concerns—issues of but sustaining the shift is equally impor- importance to women, who bear much of tant. The local externalities and the need to the burden of poor sanitation practices. In understand and draw on local conditions the approach practiced by Village Educa- and knowledge suggest that local govern- tion Resource Center (VERC) and Wat- ments are the appropriate policymaker tier. erAid in Bangladesh, an external group trig- In Vietnam and West Bengal, India, local gers community-wide recognition of the governments have supported community need for better sanitation practices. The participation and ensured its continuity by community then takes responsibility for �nancing the work of the service provider, self-regulation—motivating households to usually a not-for-pro�t organization. In strive for complete sanitation. In East and Vietnam some local governments have used South Asia this has even led to innovations a program similar to Maharashtra state’s in latrine technology and micro-credit Gadge Baba scheme to acknowledge village �nancing for investments in latrines and and individual achievements. associated infrastructure. Subsidies. The community focus also Responding to demand changes the approach to latrine subsidies. In in urban areas Bangladesh, villages in the VERC/WaterAid Households in urban settlements with high project did not require any external subsidy. population densities often show a greater To assist low-income households, higher- demand for better sanitation facilities. The income households provided resources. condominial systems in São Paulo, Brazil, Once communities focused on the need for and the community sanitation systems of collective responsibility, assisting individual Orangi in Karachi, Pakistan, and Parivartan households to reach community goals was in Ahmedabad, India, suggest that informal more readily accepted. In Vietnam the par- urban communities may be willing to man- ticipatory approach was supported by a sub- age and pay for ef�cient systems of sanitation sidy targeted at poorer households. and waste disposal. Small independent (c) The International Bank for Reconstruction and Development / The World Bank Drinking water, sanitation, and electricity 175 providers serving households directly, as in Electricity Dar-es-Salaam, show that urban households Like water, electricity has urban and rural do invest in sanitation. So what are the components—and issues of managing grid impediments to expanding these approaches? and off-grid systems. In the grid setting, the The answer may lie outside the realm of issues of separating the policymaker from water and sanitation—and in the regulatory the provider, charging for services, using domain of urban centers. First, the formal private providers, and developing effective recognition of informal communities by gov- regulatory systems are similar to water net- ernments and the provision of some form of work issues (box 9.8).477 A key difference is tenure have strongly influenced community unbundling (rather than decentralizing) willingness to invest in household infrastruc- services. ture and to work collectively on community For electricity in rural settings, the infrastructure (La Paz in Bolivia, Ahmedabad extension of the grid network provides in India). Research on garbage collection in lessons for managing non-network systems informal settlements in Indonesia provides in water. And the emerging use of off-grid empirical evidence of the negative relation- electricity systems can draw on lessons from ship between incomplete property rights and community-managed water in rural set- community investment in local public tings. goods.476 It suggests that improving tenure security increases the probability of garbage Grid systems. The experience from Latin collection by 32–44 percent. America, Eastern Europe, and South Asia Second, in dense urban areas the munici- suggests that unbundling the electricity pal government’s willingness to allocate some chain into generation, transmission, and public land to sanitation systems has enabled communities to develop community facili- ties, contracting them with a third party to maintain and operate them (Pune in India). BOX 9.8 Are pipes and wires different? Use is restricted to the community through a Electrification rates by region Electricity losses, selected countries monthly charge collected by the community and paid to the operator. Sub-Saharan 1990 OECD Africa 2000 China Third, municipal laws need to support Ethiopia flexible standards and ensure that communi- South Asia Indonesia ties and households can make arrangements Egypt East Asia with independent providers. Laws that per- The Philippines mit exclusive service provision need to be North Africa Sudan replaced by laws that permit different Algeria approaches and standards. Middle East Eritrea A concluding caveat is, however, neces- Cameroon Latin sary on the discussion about sanitation. His- America Zimbabwe Kenya torical evidence suggests that demand for Developing countries Togo water and sanitation follows a sequencing— India water �rst, followed by sanitation and then World Nigeria demand for waste water treatment. Experi- 0 25 50 75 100 0 5 10 15 20 25 30 35 ence also suggests that this sequencing is Percent Percent influenced by many factors of which service delivery arrangements in the sector is only Unlike water and sanitation coverage, electricity coverage has increased signi�cantly over the past decade. But like water and sanitation, electricity faces daunting challenges in South Asia one. In this context, policymakers must and Africa and rural areas across most regions in the world. And as in the case of water and remain realistic and patient about how far sanitation, increased electricity coverage does not automatically imply ef�cient service deliv- they can catalyze the demand for sanitation ery.The problems of theft, intermittent supply, shared access—captured broadly under the through external interventions. Unless em- heading of electricity losses—make wires no different from pipes in the context of creating accountability in service delivery. bedded in a demand-responsive approach, throwing subsidies at latrines will not resolve Source: International Energy Agency (2002). the challenge of scaling up sanitation. (c) The International Bank for Reconstruction and Development / The World Bank 176 WORLD DEVELOPMENT REPORT 2004 distribution components is critical to transfers for a part of the capital costs. This reforms in the sector—but only if the market would be similar to using regional water util- is large enough to support multiple electricity ities to support community-managed water generators, and hence genuine competi- systems. Importantly, the owner of the distri- tion.478 Unbundling provides scope for com- bution is not the policymaker but the clients. petition in the relevant sectors, primarily Unbundling the national REB into regional generation. Separating the components also REBs, with some form of benchmarking, creates scope for getting better information could support the clients in breaking a possi- about the cost structure of each part of the ble monopolistic relationship between the chain. The competition and the information REB and the cooperatives. add to client power. But transmission and distribution func- Rural off-grid. Rural provider organiza- tions are monopolies, and without effective tions—or local governments—can also sup- regulation it may be dif�cult to ensure the port off-grid systems in villages, in many separation of policymaker and provider, and cases using renewable energy to generate even reduce the scope to introduce competi- power. Donors have traditionally advocated tion in generation. Unless distribution is solar household systems—not unlike the transferred to different types of ownership, a technology push in latrines. But today’s national or regional government as a sole renewable systems can support villagewide owner will not have much incentive to sepa- grids—similar to villagewide piped water sys- rate its policymaking responsibilities from the tems—to provide AC electricity for house- operations of the distribution system. hold appliances of various types. Depending Privatizing distribution is a common pol- on local conditions the systems can also be icy approach, but decentralizing electricity wind-powered, solar, tidal, bio-gas, or hybrid, assets to local governments is not generally with fossil-fueled generators as backup. considered. Even where local governments own distribution systems—as in South Moving the reform agenda Africa—the policy discussion is about con- forward solidating into regional distribution systems. India is revolting and the Thames stinks. This is driven by economies of scale and Slogan in London, 1857 scope, and perhaps also by policy decisions to The result: Chadwick and the sanitary cross-subsidize from urban to rural settings revolution in the United Kingdom.479 and to keep the cross subsidy in the sector. Interestingly, in Mumbai and Kolkata, India, Given the weak voice relationship between where electricity is under local governments, citizens and politicians in the water, sanita- electricity provision has long been under pri- tion, and electricity sectors, deep institutional vate operation. Even in Delhi—in effect a reform often comes from broader stresses in city-state—power distribution is now pri- the economic, political, and institutional vate. machinery of a country. In London pollution was such a cause. In Johannesburg the city’s Rural grid. The extension of the grid into bankruptcy was the impetus. In cities in rural areas offer insights for rural water and Africa and Latin America a core impulse for off-grid electricity with regard to reestablish- reform of urban water and sanitation is the ing the relationship between policymaker combination of sector problems and a and service providers. A model of rural macroeconomic crisis.480 cooperatives has emerged in the United Society’s view of economic development is States and is being adapted in Bangladesh also important. In Australia, Chile, and Peru, and the Philippines. A regional or national growth-driven economic development strat- provider organization contracts with com- egy provided the impetus for improving the munity cooperatives to be village-level dis- performance of water and power markets. So tributors. In Bangladesh the Rural Electricity the possibilities for sector reform seem great- Board (REB) supports the village coopera- est when there is a confluence of natural chal- tives through technical assistance and �scal lenge, �scal crisis, and institutional reform- (c) The International Bank for Reconstruction and Development / The World Bank Drinking water, sanitation, and electricity 177 mindedness.479 Opportunities for reform may taining the separation and ensuring that the well arrive only by chance, when broader regulation of public providers is effective. In changes in turn catalyze sectoral reforms. this context, introducing private players in a What are the potential interim measures? Can few of the utilities would enhance the effec- incremental change be strategic? tiveness of benchmarking the public For urban networks, change will require providers. separating the delivery functions from those Where this broader approach of making of benchmarking and regulation. Keeping services work for all is not possible, a targeted the latter with an upper-tier government— approach for serving poor people using small central or regional—while dispersing own- independent providers is still an option. ership of water and sanitation assets to Indeed, increasingly independent providers lower-tier governments and the private sec- may, at the margin, emerge as a critical lever tor could create this separation. Without for making services work for poor people. ownership responsibilities, the upper-tier For rural systems—community-managed policymaker would have greater incentives systems and self-provision—the challenge is to use �scal instruments, benchmarking, to seek mechanisms for the policymaker to and regulation to promote improvements in support client power, using local govern- service provision. Such incentives are less ments, regional utilities, and independent inherent in a model where the regulator, providers. This is similar to the model of the provider, and owner are one and the same. rural electricity cooperatives supported by a Charging users for services strengthens this provider organization that provides a techni- separation by directly involving clients in cal and �scal hub. Where local governments the service chain through the short route of provide this hub, the voice channel is direct; accountability. where utilities are the support mechanism, Where the introduction of private sector the voice channel is indirect. Where these participation is tempered by politics or other options are not possible, the approach— factors, strategic change may have to come however unsatisfactory—of targeted com- �rst through changes in ownership and rela- munity projects remains. tionships of accountability between tiers of For sanitation, the focus is on collective the public sector. Interestingly, the history of action—to change behavior and mobilize some industrialized countries suggests that communities to invest in community infra- local ownership can trigger a more credible structure. To support this, compacts between path to private sector participation, especially policymakers and NGOs may be more if local governments are effective in strength- appropriate. In urban areas, where greater ening voice. demand for sanitation services may exist, Where local governments exist and water policymakers can support client power by and sanitation services have been devolved to allowing independent providers to function local governments, the challenge of improved and by supporting tenure in informal settle- service delivery would lie in making decen- ments. A more incremental version would be tralization work. Where local governments similar to that in the rural water sector—with do not exist, the lever of decentralized owner- a public provider organization supporting ship would be lost, but benchmarking and NGO delivery in targeted areas. regulation of the public sector would remain. But if the failure of voice is why infra- But for such a strategy to be effective, charg- structure services have failed poor people, ing for water would become even more criti- targeted intervention cannot form the basis cal. It would enable providers to achieve of institutional reform. Reforming the rela- some independence in operations, but more tionships of accountability would remain the importantly it would give clients a role in sus- policy challenge. (c) The International Bank for Reconstruction and Development / The World Bank spotlight on Johannesburg Accountability in city services In 1999 the Transformation Lekgotla, the political body directed to address the �nancial and institutional crisis of Johan- nesburg, South Africa, appointed a new city management team. The team’s task was clear: not to �x street lights but to �x the institutions that �x street lights.482 The solution was a three-year plan—“iGoli 2002�—to recon�gure city services.483 B y most developing world standards Johannesburg is not a poor city. But it faces serious development and service delivery challenges. Apartheid made sure that exclusive white suburbs were well They slashed capital and operating budgets, and even expenditures needed to maintain minimum service levels. They froze posts, causing huge increases in workloads as despairing of�cials began to drift away. And independent, single-purpose entities to overhaul larger municipal services. These operating entities were the major innovation of iGoli 2002. • Three utilities were established for user serviced, forcing black residents into they began to explore public-private part- charge–based services—water and sani- sprawling underdeveloped slums. Poverty, nerships. tation, electricity, and waste manage- unemployment, and homelessness are all ment. worsened by the deeper problem of inequality. The city of gold—iGoli 2002 • Two agencies were established—for The Johannesburg Metropolitan Munici- The new city management team realized parks and cemeteries, and for roads and pality was democratically elected in 1995 to that Johannesburg needed a new system of storm water—where expenditure would address the service imbalances. It quickly accountability for service delivery within a still have to be covered by tax revenue. found itself in a �scal and institutional crisis. dramatically different institutional archi- • Smaller corporatized units were set up Johannesburg was not one institution tecture. To address fragmentation and the for facilities like the zoo and the civic but �ve, with an overarching Metropolitan severe moral hazard, the city had to be theater. Council and four primary-level councils. reuni�ed. Political debate focused on two models of metropolitan coordination: All were established as new companies, Each could decide its priorities and approve with the council as sole shareholder. its budget. But responsibilities for key ser- • De�ning more clearly the rules of bud- Two key units would guide and oversee vices were split between the two levels, and geting, �scal transfers, and service deliv- the new entities: a corporate planning the operating budgets of the councils had to ery between the metropolitan and unit to do citywide strategic planning, and balance only in aggregate. That meant each municipal tiers, strengthening both. a contract management unit to regulate council could blissfully spend on the • Creating a one-tier metropolitan gov- the operating utilities through a range assumption that its shortfalls would be off- set by surpluses in another. ernment.484 of new instruments, including licensing agreements and annual service level The arrangement was a recipe for disas- Johannesburg chose a hybrid. It central- agreements. ter. Each municipality went on a spending ized political authority, treasury manage- spree, and ambitious infrastructure plans ment, and spatial planning under one met- One size does not �t all were rolled out without the �nance. Deteri- ropolitan government. But it organized Since the operating entities are not bound orating revenues—due to a service-payment service delivery through decentralized by overarching administrative rules, they boycott culture left over from anti- structures. This meant merging �ve sepa- have scope to differentiate. Each could set apartheid struggles, poverty, and poor credit rate councils into one overarching munici- up different management structures, control—made the situation worse. The city pality, creating integrated service delivery reporting lines, delegations, job descrip- was forced to delve into its reserves, but structures with new incentives. tions, performance management systems, these could go only so far, and by late 1997 Accountability in service delivery and operating procedures. Each could con- major creditors could no longer be paid. At �gure its internal accountability to suit a the peak of the crisis, the city had an operat- Under one metropolitan council, iGoli 2002 speci�c service delivery environment. ing de�cit of R314 million. split the institution for policy formulation Three examples: Johannesburg was in serious trouble. and regulation from the institutions for Having decentralized responsibilities, the implementation. On one side, a core admin- • The water and sanitation departments national government followed the intergov- istration remained responsible for strategic were merged into one department and ernmental rules and would not bail the city planning, contract administration, and such under the Company’s Law converted out. So Johannesburg had to dig itself out corporate services as �nance, planning, and into a city-owned utility with a board of of its own crisis. communication. On the other, two sets of directors. The assets and workers of the Two years of harsh cutbacks followed. operating entities were established: 11 new departments were transferred to the Blaming of�cials for the crisis, politicians regional administrations for libraries, utility, which was put under a �ve-year took a much tighter rein over day-to-day health, recreation, and other community management contract with a private decisions, ending management discretion. services; and �nancially ring-fenced, semi- company. (c) The International Bank for Reconstruction and Development / The World Bank Spotlight on Johannesburg 179 • The roads department was converted Figure 1 Getting back to an operating surplus—thanks to iGoli 2002 Risks and prospects into a city-owned agency with a profes- Will Johannesburg maintain the separation sional board and divided into two Millions of rands between policymaking, providers, and reg- departments—for planning and for 150 ulators? The roles of client and contractor 100 contracts. The contracts department are still evolving. Some implementation 50 operated against speci�c outcomes set capacity remains within the core adminis- 0 by the planning department, with the tration. As in the past, managers occasion- –50 threat that failing to meet benchmarks –100 ally get hauled into councilors’ of�ces to could lead to contracting tasks out to –150 explain their actions. There are also unre- the private sector. –200 solved governance debates, with the council • The gas company was sold to the private –250 arguing for a greater councilor representa- sector. –300 tion on the boards of operating entities. –350 Five factors will be critical in sustaining The reforms gave operating entities 1995 1996 1997 1998 1999 2000 2001 2002 the commitment to the principles of iGoli management independence. For example, Source: Allan, Gotz, and Joseph (2001). 2002: salaries have been adjusted to attract top- flight skills, and new systems have been • Keeping the monitoring and regulatory procured for everything from human also apparent in the city’s �nancial stand- unit of the operating entities within the resource management to remote water- ing, with dramatic improvement in both city administration; they are not legally pressure metering—increasing productiv- operating and capital budgets (�gure 1). and administratively independent. ity and service ef�ciency. And they have • Maintaining the contract management introduced innovative staff development Engaging other stakeholders unit’s operational autonomy and capac- programs and performance-linked pay Labor: Despite protracted negotiations ity—and thus the independence of the schemes. with organized labor, iGoli 2002 did not get operating entities. The entities operate at arm’s length from its endorsement. According to labor groups, • Benchmarking service delivery stan- the council, but accountability has been the city’s crisis was not a result of a failure dards, monitoring these over time, and strengthened because the primary mecha- of institutional design. Instead it was a making the information available. nism is no longer the impossible-to-digest result of “a lack of skills and experience, and management’s unwillingness to • Ensuring that �scal and �nancial decen- committee report on everyday operational [establish] functional organizations and . . . tralization remains binding. Municipali- matters. Now councilors focus on strategic �nancially unsound decisions.�485 ties relying primarily on their own rev- oversight, and of�cials are responsible for enue sources to ful�ll their democratic outcomes clearly de�ned in service-level duties without national guarantees are agreements. Reporting goes through struc- National government: The team ne- more likely to be accountable to their tured channels, either to the contract man- gotiated a R500 million restructuring grant citizens. The current intergovernmental agement unit or to company boards of with the National Treasury to support iGoli system has devolved authority and directors, which include external specialists 2002 in exchange for a commitment to accountability to the cities; this needs to capable of probing service results. timely and steadfast implementation of its remain. The operating entities have also set up key elements. It is a key accountability user forums allowing communities to com- mechanism between the national and city • Both councilors and of�cials consistently municate needs, raise complaints, and even governments and has become an incentive adhering to a clear, courageous, and far- participate actively in service provision. scheme to catalyze citywide restructuring sighted strategy. Sustaining momentum Of�cials are much more sensitive to ever- throughout the country. will require greater citizen voice at all lev- changing service delivery challenges. els. The decentralized operating entities These management improvements are Capital markets: On the strength of the and the administrative regions have already translating into better service deliv- reforms, management sought a new credit mechanisms for engaging citizens. Using ery. Waste collection has been extended to rating, aiming to win back the con�dence them will be critical for sustaining iGoli poorer neighborhoods for the �rst time. of the city’s banking community. As the city 2002. Fleets of new buses now serve outlying shifted from a large de�cit to a balanced communities. In addition, expenditure on budget, capital expenditure �nanced by the water infrastructure has increased and markets went from R300 million to well water services have expanded. Results are over R1 billion in two years. (c) The International Bank for Reconstruction and Development / The World Bank Public sector underpinnings of service reform 10 For Forms of Government let fools contest; may need to be covered in small steps— What’er is best administer’d is best. what this Report calls strategic incremen- Alexander Pope, Essay on Man. talism. Reforming basic incentives that chapter strengthen accountability and raise per- For basic services in education, health, and formance closer to formal standards is the infrastructure to work for poor people, gov- place to start. As incentives become better ernments have to be involved. Whether they aligned and internalized and as adminis- ful�ll this responsibility by providing, trative capacity grows, more advanced �nancing, regulating, or monitoring ser- reforms can be deployed to support vices or providing information about them, deeper institutional change and scaling the basic functioning of government should up. Throughout this process reforms underpin, not undermine, effective services. should be guided by the lessons of success When governments do not run well, they and failure. cannot sustain the institutional arrangements and accountability relationships that yield Strengthening the foundations good services. Looking at all that govern- ments do, the biggest payoffs to service deliv- of government ery are likely to come from a few key actions: Governments are essential to making basic spending wisely and predictably in line with services work for poor people, but a govern- priorities and coordinated across sectors; ment village school does not ensure that managing decentralization to reap the bene- children learn, or a maternity clinic that �ts of being closer to the client; developing mothers can give birth safely. Both need and deploying administrative capacity to take timely budget transfers, reliable electricity, a sound decisions at the top and to implement connecting road, probity in procurement, them well; curtailing corruption; and learn- and competent public servants. To sustain ing from success and failure. services that work, broader structures at the Public sector reforms take time and foundation of government must also work. skillful political navigation. Agreeing on Whether providing, �nancing, regulating, desirable goals is easy. Managing the tran- or monitoring services, governments focused sition is hard. When starting capacities are on outcomes for poor people must low, the road to improved performance strengthen the compact relationship between policymakers and providers along the long Figure 10.1 Strengthening public sector foundations for service delivery requires route of accountability.486 For basic services coordinating multiple compact relationships in education, health, and infrastructure, poli- cymakers must deal with multiple compact The state relationships with providers across sectors, Politicians Policymakers Compacts space, and time (�gure 10.1). Just as an ce Voi ensemble makes great music when it is well coordinated and not because it has a few vir- Citizens/clients tuoso musicians, strengthening the long Providers Coalitions/inclusion Infra- route is easier when the general business of Client power Education Health Other structure government runs well across the entire gamut Nonpoor Poor of government activities, and not just in a few 180 (c) The International Bank for Reconstruction and Development / The World Bank Public sector underpinnings of service reform 181 sectors or agencies. The more sound the basic tion, high interest rates, and burgeoning functioning of government, the stronger the current account de�cits. Despite the simple foundations for service reforms. logic of this argument—and sometimes In managing the cross-cutting activities driven by external shocks—countries slip of governments, the three institutional into macroeconomic crises that inevitably structures likely to influence service deliv- lead to belt-tightening. Countries in crisis ery the most are budgets, decentralization, may have no option but to curtail basic ser- and public administration. These are cru- vices, even if the service delivery chain nor- cial tasks for a government that wants to mally works well. Argentina is just the most make services work for people: making recent example (box 10.1). budget allocations and implementing them; Countries can instill �scal discipline by organizing and monitoring the tiers of gov- strengthening budget formulation by the ernment that provide, �nance, regulate, or �nance ministry. Constitutional or legisla- monitor services; and managing public tive restraints can rein in legislatures and employees involved in service delivery. ministries. Brazil and Chile have laws on �s- cal responsibility that limit budget de�cits. Spending wisely In Colombia, Peru, the Philippines, and Uruguay the constitution constrains or pro- When services fail poor people, a good place hibits amendments to increase budgets. to start looking for the underlying problem Sound public expenditure management is almost always how the government spends requires reliable revenue projections and money. If politicians and policymakers comprehensive budgets that do not hide spend more than they can sustain, services guarantees and other contingent liabilities. deteriorate. If budgets are misallocated, basic When budgets are not comprehensive, the services remain underfunded and frontline consequences can be harsh, as Thailand providers are handicapped. And if funds are found in 1997 when contingent liabilities misappropriated, service quality, quantity, from the banking and �nance sectors blind- and access suffer. The budget is the critical sided the government and triggered a link on the long route of accountability con- regionwide �nancial crisis. necting citizens to providers through politi- cians and policymakers. Public expenditure management—for- Allocative ef�ciency and equity mulating, implementing, and reporting For basic services in education, health, and annual budgets—is a challenging task, par- infrastructure to work for poor people, gov- ticularly when capacities are limited and the ernments have to be involved, as chapter 2 long route of accountability is weak. Chap- ter 5 discusses how citizen budget initiatives BOX 10.1 The impact of Argentina’s crisis on health and can increase voice. This chapter discusses how politicians and policymakers can education services strengthen the compact using public expen- After three years of recession, the economic care has further cut into the already low diture management to systematically and �nancial crisis in Argentina came to a resources allocated to primary care. Mater- achieve three desirable outcomes that can head at the end of 2001.The social impact of nal and child health is likely to be at risk. Epi- the crisis has been devastating. Poverty rates demiological surveillance data report an underpin effective services: aggregate �scal have jumped 40 percent.There is growing increase in some endemic diseases. discipline, allocative ef�ciency and equity, evidence of deterioration in service quality, Education has been similarly hit, with and operational impact.487 access, and use of social services. Roughly 12 salary delays and work stoppages in several percent of people with formal health insur- provinces. During 2002 roughly a third of ance discontinued or reduced their cover- provinces experienced school closings of Aggregate �scal discipline age, increasing the burden on already 20–80 days over a school year of 180 days. With no effective mechanism for resolving strapped public hospitals, the traditional Many provinces were forced to concentrate the competing budget claims of politicians, provider for the uninsured. Dif�culties with their falling resources on wages, sharply federal transfers have led to serious short- reducing �nancing for school lunches, infra- line ministries, and subnational govern- ages in medical supplies throughout the structure, and other investments. ments, public expenditures will exceed public hospital network.The pressure for available funds. The resulting unsustainable maintaining funding for high-cost curative Source: World Bank staff. �scal de�cits can translate into high infla- (c) The International Bank for Reconstruction and Development / The World Bank 182 WORLD DEVELOPMENT REPORT 2004 Table 10.1 Fallible markets, fallible governments, or both? Government failure High Low High Ambiguous, hard-to-monitor situations in which government failure Market failures keep services from bene�ting poor people. Depending may swamp market failure and so public �nancing for ef�ciency or on the nature of the market failure, public actions could range from equity reasons may not work for poor people (government primary public provision or �nancing (subsidies) to regulation or information teachers fail to show up for work, public clinical care goes only to disclosure that does not crowd out private responses or that at least the non-poor). Public expenditures should be directed to increasing takes them into account. client power through demand-side subsidies, co-payments, client Market monitoring, provider peer monitoring, and information; strengthening failure voice (through decentralization, delivery arrangements that yield more information, participatory budget analysis); and supporting altruistic providers. Market and community-led delivery should be used to strengthen public institutions over time. Low Private provision and �nancing with appropriate public regulation or Private provision with appropriate regulation, and equity-driven public education interventions informed by potential private responses makes clear. This requires sound budgeting. appropriate (table 10.1). But where govern- Good, results-oriented budget allocations ment failures outweigh market failures, are both an outcome of the long route of ignoring them can lead to large public accountability and a source of its strength, expenditures that bene�t only the non- particularly for the link between policy- poor or to services so defective that their makers and providers. How should govern- opportunity costs outweigh their bene�ts ments allocate budgets to improve educa- for most poor people. In dif�cult-to-moni- tion and health outcomes? First, the tor clinical care, if primary rural health ef�ciency rationale for government inter- clinics lack professional staff and medicines vention: are there market failures due to and the political environment is not pro- public goods or externalities? Or is redistri- poor, public provision or even subsidies for bution for equity the goal? Second, given private provision may not work for poor the rationale, what is the appropriate people. Better alternatives might be fund- instrument—public provision or �nancing, ing demand-side health subsidies or dis- or regulation, or educating the public? trict hospitals where monitoring is easier Third, what are the �scal costs over time, and peer pressure for doctors can work. and how do their expected bene�ts com- Where monitoring is easy, as in immuniza- pare with those for expenditures on other tion campaigns, contracting for private things that government should �nance? In provision may be a good solution. considering these issues, politicians and Similarly, ignoring the likely private policymakers need to pay particular atten- response to public interventions (such as tion to what is known about the multisec- the crowding out of private providers or toral determinants of health and education household income effects of government outcomes in their country (see crate 1.1). subsidies) can lead to ineffective public Reducing infant mortality may have as expenditures. Equity-seeking public expen- much to do with how the water ministry ditures can end up helping the non-poor if (clean water) or the education ministry analysis suggesting that services or money (female literacy) gets and uses its budget as never reach poor people is ignored in policy with how the health ministry does. design. There are many pitfalls in considering These questions about rationale and the rationale and instruments for govern- instruments cannot be answered without ment interventions. Focusing on market detailed information about the sector, the failures alone (information asymmetry, service, the nature and depth of market and missing insurance markets) presumes that government failures, who bene�ts (expen- government implementation failures are diture incidence), and private responses to inconsequential. Where this is actually public interventions. This information true, public provision or �nancing is needs to be developed through in-depth (c) The International Bank for Reconstruction and Development / The World Bank Public sector underpinnings of service reform 183 analytical work (in itself a public good that cymakers can focus on new programs, since governments and their external partners allocations for existing programs, decided should fund). Determining true costs and in previous years, would only need updat- impacts for allocation decisions is not easy, ing. Finance ministries can more transpar- particularly when self-serving line agencies ently require line ministries to propose cuts have strong incentives to manipulate or in ongoing activities to pay for new pro- withhold information from the ministry of grams. Line ministries would have an �nance. This information asymmetry can incentive to know the least effective pro- lead to perverse practices (such as line min- grams at any point in time, creating istries back-loading costs to later years) that demand for systematic monitoring and reduce the transparency of the budget and impact assessment capacity and for client its alignment with overall priorities and the feedback. practicalities of what works. For all their advantages however, In recent years several countries have medium-term expenditure frameworks are approached these problems of transparency not a magic bullet. Aggregate and sectoral and results orientation in budget formula- outcomes and capacity development reveal tion through medium-term expenditure a mixed picture. Some applications are frameworks. These multiyear frameworks maturing slowly (in Albania, South Africa, make tradeoffs more transparent across sec- Uganda), some are still coming together tors and time and synchronize medium- (Rwanda, Tanzania), and some are strug- term priority setting with the annual bud- gling (Bolivia, Burkina Faso, Cameroon, get cycle. They offer the promise of better Ghana, Malawi).489 In Malawi’s develop- budget management, though early imple- ment budget for 1996–97, health was allo- mentation suggests that realizing these cated at 21 percent of the total but it gains takes quite a bit of time, effort, and received only 4 percent.490 Implementing parallel improvements in budget execution medium-term expenditure frameworks is and reporting.488 dif�cult, perhaps taking a dozen years or Properly implemented, a medium-term more, as the experience of early adopters expenditure framework can reduce incen- such as Uganda demonstrates. A solid foun- tives for bureaucratic gaming and reveal dation of budget execution and reporting the true costs of the political choices being seems key, but is also dif�cult to achieve. made in the budget. It can usefully address Implementing a medium-term expenditure the information asymmetry between the framework can help build the basics, as can ministry of �nance and line agencies, participatory budgeting initiatives dis- because its forward-estimate system cussed in chapter 5. Other success factors requires line ministries to cost their pro- include carefully matching implementation grams over the medium term—essentially to capacity, keeping budget projections and a rolling three- or four-year budget. A estimates realistic, distinguishing between properly functioning forward-estimate collective ministerial responsibility in the system can induce line agencies to set cabinet and the interests of individual min- aside funding for recurrent costs and istries, and engaging line ministries in the improve the delivery of services suffering strategic phase prior to considering detailed from inadequate maintenance, such as estimates, when the rationale and instru- primary schools. ments for public intervention can be care- As the capacity to manage grows, a fully thought through. medium-term expenditure framework can offer other advantages. Sector-speci�c Operational impact expenditure frameworks can be developed Ultimately, even the best budget allocations and linked to the overall framework, are only as good as their impact on desired increasing con�dence that the budget is outcomes for poor people. After controlling becoming more results-oriented (chapter 8 for national income, comparative studies discusses this approach to health budgeting show that public spending per capita and in Mali). With a multiyear framework poli- outcomes are only weakly associated (c) The International Bank for Reconstruction and Development / The World Bank 184 WORLD DEVELOPMENT REPORT 2004 (chapter 2). Similar changes in spending are and the International Monetary Fund and associated with different changes in out- for debt relief under the enhanced Heavily comes, and different changes in spending Indebted Poor Countries (HIPC) Debt Ini- are associated with similar changes in out- tiative.491 Many countries and donors have comes. This is not to suggest that public stressed better public expenditure manage- funding cannot be successful—countries ment as a means of tracking pro-poor like Thailand have sharply reduced infant spending and increasing donor and recipi- mortality rates through commitment, good ent accountability for external assistance. A policies, and spending. But it does mean recent review of the pro-poor expenditure that unless public expenditures are results- tracking capacities of budget management oriented they will be ineffective. There has systems among HIPCs suggests that they been a major push in recent years to make have far to go.492 While recognizing that policymakers and providers accountable improvements in public expenditure man- not only for how they spend money but also agement will take time,493 both domestic for what they achieve—for intermediate stakeholders and donors have highlighted outputs and �nal outcomes. Countries are the need for developing and implementing using several instruments: single-sector and detailed plans for improvement. multisector program approaches, align- Ideally, poverty strategies should be fully ment of overall national strategies with integrated into the budget, but this is still a budgets, tools for verifying where the new approach and success has varied. For money goes, and stronger oversight con- some countries integration has been a pri- trols to reduce fraud and misuse of public mary goal (Albania). Tanzania and Uganda funds. have integrated poverty strategies with their medium-term expenditure frameworks, Programmatic approaches. Individual in- adding focus, legitimacy, and stability to vestment projects can fall short of their both. But other countries have assigned objectives if they ignore linkages or trade- responsibility for preparing their poverty offs over time and space or with other sec- strategy to a ministry not directly con- tors. Chapter 11 discusses sectorwide cerned with public expenditure planning. approaches as a way of enhancing develop- In Ghana, it was initially assigned to the ment impact, building stronger donor part- planning ministry, though more recently nerships, improving the management of the planning portfolio has been folded into sector resources, and scaling up successes. the �nance minister’s portfolio. Used in countries as diverse as Bangladesh, Bolivia, Brazil, Burkina Faso, Ethiopia, Public expenditure tracking surveys. In Ghana, Mali, Mozambique, Pakistan, Tan- judging operational impact—the quality zania, and Zambia, sectorwide approaches and quantity of service delivery, and where, show that over time strategies and objec- how, and to what effect allocated funds are tives are better articulated, and manage- spent—public expenditure tracking sur- ment information, monitoring, evaluation, veys can follow the flow of funds through and resource planning systems better estab- tiers of government to determine whether lished, in sectors that use such approaches the funds actually reach the schools or clin- than in those that do not. ics they are destined for. Tracking surveys not only highlight the uses and abuses of Poverty reduction strategies. A country’s public funds, but also give insights into poverty reduction strategy can link public capture, cost ef�ciency, decentralization, expenditures explicitly to service delivery and accountability.494 Even when little for the poor, build country ownership, and �nancial information is available, tracking strengthen citizen voice through consulta- surveys can show what money is supposed tions with civil society. In 1999 low-income to reach a community and how much actu- countries began preparing Poverty Reduc- ally does. Made public, this information tion Strategy Papers (PRSPs) as the basis for can strengthen voice and client power rela- concessional lending from the World Bank tionships (box 10.2). (c) The International Bank for Reconstruction and Development / The World Bank Public sector underpinnings of service reform 185 Financial management. Auditing helps a government hold itself accountable for the BOX 10.2 The case of the missing money: public way policymakers and providers spend expenditure tracking surveys money. Audits have traditionally focused on In the early 1990s the Ugandan government tion programs into a social fund that will basic �nancial controls and cash flows. This dramatically increased spending on primary transform Vaso de Leche into a conditional, focus reflects the control culture in public education. But school enrollments stagnated. multipurpose, cash-transfer program with �nance and the long-established view that Could it be that the money was not reaching stronger accountability. accountability for fund use supports the dis- schools? To answer this question, a public These and other tracking surveys in expenditure tracking survey started collect- Chad, Ghana, Honduras, Mozambique, ciplined use of resources as intended by bud- ing data in 1996 on government transfers to Papua New Guinea, Rwanda, Senegal, Tanza- gets. In recent years, however, accounting and schools. It found that 87 percent of the non- nia, and Zambia suggest several lessons. auditing processes have been challenged to wage resources intended for the schools was They con�rm that budget execution is a examine expenditure performance as well as diverted to other uses.This information was major problem and show that procedural made public and prompted a vigorous clarity and due process are often missing. conformance. The new performance orienta- response from the national government, They �nd that poor resource management tion of audits is particularly relevant to oper- which, along with parents, put pressure on is often a result of too much discretion in ational ef�ciency concerns in budgets and school principals to plug the leaks (see spot- resource allocation when there is limited suggests an expanded notion of accountabil- light on Uganda). Follow-up studies have information, weak controls, and strong shown that the situation has improved. vested interests.Tracking surveys reveal ity. Public �nancial managers now need to Tracking surveys can �nd problems in insights into the actual (rather than the for- consider their roles as contributors to �nal unexpected places. A survey in Peru track- mal) operation of schools and health clinics outcomes as well as controllers.495 ing a participatory food supplement and allow comparisons of public, private, program (Vaso de Leche, or “Glass of Milk�) and nongovernmental providers.Tracking Procurement. The cost and quality of gov- revealed that less than a third of each dollar surveys are highly cost-effective if the leaks transferred from the central government they detect are plugged. But they need an ernment programs are critically affected by the reached intended bene�ciaries. Most of the authorizing environment: unless there is a procurement process through which budgets leakage occurred below the municipal solid political commitment for more trans- are spent. Procurement inevitably encom- level—in the Mothers Committees and parency, government agencies may be passes an intricate set of rules and procedures, households.The results challenged the reluctant to open their books.The challenge belief underlying the program that local is to institutionalize tracking surveys within each capable of retarding or promoting trans- community organizations were always a country’s own �nancial control regime. parency, contestability, accountability, and ef�- more accountable than public agencies. ciency. Leakages, primarily through fraud and Authorities have decided to merge all nutri- Source: World Bank staff. corruption, can mean substandard equipment and infrastructure, lack of essential medical supplies, insuf�cient textbooks, unnecessary tries, China and India, have embraced decen- low-priority goods, and poor-quality public tralization. China’s phenomenal industrial services. Inef�cient procedures create higher growth took place within an institutional costs for suppliers, which are passed through framework of decentralization, and India’s as higher program costs. Improving procure- constitution was amended in 1992 to pro- ment requires extensive analysis of its rules, mote local government.497 But the extent of procedures, and institutional arrangements. decentralization varies considerably and is To support streamlining, several countries probably less than generally imagined: even have turned to information and communica- in developed countries the average subna- tions technology. Brazil, Chile, Mexico, the tional share of expenditures was just above 30 Philippines, and the Republic of Korea, among percent in recent years (�gure 10.2). others, have developed strong e-procurement Subnational authorities can be ef�cient systems that lower costs and increase trans- providers and regulators of local services parency, competition, and ef�ciency.496 under the right institutional incentives and with clarity about who does what—and with Decentralizing what.498 But greater autonomy can also to improve services increase opportunistic behavior and create In countries big and small central govern- moral hazard, resulting in costs that diminish ments are transferring responsibilities to accountability and the bene�ts of decentral- lower tiers of government, motivated in part ization.499 Good design, sound management, by the desire to bring politicians and policy- and constant adaptation by both central and makers closer to clients and to make services subnational authorities are needed to make more effective. The world’s two largest coun- decentralization work. (c) The International Bank for Reconstruction and Development / The World Bank 186 WORLD DEVELOPMENT REPORT 2004 Figure 10.2 Subnational shares of expenditures vary Decentralization and service delivery considerably Decentralization is not magic. Allocating By country, latest available year more responsibilities to subnational gov- Peru State or provincial Indonesia ernments does not itself transform service Argentina delivery. This depends on whether decen- India Panama tralization is motivated by political, �scal, Costa Rica or service reform objectives. Nicaragua Trinidad and Tobago Decentralization is often primarily a Mauritius political act aimed at greater regional Kenya Local Botswana autonomy. Decentralization of services is a Portugal by-product (box 10.3). Indonesia decentral- Thailand ized responsibility for many services in Romania Belgium 1999–2000, including schooling, as part of a Croatia larger move to greater regional autonomy. Israel Luxembourg In such cases decentralization is a fact of life France educators must cope with—not a deliberate Albania Bulgaria educational reform. New arrangements can Czech Republic always create opportunities for reform, Lithuania Poland however. Using those opportunities effec- Hungary tively depends on two conditions. First, Latvia Netherlands there must be relevant information about Italy performance across jurisdictions so that cit- United Kingdom Iceland izens can bring justi�ed pressure to bear on Norway politicians and policymakers if their area is Sweden lagging. Second, there must be an environ- Mongolia Belarus ment in which local jurisdictions can exper- Finland iment and evaluate new approaches. Russian Federation Denmark Decentralization may also be driven by Netherlands Antilles �scal concerns to align responsibility for Malaysia Austria services with the level of government best Bolivia able to manage and mobilize resources for Spain Germany them. One danger is that the central gov- Brazil ernment uses this as an excuse to off-load United States Australia expenditure responsibilities onto jurisdic- Switzerland tions that cannot have recourse to poten- South Africa tially inflationary �nancing. While this 0 10 20 30 40 50 60 70 80 could lead to a greater willingness to pay Percent more local taxes (because citizens perceive a By region, latest available year direct link between taxes and service qual- Sub-Saharan Africa (4) ity), there is no reason to believe that this is East Asia and the automatic. Fiscally motivated decentraliza- Pacific (4) tion is particularly worrisome where special Latin American and the equalization efforts for lagging regions or Caribbean (9) Europe and Central safety nets for poor families must be sus- Asia (13) tained by the center. High Income, Decentralization can also be driven by a OECD (18) desire to move services administratively South Asia (1) closer to the people. But success depends on 0 20 40 60 how decentralization affects relationships Percent of accountability. If decentralization just Note: Simple averages of most recent observations for countries replaces the functions of the central min- with available data. Numbers in parentheses indicate number of countries represented. South Asia refers only to India. istry with a slightly lower tier of govern- Source: IMF, Government Finance Statistics; World Bank staff. ment (a province or state), but everything (c) The International Bank for Reconstruction and Development / The World Bank Public sector underpinnings of service reform 187 else about service delivery remains the same, there is little reason to expect positive BOX 10.3 Decentralization as a political imperative: change. The assumption is that decentral- Ethiopia ization works by enhancing citizens’ voice In Ethiopia decentralization has been a the primary education curriculum and in a way that leads to improved services. But response to pressures from regional and eth- teacher training.The syllabus remained cen- on both theoretical and empirical grounds nic groups for greater political participation. trally controlled, with input from the this could go either way. The crucial ques- When the Ethiopian People’s Revolutionary regions. Previously, Amharic had been the tion always is whether decentralization Democratic Front (EPRDF) defeated the sole language of instruction, but the new Mengistu dictatorship in 1991, the new gov- policy gave all children the right to receive increases accountability relative to its alter- ernment faced a complex political landscape. primary education in their mother tongue. natives. If local governments are no more Ethnicity was extremely politicized, and the At least 18 languages are now being used vulnerable to capture than the center is, struggle against Mengistu had been spear- as the medium of instruction, although decentralization is likely to improve both headed by organizations promoting ethnic Amharic remains the national language. nationalism in Ethiopia’s diverse population. Politically motivated decentralization ef�ciency and equity.500 The single-party EPRDF government needed carries implications that are critical for the The impact of decentralization on ser- to establish control over the entire country, impact of reform.The education policies vices is further complicated when, as is usu- legitimize its authority, and include other adopted along with political decentraliza- ally the case, political, �scal, and administra- groups in the political system. tion may well be good ideas for improving The 1994 constitution transformed the quality of instruction and learning. But if tive goals are not followed simultaneously or Ethiopia into an ethnicity-based federation improving quality is not a central objective in a supportive sequence. Decentralization and decentralized administrative responsi- of decentralizing Ethiopian education, the in eight Latin American countries suggests bilities to nine regions.The accompanying resulting lack of commitment to ensuring education reforms were laid out in the “Edu- that outcome could become the most dif�- that political objectives were often the trig- cation and Training Policy of 1994.� Regions cult obstacle to overcome. ger, but paths diverged thereafter (box 10.4). were given responsibility for planning, Only some countries moved on to �scal and designing, implementing, and monitoring Source: Pritchett and Farooqui (2003). administrative decentralization as primary objectives. Such variation, inevitable as countries adapt, makes it hard to predict the Decentralization course of decentralization and to measure and accountability for services its costs and bene�ts.501 Given its many Decentralization must reach the clinic, the paths, the record of service improvements is classroom, and local water and electricity util- mixed—including some notable successes ities in ways that create opportunities for (decentralizing education in Central Amer- strengthening accountability between citi- ica, devolution in Bolivia, municipal zens, politicians/policymakers, and providers. reforms in South Africa), some reversals (in Depending on its degree—deconcentration, the Russian Federation and parts of Latin delegation, and devolution—and its imple- America), and some cases too new to assess mentation, decentralization offers opportuni- (initiatives in Indonesia and Pakistan).502 ties for strengthening different parts of the BOX 10.4 Many roads to decentralization: Latin America Decentralization in Latin America shows how objec- This experience in Latin America shows that the tives changed over time in each country and shaped transfer of political, �scal, and administrative power outcomes and the path of decentralization. Where does not necessarily occur simultaneously or in a decentralization was driven mainly by political supportive sequence. In fact, only in Bolivia’s reform objectives (as in Ecuador, Peru, and Venezuela), the effort in 1994 were these powers transferred transfer of resources was often signi�cant, but the together. Chile democratized in 1990, introduced transfer of responsibilities was more dif�cult to pur- popular participation but not regional elections and sue. Where political decentralization was joined and devolution, and in the mid-1990s further deepened driven by sophisticated but misaligned regional �s- the administrative delegation that had marked its cal autonomy (as in Argentina and Brazil), cyclical earlier military regime. Of these countries Chile may economic and political crises erupted because of the now be best placed to attempt deeper administra- inability of the center to impose �scal discipline on tive and political devolution because of the growth subnational governments. In Colombia, though of local capacity and the absence of the regional �s- decentralization was initially driven by political cal crises that struck many of its neighbors on their motives, �scal and administrative adjustments ran road to decentralization. deeper, and cyclical adjustments in the �scal and administrative systems were common. Source: Frank, Starnfeld, and Zimmerman (2003). (c) The International Bank for Reconstruction and Development / The World Bank 188 WORLD DEVELOPMENT REPORT 2004 Figure 10.3 Decentralization and the service delivery framework Center Politicians Policymakers on oluti Subnational government Dev on ati n Politicians Policymakers g atio Dele entr onc Co mp De c ice a ct Vo Citizens/clients Providers Client power Nonpoor Poor Frontline Organizations Services service delivery chain (�gure 10.3). Deconcen- ter, three areas are key: subnational �nance, tration affects primarily the compact relation- the division of administrative responsibilities ship between central policymakers and their between center and subnational govern- local frontline providers and may have little ments, and local capacity. influence on local voice. At the other end, devolution implies the handing over of greater Getting �scal incentives right power and resources to local politicians and A subnational government will have weaker therefore greater scope for strengthening local incentives to deliver cost-effective services that voice, their compact with local providers, and meet minimum standards if it can manipulate local client power. Delegation falls in between. funding (from the center or from market bor- The degree of decentralization thus impacts rowing) to shift its liabilities to the center differently on the short and long routes of (called a soft budget constraint).504 Subna- accountability (table 10.2). In practice, decen- tional liabilities can be contractual, �scal tralization inevitably involves a mix of decon- de�cits, or public goods that are underpro- centration, delegation, and devolution. vided. A soft budget constraint weakens Particularly when local taxing and spend- accountability, creates moral hazard, and ing powers and central �nancing are well threatens macroeconomic stability by creating matched, decentralization can create checks contingent liabilities for the center that it may and balances that can motivate both central �nd hard to refuse to pay. Underdeveloped and subnational governments to make local capital markets and elections that do not services work. But accountability may not penalize local politicians for cost and de�cit improve, and the potential gains of decentral- shifting are part of the problem. A hard budget ization may be lost, if the �scal and other constraint strengthens accountability but incentives underlying the center-subnational requires a sound intergovernmental �scal sys- relationship are misaligned so that checks and tem. The center, having devolved responsibil- balances do not work. A study of the transfer ity and resources, is prodded by a hard budget of responsibility for secondary schools to constraint to support effective subnational provinces in Argentina in 1994–98 found that management and service delivery, thereby while average test scores improved, the gains avoiding �scal problems and unhappy citizens. were much lower when schools were trans- ferred to severely mismanaged provinces (as Getting the intergovernmental �scal system measured by provincial �scal de�cits).503 right. Standard welfare economics suggests To allow decentralization to reach local the ef�ciency and equity grounds for assign- classrooms, clinics, hospitals, and public ing expenditure responsibilities, revenues, works departments in a way that increases and grants to lower tiers of government.505 accountability and makes services work bet- Service decisions and expenditures should be (c) The International Bank for Reconstruction and Development / The World Bank Public sector underpinnings of service reform 189 Table 10.2 Decentralization is never simple Key political, �scal, and administrative features of decentralization and the accountability for service delivery Degree of decentralization Political features Fiscal features Administrative features Deconcentration • No elected local government • Local government is a service delivery arm • Provider staff working at local level are (minimal change) • Local leadership vested in local of�cials, of the center and has little or no discretion employees of center, and accountable to such as a governor or mayor, but appointed over how or where services are provided center, usually through their ministries; weak by and accountable to the center • Funds come from the center through local capacity is compensated for by central • Voice relationships are remote and individual central ministry or department employees possibly weak budgets • Accountability remains distant: the short route • No independent revenue sources of accountability may be weak if provider monitoring is weak and citizens may have to rely on a weak long route stretching to politicians at the center; a strong compact between policymakers and providers can compensate to some extent Delegation • Local government may be led by locally • Spending priorities are set centrally, as well • Providers could be employees of central or (intermediate change) elected politicians, but it is still accountable, as program norms and standards; local local government, but pay and employment fully or partially, to the center government has some management conditions are typically set by center • Voice relationships are more local and authority over allocation of resources to • Local government has some authority over proximate, but can be overruled by center meet local circumstances hiring and location of staff, but less likely to • Funding is provided by the center through have authority over �ring transfers, usually a combination of block • Both long and short routes of accountability and conditional grants potentially stronger; greater local knowledge • No independent revenue sources can allow better matching and monitoring of supply with local preferences, strengthening both the compact and client power Devolution • Local government is led by locally elected • Subject to meeting nationally set minimum • Providers are employees of local government (substantial change) politicians expected to be accountable to standards, local government can set • Local government has full discretion over the local electorate spending priorities and determine how best salary levels, staf�ng numbers and allocation, • Voice relationships can be very strong, but to meet service obligations and authority to hire and �re also subject to capture by elites, social • Funding can come from local revenues and • Standards and procedures for hiring and polarization, uninformed voting, and revenue-sharing arrangements and transfers managing staff may still be established within clientelism from center an overarching civil service framework • A hard budget constraint is imperative for covering local governments generally creating incentives for accountable service • Potentially strongest long and short routes of delivery accountability, but now also more influenced by local social norms and vulnerable to local capacity constraints and politics Note: See the glossary in chapter 3 of this Report for de�nitions of accountability terms (in italics). Source: Based on Evans (2003). devolved to the lowest tier of government Also important are simple, transparent, for- that can internalize the costs and bene�ts of mula-based transfers from the center that the service—the so-called subsidiarity princi- are predictable over several years. If made ple. The principle suggests that subnational contingent on service outputs, lump-sum governments should administer basic health grants can ensure a minimum level of ser- and education services. But setting minimum vice delivery for poor people, equalize �scal standards (for quantity, quality, and access) capacity across jurisdictions, and create per- and �nancing minimum access should be formance incentives. Ideally, expenditures, central responsibilities on grounds of inter- revenue assignments, and transfers should jurisdictional equity. In practice, things get be designed jointly so that once they are set, more complicated. Expenditures are often any additional expenditure demands could not assigned carefully to subnational govern- be met through taxes rather than grants.507 ments.506 Central governments delay trans- The more these principles are violated, the fers. Shared expenditure responsibilities are greater the informality around transfers, the trickiest to handle and can lead to free- and the lower their predictability and stabil- rider problems and de�cit- and cost-shifting ity, the softer the budget constraint gets.508 behavior that softens the budget constraint. To increase responsiveness to local citi- Getting subnational borrowing right. Cap- zens, subnational governments need a local ital markets, where suf�ciently developed, tax instrument and the freedom to set rates. can bolster subnational accountability. (c) The International Bank for Reconstruction and Development / The World Bank 190 WORLD DEVELOPMENT REPORT 2004 Where markets are underdeveloped and mar- Though the 1992 landmark amendments ket discipline is weak, a prior question is to the Indian constitution require each whether subnational governments should state to create urban and rural local gov- borrow at all. Effective �scal decentralization ernments and assign functions and rev- should certainly precede �nancial decentral- enues, virtually all staff at the local level ization to avoid giving the signal that the cen- remain state employees. In contrast, ter is underwriting subnational debts.509 Indonesia recently adopted a “big-bang� Allowing subnational borrowing from public approach, moving quickly to transfer �nancial institutions can unintentionally send roughly 2.1 million civil servants to sub- this signal. In Argentina, Brazil, India, and national district governments.512 Uganda, Ukraine, specialized development banks and in shifting from deconcentration to devo- institutions have provided a backdoor route lution in the 1990s, established district to central subsidies when transfers would service commissions with the authority to have been simpler and more transparent. hire and �re personnel—though in prac- tice central policy and administrative Getting subnational regulation right. rules have tightly controlled the process so Governments �nd it hard not to bail out that it has resembled delegation more lower-tier governments when �nancial than devolution. That may change as local profligacy threatens basic services, risks capacities grow. Pakistan’s recent three- spreading to other jurisdictions, or threat- tier devolution envisages the creation of ens monetary policy or the country’s credit district and subdistrict cadres: district rating. This has led to the imposition of health and education cadres have been top-down regulation, either administrative created in some provinces, but adminis- controls or rule-based debt restrictions that trative decentralization still has a long way mimic the market. Regulation, because it is to go. vulnerable to political bargaining, usually National pay scales, rigid collective bar- needs to be supplemented by checks and gaining agreements, and disagreements with balances on the center itself so that its national labor unions can severely circum- stance remains credible. In South Africa scribe the flexibility that subnational gov- these are provided by the constitution, the ernments have in rationalizing employment, constitutional court, and international cap- as seen in many Asian, African, and Latin ital markets.510 Subnational bankruptcy American countries.513 Centralized labor arrangements can help. A control board negotiations and bargaining agreements can (that can be invoked only by an indepen- act as unfunded mandates that undo �scal dent court) to �nance minimum, nationally decentralization (as in South Africa). Engag- set service levels can protect the center from ing public sector workers and unions in dis- having to step in. Where bailouts are cussions about different aspects of decen- unavoidable, the center can use the oppor- tralization can increase local flexibility and tunity to make regulation more effective. A improve provider compacts. At the same comprehensive �scal monitoring and evalu- time, administrative devolution needs to ation system that works consistently across strike a balance between autonomy and uni- jurisdictions can help greatly in implement- formity to allow for desirable features such ing no-bailout and regulation strategies. as interjurisdictional mobility for highly skilled staff in short supply. It is important Getting administrative to align the structure of the civil service with responsibilities right the assignment of service responsibilities to Political and �scal considerations gener- different tiers—misalignment confuses ally claim far greater attention than incentives, weakens accountability, and cre- administrative decentralization does.511 In ates conflicts of interest instead of checks many instances decentralization has pro- and balances. In practice this is not easy, and ceeded without explicit staf�ng strategies, it takes time. and a central civil service typically coexists The twin tasks of devolving administra- with subnational and local governments. tion and building local capacity can be (c) The International Bank for Reconstruction and Development / The World Bank Public sector underpinnings of service reform 191 daunting even under ideal conditions of budget constraint and the relationships of budget and stakeholder support. When voice and client power are weak, subna- budgets are constrained and support is tional governments will have little incentive mixed, public administration reform is to develop local capacity and perform well, inevitably drawn out, falling behind politi- which will make local capture by elites cal and �scal decentralization. So the earlier more likely. Ultimately, the center is both the start in building local capacities, the regulator and facilitator of decentraliza- smoother the process of decentralization is tion. Its challenge is to balance these roles likely to be. as it makes and manages the policy frame- work for the public sector and for service Building local capacity with delivery. autonomy Making, managing, Decentralize or build local capacity: which �rst? In an ideal world subnational govern- and implementing good policies ments would be made fully accountable When the policy decisions of politicians and before they were given authority and auton- policymakers at the center of government— omy. Decentralization in the absence of ade- senior decisionmakers and veto holders in quate local capacity was once considered the executive, council of ministers, or cabi- undesirable,514 but that view is changing as net—are uncoordinated, inconsistent, or experience shows that local capacities badly implemented, the long route of expand best as decentralized systems mature, accountability and service delivery are likely even though sequencing remains dif�cult. to suffer. Breakdowns in policy management The challenge is to balance political, �scal, can include a range of failures (box 10.6). and administrative considerations even Sure signs of breakdown? When political pol- when capacity mismatches occur. Where icy decisions are not implemented, partially local institutions already exist, even informal implemented, or reversed.515 A study of two ones, the challenge is to de�ne their respon- African nations revealed that more than sibilities and legal status and move the infor- mal closer to the formal. Where local institu- tions do not exist, the challenge is to BOX 10.5 Building local capacity: the role of the center construct the underlying legal and political Devolution is dif�cult when subnational gov- on a competitive or matching basis to sup- framework for new institutions. ernments lack skills and institutional capac- port local governments that achieve perfor- Fostering capacity is best done in part- ity.The central government can provide mance benchmarks (implementing a bud- nership between the center and subnational training in top-down ways. Or it can create geting system, attaining service targets). governments, with the center providing an enabling environment, using its �nance Monitoring and evaluation capacity can and regulatory powers to help subnational also be facilitated by performance-based incentives for subnational governments to governments de�ne their needs (making the incentive grants. Monitoring efforts should match demand-driven capacity growth process demand-driven), to deploy training feed into stronger public communication with supply-side assistance and �nancing from many sources (local or national private and outreach efforts so that subnational (box 10.5). In this partnership the functions sector), to learn by doing as decentralization governments can bene�t from better client proceeds, and to establish learning networks feedback. of central staff also change, from line man- among jurisdictions.This second approach is Successful capacity building requires a agement to policy formulation, technical more consistent with devolution and more phased strategy, starting with the stabiliza- advice, and monitoring. Central staff likely to produce capacity tailored to the tion of core responsibilities. Next comes a require incentives and training to do their many cross-sector responsibilities of subna- transformation phase with restructuring tional governments. It also avoids the pitfalls plans based on a critical examination of ser- new jobs effectively. of a supply-driven approach. vice responsibilities and priorities, institu- The center may need to provide capac- tional arrangements, and �nancial and ity support, through both a demand-driven human resources. Finally, a consolidation Pulling the pieces together grant facility (for example, to help phase seeks to internalize capacity growth Decentralization fails or succeeds in the subnational governments contract local based on constant learning by doing and interplay of its �scal, administrative, and and other expertise) and a supply window adaptation.This is inevitably a drawn-out (for example, mobile teams with �nancial process marked by the constant need to local capacity attributes. The center’s role is management, technical, and community balance greater autonomy and capacity. crucial for all three elements and, more mobilization skills). Fiscal support through broadly, for the design and implementation block grants or challenge funds can work Source: World Bank staff. of decentralization. When there is a soft (c) The International Bank for Reconstruction and Development / The World Bank 192 WORLD DEVELOPMENT REPORT 2004 strong. Policy management in these sectors BOX 10.6 “Yes, Minister� is often an outcome of well-informed bar- gaining between competing domestic inter- Breakdowns in policy management cover a bined with a failure to consult all wide spectrum: ministries with a stake in a particular ests, so accounting for domestic political decision. concerns is important. By contrast, macro- • Failure to set major policy priorities, to understand tradeoffs and make tough • Failure to consult external stakeholders, economic management tends to be the pre- choices between conflicting objectives, anticipate opposition, and build electoral serve of a few relatively insulated tech- or to translate priorities into concrete support through the relationship of voice. nocrats, with the central bank and �nance operational decisions, most typically ministry as key veto players, crises having to through the budget process. • Poorly drafted and inadequately costed be dealt with expeditiously, and domestic • A policy vacuum, because of government submissions (particularly ignoring down- stream expenditures), and proposals not political concerns often not included in discontinuity or weak or poorly articulated policies. vetted thoroughly for their legality and decisionmaking.518 consistency with previous policies. • Lack of trust between politicians and pol- How a cabinet secretariat or presidential icymakers, leading to frequent end runs • Parallel groups, often invisible and unac- staff that links politicians with policymak- around formal decision structures. countable, influencing policy from out- side formal government. ers plays its role can be crucial to the ef�- • Unclear organizational roles or conflict- cacy of policy management in these sectors. ing agendas among line ministries, com- Source: Beschel and Manning (2000). Members of these staffs, which often include elite advisory groups that provide two-thirds of cabinet decisions were never high-quality policy advice, can be vital gate- implemented.516 In Zambia genuine support keepers. They can use contestability—or the for reforms introduced by the multiparty careful evaluation of alternatives—to democratic government in the early 1990s sharpen policy advice. In Thailand the never extended beyond a few cabinet minis- National Economic and Social Develop- ters. As a result, special interest groups, who ment Board in the prime minister’s of�ce had not been consulted, slowed implementa- provides independent �scal analysis of tion to a crawl.517 Such missteps are possible social sector initiatives and has promoted a at each stage of the policy management coordinated and participatory institutional process (�gure 10.4). response to Thailand’s HIV/AIDS crisis. Cabinet committees, consisting of subsets Getting good policies in education, of ministers, their representatives, policy- health, and infrastructure makers, and sometimes outside experts, can Policy management is particularly dif�cult be particularly effective for intersectoral in health, education, and infrastructure coordination and implementation and for because outcomes such as reduced infant identifying contending views and resolving mortality have multiple determinants that them before the formal decision process. cross sectors and jurisdictions (see crate Research on cabinet functioning suggests 1.1); costs come early and impacts much the conditions that favor high-quality policy later; and the spillover effects of services are management in dealing with complex multi- Figure 10.4 The anatomy of policy mismanagement at the top Policy mismanagement by politicians Common veto/delay point for external actors 1 2 Departments prepare 3 4 Departments Politicians make Politicians fail to Weak compact and uncoordinated implement poor-quality unrealistic, unaffordable provide adequate confused providers or poorly costed or unauthorized policy commitments budgets policies initiatives Policy mismanagement by policymakers Source: Adapted from Blondel and Manning (2002). (c) The International Bank for Reconstruction and Development / The World Bank Public sector underpinnings of service reform 193 sectoral issues: discipline (decisions are real- accountability of public of�cials and agen- istic and can be implemented), transparency cies. Analysis suggests that the basic drivers (systematic procedures that cannot be of performance are merit-based recruitment manipulated by individual members and that and promotion, adequate compensation, emphasize collective responsibility), stability and reasonable autonomy from political (no flip-flopping), contestability (considera- interference.520 As a result of the different tion of alternatives), and structured choice approaches, the share of public employment (only core issues come before the cabinet).519 in total employment varies widely: for the Of these conditions, discipline seems most period 1997–99 it averaged 38 percent in important. Practices vary. In the Netherlands transition economies (16 countries), 24 per- all items requiring cabinet approval are spec- cent in industrial countries (20), and 21 per- i�ed in the rules of business, and in Finland cent in developing countries (23).521 General almost every government decision requires government employment in education and cabinet approval. In Australia the Cabinet health and in central and subnational gov- Expenditure Review Committee ensures col- ernment also varies considerably (�gure lective responsibility and contestability for 10.5). The differences reflect the different spending proposals—ministers have every roles of the state and public administration incentive to test the new spending proposals in individual countries, institutions that have of their colleagues so as to maximize the pool historic roots and cannot be changed of uncommitted budget funds available for overnight. their own proposals. More realistic �scal forecasting and dis- cussion rules that allow sensible tradeoffs to Figure 10.5 Working to keep citizens educated, healthy, and safe General government employment, mid- to late-1990s emerge between key service sectors may be needed to avoid overcommitment at early Australia stages of the policy management process. A Bolivia cabinet of�ce that can negotiate feasible pol- Botswana icy and legislative programs with line departments, analyze policy proposals, and Brazil Central government coordinate without itself developing policy Cambodia Subnational government (to avoid conflicts of interest) can thereafter Canada Education help ensure delivery on policy and budget Chile Health proposals on these commitments. In estab- Armed forces China lishing budgets, a multiyear budget frame- work may help ensure adequate funding and Ecuador reduce budget instability if supported by Finland politicians and the requisite implementation Hungary capacity. Communication, outreach, and India consultation can forestall opposition and Morocco improve implementation plans. Finally, the compact may need to be strengthened so Poland that neglect, incompetence, misapplication, Russian Federation or malfeasance does not prevent executive South Africa decisions from being implemented or cause Thailand those that are implemented to be flawed. United Kingdom Making strategic choices in public United States administration and management Zambia Choosing how to implement good policies is 0 5 10 15 20 25 30 as important as making them. Countries Percent of total employment have experimented in the past two decades Note: General government excludes employment in state-owned enterprises. Central and sub- with different public administration ap- national government totals exclude health, education, and police personnel. Armed forces excludes police. proaches to improving the performance and Source: World Bank (2002e). (c) The International Bank for Reconstruction and Development / The World Bank 194 WORLD DEVELOPMENT REPORT 2004 The New Public Management philoso- necessary when the policymaker-provider phy has dominated the debate on public relationship is weak and agencies lack com- administration reforms in recent years. petence and effective internal controls. But as Implemented principally in Australia, New long as external controls are in place, line Zealand, and the United Kingdom, it recog- agencies lack the incentive to acquire compe- nizes the government’s special role in ser- tency and establish internal controls. vice provision, �nancing, or regulation, and The answer? Choosing and sequencing the resulting incentive problems. It seeks to public sector reforms carefully, in line with strengthen accountability by exchanging initial capacities, to create �rmer ground for management flexibility for internal con- further reform. Pragmatic, incremental tracting among policymakers and between reforms in weak institutional environments— policymakers and providers. New Public strategic incrementalism—can alleviate, if not Management also seeks to provide a more fully resolve, accountability problems while transparent accounting system and tighter, creating the conditions for deeper change by private sector–like �nancial management modifying incentives and building capacity to controls. In its extreme form civil servants respond to the next stage of reforms. Thailand have no tenure, and their term in of�ce and is considering a “hurdle� approach to reform- promotion depend on successful comple- ing its centralized budget system. Line agen- tion of contract-speci�ed deliverables. cies would clear a series of hurdles to qualify Experience in developing countries has at each level for greater budget autonomy. In been mixed, with some improvements in this incremental approach to budget reform, ef�ciency and uneven effects on equity.522 the dismantling of external controls has to be In the weak institutional settings of many synchronized with the building of internal developing countries, New Public Manage- controls.524 Other countries can follow a more ment reforms may impose high transaction traditional but still sequenced path of budget costs that may outweigh ef�ciency gains. reforms, differentiating between short- and As the experience with New Public Man- medium-term measures and building infor- agement suggests, often the problem in mation channels for accountability as the implementing public sector reforms is not reforms unfold—another form of strategic deciding on reform objectives but on how to incrementalism (�gure 10.6). get there. In Bolivia agencies were given man- agement flexibility in the early 1990s, but Formality in public sector institutions. there were no central controls to enforce Many aspects of government performance accountability. Public administration prob- rest on an ingrained institutional discipline lems remained.523 The Bolivian experience or formality. Actual behavior follows written highlights the “catch-22�: central controls are rules, or actual budget outcomes bear a close resemblance to the legislatively agreed bud- Figure 10.6 No straight roads to success: sequencing budget reforms get.525 Informality emerges in weak institu- tional settings where incentives and proce- Short term dures do not match formal rules, rewards, 1 Establish financial management and procedures.526 criteria in performance contracts. Fiscal data and This formality gap is most evident in per- information 2 Deploy systems to improve the quality of fiscal data in the sonnel and budget problems. Teacher absen- 2 budget. teeism in many countries exposes the stark 3 Strengthen internal and external differences between explicit rules on recruit- 3 audit. ment, promotion, pay determination, and 1 monitoring and the actual, informal arrange- Medium term ment of connections and patronage that Motivated Checks and staff 4 balances 4 Implement pay reform and determine who gets hired and even whether restructure line agencies. 5 they have to show up at all. Lateral entry to 5 Deepen accountability, including parliamentary and civil society the civil service, intended to provide flexibil- involvement in budget scrutiny. ity and contestability, becomes a window for Source: Girishankar (2002). patronage and nepotism when of�cial posts (c) The International Bank for Reconstruction and Development / The World Bank Public sector underpinnings of service reform 195 are bought and sold by politicians for private citizens can give budget management a basic pro�t.527 Checks and balances on policymak- performance orientation (though well short ers are misused when reformers who arouse of performance contracting). In personnel opposition are transferred capriciously.528 In management �rst-stage reforms might budget management as well a formal process include enhancing job security to strengthen of policy choices disciplined by budget rules protection from political interference. can differ greatly from the informal process Second-stage reforms build on a culture of in which the budget is made and remade con- following rules and offer more choices. In stantly during execution. As noted earlier, budget management second-stage reforms several African countries suffer from this for- include a much stronger orientation toward mality gap in implementing their medium- results and performance auditing, building term expenditure frameworks. on good budget execution capacities in gov- What are the practical implications of ernment. In personnel management second- formality in public sector performance? The stage reforms include reducing job security, experience of countries with public manage- harmonizing individual rewards with perfor- ment reforms suggests that the presence or mance targets, and aligning broader terms absence of formality should influence the and conditions with those in the private sec- direction of reform, even if the objectives are tor. But a greater contract orientation is not the same. Where there is no strong tradition the only way to go. Countries such as Canada of merit-based civil service employment, the and Germany have adopted a process of con- direction of reform has been to set up checks tinuous adaptation of their existing systems and balances to legally de�ne entry to the that relies on granting greater flexibility to civil service and the responsibilities of civil achieve stronger results. servants, and to build a distinct and uni�ed corps. Security and stability of tenure and objectivity in promotion are used to protect Curbing corruption against political interference. Where formal- in service delivery ity is the norm, the ambition has been to Many reforms to improve public sector per- move in the opposite direction—to reduce formance and its results orientation cut the security of tenure and seniority in pro- across multiple sectors. Curbing corruption motions and increase individual perfor- is one. Service delivery is weakened by cor- mance contracting, lateral entry, and rewards ruption, and poor people suffer its conse- for results. This experience points to an quences more than others do. important formality threshold in making reform choices. Understanding the economic and social costs of corruption Corruption—the abuse of public of�ce for First-stage and second-stage reforms. This private gain—is a symptom of weak rela- threshold suggests a useful distinction tionships in the service chain. Both grand between �rst-stage or basic reforms and sec- ond-stage or more advanced reforms (�gure 10.7). First-stage reforms provide incentives Figure 10.7 From weak basics to strong foundations in public sector institutional reforms to achieve or strengthen formality when the starting point is a weak institutional setting. First-stage Second-stage reforms reforms Second-stage reforms build on a foundation of formality in stronger institutional envi- to achieve or strengthen to strengthen flexibility, formality, discipline, and discretion, and a focus Greater ronments (table 10.3 suggests illustrative compliance with rules on results contract examples of such reforms). In budget man- Weak basics Formality orientation Informal threshold agement the basics include hardening the public sector Entrenched budget constraint as a more top-down behavior tradition of rule- Continuous compliance adaptation approach to budget formulation and strengthening implementation of input- Strategic oriented line item budgeting. Disseminating incrementalism performance information internally and to Source: Adapted from World Bank (2002e). (c) The International Bank for Reconstruction and Development / The World Bank 196 WORLD DEVELOPMENT REPORT 2004 Table 10.3 Walk before you run Objective First-stage reforms Second-stage reforms Budget management Greater ef�ciency and impact Introduce input-oriented line-item budgeting Change budget formulation and format to link reforms with some performance information the budget to program performance and out-year plans Aggregate cost management Harden budget constraints and focus on Use block or frame budgeting implementation and reporting Accounting reforms Strengthen cash accounting Introduce double-entry bookkeeping and accrual accounting Auditing reforms Strengthen traditional �nancial and Institutionalize performance auditing in a compliance audit and introduce some supreme audit institution and in internal audit performance auditing Personnel Career management Enhance job security and protection from Decrease tenure and link to continuous management reforms political interference performance assessment Unity of the civil service Create a legally de�ned cadre with common Devolve and diversify pay arrangements to terms and conditions provide flexibility to employers Individual incentives Apply standard merit promotion and reward Establish annual performance targets rules consistently Openness Encourage career development within a Move toward “position-based� systems and closed system and avoid nepotism encourage lateral entry Source: Adapted from World Bank (2002e). corruption (involving politicians, senior of�- show that the poor are the least likely to cials, and state capture) and petty corruption know how to get redress when of�cials (involving lower-level of�cials, administra- abuse their position. Transaction-intensive tive procedures, and routine public services) discretionary services that are hard to mon- weaken services. The avenues for corruption itor offer particularly broad scope for cor- in education, health, and infrastructure are ruption because providers have a strong many; they include absenteeism, patronage, information advantage over clients. construction kickbacks, procurement fraud, Corruption in its broader sense of the sale of lucrative of�cial positions, false certi- capture of public resources and decision- �cation, misuse of facilities, unwarranted making affects public spending decisions. services (unjusti�ed caesarian deliveries, pri- The loss of revenue, diversion of public vate payments to government teachers for funds, and evasion of taxes associated with after-school tuition), and bribes at the point such corruption mean that governments of service.529 Bribes are the most common have less to spend on education, health, and face of corruption for poor people, as pay- infrastructure. Studies have found that cor- ments to providers to evade approved proce- ruption is negatively associated with the dures or to perform stated duties. Once share of public expenditures on health and entrenched, corruption reduces the ability education532 and with health and education and incentives of policymakers to monitor outcomes. Politicians may prefer to spend providers, of citizens to monitor politicians, less on ensuring that primary health and and of clients to monitor providers. education services work and more on new Many recent studies present empirical construction and infrastructure, which evidence on the costs of corruption.530 Cor- offer greater opportunities for corrup- ruption is a regressive tax, penalizing poor tion.533 And corruption is empirically asso- people more than others.531 Poor people ciated with lower economic growth rates. often pay bribes to receive basic public ser- vices in education and health, whereas Dealing with corruption richer households tend to pay bribes to There has been rapid growth in diagnostic receive special treatment in courts, customs, tools to measure corruption, assess service and tax authorities. Household surveys delivery, and make informed judgments (c) The International Bank for Reconstruction and Development / The World Bank Public sector underpinnings of service reform 197 about entry points for reform.534 Diagnostic ing piecemeal with corruption risks treating surveys, already implemented in some 20 the symptom and not the malaise. Curtailing countries, usually consist of three separate but corruption requires a multipronged strategy linked instruments covering households, that addresses a number of concerns—politi- �rms, and public of�cials. This allows trian- cal accountability, institutional restraints, cit- gulation of perspectives on the extent, inci- izen voice, effective media, public disclosure dence, locus, and causes of corruption.535 laws, competition, and good public sector Public expenditure tracking surveys and performance (�gure 10.8). A multipronged quantitative service delivery surveys of spe- strategy is dif�cult anywhere, but particularly ci�c facilities can yield useful information on where corruption is widespread and the insti- the contours of corruption and identify entry tutional setting is weak. Anticorruption diag- points for reform. Service delivery surveys can nostics can shed light on the patterns and measure staff incentives and ef�ciency, pro- root causes of corruption, thereby helping to viding information on the determinants of sort reform priorities and suggest suitable service quality and qualitative data on corrup- entry points. In transition economies that are tion. Together, they can provide a cross-check building new public institutions while mas- on the causes and consequences of corruption sively redistributing state assets, opportuni- and provide information that strengthens the ties arise for both administrative corruption voice relationship and client power. and state capture.536 Where administrative Corruption in service delivery is a symp- corruption is high but state capture at the tom of an underlying systemic malaise. Deal- center is not, strengthening accountability Figure 10.8 Many forces at play in curbing corruption in service delivery Curbing corruption Institutional restraints • Independent and effective judiciary Competitive private sector • Independent prosecution, enforcement • Incentive framework and policies Effective public sector management • Parliamentary oversight • Competitive restructuring of monopolies • Watchdog enforcement agencies • Sound public expenditure management • Regulatory reform • Merit-based civil service with monetized, adequate pay • Transparency in corporate governance • Decentralization with accountability and local capacity • Effective business associations • Public services that work • Access to redress mechanisms and legal system • Tax and customs administration Political accountability • Political competition, credible political parties • Transparency in party �nancing, public scrutiny • Open parliaments, courts, and sunshine rules • Asset declaration, conflict-of-interest rules Voice, civil society participation • Freedom of information • Public hearings on draft laws • Role of media and NGOs • Governance monitoring Source: World Bank (2000a). (c) The International Bank for Reconstruction and Development / The World Bank 198 WORLD DEVELOPMENT REPORT 2004 within public administration and deploying corrupt arm of government. Even if judges expenditure tracking surveys and other are above reproach, lawyers, court clerks, and tools for �nancial accountability might be other court of�cials on the take can add to the place to start. But where state capture at the web of corruption. The ingredients of the center is high, political accountability reform are many—freedom of information, and decentralization might be the better greater transparency and sunshine laws, self- entry points. regulation through reform-minded bar asso- An independent, well-functioning judi- ciations and law societies, updating of anti- ciary is vital for combating corruption and quated laws and court procedures, and the often offers a viable entry point. In enforcing independence, competence, and integrity of laws and providing checks and balances on judicial personnel—but they are compli- the power of policymakers and providers, cated to assemble and need time to take root. courts directly strengthen voice. In many Experience suggests that important progress countries, however, courts are themselves a can be made if reforms focus on incentives, institutional relationships, and information access rather than only on formal court BOX 10.7 Managing the thorny politics of pro-poor rules, procedures, and court expansion. Anti- service delivery reforms corruption legislation that matches the enforcement capacity of the country, inde- Reform means change and therefore oppo- the media to empower supporters, and pendent supreme audit organizations, and sition, often political.This is particularly the seeking strategic entry points. legislative oversight can help. case for reform of basic services in which governments are involved as providers, • Striking sensible tradeoffs between com- prehensive and incremental reforms, �nanciers, or regulators, and for which the seeking early wins for stakeholders, and long route of accountability therefore supporting reform champions and cross- Managing transitions: comes into play. agency teams that can bring along oth- overcoming reform hurdles Institutional reforms in education, ers of like mind. health, and infrastructure service delivery Public sector reforms can arouse stiff opposi- are particularly complex. Multiple actors, • Welcoming policy contestability as tion from groups that bene�t from existing inevitable, but using it to mobilize stake- long timetables, early costs, and late bene- relationships. How can this opposition be holders, build coalitions, and gain elec- �ts create many known and unknown veto players and risks. Support for expanding toral credibility. softened? And how to explain the dilemma of access is easy to organize (new jobs, new • Ensuring broad, sustainable support as “considerable reform in political landscapes early as possible, and avoiding a backlash contracts, new patronage), but improving seeded with the potential for failure�— quality is hard. by aiming at universal services that bene- �t all users, including the poor, rather exempli�ed, for example, by contentious Reforms can be implemented more eas- ily in pro-poor settings because of the con- than special groups. education reforms in Latin America.537 There sensus on social equity. Managing politics • Marginalizing opponents before, during, is no easy answer, but a major factor is how then often implies curbing unsustainable and after implementation, particularly politicians and policymakers manage the populism, promoting universal public ser- those with veto power, and exploiting splits in their ranks to move beyond the numerous transitions in public sector and vices, and building coalitions among the poor and the middle class so that there is static arithmetic of winners and losers. service delivery reforms, engaging with citi- broad support for reforms. Managing the politics of reform is often zens and frontline providers to promote But in clientelist settings consensus on a top-down technocratic process led by change. Experience suggests that dealing with reforms may be dif�cult to reach among central design teams and lacking participa- the political economy of such transitions may politicians, policymakers, and potential veto tion, transparency, and occasionally even players (government bureaucracies, busi- be the hardest task for reformers. While each legitimacy.This is usually a mistake. Without ness associations, labor unions, nongovern- good feedback it is dif�cult to master country’s experience is unique, some general mental organizations). Vested interests and changing areas of conflict. Making services principles provide a starting point (box 10.7). patron-client relationships may have co- work for poor people requires strengthen- But knowing what to push and what to hold opted many institutions. Reformers must ing their voice in order to strengthen then create political room by more back is an art not easy to learn or teach. accountability.This reduces, in a positive purposefully managing the politics of way, the room for maneuver by reformers. Policy managers must choose appropri- reform. Inclusive decisionmaking and implementa- ately between �rst- and second-stage Though each situation is different, expe- tion processes are both a means and an end rience suggests some basics: reforms. Even reforms such as implement- in the management of the politics of pro- ing the budget require considerable leader- • Setting the terms of the debate, controlling poor service reforms. ship, capacity, and coordination across the agenda, and taking the high road, including getting ahead of the opposi- Sources: Grindle (forthcoming), Nelson (2000), many parts of government. Choosing sec- tion, using information disclosure and Weyland (1997), and Olson (1971). ond-stage reforms in a weak institutional setting can be doubly dif�cult. Not only is (c) The International Bank for Reconstruction and Development / The World Bank Public sector underpinnings of service reform 199 there likely to be signi�cant opposition (sec- necting policymakers, providers, and clients. ond-stage reforms represent a greater depar- A results-based monitoring and evaluation ture from the status quo than �rst-stage system that joins information from more tra- reforms) but supporters may favor the ditional monitoring efforts with information reforms for the wrong reasons (anticipating from the service delivery framework can pro- the possibility of private gain when complex vide guidance on the institutional reforms reforms fail in an informal institutional needed to improve service delivery. It can be environment). So a mismatch of reforms particularly useful to embed an evaluation and initial conditions can lead to the subver- regime within a poverty reduction strategy so sion of the reforms from outside and inside. that it is possible to see what the strategy is Even if reformers recognize the need to doing for services for poor people. start with �rst-stage reforms, there is the The technology of monitoring and eval- problem of reform traction.538 Embarking uation is widely known and usually speci�c on reform is less of a challenge where trac- to the service and delivery mechanism.539 tion is high—reformers have considerable What is more important to focus on are the leverage in society and politics, are good underlying incentives for monitoring and communicators who have sold their vision evaluation, and how demand for informa- to the majority of the population, and the tion can be made to drive the supply. Three institutions to be reformed are amenable to issues stand out: the institutional frame- change and salvageable. But in settings work for monitoring and evaluation, the where traction is low, reformers must deal role of systematic program assessment and with their slippery grip on reforms, which its links back into policymaking, and the can make it hard to shape the implementa- importance of dissemination. tion of even �rst-stage reforms. Creating a new information system that How then should reformers in low- results in greater transparency, accountabil- traction settings initiate and implement ity, and visibility will alter political power reforms? The answers, clearly country- equations. It can challenge conventional speci�c, go beyond the simple principles wisdom on program performance, drive enunciated in box 10.7. Above all, initiating new resource allocation decisions, and call reforms in low-traction settings is a matter into question the leadership of those of opportunity and patience. To take advan- responsible. Box 10.8 highlights the impor- tage of opportunities as they arise, reform- tance of understanding the institutional and ers need to build alliances with key stake- political dimensions of a results-based mon- holders in advance. They need to encourage itoring and evaluation system and how diversity and experimentation and to learn demand for monitoring and evaluation quickly and systematically from the results. should drive the supply, rather than the And they need to create their own opportu- other way around. Efforts to improve statis- nities; building on what traction does exist tical systems, for example, have often in their settings. focused on �xing supply problems by strengthening national statistical systems to Evaluating and learning collect, process, and disseminate data rather Monitoring and evaluation give meaning to than on understanding the sources of the accountability relationships between ser- demand. This has led in some cases to an vice clients, policymakers, and providers. Tra- oversupply of information: in Tanzania, for ditionally, governments have associated example, health information systems monitoring and evaluation with individual abound, but it is still dif�cult to obtain accu- areas of the core public sector—the audit sys- rate estimates of service delivery coverage. tem, discussion of audited �nancial state- Without some understanding of how ments by the legislature—but these have information is used, those who collect it tended to remain unconnected and myopic. may see the process as time consuming and What has been missing is the feedback on unrewarding, leading to poor compliance outcomes and consequences of actions at and low quality. As decentralization pro- each stage of the service delivery chain con- ceeds in many countries, it is important to (c) The International Bank for Reconstruction and Development / The World Bank 200 WORLD DEVELOPMENT REPORT 2004 what does not. Given the complexity of BOX 10.8 Ready for results? public sector reforms and the dif�culty of choosing entry points and appropriate existing monitoring and evaluation • Are champions of results-based monitor- information? sequencing, governments are constantly ing and evaluation evident within the • Who regularly collects and analyzes mon- trying new policy and program approaches. country? itoring and evaluation information to Some of them work well, many produce • What reforms are underway or planned assess government performance, either mediocre results, and many fail. But unless to which a results-based monitoring and inside or outside the government? evaluation initiative might be linked? there is systematic evaluation of reforms, • Who will use results-based monitoring • Where can local capacity be found in there is no way to be sure that they worked public management, surveying, evalua- and evaluation information to assess ser- because of the policy or program or because tion, and data management to support vice delivery performance? the supply of and the demand for results- of other reasons.540 And unless the results • What management framework within the based monitoring and evaluation? play a major part in the design of subse- government will oversee the introduction and operation of a results- • Are there proposed or existing donor- quent delivery mechanisms, there is no way supported initiatives, such as a PRSP, to based monitoring and evaluation which a results-based monitoring and to be sure that governments can succeed system? evaluation initiative might be linked? when they decide to scale up. • Are there links between budget and Finally, as chapter 5 and the rest of the resource allocation procedures and Source: Based on Kusek, Rist, and White (2003). Report emphasize, wide dissemination of the results of monitoring and evaluation activities is crucial to improvements in ser- build decentralized monitoring and evalua- vice delivery. If not widely disseminated tion capacity so that central and local sys- inside and outside government through tems are complementary. mechanisms tailored to speci�c audiences, As emphasized elsewhere in this Report, the results of monitoring and evaluation systematic program evaluation can be a activities may not live up to their potential powerful tool for showing what works and for improving service delivery. (c) The International Bank for Reconstruction and Development / The World Bank spotlight on Ceará 170,000 community health agents reaching 80 million Brazilians One of Brazil’s poorest states, Ceará reduced infant mortality dramatically in the late 1980s and 1990s. A major effort of the local government motivated health workers, municipalities, local communities, and families to work for better health. I n the 1980s the socioeconomic indica- tors in Ceará, a state of about 7 million people in northeast Brazil, were among the worst in the country. The infant mortal- ity rate was around 100 per 1,000 live health agents were gradually expanded to include a doctor, a nurse, a nurse’s aid, and �ve to six community health agents for every 800 families. This Family Health Program was based the 1990s, and in Palmas the incidence of diarrhea fell by half, with antenatal care coverage doubling between 1997 and 1998.542 births. Fewer than 30 percent of municipal- on the success of the São Paulo, Porto Ale- ities had a nurse. And essential health ser- Balancing decentralization gre, and Niterói municipalities with “family vices reached only 20–40 percent of the physicians.� It added follow-up of at-risk with a results orientation population. In 1986 the state government families and home care for chronic diseases Mobilizing actors began a massive effort to reduce infant to the existing services. The family physi- Using matching funds to motivate munici- deaths. It succeeded. By 2001 infant mortal- cians and nurses’ aides also provide curative palities to implement new programs, ity was down to 25 per 1,000 live births. care and referrals to hospitals. By 2002, Ceará state policymakers struck a balance 150,000 Family Health teams were reaching between decentralizing responsibilities to Sending health workers to poor 45 million people. the municipalities and keeping a results households focus through state control over key The Ceará state government began in 1987 Health outcomes, 1987–2001 aspects of the program. to recruit, train, and deploy community Some of the decreases in infant mortality Strategies were also developed to health agents. By the early 1990s health and malnutrition can be attributed to the strengthen community leverage over agents were visiting 850,000 families a increased coverage of immunization, oral health providers and to strengthen com- month, the �rst public service to regularly rehydration therapy, and breastfeeding (�g- munity voice. The widely publicized selec- reach nearly all local communities. ure 2). Socioeconomic inequalities in cov- tion of a large number of community The monthly family visits and family erage were also reduced, and the greatest health agents from the communities records improved oral rehydration therapy, improvements were made among the poor- helped to “socialize� the program. Com- breastfeeding, immunization, antenatal est of the population.541 Output mea- munity organizations were involved in the care, and growth monitoring—as well as sures—such as immunization, oral rehy- second round of assessments for the treatment of pneumonia, diarrhea, and dration therapy, breastfeeding, and child Municipal Seal of Approval—a program to other diseases. weighing—have also improved. give incentives to municipalities to By 2001 more than 170,000 community Anecdotal evidence points to impacts in improve outcomes (box 1). health agents covered 80 million Brazilians other states. Implementing the Family (�gure 1). In 1994 the teams of community Health Program in the town of Camaragibe Financing brought infant mortality down from 65 per Several �nancing mechanisms covered Figure 1 The number of community health 1,000 live births in 1993 to 17 at the end of agents increased dramatically annual program costs of roughly $1.50 per bene�ciary. In line with the 1988 constitu- Thousands 100 Figure 2 Changes in health and nutrition tion and 2001 health funding laws, munici- indicators in Ceará 1987–94 and 1997–2001 palities can retain tax revenues but must 90 spend 25 percent on education and 10 per- 80 Percent 80 cent on health. The salaries of community 70 2001 70 health agents ($60 a month), and the costs 60 of supervision and drugs are paid directly 60 50 by the state. Municipalities are required to 50 40 1994 40 IMR cover only the salaries of nurse-supervisors 30 30 ($300 a month), but many voluntarily sup- 20 20 port other costs. Malnutrition 10 10 The national government offers match- 0 0 ing block grants to municipalities for edu- North- North Center- South South- 1987 1994 1997 2001 cation and health as an incentive to imple- East East East ment priority programs. The grants for Source: 1987 to 1994 from Victora and others (2000a); 1997 to Source: Brazil Ministry of Health (2001). 2001 from Fuentes and Niimi (2002). minimum basic health care amount to 10 (c) The International Bank for Reconstruction and Development / The World Bank 202 Spotlight on Ceará mayors,� and scorecards of municipal indi- BOX 1 The Ceará Municipal Seal of Approval cators. The Seal of Approval required that municipalities have better-than-average In 1990 Brazil enacted the Statute for Children performance indicators of child survival and health indicators for the group in which and Adolescents, one of the world’s most development and on administrative manage- the municipality was classi�ed, based on advanced laws on child rights, introducing ment of health, education, and child protection. local rights councils and guardianship councils No monetary award is attached to the socioeconomic criteria. Color-coded maps to help de�ne, implement, and monitor public seals, but the municipality may display the seal facilitated monitoring and recorded the policies for children. on of�cial stationary and in health centers, evolution of indicators. In 1997 Ceará introduced Municipal Seals schools, and other of�cial services. Mayors, of Approval with support from UNICEF.The showing interest in the seal, like being viewed Enforcement through hiring and �ring seals were awarded to municipalities based on as “child friendly� and good managers. Although the program was decentralized to municipalities, a special team attached to the state governor had control over the hir- reals per person per year, 2,400 reals per its bene�ts, and they lobbied mayors to join ing and �ring of the community health health agent per year for municipalities the program. Implementation was phased workers, and over a special fund created for implementing the Community Health in, beginning with municipalities that the program. Worker Program, and 28,000 to 54,000 reals demonstrated interest and readiness, stimu- Many community health agents were per year per team when the municipalities lating competition among municipalities. recruited from the community through a implement the Family Health Program. Innovative social mobilization strate- high-pro�le selection process that con- gies expanded public awareness of the Seal tributed to a sense of ownership and Monitoring and information dissemination of Approval and broadened understanding empowered communities to demand better To encourage municipalities to participate, of the social indicators needed for certi�ca- services from the mayors. Candidates not Ceará state of�cials tried to create a strong tion. These included compact discs to selected become public monitors of the per- “image� program. Citizens were informed of guide radio coverage, elections of “child formance of the community health agents. (c) The International Bank for Reconstruction and Development / The World Bank Donors and service reform 11 Timeo danaos et dona ferentes.—I fear the Greeks even when they bear gifts. • Donors should support recipient institu- tions by evaluating innovations system- Laocoön on the Trojan Horse chapter In many developing countries, external atically, by harmonizing and realigning their �nancial assistance and knowledge donors support service reform. In middle- transfers with the recipient’s service income or large low-income countries, they delivery (particularly where aid’s share mostly pilot innovations or implement of spending is large), and by focusing on demonstration projects. If chosen strategi- outcomes and results. cally and evaluated properly, these projects can be powerful. In other low-income coun- • In good country environments where there are genuine reformers, donors tries the story is quite different. Donors sup- should also integrate their support in the ply 20 percent or more of public resources recipient’s development strategy, budget, in more than 60 low-income countries. And and service delivery system. they supply more than 40 percent of public resources in at least 30 poor countries—such • In low-income countries coming out of conflict or with weak institutions, donors as Bolivia, Madagascar, Nepal, and Tanzania. should support urgent social and other For these countries aid flows are obviously services, while identifying mechanisms important for service delivery. that build transparent public institutions The international community has come in the longer term. Pooling of aid will a long way in understanding what makes reduce transaction costs. aid more effective, focusing on the selection of recipient countries.543 This chapter sug- This is all �ne, but the multiple objectives gests that, along with country selectivity, the of foreign aid create incentives for donors to way donors provide their aid matters a lot. control their interventions directly rather Donors still underestimate how dif�cult than to align them with the recipient’s ser- it is to influence reform without undercut- vice delivery systems. Because of these ting domestic accountabilities. Too aware of incentives, reforming aid will not be easy. Yet failures in the key relationships of account- for service reform to succeed, donors have to ability in recipient countries, donors often attach an even higher priority to aid effec- bypass them. This can produce good iso- tiveness and development outcomes. lated projects, but it can also weaken the aid recipient’s internal systems and account- Aid and accountabilities ability relationships (chapters 3 to 6). This Aid differs in important ways from domes- chapter suggests that: tically �nanced services. The bene�ciaries • Donors need to pay more attention to and �nanciers are not just distinct—they the problems in influencing service live in different countries, with different reform in recipient countries. political constituencies.544 This geographi- • They should strengthen the critical rela- cal and political separation—between ben- tionships among policymakers, providers, e�ciaries in the recipient country and tax- and clients. In circumventing those rela- payers in the donor country—breaks the tionships, they can undermine the deliv- normal performance feedback loop in ser- ery of services. vice delivery (�gure 11.1). For example, 203 (c) The International Bank for Reconstruction and Development / The World Bank 204 WORLD DEVELOPMENT REPORT 2004 Figure 11.1 The feedback loop between bene�ciaries and donor country taxpayers is broken Donor Recipient Politicians and Programmatic aid Politicians and policymakers policymakers Co mp ice act Vo Taxpayers and Aid agency Poor people Client power Providers interest groups Self-help projects Investment projects bene�ciaries in a recipient country may be pacts between policymakers and provider able to observe the performance of aid organizations (chapter 6) in many ways. By agencies. But they cannot reward or punish influencing spending patterns and bud- the policymakers responsible for this per- getary processes, donors interfere directly formance in donor countries. The broken with the design of the compact. And by feedback loop induces greater incentive going straight to provider organizations, biases in aid than in domestic programs. So donors sidestep the policymaker as well as aid effectiveness is determined not only by the compact. the performance of the recipient but also by Donors affect the recipient’s spending the incentives embedded in the institutional patterns and budgetary processes in many environment of aid agencies. Understand- ways:545 ing these incentives is central to any reform of aid to support service delivery better. • Donors may support only capital spend- The divergence and distance between ing (construction) and expect the gov- constituencies and clients may be impor- ernment to supply complementary tant—but there is more. Even if donor con- inputs (staf�ng, maintenance). Govern- stituencies adopted client feedback as a ments often fail to �nance the comple- paramount criterion for aid, there would mentary inputs. still be dif�culties in exercising external • Donors may fund projects that govern- influence without undermining local ments are not interested in. This contra- accountability relationships. To illustrate dicts ownership, though it can work the inherent problem of external actors, where a good pilot project encourages a consider enterprise �nance. When new approach through its demonstra- �nanciers or venture capitalists want to tion effect—or where a one-time inter- influence an enterprise they are investing in, vention is needed. they become an equity holder and perhaps • Donors may give aid to a priority sector and request a seat on the company’s board. assume that government spending from its Clearly it would be politically infeasible for own resources remains unchanged. This donors to request seats in the recipient’s runs into fungibility because governments cabinet. Yet the influence that donors exer- attempt to smooth spending by adjusting cise on the recipient’s public spending often their own allocations.546 resembles that of an equity �nancier. • Donors may set targets for the share of spending in particular sectors as condi- Strengthen—don’t weaken— tions for aid flows. Consider the current the compact donor preference for social sectors, which When aid flows are substantial relative to the appears to have increased both recipi- recipient’s resources, donors affect the com- ents’ public spending on these sectors (c) The International Bank for Reconstruction and Development / The World Bank Donors and service reform 205 and the social sector’s share of aid (from directly to frontline providers, such as 14 percent of the aid flows in 1991 to 34 health clinics or schools. Sectoral ministries percent in 2000).547 But strong donor independently lobby donors for funding. preferences can leave other important From the donor perspective competition areas underfunded or set perverse incen- among ministries, departments, and other tives in the privileged sector. In Zambia organizations permits a better selection protecting social spending led to deep process—because hopeful recipients will do cuts in rural infrastructure spending— their best to reveal as much information as possibly creating more rural poverty.548 possible to attract donors. The result: recip- ients’ policymakers lose control of the To avoid such distortions, donors can expenditure program, because the �nance is discuss priorities with policymakers and off-budget and the activities bypass the work to shape public expenditure during compact. Incoherent spending allocations the annual budget cycle. But the recipient and uneven coverage of services ensue. has to have a budget process that functions Similar competition can occur among fairly well. donors, making incentive problems worse. Many donors see a need to align aid with When the recipient agency knows that if the recipient’s compact between policy- one donor threatens to withdraw due to the makers and providers. But there are other recipient agency’s poor performance other tendencies as well. Global funds, which are donors will step in, few incentives exist for private-public partnerships at the global improving its performance.550 level, have chosen to provide funding on a Some donors, including the World Bank, project basis directly to service providers in even circumvent provider organizations by poor countries.549 The new health-related setting up autonomous or semi-autonomous global funds also develop policies for global project implementation units for their inter- procurement and distribution of commodi- ventions. Advocates of project implementa- ties, such as mosquito nets, vaccines, and tion units recognize that the arrangements essential medicines. can undermine local capacity building, create In many ways, the delivery of global salary distortions, and weaken the compact funds—from a global source of �nance between the policymaker and the provider directly to the local provider—reflects the organization. But they argue that the better need for donors to demonstrate that the results outweigh the costs. A study of about funds are additional to what otherwise 100 World Bank projects in the Latin Amer- would have been given. But it might also ica and Caribbean Region shows otherwise: reflect dissatisfaction with the functioning that project implementation units have no of the recipient’s relationships of account- ability and with aid agencies. But it is not clear that this is a sustainable solution to the institutional problems. Evidence from BOX 11.1 The debate over global funds: Uganda Uganda indicates that global funds can pit It is like a hungry boy who sees ripe man- below floors and that it costs much more to the recipient’s policymakers—in charge of goes hanging abundantly from a tree, but deliver health services than the budget allo- the overall spending program—against its he is not allowed to pick the fruits.The Min- cation it receives.The Ministry of Finance istry of Health remains needy, while donor counters by saying that the country’s attrac- provider organizations, who directly lobby money hangs around. As part of the budget tiveness to donors depends on its reputa- for off-budget funds at the international and medium-term expenditure framework, tion for sound macroeconomic level (box 11.1). Parallel �nancing mecha- the Ministry of Finance sets a limit to the management. Global funds risk undermin- nisms can also undermine efforts to ratio- amount of money that can be spent on ing this by providing resources outside the health—as it does for all sectors—refusing normal budget process. It is not nalize expenditures, reform government to earmark excess funds for health from government’s intention to turn away addi- systems, and increase transparency at the global funds.The Ministry of Finance argues tional resources, �nance of�cials say, but it is country level. that there are a whole host of important important that such resources be Donors interact directly with provider things that poor people need and that channeled through the regular budget there are not enough resources around to process. organizations at various levels. Some aid provide all of them. agencies choose to work with line min- But health of�cials insist that the Min- Source: Adapted from The New Vision, Uganda’s istries. Others choose to engage providers istry of Finance is constructing ceilings main daily newspaper. under local governments. And others go (c) The International Bank for Reconstruction and Development / The World Bank 206 WORLD DEVELOPMENT REPORT 2004 signi�cant positive impact on project out- The costs of aid fragmentation comes, while the likely sustainability of The problem with aid fragmentation is not results clearly suffered.551 A parallel study in that individual projects are misconceived— the Eastern Europe and Central Asia Region it is that there are too many projects for any produced similar �ndings.552 to work ef�ciently. When a project’s �xed In Bangladesh donors responded to a cri- costs are high and there are returns to scale, sis by setting up a separate project manage- fragmented aid can be wasteful. Further- ment unit for the Bangladesh Arsenic Miti- more, when donors each have only a small gation Water Supply Project—to speedily share of the total aid in a recipient country, address arsenic contamination in drinking their stake in the country’s development, water (chapter 9). The unit bypassed the tra- including capacity building, may be ditional water engineering departments, reduced relative to their concern for the deemed too inflexible to respond to the success of their own projects. Fragmenta- emergency. Two years later it has fallen far tion also imposes high transactions costs on short of expectations. The government is recipients, with large amounts of of�cials’ expected to close the project shortly, arguing time taken up by donor requirements. that the unit, having bypassed government, Little systematic evidence is available on was unable to deliver on the ground. fragmentation and its effect on the manage- To staff project implementation units, ment of provider organizations. One source, donors tend to hire the most highly skilled though limited, is the Development Gate- civil servants, often at salaries many times way database, with records on about 340,000 what they could earn from the government.553 aid projects and programs across the devel- In Kenya a World Bank agricultural project oping world.556 Using the database to quan- paid eight local staff between $3,000 and tify the extent of donor fragmentation yields $6,000 a month, many times the $250 avail- a mean index value for donor fragmentation able to a senior economist in the civil ser- across recipients of 0.87.557 (Index values vice.554 Another study found that of 20 increase with the number of donors active in Kenyan government economists receiving the country and with greater parity among master’s degree training in a donor-funded donors. Low values indicate a smaller num- program between 1977 and 1985, 15 were ber of donors, or that some donors domi- working for aid agencies or nongovernmental nate.558) For example, Tanzania has a high organizations (NGOs)—or for their projects index value of 0.92, with more than 80 aid by 1994. The study concluded: “elite external agencies having funded 7,000 projects over master’s degrees are, in effect, passports out of time. A similar index computed from the public sector.�555 In countries with many another data set—annual aid disburse- donors, salaries are likely to be bid up even ments—suggests that donor fragmentation more, as donors compete for quali�ed staff. is on the rise (�gure 11.2).559 A better choice to improve aid effective- High fragmentation indices could reflect ness is to phase these units out and to work donor specialization in different sectors, so Figure 11.2 Donor fragmentation: on with the recipient’s provider organizations, that fragmentation would be low in each the rise building their capacity. And it should take sector. But mean levels of the index are only Donor fragmentation index place within the compact between the recip- slightly lower within individual sectors: 0.85 0.70 ient’s policymakers and service providers. for education, 0.77 for health, and 0.78 for But this requires changes in incentives in aid water projects. High fragmentation values 0.65 agencies (see the last section of this chapter). for most recipients show that donors do not specialize very much, either by sector or by Let provider organizations country. Most donors are active in many 0.60 manage sectors, in most countries: a typical recipient Donors affect management of provider orga- nation in 2000 received aid from about 15 0.55 nizations in recipient countries in at least bilaterals and 10 multilaterals (table 11.1). 1975 1980 1985 1990 1995 2000 three ways: by the fragmentation of aid in a How does donor fragmentation affect Note: The higher the index, the greater the large number of donor projects, by the choice the recipient’s provider organizations? As degree of donor fragmentation. Source: Knack and Rahman (2003). of activities, and by the choice of inputs. mentioned, little systematic evidence is (c) The International Bank for Reconstruction and Development / The World Bank Donors and service reform 207 available. One study �nds an association Table 11.1 So many donors . . . between rising fragmentation and declin- Type of aid donor ing bureaucratic quality in high-aid coun- Bilateral donors only Bilateral and multilateral donors tries and in Sub-Saharan Africa, control- Number of recipients with 1–9 donors 3 13 ling for changes in per capita income and Number of recipients with 10–19 other variables (�gure 11.3).560 This �nd- donors 93 27 ing suggests that donors with a small share Number of recipients with 20–29 of the aid in a country may focus more on donors 22 69 delivering successful projects, even at the Number of recipients with 30–39 donors 0 40 expense of government capacity—for example, by hiring the most quali�ed gov- Average number of donors per ernment administrators to run their pro- recipient 14 26 jects. This collective action problem may be Median number of donors per recipient 16 23 less severe where there is a dominant donor, who has a greater incentive to take a Note: The number of donors was calculated by using �gures for total of�cial development assistance (ODA) in 2000, provided by the OECD DAC. The number of recipients takes into consideration only the independent countries accord- broader and longer-term view of the coun- ing to the list of member states of the United Nations. Source: Acharya, de Lima, and Moore (2003), from OECD DAC data. try’s development. High fragmentation means high transac- tion costs for recipients. Tanzanian govern- ment of�cials have to prepare about 2,000 behavior to the procedures used by the reports of different kinds to donors and recipient’s service providers in their report- receive more than 1,000 donor delegations ing to domestic policymakers. Exceptions each year. These requirements tax rather are beginning to emerge, including Tanza- than build provider organizations’ limited nia, Bolivia, Vietnam, and Ethiopia, where capacities, diverting efforts toward satisfy- donors are planning to help the government ing donor obligations rather than reporting develop a harmonization program rather to domestic policymakers. Recognizing the than limit it to the donor community. adverse effects, donor agencies have recently initiated measures to curb compliance costs Donor influence on choice and streamline operational policies, proce- of activities and inputs dures and practices, focusing on �nancial Donors also influence the choice of activi- management, procurement, environmental ties within a sector. They tend to be gener- assessment, and reporting and monitoring. ous with training. In Malawi training High fragmentation may have an even accounts for a staggering $4.5 million, or 10 stronger impact in low-income countries percent of donor spending on health care a with weak policies and institutional environ- year.561 It is hard to believe that the return ments. This is because the domestic capacity on this investment matches the cost or that to implement reforms is typically highly the government would spend this much on constrained—both the political capital of training if it had the choice. And the real reformers and the technical capacity of the cost appears to be even higher: staff may be administration. Fragmented donor interven- absent from work for long periods on train- tions create pressure on existing capacity, by ing courses. Training opportunities are demanding both political and administrative often a form of incentive for staff. If so, the efforts to implement change across a wide funds would likely be better used if the variety of areas at the same time. Aid flows sponsoring donors provided them directly are often at low per capita levels, so a large to supplement salaries through the budget. number of projects may also mean that the The $4.5 million spent on training health average value of each project is small, leading workers in Malawi would translate on aver- to high overhead and transaction costs. age to a 50 percent increase in salary for all But change has been slow. The emphasis health care staff. has so far been to �nd common interna- The input mix in aid-�nanced public tional standards and principles at the aid spending often differs from that in recipient agency level, rather than to adapt donor spending. For example, donors provide far (c) The International Bank for Reconstruction and Development / The World Bank 208 WORLD DEVELOPMENT REPORT 2004 Figure 11.3 Bureaucratic quality more technical assistance (and project vehi- systems do better at identifying poor com- declines with donor fragmentation in Sub-Saharan Africa cles) than the recipient would buy if it had munities than at identifying poor households the money. In Malawi technical assistance or poor individuals.563 The effectiveness of Change in bureaucratic quality index* accounts for 24 percent of donor spending on targeting varies widely, which suggests the 2 health.562 A major obstacle to addressing this importance of unobserved attributes of com- issue is the shortage of data. Many recipient munities. Some studies show that public ser- budgetary systems have much better data on vice delivery—measured by access to infra- the input mix for domestically �nanced structure or outcomes—improved through 0 expenditures than they do on donor projects, community involvement,564 and others, that which are sometimes treated as single lines in performance could be better.565 the budget. Donor of�cials have weak incen- Yet social funds and self-help interven- tives to provide full information to recipient tions continue to face serious challenges of –2 governments. Public expenditure reviews, sustainability. One challenge arises from the –0.05 –0.025 0 0.025 0.05 often critical of government spending, let cultural and social context of communities Change in donor fragmentation index* donors off lightly. To improve scrutiny, better and their capacity for collective action. It is *After controlling for the effect of initial level of data are urgently needed. not clear that the self-help approach can bureaucratic quality, growth of per capita bene�t fractured, heterogeneous communi- income, growth of population, and level of aid in relation to GDP (averaged over the period). Increase client power ties that have little capacity for collective Note: The scatter plot shows the partial relation- ship between the change in bureaucratic quality Client power—the relationship between action. Alternative methods of service deliv- (1982–2001) and donor fragmentation (based on bene�ciary and service provider—tends to ery may suit such poor communities better. project counts). Source: Knack and Rahman (2003). be weak in many developing countries. This But there is little factual evidence on this has long presented donors with a dilemma. because evaluations typically do not com- Should they help strengthen the links pare social funds and self-help projects with between users and existing providers? Or conventional service delivery mechanisms. should they �nd a way around the recipi- Nor do they take into account the negative ent’s service delivery system to ensure that side effects. The sustainability of self-help aid-funded services reach poor people? This projects can be in jeopardy if line ministries becomes even more complicated in heavily or local governments ignore them once they HIV/AIDS-affected countries, those com- are completed. Unless communities can ing out of conflict, and those with weak and ensure continuing support for recurrent corrupt public institutions. Of�cial donor costs and staff, they may not be able to sus- agencies have followed the example of their tain their project. nongovernmental counterparts, approach- Donors need to disburse funds fast and to ing communities and user groups directly, show visible results quickly to supporters or through sharply increased funding for taxpayers. A recent study in a Sahelian coun- social funds and self-help projects. Three try, also applicable elsewhere, shows that main problems surface in these activities: an these needs may be incompatible with undermining of government and other reducing poverty.566 When donors are impa- local capacity, weakened prospects of sus- tient, when they compete with similar agen- tainability, and the capture of bene�ts by cies for good projects, when they do not have elites (chapter 4). the capacity to monitor activities on the In principle, social funds and self-help ground, they may choose particular groups projects could operate within the recipient’s to work with—risking the capture of donor service delivery system. They could also funds by elites. Impatient donors may even serve as entry points for the policy dialogue make the patient donors attach greater with policymakers and providers and hence weight to quick results, undermining the build local government capacity rather than prospects for poverty reduction. This undermine it. But like many other projects, becomes a serious problem when malevolent they tend to be operated directly by donors elites capture donor funds for private gain. with little integration. But that need not be the case (box 11.2). Most assessments of social funds and self- Social funds and self-help projects should help projects focus on poverty targeting. be designed for each context, with best-prac- Overall, the evidence suggests that centralized tice templates as initial guides only. Rapid (c) The International Bank for Reconstruction and Development / The World Bank Donors and service reform 209 expansion of such projects by donors with for donors. Yet donors attempt to do it in little experience may not be feasible. Rather many ways—a testament to the importance than implementing numerous enclave oper- of voice in service reform. The attempts ations in a single recipient country, donors include imposing conditions and setting per- could pool support openly and transparently formance criteria on aid flows where voice is to achieve better results in scaling up and weak, providing direct support to democra- preventing elite capture—even when bypass- tic governance, and actively promoting ing the policymaker-provider relationship in transparency and participatory processes. a failed state or low-income country under By imposing conditions donors try to stress.567 Where conditions are right, the replace the weak voice of citizens in disci- pooling of aid should not stop with the plining policymakers (chapter 5). Yet donor donors—it should extend to national and conditions are fundamentally different from local governments and other providers, pri- citizen voice, which is diffuse, after-the-fact, vate for-pro�t and not-for-pro�t. and a long-term process. In the 1980s struc- Donors can also promote other initia- tural adjustment loans extended conditions tives to enhance client power. They can in projects to a wide spectrum of govern- encourage citizen monitoring of service ment economic policies, processes, and pub- providers, such as report cards and public lic spending. There is ample evidence today expenditure tracking surveys (chapters 5 that conditions based on promises do not and 10, and see spotlight on Uganda). They work well, because they undermine owner- can help monitor the use of services and ship of the reform program.568 support bene�t-incidence analyses to iden- When policymakers are not encouraged tify the groups missing out. Keep in mind, to develop their own positions on, say, pri- however, that involving providers in the vatization of water supply or other services, design of the monitoring process is critical but rely on donor conditions in taking to ensuring buy-in for the results. action, they can more easily deny responsi- bility for a later failure. It is not the quantity Promote voice of aid that makes the recipient’s policies Promoting citizen voice through formal good or institutional reforms happen. political mechanisms or through informal Empirical studies show that aid �nance is advocacy groups or public information cam- ineffective in inducing policy reform in a paigns is one of the most dif�cult endeavors bad policy environment.569 What works BOX 11.2 Social Investment Fund: Jamaica There are more questions than answers on how motivate a larger group to contribute to the ties met.The Social Investment Fund also self-help projects really work. Do they improve project. And once completed, the service facility appears to have improved trust and capacity for participation and targeting? Do they build is generally viewed as belonging to the commu- collective action, but the gains are greater for capacity for collective action? One way to nity and there seems to be wide satisfaction more educated and networked individuals. So answer these questions is to analyze how the with the outcome. But it also appears that the the process might be characterized as “benevo- process works in a particular political, social, and positive social bene�ts from a community- lent capture�: elites dominate the process but in cultural setting. based intervention may be dif�cult to sustain in a way that eventually bene�ts the community. A case study of the Jamaica Social the long term, particularly in communities beset Both participating and nonparticipating com- Investment Fund integrates quantitative and by deep divisions. munities show more community-based qualitative data from �ve pairs of randomly Quantitative data on 500 randomly chosen decisionmaking, indicative of a broad-based selected communities. Each pair has similar households from the same �ve pairs of commu- effort to promote participatory development. social and economic characteristics, but only nities mirror these qualitative �ndings. Within- The Jamaica Social Investment Fund shows one of the pair participated in the social fund. community “preference targeting� is poor, with that self-help does not necessarily “empower The fund typically uses NGOs to mobilize com- three of the �ve participating communities not the poor� and can be either supply or demand munities to participate.The NGOs work closely obtaining the project preferred by a majority. By driven. But community involvement does seem with local elites, such as pastors and teachers. the end of construction, however, 80 percent of to make service delivery more effective by Project selection is not generally participatory the community members expressed satisfaction increasing ownership and participation and by but is driven by this small, motivated group. with the outcome. More educated and improving the capacity for collective action. Once construction of a service facility networked individuals dominate the selection commences, however, the group is often able to process and are more likely to have their priori- Source: Rao (2003) and Rao and Ibáñez (2003). (c) The International Bank for Reconstruction and Development / The World Bank 210 WORLD DEVELOPMENT REPORT 2004 better is choosing recipients more carefully, to allow donors to signal the conditions for a based on performance (country selectivity), recipient to expect an increase or reduction and setting conditions that reward reforms in aid. Initiatives to improve the measure- completed rather than those promised.570 ment of results are under way, but it will take Traditional conditionality does not work time to establish an effective link between well. How can donors then allocate aid so that the volume of aid and performance. it provides a strong incentive for the recipient There is also tension between the moni- to promote citizen voice and undertake ser- toring and incentive functions of perfor- vice reform and thus increases aid effective- mance indicators. Donors, preoccupied with ness? In principle, there is broad agreement �duciary concerns, tend to keep a close watch today that, instead of conditions, the aid on the programs they support—hence the compact needs to contain veri�able indica- focus on short-term process undertakings tors that can measure performance. But, in rather than genuine outcome measures as practice, the use of performance indicators triggers for performance evaluation.572 But has not yet changed the incentives underpin- this can lead instead to micromanagement, ning the relationship between recipients and exactly what the new system of performance- donors.571 based conditions is intended to avoid. This Few performance indicators used today tension, if unresolved, makes aid compacts measure outcomes; most still measure incoherent. inputs and processes. So far the link between Many bilateral donors go further and these operational performance indicators support electoral participation and democ- and the outcome targets articulated in racy directly and use aid to induce and poverty reduction strategies remains vague. reward such reforms (box 11.3). For exam- And there are few transparent mechanisms ple, donors rewarded Ghana for holding free elections in 1992, despite the excessive public spending prior to the elections that resulted BOX 11.3 Donors support democratic governance in poor macroeconomic performance. Donors also support informal mecha- Most bilateral donors explicitly include pro- donors could more actively nurture core nisms to strengthen citizen voice. One is to moting democracy among the goals of their political processes and values, such as repre- promote participatory processes in the aid programs.The U.S. Agency for Interna- sentation, accountability, tolerance, and tional Development alone spends more openness. Legislative aid programs have development of poverty reduction strategies than $700 million a year on programs— often failed due to donors’ lack of knowledge and budgetary processes (box 11.4). But aid supporting free elections, fostering civil soci- about the political and personal dynamics of agencies and recipient governments some- ety organizations, and strengthening par- the institutions they are trying to reshape, times have different views on what form the liaments, judiciaries, and political parties. their determination to apply models that do Election assistance is the highest-pro�le not �t the local situation, and their focus on participation should take. Aid agencies sel- component of democracy promotion. Partic- technical solutions (such as new rules for dom hold a dialogue with parliamentarians, ularly in postconflict situations, numerous staffers or Internet access) for deeply political stressing instead the extragovernmental bilateral donors, international organizations— problems. such as the United Nations, the Organization To overcome some of these problems, aspects of participation, involving a wide of American States, and the Organization of the Swedish International Development range of civil society. Members of parlia- Security and Cooperation in Europe—and Agency (SIDA) has initiated a new type of ment sometimes view the donor emphasis private organizations send observer analysis of the underlying interests and on civil society as undermining the legiti- missions and provide assistance to election power relationships as part of a program to administrators. More than 80 international support political institutions in Burkina macy of elected representatives, particularly groups observed the 1996 elections in Faso, Ethiopia, Kenya, and Mali. In electoral in emerging democracies. They also ques- Nicaragua. Aid for promoting democracy has assistance SIDA has moved to longer-term tion the legitimacy of NGOs selected to increased from 0.5 percent of total of�cial programs and closer donor collaboration in speak for “the people.� development assistance in 1991 to 5 percent South Africa and Zambia. In Cambodia it in 2000. supports a “national issues forum� to air Donors have encouraged many low- The donor approach to democratic gov- public debates on topics like corruption and income countries to open policy debates ernance—democracy, participation, human traf�cking in women and children on televi- and discussions when preparing poverty rights, and the rule of law—has striking simi- sion and radio across the country. reduction strategies. Governments are larities to the rest of aid: a heroic short-term effort to get countries through a sudden working to bring the views of a wider range takeoff to democracy. Rather than try to Sources: Carothers (1999), Ottaway and Chung of stakeholders into discussions. Madagas- reproduce certain types of institutions, (1999), Knack (2001), and SIDA. car, Rwanda, and Vietnam now have more timely information, make greater use of (c) The International Bank for Reconstruction and Development / The World Bank Donors and service reform 211 local languages, and use participation more than consultation. There is also evidence BOX 11.4 Donors support transparent budget processes: that these processes are influencing a shift Tanzania toward broader consultation on govern- Since 1997 Tanzania has conducted an cedures that open the possibility for ment decisions beyond the poverty reduc- annual public expenditure review, led by the domestic constituencies to use the tion strategy. But major challenges remain government with the participation of donors process increasingly for their own pur- in regularizing greater openness in govern- and a wide range of civil society. A primary poses of legislative scrutiny, feedback, ment decisionmaking. objective is to review the government’s for- public comment, and lobbying. Poverty reduction strategies seek to pro- ward expenditure plans in its medium-term expenditure framework, to discuss the pro- • Expanding the scope of the review over time, from immediate donor concerns— mote stronger citizen voice, with an effec- gram with donors, and to con�rm the exter- how donor interests have been tive link to public spending. But they also nal �nancing. From its inception the initiative addressed in budget plans or how seek to change the relationship between also had the objective of developing a public donor �nance features in the consultative process to engage a wide range framework—to broader concerns of recipients and donors by stressing the recip- of domestic constituencies on the govern- policy and performance, such as the ient’s ownership of the reform agenda. ment’s performance and forward plans.The government’s overall strategy and how There are often tradeoffs when one instru- working group overseeing the process is led it is reflected in budget plans, ment is used to achieve multiple goals. by the Ministry of Finance but includes mem- performance record, and efforts to bers from donors and a range of nongovern- strengthen service delivery. The review Countries preparing the early poverty mental bodies. also provides a national forum for reduction policy papers (PRSP) faced many Several features of the process show attending to various sectors and lower challenges in managing the participatory how donor accountability requirements can levels of government. be met in a way that promotes a sustainable process and linking their strategy to the domestic system of accountability: • Having domestic players take on a budget, while adhering to tight timetables greater role over successive budget for debt relief for which a PRSP was a con- • Not tying the consultations to a single cycles, both in government and among donor or �nancing instrument. Individual constituencies outside the executive, dition. Experience in many countries sug- donors can use the process in their own including the legislature and civil society gests that when a government presents a monitoring and review procedures. groups. national development strategy, supported • Grounding the review process in the by broad ownership and well-de�ned sector domestic policymaking and budgetary cycle, rather than in donor review pro- Source: World Bank staff. priorities, this contributes in no small mea- sure to attracting broad donor support. Moreover, it also offers a framework to bet- ter align and harmonize donor support. around the recipient’s own systems. Where country systems are weak, they Align aid delivery need to be strengthened to meet good with service delivery practice standards, not bypassed and substituted with ring-fenced donor sys- For donors that want to align their aid tems and procedures. This is crucial for delivery with service delivery in recipient aid effectiveness in low-income coun- countries, this Report has three important tries that receive a substantial part of messages. their public resources as foreign aid. But • First, evaluate interventions and aid pro- it is also relevant in middle-income jects for impact. More systematic evalua- countries in sectors where donors are tions of, say, an intervention’s effects on especially active, as in social protection student learning or health status are crit- in Latin America and elsewhere. ical for scaling up in both middle- and • Third, harmonization and realignment low-income countries. Evaluation linked are best done at the country level and by to early steps toward rebuilding state strengthening the recipient’s existing insti- capacity is important in situations where tutions. In countries with fairly good donors are working through alternative expenditure management and genuine service providers due to conflict or state service reforms—where donors and recip- failure. ients trust each other—budget support • Second, to reduce the costs of aid frag- should be considered a viable tool. mentation and to build capacity, work Resource pooling can also be effective in with other donors to harmonize and scaling up service delivery and reducing align policies, procedures, and practices transaction costs in low-income countries (c) The International Bank for Reconstruction and Development / The World Bank 212 WORLD DEVELOPMENT REPORT 2004 that are under stress—for example, due doing so may produce misleading results. to past or current conflict.573 Schemes that select participants randomly provide the best opportunities for unbiased Innovate and evaluate impact evaluation. An example is the Edu- Large public sector organizations—in both cation, Health, and Nutrition Program of donor and recipient countries—focus on Mexico (Progresa), a large government inputs and process evaluation rather than transfer program (see spotlight). Strong outputs and outcomes (box 11.5). The evidence of its high impact led to an expan- incentive to do this in aid agencies is even sion of the program in Mexico and the stronger because of the broken feedback adoption of similar programs elsewhere. loop between taxpayers in the donor coun- Another example is the secondary school try and bene�ciaries in the recipient country voucher program in Colombia, which (�gure 11.1). Outcome and impact evalua- assigned bene�ciaries by lottery, making it tions are seldom built into aid projects. Spe- feasible to compare those receiving vouch- cial attention is required to counter the ten- ers with those who did not.574 A randomly dency of aid agencies to be input-oriented assigned pilot program for treating intesti- and to increase the share of interventions nal worms in Kenyan schoolchildren has subject to rigorous impact evaluation. Out- been similarly evaluated.575 come-oriented international targets, such as In many operational settings, however, the Millennium Development Goals, can randomization cannot be applied, and add to the incentives for aid agencies to other methods must be found to create a overcome their focus on inputs. matched comparison. Even when data on A major dif�culty in assessing the bene�ciaries before and after implementa- impacts of any public program is that bene- tion exist, determining the real effects of a �ciaries are rarely selected randomly. program or policy change requires data on Indeed, most programs are purposely tar- matched comparison groups to get at the geted to speci�c groups or regions. Isolating counterfactual of what would have hap- the impacts from the circumstances that led pened without the policy. For some to participation is then tortuous. Yet not schemes that do not have baseline data, it is possible to construct an adequate control group from the postintervention data. But baseline data are needed for others, such as BOX 11.5 Why aid agencies focus on inputs rural roads with far-reaching impacts on poverty, health, and education outcomes.576 The disruption in the feedback loop the results of aid projects or to mainstream between donor country taxpayers and rigorous impact evaluation.The input bias Not every program can be evaluated for recipient bene�ciaries, combined with the and the need to handle political problems or impact, so governments and donors should complexity of measuring performance, pressures tilt staf�ng toward generalists and select programs for evaluation carefully, results in a focus on inputs in aid agencies. administrators. focusing on areas where new knowledge is Budgets, contracts, and expenditures on Yet most aid projects include a formal projects are easier to monitor and assign evaluation requirement. Process needed. Interventions rolled out in phases than are the outputs and impacts that the evaluations—such as audits, monitoring, because of budget and other constraints aid projects produce. and veri�cations that the intended action offer good opportunities for effective Elected policymakers in donor countries took place—are often conducted well. Out- impose administrative procedures on aid come evaluations are seldom built in, and if impact evaluation. Similarly, when a pilot is agencies to restrict their discretionary deci- done at all are contracted later to required before a large-scale rollout, an sionmaking. In addition, bureaucracies tend consultants. If evaluations are a small part of impact evaluation will generate important to develop their own procedures, which these consultants’ foreign aid market, they information for decisionmakers. increases the complexity. In this sense aid also tend toward input bias and may avoid agencies are like any other large public orga- revealing results that can affect their main nization. Extensive consultations with non- market. Consultants may not consciously Harmonizing donor support governmental organizations diffuse the risk misreport; the market pressure is usually around recipient systems of political pressure in donor countries—and more subtle. Contractors who work mainly Harmonizing is easier where the recipient complex tender and contract procedures for aid agencies often become generalists defuse criticism by aid service suppliers.With themselves. has a well-functioning national develop- such an incentive structure, aid agencies allo- ment strategy and budget process that can cate comparatively few resources to verify Source: Martens and others (2002). serve as the common framework. But these are not prerequisites—for even in their (c) The International Bank for Reconstruction and Development / The World Bank Donors and service reform 213 absence harmonization and pooling of aid Sectorwide approaches can offer signi�cant bene�ts and reduce Since the mid-1990s many countries have transaction costs.577 In low-income coun- worked to integrate government and donor tries coming out of conflict or in situations activities within a sector. In the ideal of very weak public institutions, a “budget approach, outlined in early documents on within the budget�—with separate account- sectorwide approaches, the government ability mechanisms for donors—or an inde- and its partners would agree on a pre- pendent service agency funded by donors, dictable resource envelope and on a policy or a consortium of nongovernmental environment consistent with the national providers chosen jointly through a transpar- budget and economic strategy.579 They ent process can be relevant options.578 Both would then agree on how to assign arrangements can be linked to public sector resources within this envelope. Procedures reform efforts, which emerge over time. But for disbursement would be harmonized, to be effective, harmonization requires a and funds would be pooled. All activities radical realignment of procedures and oper- would reflect a shared view of the priorities ational policies in donor agencies, as the and costs of activities. Differences would be health project in Bolivia shows (box 11.6). resolved by compromises in the design of For donor-�nanced services shifting programs, not in the activities undertaken. responsibility—and hence accountability— There would be no detectable difference toward the recipient would reduce duplica- between the approach taken on govern- tion, waste, and transaction costs. It would ment-funded and that taken on donor- also keep donors from crowding into a few funded activities; indeed, that distinction fashionable sectors—and thus lessen con- would wither away. cerns about absorptive capacity. And it Sectorwide approaches have been estab- would build capacity, improve collective lished in several sectors in many low- learning, and create stronger incentives for income countries—health, education, agri- monitoring and evaluating impacts and culture, transport, energy, and water.580 results. Efforts so far are only partial realizations A common refrain is that recipients need of the ideal. To some extent determining to improve their �nancial management and what constitutes a sectorwide approach is public procurement practices before donors still an arbitrary decision. There has been can align their support around the recipi- progress toward pooling funds in recent ent’s systems. But reality is more nuanced. First, donors need to ask whether relying on country systems is riskier than the alterna- BOX 11.6 A case for harmonization in Bolivia tive of ring-fencing. Currently, developing- country borrowers must produce 8,000 In Bolivia three donors in the health sector particular floors, procuring the materials audit reports every year for multilateral agreed to co�nance construction of a build- and hiring builders according to its own ing. But the fact that each donor had its own standards and procedures.The idea was development banks—5,500 of such reports procurement procedures made it dif�cult to that one donor agency would �nish the �rst for the World Bank. Such a fragmentation of �nd a common approach.The donors could two floors, after which the second would activities cannot increase accountability. not pool their contribution in a common build the third floor, and the third would �n- fund because the agencies’ rules prohibited ish the building. Second, even if it is riskier, this needs to be channeling money to another agency. None After long debates, one of the donors set against the more sustainable bene�ts of of the agencies could accept the procedures withdrew from the project. Of the remain- helping to build the recipient’s institutions of the others, and two of the agencies were ing two, the one contributing the smaller and systems. Third, donors need to avoid unwilling to adopt Bolivian rules. amount accepted the rules of the donor A thematic approach was considered putting up the bulk of the funds. Only one the trap of making a given level of capacity a �rst. One donor would pay for the design, contractor would be hired, not the three condition for aligning or pooling aid, when another the construction works, and the previously envisaged, and just one engineer in many cases the pooling launches efforts third would contribute paint, air condition- would supervise construction. One agency that can get capacity closer to where it needs ing, electrical apparatus, and lavatories. would oversee the entire process.The Then, for practical and administrative rea- process took two years, and the foundation to be. That said, there will still be cases in sons and to avoid blaming the other agen- stone has yet to be laid. which donors judge (rather than scienti�- cies if something went wrong, it was cally determine) that it would be inappro- suggested that each donor would pay for Source: World Bank (2002i). priate to pool, given the �duciary risks. (c) The International Bank for Reconstruction and Development / The World Bank 214 WORLD DEVELOPMENT REPORT 2004 years, however. For the 24 programs tracked also reflect the ambitious agenda for sector- by the Strategic Partnership with Africa in wide approaches. A tentative conclusion is 2002, 41 percent of assistance came through that sectorwide approaches are an important projects (down from 56 percent two years part of a poverty reduction strategy, not an earlier), 13 percent through NGOs, 11 per- alternative, and the full bene�ts will not be cent as common basket, and 35 percent as realized until �nancing mechanisms become budget support. In Ghana’s health program more flexible. the pooling arrangements started with “one donor and minimal funding� but later reached 40 percent of program resources.581 Budget support Preparing and implementing sectorwide The focus on budget support was sharp- approaches can be a long, drawn-out ened by the debt relief for heavily indebted process. It can also weaken rather than poor countries, allocating relief to priority strengthen the recipient’s compact between sectors through the recipient’s budget. Bud- its policymakers and provider organizations get support restores the compact between by taking the sector out of the domestic policymakers and providers. It allows con- decisionmaking process, particularly the testability in public spending. And it budgetary process, �nancial management reduces the costs from fragmentation and and public procurement.582 Four lessons separate project implementation units. Pro- have emerged from sectorwide approaches: viding funds to the general budget also offers a better framework for discussing • An institutional analysis of the sector is intersectoral allocations. Advocates of more recommended beforehand, including funding to one sector have to show that the the sector’s relation to the rest of the sector has higher returns than others at the public sector. margin.584 If funds go to sectors that • If the capacity constraint is in the lead demonstrably reduce poverty—directly or ministry rather than the country, new indirectly—donors should be flexible about personnel can sometimes be found budget allocations. quickly, as in the Ugandan education Budget support, like basket funding for a and health sectors.583 sectorwide approach, raises questions of • Procedures need to be designed with �duciary risk. But there is no clear evidence capacity limitations in mind, particu- that the risk is greater for budget support larly at decentralized levels. This will than for project aid.585 Needed are transpar- often involve encouraging public trans- ent systems for procurement and public parency and bottom-up monitoring to information to ensure that the movement bolster simple but rigorously enforced of funds through the system can be publicly upward-reporting requirements. Pro- observed and that charges paid for services curement procedures—often a major are clearly de�ned. Donors can contribute dif�culty—need to balance rigor with best by promoting these systems in the simplicity. recipient country. The Utstein group of donors—the United Kingdom, the Nether- • Capacity constraints are not a reason to lands, Norway, and Germany—has been delay a sectorwide approach. Few coun- tries achieve the ideal, but most can ben- developing monitoring arrangements along e�t from some aspects of the process. these lines. The European Union links part of its budget support to performance, using Assessments of sectorwide approaches a small set of indicators (box 11.7). have reached mixed �ndings. Ratings by the What does all this suggest? That aid will Strategic Partnership with Africa in 2002 work best where it is provided flexibly to show an average implementation rating for recipients with sound overall strategies and programs of between 0.42 and 0.58, depend- well-designed sectoral programs. Flexible ing on the sector (on a scale where 0 is poor, aid can catalyze processes within govern- 0.33 fair, 0.66 good, and 1 very good). But it ments to produce sound strategies, rational is possible that these relatively poor ratings spending programs, and effective services. (c) The International Bank for Reconstruction and Development / The World Bank Donors and service reform 215 Knowledge transfers Donor competition for new ideas can be BOX 11.7 Linking budget support to performance good for the recipient. It can also create The European Commission is explicitly linking indicator: one point if the agreed objective confusion, particularly in low-income part of its budget support to performance. is attained, half a point if there is evidence countries that have weak capacity. One The amount to be disbursed is based on of “considerable positive development,� progress in social service delivery, notably and zero if there is no progress. The budget solution is to pool knowledge transfers health and education, and in public expendi- support provided is the maximum amount and joint analytical work at the country ture management. Progress is measured by a available multiplied by the (unweighted) level, including impact evaluations of small number of performance indicators average performance score (ranging from agreed to by the recipient and the European zero to one). The approach is not mechani- interventions and programs. All analytical Commission. Indicators are typically drawn cal but also takes into account external work supported by donors should draw from the recipient’s poverty reduction strat- factors. on in-country capacity, including univer- egy. For the �rst set of countries, the most fre- The performance-based system high- sities, government, and the private sector quently used indicators are: lights the quality of data. According to the European Commission, the system is not an (box 11.8). • Planned and actual expenditures in the end but a means: getting policymakers and social sectors. the public in developing countries to pay Poor institutions—post-conflict • Differences in unit costs of key inputs more attention to results than to declara- between the public sector and the market. tions of intentions and conditions set by and “failed� states • Use of primary and antenatal health care donors. In poor institutional environments—as in services. So far, 30 percent of the European Com- post-conflict countries or “failed� states— • Immunization rates. mission’s budget support is linked to perfor- mance indicators.This is deliberate, donors may not be able to rely on conven- • Births assisted by medical personnel. motivated by the desire to introduce a new tional channels of service provision because • Enrollment rates for boys and girls. approach gradually and to balance perfor- policymakers and providers lack the capac- • Cost of primary education (private and mance rewards and the recipient’s need for public). predictable budget �nance. ity or the intent to use resources well.586 After a joint evaluation by government Whatever the short-term or even medium- and donors, a score is calculated for each Source: European Commission. term delivery vehicle, aid should contribute in the longer term to rebuilding an effective service delivery system and public sector. The temptation to avoid the government is ensure high standards of accountability understandable. But without some clear directly to donors. To deliver on such stan- and shared donor strategy for rebuilding a dards, hiring may have to be outside the responsive and effective state, the prolifera- public sector. Again, donors should coordi- tion of nongovernmental and community- nate and pool their support to reduce waste based organizations—and self-help and and duplication, �nancing both recurrent social fund initiatives—will lack breadth of and capital expenditures. impact and sustainability. Civil society Donors face the challenge of �nding a organizations cannot design national poli- balance between short- to medium-term cies or standards. Nor can they substitute in institutional failures and the long-term cre- the long term for the citizen-policymaker ation of an effective state that can deliver on relationship. its public responsibilities. Donors have The options for donors range from sought to address short-term service needs selectively supporting existing programs, through national and international NGOs, such as immunization programs operated social funds, United Nations agencies, or a by the government or private providers, to combination of these providers. What establishing an independent service author- needs to be given equal emphasis is the ity with a temporary mandate to deliver or identi�cation of country-speci�c paths for regulate basic services.587 In between are strengthening capacity and reduction in self-help projects and social funds. An inde- patronage and corruption. Nongovernmen- pendent service authority and a social fund tal channels for service delivery can thus are ways to deliver services in dif�cult cir- play a very important role but should be cumstances, perhaps by wholesaling to a seen as a transitional strategy to strengthen local consortium of NGOs, religious orga- state capacity in the long term. nizations, and private �rms. Such organiza- Even in the weakest states donors and tions require institutional autonomy to domestic stakeholders would bene�t from (c) The International Bank for Reconstruction and Development / The World Bank 216 WORLD DEVELOPMENT REPORT 2004 BOX 11.8 Pooling knowledge transfers To improve access to infrastructure services for poor people, several donors have pooled funds • Access to ef�cient and affordable energy With its strong �eld presence in Africa, South and East Asia, and the Andes, the to augment and disseminate knowledge on • Environmentally sustainable energy produc- program has a well-established network of sec- tion, transportation, distribution, and use. infrastructure services.These facilities, adminis- tor specialists who can respond quickly to the tered by the World Bank, provide a source of changing demands of clients. knowledge and advice that is “�re-walled� from The Water and Sanitation Program assists the Bank’s lending activities. Donors in the pool central governments, municipal agencies, local The Public-Private Infrastructure Advisory are Canada, Japan, the Netherlands, Sweden, authorities, NGOs, community organizations, pri- Facility is a technical assistance facility helping Switzerland, the United Kingdom, and the vate service providers, and external agencies in developing countries to improve their United Nations Development Programme. helping poor people gain sustained access to infrastructure through private sector better water and sanitation services. It focuses involvement. It pursues its mission through: The Energy Sector Management Advisory on: Program, a global technical assistance program, • Policy, strategy, and institutional reform advi- • strategies for tapping the full potential of pri- Technical assistance to governments on focuses on energy in economic development, with the objectives of contributing to poverty sory services vate involvement in infrastructure alleviation and economic development, improv- ing living conditions, and preserving the envi- • Innovative solutions to problems, including • Identifying, disseminating, and promoting pilot and demonstration projects best practices related to private involvement ronment. It focuses on: • Strategically selected investment support ser- in infrastructure. • Market-oriented energy sector reform and vices, including networking and knowledge restructuring sharing. Source: World Bank staff. pooling efforts for a better results. Har- donors were driven solely by the motive to monization of aid may not be feasible reduce poverty—and if recipients were per- through budget support, but agreement on ceived to be committed to the same goal. a common framework and implementation But the world is more complex. Incentives arrangements for service delivery will help in aid agencies and the political economy of avoid too heavy a burden on limited domes- aid in donor countries work against this: tic capacity. Further, it is important to share lessons and to recognize that aid may be less • Aid agencies want to be able to identify effective when modalities that work in good their own contributions, often through policy environments are transferred to distinct “projects,� to facilitate feedback other situations. The flexibility of “eight to taxpayers and sustain political sup- sizes �t all� includes a strong emphasis on port for aid flows. A new hospital is eas- country context. ier to showcase than the outcome of pol- icy reform or budget support. Why reforming aid is so dif�cult • Aid agencies, facing disbursement pres- The unintended negative effects of donor sures, need to show quick results to tax- behavior are not a recent discovery. World payers—and NGOs, to their contribu- Development Report 1990: Poverty (World tors. This is easier when donors are in Bank (1990)) discussed the role of aid in charge of interventions. poverty reduction, drawing attention to • Politicians and policymakers in donor many similar problems. Why has there not countries cannot dismiss the interest been more reform? Why, for instance, are groups that support them, groups that donors so reluctant to channel aid as part of may place a high priority on funding like- the recipient’s budget? minded groups in developing countries. Simplifying donor policies, procedures, • Many donors limit the market for aid and practices and directing aid flexibly services and supplies to their own through sectorwide approaches or the bud- nationals (tied aid). Foreign aid sustains get process would lower the high transac- a large consultancy industry in OECD tion costs in low-income countries and countries—estimated at $4 billion a year allow recipients to pursue their objectives for Sub-Saharan Africa, or 30 percent of more ef�ciently. That could be done if aid to the continent. (c) The International Bank for Reconstruction and Development / The World Bank Donors and service reform 217 • Preferences for spending differ among siderations help explain why the typical donors and between donors and recipi- bilateral donor in 2000 provided of�cial ents. Donors often are most comfortable development aid (ODA) to about 115 with service delivery systems of the type independent nations. Even omitting operating in their own country. For recipients that received less than instance, British and Nordic advisers are $100,000, the mean number of ODA familiar with a clinic-based free health recipients for each of the 22 major bilat- service and so prefer to support those eral donors was 95. systems in low-income countries too. These multiple objectives create incen- • Fiduciary concerns and incentives in aid tives for donors to �nance and directly con- agencies cause donors to focus on moni- trol their aid interventions. That creates toring inputs and processes. Again, the problems for recipient countries: donors monitoring is easier in project aid where often do not know (or don’t care) what the donor controls the design and imple- other donors and the recipient are doing, mentation of each intervention. which results in duplication, waste, and • Donors may want to persuade aid recipi- gaps in services.589 These days donors tend ents of the value of a different approach to favor social sector projects over other through a pilot project, to show success. public expenditures. If they do not pay • Bilateral donors distribute their aid bud- attention to what the others are doing, they gets across a large number of recipients may concentrate too much on higher-prior- and sectors, to increase the visibility of ity sectors, leaving sectors with a lower pri- their programs or to leverage or reward ority, such as rural roads in Zambia, short diplomatic support from recipient of funds. Or there may be gaps in the prior- nations.588 More specialization among ity areas simply because nobody is looking sectors or recipients, however ef�cient, at the big picture. But priorities among could expose a donor to charges that it is donors vary, and their approaches change neglecting, say, a global health crisis or a over time.590 So there is some scope—and regional humanitarian crisis. Such con- hope—for improvement. (c) The International Bank for Reconstruction and Development / The World Bank Bibliographic note This Report draws on a wide range of World Bank documents and Daniela Gressani, Charles Grif�n, Merilee S. Grindle, Jan Willem on numerous outside sources. Background papers and notes were Gunning, Christopher Hall, Kirk Hamilton, Clive Harris, Robert prepared by Abdelwahid El Abassi, A. Aghajanian, S. Ahmadnia, Hecht, John Hellbrunn, Susanne Hesselbarth, Norman Hicks, Dale Harold H. Alderman, James Anderson, Matthew Andrews, Aida Hill, James Keith Hinchliffe, Karla Hoff, Mary Kathryn Holli�eld, Atienza, Suresh Balakrishnan, Nabhojit Basu, Paolo Carlo Belli, Sur- Robert Holzmann, Timothy Irwin, Jaime Jaramillo-Vallejo, Abhas jit Bhalla, Gerry Bloom, Ronelle Burger, J. Edgardo L. Campos, Indu Kumar Jha, Anne Johansen, Olga Jonas, Ruth Kagia, Satu Kähkö- Bushan, Yero Boye Camara, Jonathan Caseley, Prema Clarke, Dave nen, Jeffrey A. Katz, Philip Keefer, Damoni Kitabire, Homi Kharas, Coady, Alberto Diaz-Cayeros, Richard Crook, Monica Das Gupta, Stuti Khemani, Jeni Klugman, Steve Knack, Valerie Kozel, Dan Antara Dutta, Dan Erikson, Paulo Ferrinho, Angela Ferriol, Varun Kress, Jody Kusek, Karen Lashman, Frannie Leautier, Danny Gauri, Anne Marie Goetz, Kelly Hallman, Maija Halonen, Susanne Leipziger, Brian Levy, Samuel Lieberman, Soe Lin, Magnus Lin- Hesselbarth, Rob Jenkins, Anuradha Joshi, Henry Katter, Daniel delöw, Marlaine Lockheed, Elizabeth Laura Lule, Mattias Lundberg, Kaufmann, Philip Keefer, Peyvand Khaleghian, Stuti Khemani, Akiko Maeda, Wahiduddin Mahmud, Nick Manning, Bertin Stephen Knack, Rudolf Knippenberg, Kenneth Leonard, Bernard Martens, Om Prakash Mathur, Subodh Mathur, Aaditya Mattoo, Liese, Angela Lisulo, Annie Lord, John Mackinnon, Beatriz Maga- Elizabeth McAllister, Judith McGuire, Oey Astra Meesook, Vandana loni, Nick Manning, James Manor, Melkiory Masatu, A. Mehryar, Mehra, Alain Mingat, Mick Moore, Christopher Murray, David Anne Mills, Mick Moore, Joyce Msuya, Fatoumata Traore Nafo, Nabarro, Raj R. Nallari, Deepa Narayan, W. Paatii Ofosu-Amaah, Joseph Naimoli, Andrew Nickson, Rami Osseni, C. Torres Parodi, Peter O’Neill, Elisabeth Page, Elisabeth Pape, Puspa Pathak, Harry Harry Patrinos, Mark Pearson, Victoria Perez, Janelle Plummer, Patrinos, Judith Pearce, Ronald F. Perkinson, David Peters, Guy Benjamin Powis, Didio Quintana, Carole Radoki, Aminur Rahman, Pfeffermann, Tomas Philipson, Janelle Plummer, Alexander Preker, Francesca Recanatini, John Roberts, James Robinson, F. Halsey Robert Prouty, Firas Raad, Anand Rajaram, Mamphela Ramphele, Rogers, Pauline Rose, Suraj Saigal, R. Sarwal, Parmesh Shah, Mau- Vijayendra Rao, Ray Rist, Peter Roberts, F. Halsey Rogers, David reen Sibbons, Janmejay Singh, Hilary Standing, David Stasavage, Rosenblatt, Alex Ross, James Sackey, Mauricio Santamaria, Sarosh Jonas Gahr Støre, Denise Vaillancourt, Servaas van der Berg, Wim Sattar, William Savedoff, Eugen Scanteie, Norbert Schady, George van Lerberghe, Ayesha Vawda, Emiliana Vegas, and Peter Wolf. Schieber, Ruth Ingeborg Schipper-Tops, Supriya Sen, Nemat Talaat Background papers for the Report are available either on the Sha�k, Monica Singh, John Snow, Lyn Squire, Lynn Stephen, Mark World Wide Web via http://econ.worldbank.org/wdr/wdr2004/ Sundberg, M. Helen Sutch, Jakob Svensson, Jee-Peng Tan, Judith or through the World Development Report of�ce. The views Tendler, Gregory Toulmin, Emmanuel Tumusiime-Mutebile, Brian expressed in these papers are not necessarily those of the World van Arkadie, Caroline van den Berg, Dominique van de Walle, Bank or of this Report. Rudolf van Puymbroeck, Hema Visnawathan, Adam Wagstaff, Jef- Many people inside and outside the World Bank gave comments frey Waite, Wendy Wakeman, Christine Wallich, Maitree Wasuntin- to the team. Valuable comments and contributions were provided wongse, Hugh Waters, Dana Weist, Michel Welmond, Richard by Christopher Adam, James Adams, Orville Adams, Olosodji Westin, Howard White, Mark Williams, James D. Wolfensohn, Adeyi, Sha�ul Azam Ahmed, Asad Alam, Aya Aoki, Omar Azfar, Michael Woolcock, Alan Wright, Ian P. Wright, Salman Zaheer, Raja Rehan Arshad, Yvette Atkins, Melvin Ayogu, Raja Bentaouet Abdo Yazbeck, and Jürgen Zattler. Kattan, Peter Berman, Paul Berminghan, Markus Berndt, John Other valuable assistance was provided by Mary Bitekerezo, Besant Jones, Robert Beschel, David Bevan, Anil Bhandari, Helena Soucha Borlo, Johanna Cornwell and staff of the World Bank Bjuremalm, John Briscoe, Colin Bruce, Barbara Bruns, Donald A. P. libraries, John Garrison, Phillip Hay, Rachel Winter Jones, Agnes Bundy, Pronita Chakravarty, Vandana Chandra, Mae Chu Chang, Kaye, Emily Khine, Zenaida Kranzer, Angela Lisulo, Precinia Robert Chase, Marian Claeson, Paul Collier, Michael Crawford, Lizarondo, Joaquin Lopez, Jr., Jimena Luna, Karolina Ordon, Car- Jishnu Das, Angelique de Plana, Jean-Jacques Dethier, Annette olyn Reynolds. The Water and Sanitation Program (WSP) of South Dixon, Paula Donovan, William Dorotinsky, Mark Dumol, Ibrahim Asia provided support for the consultation in Bangladesh and Elbadawi, Poul Engberg-Pedersen, Gunnar Eskeland, Antonio access to on going research and policy work of the WSP. Estache, Barbara Evans, Shahrokh Fardoust, Armin Fidler, Ariel Despite efforts to compile a comprehensive list, some who con- Fiszbein, Jonas Frank, Ahmad Galal, Marito Garcia, Varun Gauri, tributed may have been inadvertently omitted. The team apologizes Alan Gelb, Ejaz Ghani, Elizabeth Gibbons, Indermit Gill, Daniele for any oversights and reiterates its gratitude to all who contributed Giusti, Philip S. Goldman, Mark Gradstein, Vincent Greaney, to this Report. 218 (c) The International Bank for Reconstruction and Development / The World Bank Endnotes 1. Taking the world as a whole hides the fact that Sub-Saharan 33. World Bank (1998a) and World Bank (2002a). Africa is off track in reaching the income poverty goal. 34. Even a recommendation to apply interventions that pass a 2. Walker, Schwarlander, and Bryce (2002). social bene�t-cost analysis test will not be enough. Social bene�t- 3. Devarajan, Miller, and Swanson (2002). cost analysis is concerned with valuing an intervention’s outputs 4. Peters and others (2003), p. 218. and inputs at the right set of shadow prices (Bell and Devarajan 5. Reinikka and Svensson (2001). (1987) and Dreze and Stern (1987)). Yet the problem is that the 6. Chaudhury and Hammer (2003). inputs often do not translate to the desired output because of weak 7. Jaffré, Olivier, and de Sardan (2002). incentives. The same point applies to recommendations of using 8. PROBE Team in association with Centre for Development “cost-effective� interventions in health (World Bank (1993). Economics (1999); Rosskam (2003). 35. Spotlight on Cambodia. 9. Analysis of Demographic and Health Survey data (see table 36. Schick (1998). 1.1 of the Report). U.K. Department of International Development 37. Realizing that the central education system has led to under- (2002). representation of students from low-income families, one of the 10. Bhushan, Keller, and Schwartz (2002). prestigious French grandes écoles, L’Institut d’Etudes Politiques de 11. Ahmad (1999). Paris (“Sciences Po�) has begun to use separate admissions criteria 12. World Bank (2002n). for students from poor neighborhoods. 13. Behrman and Hoddinott (2001) and Gertler and Boyce 38. Leonard (2002). (2001). 39. Another reason is that most project managers are not inter- 14. Spotlights on Educo and Bamako Initiative. ested in investing in knowledge that might show their program to 15. Glaeser and Shleifer (2002). have been a failure. 16. Diaz-Cayeros and Magaloni (2002). 40. This account from the New York Times, excerpted from 17. Spotlight on Costa Rica and Cuba. Brooke (1993), describes the unprecedented joint effort by politi- 18. Besley and Burgess (2002). cians, health workers, and communities to put in place a program 19. Spotlight on Kerala and Uttar Pradesh. to substantially reduce infant mortality in the state of Ceará in 20. When asked why he did not complain, one villager replied, “I Brazil. The infant mortality rate for children born in Ceará between could meet with an accident on the road. I could be put in the brick 1981 and 1985 was 142 deaths per 1,000 births, for children born kiln oven. My bones could be broken.� (Spotlight on Kerala and between 1986 and 1990 the rate had fallen to 91, almost a 40 per- Uttar Pradesh). cent reduction. Infant mortality in the poorest �fth of the popula- 21. Spotlight on Johannesburg. tion fell from 154 to 113—almost 30 percent. The decline in infant 22. International Labor Organization (ILO) (2002). mortality in neighboring states of Northeast Brazil was 20 percent 23. Scott (1998). over the same period (Analysis of Demographic and Health Survey 24. Reinikka and Svensson (2003b). data). 25. Chomitz and others (1998). 41. Department for International Development and Water and 26. Computerization of land registration in Karnataka, India, Environmental Health at London and Loughborough (1998). In reduced the transaction time to 30 minutes and eliminated the pay- Ethiopia more than 70 percent of households use an open spring or ment of bribes, which had risen to 25 to 50 times the registration fee. river as their main source of drinking water, and about 80 percent 27. Koenig, Foo, and Joshi (2000). of households have no toilet facilities (Analysis of Demographic 28. Jimenez and Sawada (1999). and Health Survey data). 29. Hsieh and Urquiola (2003). 42. Kunfaa and Dogbe (2002). 30. Gauri and Vawda (2003). 43. Lewis, Eskeland, and Traa-Valerezo (1999). 31. Angrist and others (2002). 44. Mtemeli (1994). 32. Interview by John Briscoe. 45. See also Gwatkin and Others (2000) and Wagstaff (2000). 219 (c) The International Bank for Reconstruction and Development / The World Bank 220 WORLD DEVELOPMENT REPORT 2004 46. See also Filmer and Pritchett (1999a) and Filmer (2000). 85. Narayan and others (2000a). 47. UNESCO (2002). 86. Knippenberg and others (1997). 48. UNICEF (2001). 87. King and Ozler (2002). 49. The multiple determinants of child health are discussed in 88. Lewis, La Forgia, and Sulvetta (1996). Wagstaff and others (2002). 89. WHO (World Health Organization) (1998). 50. See Deaton (1997). 90. Langsten and Hill (1995). 51. For more on this approach see Filmer and Pritchett (2001). 91. Rowe and others (2001). 52. For example, see World Bank (2001k). 92. Lakshman and Nichter (2001). 53. Papua New Guinea Of�ce of National Planning (1999). 93. For example, see Bruns, Mingat, and Rakatomalala (2003) or 54. Gibson (2000) based on a survey undertaken in 1996. Aver- Pritchett and Filmer (1999). age distance may be lower today since a subsequent education 94. Millot and Lane (2002). reform expanded the number of elementary and primary schools. 95. World Bank (1998b). 55. International Forum for Rural Transport and Development, 96. Waitzkin (1991). 2002 input to WDR team. 97. Betancourt and Gleason (2000) and Koenig, Foo, and Joshi 56. Estimates for 2000 from WHO UNICEF Joint Monitoring (2000). Programme for Water Supply and Sanitation (2001). “Improved� 98. Lewis, Eskeland, and Traa-Valerezo (1999). water source is de�ned as sources that provide adequate quality and 99. Jaffré and Prual (1994). quantity of water (i.e., a household connection or a protected well 100. WHO (World Health Organization) and World Bank and not an unprotected well or bottled water). “Improved� sanita- (2002). tion covers flush toilets and private latrines. 101. Schneider and Palmer (2002). 57. “Improved� water source is at best a crude proxy for access to 102. Narayan and others (2000b). safe water. For example, in Bangladesh access to water through tube- 103. Haddad and Fournier (1995). wells—an “improved� source—is extremely high although the water 104. Davis and Patrinos (2002). so accessed is frequently contaminated with arsenic (see chapter 9). 105. Rao and Walton (forthcoming). 58. Filmer, Lieberman, and Ariasingam (2002). An evaluation of 106. Dutta (2003). the enrollment and labor market outcomes of the program are in 107. This spotlight is based on Coady (2003) and Levy and Duflo (2001). Rodríguez (2002). 59. See the discussion in Alderman and Lavy (1996). 108. Percentages from World Development Indicators database. 60. Based on World Bank (2002s). 109. Bruns, Mingat, and Rakatomalala (2003). 61. Radoki (2003). 110. Indonesia: Ministry of National Education (2002); Indone- 62. Leonard, Mliga, and Mariam (2002). sia: Ministry of Religious Affairs (2002); and Filmer, Lieberman, 63. Yip and Berman (2001). and Ariasingam (2002). 64. Leonard, Mliga, and Mariam (2002). 111. Hutchinson (2001). 65. Samrasinghe and Akin (1994) and Akin and Hutchinson 112. 137,000 health subcenters, 28,000 dispensaries, 23,000 pri- (1999). mary health centers, 3500 urban family welfare centers, 3000 com- 66. Pakistan Institute for Environment Development Action and munity health centers, and an additional 12,000 secondary and ter- Project Management Team (1994). tiary hospitals (Peters and others (2003)). The populations of 67. Alderman and Lavy (1996) and Lloyd and others (2001). Uganda, Indonesia, and India are 22 million, 210 million, and 1,015 68. Chaudhury and Hammer (2003). million, respectively. 69. NRI and World Bank (2003). 113. Hutchinson (2001). 70. Chomitz and others (1998). 114. Peters and others (2003). 71. PROBE Team in association with Centre for Development 115. A comprehensive exposition of these ideas is in Stiglitz Economics (1999). (2000). 72. Schleicher, Siniscalco, and Postlewaite (1995). 116. Articles 25 and 26 of the Universal Declaration of Human 73. World Bank (2001e). Rights (http://www.un.org/Overview/rights.html). 74. Thomas, Lavy, and Strauss (1996). 117. See, for example, WHO (World Health Organization) 75. Alderman and Lavy (1996). (2002). 76. Thomas, Lavy, and Strauss (1996). 118. Articles 23 and 24 of the Universal Declaration of Human 77. Schleicher, Siniscalco, and Postlewaite (1995). Rights. (http://www.un.org/Overview/rights.html). 78. PROBE Team in association with Centre for Development 119. Hunt (2002). Economics (1999). 120. See discussions in Green (1990), Pritchett (2002), and Kre- 79. Schleicher, Siniscalco, and Postlewaite (1995). mer and Sarychev (2000). 80. World Bank (2002m). 121. For example, see the theoretical discussion in Gradstein 81. GfK Praha—Institute for Market Research (2001). and Justman (2002) and empirical exploration in Ritzen, Wang, 82. World Bank (2000c). and Duthilleul (2002). 83. McPake and others (2000) and Levy-Bruhl and others 122. Appleton (2001). (1997). 123. In Madagascar GDP per capita was about $250 averaged 84. Di Tella and Savedoff (2001a). over the 1990s and mortality 156 in 2000. In Burundi GDP per (c) The International Bank for Reconstruction and Development / The World Bank Endnotes 221 capita was about $160 averaged over the 1990s and mortality 190 in all users. Fourth, they do not include the incidence of raising 2000. These two countries fall very close to the cross-country funds—that is, a fairly regressive pattern of spending might still be regression line between income and mortality. These data on child pro-poor if it is �nanced through a very progressive tax system. mortality are from UNICEF (2002). Fifth, it is hard to know what a “good� allocation is without com- 124. See in particular: Barro (1991), Bhargava and others paring it to other types of social spending. (2001), Bils and Klenow (2000), Pritchett and Summers (1996), and 141. Reinikka and Svensson (2001). Savedoff and Schultz (2000). 142. Foster (1990). 125. Dollar amounts in this paragraph are in 2001 U.S. dollars 143. World Bank (1994a). and refer to averages for the 1990s. 144. Filmer, Hammer, and Pritchett (2000). 126. Dollar amounts are in 2001 U.S. dollars. 145. Gertler and Molyneaux (1995). 127. Moreover, these cross-national estimates likely overstate 146. Sixty-six percent is the amount recommended by Bruns, the association between income and outcomes as they do not take Mingat, and Rakatomalala (2003) based on a review of countries into account speci�c country attributes. The growth rates discussed that have made substantial progress toward universal completion. here are at best underestimates of those necessary. 147. Bruns, Mingat, and Rakatomalala (2003). 128. Dollar amounts in this list are expressed in 1995 U.S. dol- 148. Devarajan, Miller, and Swanson (2002) use a similar lars. approach to costing the �rst Millennium Development Goal, halv- 129. Between 1980 and 2000 annual average growth of GDP per ing income poverty between 1990 and 2015. capita was: Ethiopia –0.55 percent; Malawi 0.25 percent; Thailand 149. Devarajan, Miller, and Swanson (2002) avoid some of the 0.046 percent; Peru –0.41 percent; Mexico 0.74 percent; Jordan double counting by calculating the cost of the health, education, –0.57 percent; Côte d’Ivoire –0.017 percent; Haiti –0.025 percent and environmental goals independently of the income poverty goal (based on World Development Indicators database). and then calculating the cost of the income poverty goal indepen- 130. Hammer, Nabi, and Cercone (1995). dently of all the others. 131. van der Berg and Burger (2003). 150. This account is drawn from Paul (2002). 132. Duflo (2001). Filmer, Hammer, and Pritchett (2000) dis- 151. Drawn from Community Driven Development (2002). cuss the within-country evidence further. 152. This spotlight relates to Uttar Pradesh as it existed before 133. The result holds for other outcomes as well. For example, its hill districts were separated out into a new state, Uttaranchal, in the Organisation for Economic Co-operation and Development’s late 2000. Program for International Student Assessment found that more 153. Ramachandran (1996) and Dreze and Gazdar (1996). spending on education was associated with better test results in a 154. Mencher (1980), Nag (1989), and Antia (1994). sample of largely upper income countries (Organization for Eco- 155. Dreze and Gazdar (1996) and PROBE Team in association nomic Cooperation and Development (2001)). However, the asso- with Centre for Development Economics (1999) ciation becomes almost zero (and insigni�cant from it) once GDP 156. See Shah and Rani (2003). per capita is controlled for. 157. Dreze and Sen (2002). 134. Filmer and Pritchett (1999b). 158. Keefer and Khemani (2003), Shah and Rani (2003). 135. Bidani and Ravallion (1997) and Wagstaff (2002). 159. Chandran (1999). 136. Gupta, Verhoeven, and Tiongson (2002). 160. Dreze and Gazdar (1996), p. 111. 137. Rajkumar and Swaroop (2002); Gupta, Verhoeven, and 161. Ramachandran (1996), p. 268. The Tranvancore rescript Tiongson (2002) �nd corruption to be important, but Jayasuriya was issued 55 years before the similar Meiji Educational Law of and Wodon (2002) do not. 1872 in Japan. 138. For example, the number of countries, and country cover- 162. Narayan (2002). age, in cross-national studies of spending and mortality are: 98 in 163. World Bank (2001h). Filmer and Pritchett (1999b); 22 in Anand and Ravallion (1993); 76 164. Hammer and Jack (2001) and Gertler and Hammer (1997b). and 56 in Jayasuriya and Wodon (2002); 22 in Gupta, Verhoeven, 165. Das and Hammer (2003). and Tiongson (2002); 32 in Gupta, Verhoeven, and Tiongson 166. Cornell and Kalt (1995); Cornell and Kalt (1997); Cornell (forthcoming);116 in Gupta, Davoodi, and Tiongson (2002); 32 in and Kalt (2000) Wagstaff (2002); 35 in Bidani and Ravallion (1997). There is a par- 167. Mamdani (1996). allel, although somewhat less developed literature on education 168. Cohen (1957). outcomes and spending: for example, Wößmann (2003); Gupta, 169. de Soto (2000). Verhoeven, and Tiongson (2002). 170. Public Services International and Education International 139. This discussion is based on Lieberman (2003). (2000). 140. Such expenditure incidence studies of health and educa- 171. Stasavage (2003). tion spending provide a valuable description, but they cannot tell 172. Tumusiime-Mutebile (2003). the full story. First, they provide a cross-sectional snapshot that is 173. Reinikka and Svensson (2001), Reinikka and Svensson not the same as who would bene�t from the marginal resources (2003a). devoted to the sector. Second, while the data are often based on the 174. Shreenivasan (2002). best available they are limited—especially when it comes to assess- 175. Akin, Guilkey, and Denton (1995) and Peters and others ing the costs of each unit of the service provided. Third, the studies (2003). implicitly assume that the value of the expenditure is equal across 176. Pritchett and Woolcock (2002). (c) The International Bank for Reconstruction and Development / The World Bank 222 WORLD DEVELOPMENT REPORT 2004 177. Appadurai (2001). 220. See Glossary for explanations of terms related to the service 178. In the literature this is known as equalizing their agency delivery framework. (Rao and Walton (forthcoming). Empowerment also refers to poor 221. See http://www.hinso.moph.go.th/30baht_English/index. peoples’ abilities to influence the political power structure, but that htm. is the subject of chapter 5, “Citizens and Politicians.� 222. See http://www.cabinet-of�ce.gov.uk/pmdu/. 179. Conning and Kevane (2002) discuss this in the context of 223. International Budget Project (2000). community-based targeting programs. See also Mansuri and Rao 224. Hirschman (1970) shaped understanding of “voice� as (2003). directed protest, both in its electoral (voting) and nonelectoral 180. This does not apply to technical quality, which can be quite (advocacy, lobbying, naming/shaming, participation in policymak- low—and more variable—in the private sector. ing) sense. 181. Probe Qualitative Research Team (2002). 225. Goetz and Jenkins (2002) and Schedler (1999). The many 182. Ibid Probe Qualitative Research Team (2002). See also meanings given to accountability—an overused term—often blur. Leonard (2002). So, “vertical� accountability (citizens individually or collectively 183. Becker (1971). holding the state to account, as in elections) is sometimes distin- 184. Lewis (2000). guished from “horizontal� accountability within government (a 185. Tan, Soucat, and Mingat (2000). minister or senior civil servant formally holding another civil ser- 186. Wolfensohn (1997). vant accountable). Authoritarian states may manifest horizontal 187. Gertler and Hammer (1997b). accountability, but not offer much vertical accountability. 188. Personal communication with Dr. Zafrullah Chowdhury of 226. Shah (2003a); also see http://www.mampu.gov.my/ Gonoshasthaya Kendra. Circulars/Clients_Charter.htm. 189. Nichols, Prescott, and Phua (1997). 227. Hossain and Moore (2002). 190. Case (2001). 228. Jenkins and Goetz (2002) discuss civil engagement with 191. Das and Hammer (2003). India’s public distribution system for basic goods targeted to poor 192. Werner, Thuman, and Maxwell (1992). people. When the system was exploited as a source of patronage, 193. Shleifer and Vishny (1993) call this corruption with theft. civil society groups advocating more ef�cient delivery had no trac- 194. Goetz and Gaventa (2001). tion for their equity-led agenda, and the poor suffered. 195. Glinskaya and Jalan (2003). 229. For a review of clientelism and how core and swing voting 196. Grindle (forthcoming). can impact services, see Diaz-Cayeros and Magaloni (2003). 197. Chandran (1999). 230. Joshi and Moore (2000) discuss the role of the right to 198. Ostrom (1990). guaranteed work in the Maharashtra Employment Guarantee 199. Scott (1998), and Mackey (2002). Scheme in India and its implications for the mobilization and voice 200. Wade (1987) and Blomquist and Ostrom (1958). of the poor. Jenkins and Goetz (1999) examine the role of the right 201. World Bank Economic Review, Special Issue (2002) to information in the state of Rajasthan in India. 202. Hino (1993). 231. Putnam, Leonardi, and Nanetti (1992) and Boix and Pos- 203. Water and Sanitation Program (2001). ner (1998). 204. Alatas, Pritchett, and Wetteberg (2003). 232. Freedom House (2002). Democracies are de�ned as politi- 205. Platteau and Gaspart (2003). cal systems whose leaders are elected in competitive multi-party 206. Agarwal (2001). and multi-candidate processes in which opposition parties have a 207. World Bank (1998c). legitimate chance of attaining power or participating in power. 208. Mansuri and Rao (2003) and Kleemeier (2000). 233. See Moore and Putzel (2001) for a discussion of democracy 209. Mehrotra and Jarrett (2002). and poverty outcomes. 210. Gilson and others (2001). 234. Keefer (2002). 211. Gilson and others (2001). 235. This draws on Keefer and Khemani (2003). 212. Knippenberg and others (1997). 236. Keefer (2003), based on countries with available education 213. In the early 1980s, successes in delivering primary care ser- expenditure data from among the 117 countries in the Database of vices were �rst analyzed based on the experiences of the Narangwal, Political Institutions, 1975–95 (Beck and others (2001)). Lampang and Bohol projects in Asia, as well as the Danfa, Kin- 237. See various articles in Ferejohn and Kuklinsky (1990). tampo, Kisantu, Kasongo and Institute of Child Health Nigeria pro- 238. Fiorina (1990). jects in West Africa. These best practices were translated into a 239. See the literature on political cycles in developing coun- coherent set of service delivery strategies, management systems, tries, including Shi and Svensson (2003), Khemani (forthcoming), and instruments in the Pahou pilot project in Benin (1982–86). Block (2002), and Schuknecht (1996). 214. Ministry of Health Guinea (2002). 240. Fiorina and Shepsle (1990) and Chappell and Keech 215. Ministère de la Santé de Bénin (2003). (1990). 216. Zhao, Soucat, and Traore (2003). 241. Grossman and Helpman (1999). 217. Soucat, Gandaho, and Levy-Bruhl (1997). 242. Easterly and Levine (1997); Alesina, Baqir, and Easterly 218. Gilson (1997) and Gilson and others (2000). (1999); Betancourt and Gleason (2000). 219. Narayan and Pettesch (2002) and Narayan and others 243. Alesina, Baqir, and Easterly (1999). (2000a). 244. Ferejohn (1974) and Persson and Tabellini (2000). (c) The International Bank for Reconstruction and Development / The World Bank Endnotes 223 245. Gazdar (2000). 288. Botchway (2001). 246. Diaz-Cayeros and Magaloni (2003). 289. Lazear (2000). 247. Medina and Stokes (2002). 290. Boston (1996) and Stewart (1996). 248. Diaz-Cayeros and Magaloni (2003). 291. Dixit (2000) and Holmstrom and Milgrom (1991). 249. Miguel (1998), as noted by and consistent with the �ndings 292. Hammer and Jack (2001) and Gertler and Hammer (1997a). of Diaz-Cayeros and Magaloni (2003). 293. Glewwe, Ilias, and Kremer (2000). 250. Keefer (2002) and Robinson and Verdier (2002). 294. Dixit (2000) and Burguess, Propper, and Wilson (2002). 251. Keefer (2002). 295. Fitz-Gibbon (1996). 252. Keefer and Khemani (2003). 296. The City of Johannesburg Council (2001). 253. Alatas, Pritchett, and Wetteberg (2003). 297. Frey (1997). 254. Putnam, Leonardi, and Nanetti (1992). 298. Irwin (2003). 255. Goetz and Jenkins (2002); see also Narayan (2002). 299. Hammer, Nabi, and Cercone (1995). 256. Masud (2002). 300. Hammer and Jack (2001). 257. See http://www.sdinet.org/ and Appadurai (2001). 301. Klein and Roger (1994). 258. Jenkins and Goetz (2002). 302. Aitken (1994). 259. See http://www.poderciudadano.org.ar/. 303. Reinikka and Svensson (2003b). 260. Boix and Posner (1998). 304. World Bank (2001l). 261. NGOs can make huge contributions to human develop- 305. Leonard (2002). ment by stepping in to provide local community-based services 306. Bierschenk, Olivier de Sardan, and Chauveau (1997); Beb- where there is little public presence. But these NGOs may lack a bington (1997); Meyer (1995); Chabal and Daloz (1999); Platteau credible voice in reforming public services because they may be and Gaspart (2003). perceived as having a vested interest in the existing service delivery 307. Chaudhury and Hammer (2003). arrangements. 308. This is discussed in greater detail in Filmer, Hammer, and 262. Goetz and Gaventa (2001). Pritchett (2000), Filmer, Hammer, and Pritchett (2002). 263. Manor (2002). 309. Cambodia National Institute of Statistics and ORC Macro 264. Platteau (2003) and Crook (2002). (2001). 265. World Health Organization (2003). 310. The operations research was funded by the Asian Develop- 266. See spotlight on Uganda in this Report. ment Bank. 267. Besley and Burgess (2002) 311. Bhushan, Keller, and Schwartz (2002). 268. National Democratic Institute for International Affairs and 312. There is only one instance, that of vitamin A coverage, in World Bank (1998); Di Tella and Schargrodsky (forthcoming) which one district had not increased coverage at the time of the 269. It is not possible, of course, to know whether the actual mid-term evaluation. outcomes in better-informed states were socially superior. It could 313. Soeters and Grif�ths (2003). just as easily be the case that state governments in which media cov- 314. PROBE Team in association with Centre for Development erage of food crises was widespread devoted more resources to Economics (1999). assistance than they should have, including providing assistance 315. Galabawa, Senkoro, and Lwaitama (2000). not only to those who needed it but also to those who didn’t but 316. Greaney, Khandker, and Alam (1999). In reading, “mini- were core or swing supporters and voters. mally competent� means able to answer three of �ve questions 270. Sen (2002) and Dreze and Sen (1991). based on a literal passage; in writing, “minimally competent� means 271. Paul (2002) and Balakrishnan (2002). able to write a short (12-word) passage based on a picture. 272. For information on the Millennial Surveys, see http://www. 317. And testing 15-year-olds still in school overstates Brazil’s pacindia.org. performance relative to that of OECD countries because larger 273. Deichmann and Lall (2003). numbers of Brazilian teens have already dropped out. 274. For a recent discussion of the role of the media in develop- 318. Narayan and Pettesch (2002). ment, see World Bank (2002q). 319. Daramola and others (1998). 275. Faguet (2001). 320. Gundlach and Wößman (2001) and Gundlach, Wößman, 276. Wetterberg and Guggenheim (forthcoming). and Gmelin (2001). The key empirical insight of these studies is 277. Evers (2003). that the evolution of learning achievement can be inferred for the 278. Evers (2003). countries that do not themselves maintain comparability over time 279. Molyneaux and Gertler (1999) and Alatas (1999). by linking their performance relative to that of the United States at 280. Schiotz (2002). a point in time based on the internationally comparable exams and 281. Pritchett (forthcoming). then linking those to the U.S. National Assessment of Education 282. Wilson (1989). Progress results, which are comparable over time. 283. Jaffré and Prual (1993). 321. Lewis (1961). 284. Chomitz and others (1998). 322. The �ve briefly: Belief in one supreme God; just and civi- 285. Chaudhury and Hammer (2003). lized humanity; the unity of Indonesia democracy is the wisdom 286. Shleifer and Vishny (1993). of the deliberation; social justice for the whole of the Indonesian 287. Vasan (2002). people. (c) The International Bank for Reconstruction and Development / The World Bank 224 WORLD DEVELOPMENT REPORT 2004 323. Sweeting (2001). 334. Eskeland and Filmer (2002). 324. Lanjouw and Ravallion (1999). 335. King, James, and Suriyadi (1996) and Pritchett and Filmer 325. Lott (1999). (1999). 326. Cited in Madaus and Greaney (1985). 336. Birdsall and Orivel (1996). 327. Much of the debate is about how to properly isolate the 337. Case (2001). causal impact of variations in class size, mostly from nonexperi- 338. See Progresa spotlight. mental data. This is a problem because if class size is consciously 339. Wodon (1999). chosen in ways that cause a correlation between performance and 340. Cameron (2001). class size—say, by school administrators who make classes with dis- 341. Angrist and others (2002). ruptive children (who would cause low performance) smaller (so 342. Carnoy (1997) and Ladd (2002). the teacher can better handle the situation) or by students, who, 343. World Bank (1996). given choice within a school will choose teachers with better repu- 344. Grindle (forthcoming). tations—then the observed, nonexperimental data might show a 345. World Bank (2002s). negative or zero correlation between class size and performance 346. Eriksson, Kreimer, and Arnold (2000). even though a truly exogenous shift in class size would improve per- 347. This assessment of the position of teachers and school clos- formance. There is evidence of a reasonably large effect of class size ings is from Reimers (1997). from a randomized experiment in Tennessee, and “quasi-experi- 348. Action learning program on participatory processes for mental� evidence from Israel (Angrist and Lavy (1999), South PRSP (2003). Africa (Case and Deaton (1999), and Bolivia (Urquiola (2001). But 349. Action learning program on participatory processes for critics of this evidence argue that reported results are “hit and PRSP (2003). miss�—in that, if class size effects are measured in two subjects in 350. For an example of this critique see Davies (2000) and a dis- three grades, there are class size effects in some grades and subjects cussion in Reimers (1997). and not others—with no particular pattern; that the literature is 351. Initial studies suggested that few of these “Parent School� subject to enormous “publication bias� in that statistically signi�- programs took hold. But they were made an of�cial program— cant results are much more likely to be written up and published, with �nancial support—in the past �ve years, and they appear to even if they are in fact rare; and that randomized experiments in have expanded since then. which the teachers know the purpose of the experiment are not in 352. Jimenez and Sawada (1999). fact a clean test, as teachers will attempt to perform well to justify 353. Jimenez and Sawada (2002). smaller class sizes (Hoxby 2000). Hanushek (2002) continues to 354. Indeed, one early assessment based on a survey of 140 emphasize the huge literature in which there is a general lack of a schools in 1993 found little difference between different types of correlation—with “better� studies less likely to �nd effects—and schools (Reimers (1997)). points to the “big picture� evidence unlikely to be affected by the 355. El Salvador Evaluation Team (1997). “endogeneity arguments�—the time series in the United States and 356. Jimenez and Sawada (1999). OECD countries in which class sizes have fallen substantially while 357. Sawada (1999). scores have stagnated, and the lack of cross-national evidence. 358. Reimers (1997). Hoxby (2000) produces quasi-experimental evidence from the 359. A detailed bibliography for this chapter can be found in United States (Vermont) showing no class size effects and argues Soucat and Rani (2003a). her results are more typical and representative than others. 360. UNAIDS and WHO (2003). 328. Vegas (2002). 361. Gwatkin and others (2000). 329. Banerjee and others (2003). 362. Victora and others (2000a). 330. Crouch and Healey (1997). 363. Gwatkin and Guillot (2000) and Bonilla-Chacin and Ham- 331. Sillers (2002). mer (2003). 332. For instance, empirical studies that run standard wage (or 364. Haddad and Gillespie (2001) and Wang, Monteiro, and earnings) regressions with a few characteristics (age, gender, educa- Popkin (2002). tion) and include a dummy variable (or interaction terms) for 365. Das Gupta (1987) and Claeson and others (2000). teachers provide a purely statistical answer to the question, “Does 366. Victora and others (2000a); Mehryar, Aghajanian, and the wage regression over- or underpredict wages (or earnings) of Ahmadnia (2003); Suwal (2001); Bhuiya and others (2001); Schel- teachers?� But even this answer is without a clear interpretation and lenberg and others (2001); Bang and others (1999); Pathmanathan these studies do not, in themselves, answer the question, “Are teach- and others (2003); Rojanapithayakorn and Hanenberg (1996); Vic- ers underpaid?’ (Psacharopoulos, Valenzuela, and Arends (1996); tora and others (2000b). Liang (1999); Filmer (2002); Vegas, Pritchett, and Experton (1999). 367. Diop, Yazbeck, and Bitran (1995); Soeters and Grif�ths In some situations in which these regressions suggested that teach- (2003); Bhushan, Keller, and Schwartz (2002); Saadah, Pradhan, ers were “underpaid� the annual output of teachers colleges and Sparrow (2001). exceeded available positions by several-fold (suggesting teacher pay 368. Evans (1996) and Moens (1990). was adequate), while in others where the regressions suggested 369. Hart (1971). teachers were “overpaid� there were few new teachers and wages 370. Das Gupta, Khaleghian, and Sarwal (2003). were being increased (suggesting teacher pay was inadequate). 371.As studies from Madagascar, Ghana, Georgia and the Kyr- 333. Murnane and Cohen (1986). gyz Republic show Makinen and others (2000), Pannarunothai and (c) The International Bank for Reconstruction and Development / The World Bank Endnotes 225 Mills (1997), Peters and others (2003). Castro-Leal and others ancillary or security services, consumables, equipment—which is (2000), Chawla (2001), Lewis (2000). part of the management function as per the framework presented 372. As as studies in China, Egypt, Lebanon, Peru, and Vietnam in chapter 3. show. Carrin and others (1999), Cotteril and Chakaraborty (2000), 417. Manning (1998). Preker and others (2001), Wagstaff and van Doorslaer (forthcoming). 418. Nieves, La Forgia, and Ribera (2000); Eichler, Auxilia, and 373. Cebu Study Team (1991) and Glewwe (1999). Pollock (2001); Chowdhury (2001). 374. Wagstaff, van Doorslaer, and Watanabe (2001). 419. World Bank (2002j). 375. Schieber and Maeda (1997). 420. Naimoli and Vaillancourt (2003). 376. Cai and others (1998). 421. Brenzel and Claquin (1994). 377. Chaudhury and Hammer (2003). 422. Holmstrom and Milgrom (1991) and Mills and Bromberg 378. WHO (World Health Organization) (1998). (1998). 379. Bennet and McPake (1997). 423. Hughes (1993). 380. Waters and Aselsson (2002), Soucat and Rani (2003c), and 424. Mills, Broomberg, and Hongoro (1997); Commission on Peters and others (2003). Macroeconomics and Health (2002); Taylor (2003). 381. Bloom and Standing (2001). 425. World Bank (2001d). 382. Peters and others (2003). 426. Save the Children (2002). 383. Lewis (2000). 427. Hanson and others (2002) and McPake (1996). 384. Nahar and Costello (1998). 428. Hanson (2000) and Brinkerhoff and McEuen (1999), 385. World Bank (2003b) 429. Tendler (1998). 386. Soucat and Rani (2003b) and Schieber and Maeda (1997). 430. Van Lerberghe and others (1997) and Tangcharoensathien 387.Mills and others (2002) and Macy and Quick (2002). and Nittayaramphong (1994). 388. Gilson and others (2000); Soucat, Gandaho, and Levy- 431. as in Bolivia, Cambodia, and Matlab (Bangladesh), Nieves, Bruhl (1997); Diop, Yazbeck, and Bitran (1995); Litvack and Bodart La Forgia, and Ribera (2000); Mintz, Savedoff, and Pancorvo (1993). (2000); Bhuiya, Rob, and Quaderi (1998); Bhushan (2003). 389. Price (2001). 432. Mills and others (2002). 390. Rojanapithayakorn and Hanenberg (1996). 433. as in Vietnam World Bank (2001j). 391. UNICEF (2002). 434. Lindelow, Ward, and Zorzi (2003) and Mozambique Health 392. Population Services International (2003). Facility Survey, Ferrinho and Van Lerberghe (2003). 393. Mintz, Savedoff, and Pancorvo (2000); Cuellar, Newbran- 435. Nitayaramphong, Srivanichakom, and Pongsupap (2000). der, and Timmons (2000); Soucat, Gandaho, and Levy-Bruhl 436. Lewis, Eskeland, and Traa-Valerezo (1999). (1997); Diop, Yazbeck, and Bitran (1995). 437. Onyango-Ouma and others (2001). 394. Castañeda (1999), 438. Ferrinho and Van Lerberghe (2003). 395. Mesoamerica Nutrition Program Targeting Study Group 439. World Bank (2001d). (2002). 440. Cochi and others (1998). 396. Institute For Health Sciences and World Bank (2001). 441. Ferriol and others (2003). 397. Marchant and others (2002). 442. United Nations (1961). 398. Jongudomsuk, Thammatuch-aree, and Chittinanda (2002). 443. The references to Costa Rica are based on Lisulo (2003). 399. World Bank (2002j) and Gertler and Boyce (2001). 444. Erikson, Lord, and Wolf (2003). This corresponds to 2.75 400. Van Lerberghe and Ferrinho (2003). family doctors per 1,000 people—in the Latin America and 401. Cotlear (2000). Caribbean region as a whole there are 1.5 doctors (of any kind) per 402. Porignon and others (1998). thousand people (World Development Indicators 2002. 403. Maiga, Nafo F., and El Abassi (1999). 445. Ferriol and others (2003) and World Bank (2002s). 404. Criel (1998). 446. Ferriol and others (2003). 405. Barnighausen and Sauerborn (2002) and Baris (2003). 447. Ferriol and others (2003). 406. Dror and Preker (2003). 448. Uriarte (2002). 407.Domenighetti and others (1988); Ainsworth, Beyrer, and 449. Uriarte (2002). Soucat (2003); Lamboray (2000); Haddad and Gillespie (2001); 450. Ferriol and others (2003). Hadi (2001). 451. Erikson, Lord, and Wolf (2003) and Uriarte (2002). 408. Platteau and Gaspart (2003). 452. See the World Bank (1994c) for a full discussion of eco- 409. Studdert and others (2000) and Bhat (1996). nomic infrastructure. 410. Jonsson and Musgrove (1997). 453. International Monetary Fund and World Bank (2003). 411. Janovsky (2002). 454. Human Settlements Program (2003). 412. Akin, Hutchinson, and Strumpf (2001). 455. Parker and Skytta (2000). 413. Pannarunothai and others (2000). 456. Water and Sanitation Program (WSP-AF) (2003). 414. Van Lerberghe and Ferrinho (2003). 457. Schleifer and Vishny (1994). 415. Perry and others (1999). 458. Savedoff and Spiller (1999). 416. We focus here on how governments can buy health services 459. Foster (2002). outputs and outcomes, in contrast to the purchasing of inputs— 460. Foster (2002). (c) The International Bank for Reconstruction and Development / The World Bank 226 WORLD DEVELOPMENT REPORT 2004 461. Smith (1997a), Smith (1997b), Irwin, personal communi- (2001) for Bolivia; Ahmad (1999) on South Africa; World Bank cation. (2001a) on transition economies; World Bank (2002f) on Indone- 462. Guasch (2003). sia; and World Bank (2002k) and Lundberg (2002) on Pakistan. 463. Apoyo Opinión y Mercado S.A. (2002). 503. Galiani and Schargrodsky (2002). 464. Dumol (2000). 504. Rodden, Eskeland, and Litvack (2003). An expectation of a 465. Shirley (2002). bailout reflects a soft budget constraint. 466. Smith (2003). 505. Musgrave (1959). 467. Raghupati and Foster (2003). 506. Political expediency led the Indonesian parliament to hastily 468. Brook and Locussol (2001). pass laws in 1999 to implement a “big-bang,� rapid decentralization, 469. Gómez-Lobos and Contreras (2000); David Savage, per- but left the expenditure law unclear on expenditure assignments. sonal communication. The laws are now being revised; see World Bank (2002d). 470. Plummer (2003). 507. In Indonesia, the 1999 expenditure law was passed inde- 471. Nickson and Vargas (2002). pendently of the law governing revenue assignments; see World 472. Parker and Skytta (2000) and World Bank (2002o). Bank (2002d). 473. Tremolet (2002). 508. Khemani (2003). 474. Iyer (2002). 509. Ahmad (1999). Financial decentralization (ability to bor- 475. Term suggested by Peter Kolsky based on his work in this row) is usually subsumed into �scal decentralization. Separating area. them conceptually can shed more light on the interactions between 476. Hoy and Jimenez (2003). them. 477. World Bank (2001i). 510. Ahmad (2003). 478. von der Fehr and Millan (2001). 511. See Evans (2003) for a recent review of staf�ng practices in 479. Briscoe (1997). decentralization in Benin, India, Indonesia, Mexico, Pakistan, the 480. Shirley (2002). Philippines, Poland, and Uganda. 481. Briscoe (1997). 512. Much of the government’s service delivery had already 482. Allan, Gotz, and Joseph (2001). been deconcentrated, so even though reporting arrangements 483. iGoli means “city of gold.� changed, most employees moved physically just from one of�ce to 484. Ahmad (1996). another within the same city. 485. Allan, Gotz, and Joseph (2001). 513. Grindle (forthcoming). 486. See the Glossary in this Report for explanations of terms 514. Bahl and Linn (1992). relating to the service delivery framework. 515. Blondel and Manning (2002). 487. This draws on Andrews and Campos (2003). See also Cam- 516. Schacter, Haid, and Koenen-Grant (1999); Koenen-Grant pos and Pradhan (1997). and Garnett (1996). 488. See Holmes (2002), Roberts (2002), and Le Houerou and 517. Devarajan, Dollar, and Holmgren (2001). Taliercio (2002) for reviews. 518. Collier and Pattillo (2000). 489. Holmes (2002). 519. Beschel and Manning (2000). 490. Le Houerou and Taliercio (2002). 520. Evans and Rausch (forthcoming) 491. IMF and IDA (2002). 521. International Labour Organization (2001). 492. IMF (2002). 522. There is an ongoing debate on whether New Public Man- 493. See Shah (2003b) for a discussion of the importance of agement should be attempted in developing countries: Schick public expenditure management in PRSPs and the large challenges (1996), Bale and Dale (1998), Schick (1998), Batley (1999), Man- in governance reforms that the early PRSPs show. ning (2001). 494. Dehn, Reinikka, and Svensson (forthcoming). 523. World Bank (2000b). 495. Andrews (2001). 524. Dixon (2002). 496. See Talero (2001). For a country perspective, see Chile’s 525. Schick (1998). comprehensive 2002–04 e-procurement strategic plan, ChileCom- 526. This draws on Manning and Parison (2003) and World pra (2002), and other materials on the same website. See Bank (2002e). http://wbln0018.worldbank.org/OCS/egovforum.nsf/Main/ccp for 527. Wade (1982); Wade (1985). an e-procurement pro�le of Australia, Brazil, Canada, Chile, Den- 528. Even in countries with a strong civil service tradition, the mark, Mexico, and the United States. problem of political interference can be pernicious, as the huge 497. See Bardhan (2002) for a recent review. problem in India of ad hoc transfers of civil servants to “punish- 498. Litvack, Ahmad, and Bird (1998) and Burki, Perry, and ment postings� demonstrates; see Sundaram (2001). Dillinger (1999). 529. Azfar (2002). 499. Prud’homme (1995) and Rodden, Eskeland, and Litvack 530. For a recent list of these studies, see, for example, World (2003). Bank (2000a) and Abed and Gupta (2002). 500. Bardhan and Mookherjee (2002). 531. Anderson, Kaufmann, and Recanatini (2003). 501. von Braun and Grote (2000). 532. Mauro (1998). 502. See Grindle (forthcoming) on education and Burki, Perry, 533. Rajkumar and Swaroop (2002), Abed and Gupta (2002), and Dillinger (1999) more generally on Latin America; Faguet Azfar and Gurgur (2001), and Di Tella and Savedoff (2001b). (c) The International Bank for Reconstruction and Development / The World Bank Endnotes 227 534. See Kaufmann, Pradhan, and Ryterman (1998) for a dis- funds has also increased, because the index calculated on the basis cussion of the early diagnostic approach. of disbursements does not distinguish pooled funds from non- 535. Anderson, Kaufmann, and Recanatini (2003) highlight the pooled funds. �ndings of these diagnostic surveys for service delivery. 560. Knack and Rahman (2003). The Bureaucratic Quality 536. World Bank (2000a). Indexes are subjective assessments from the International Country 537. Grindle (forthcoming). Risk Guide (ICRG). High ratings reflect the “strength and expertise 538. World Bank (2002e). to govern without drastic changes in policy or interruptions in gov- 539. For example, see chapter 3 of World Bank (2002b) for guid- ernment services.� Ratings are strongly correlated with more ance on designing a poverty monitoring system. detailed, independent assessments of “Weberian� bureaucratic 540. Kremer (2002). structure and stability (Evans and Rauch (1999)), available for a 541. Victora and others (2000a). subset of countries covered by the ICRG. 542. Davey (2000). 561. Picazo (2002). 543. World Bank (1998a), Burnside and Dollar (2000a), Burn- 562. Picazo (2002). side and Dollar (2000b), and Collier and Dollar (2002). 563. Mansuri and Rao (2003). 544. Martens and others (2002) and Ostrom and others (2001). 564. Chase (2002), Newman and others (2002), Paxson and 545. Mackinnon (2003). Schady (2002), and Van Domelen (2002). 546.The literature on fungibility—including Devarajan and 565. World Bank (2002p). Swaroop (1998); Devarajan, Rajkumar, and Swaroop (1999); and 566. Platteau and Gaspart (2003). Feyziogly, Swaroop, and Zhu (1998)—�nds that only a portion of 567. World Bank (2002r). aid stays in the sector: when the government receives sector-speci�c 568. Gunning (2001). aid, it shifts its own resources partially to other sectors. Fungibility 569. Collier (1997), Kapur and Webb (2000), Devarajan, Dollar, suggests that donors should take a more holistic approach to recip- and Holmgren (2001), Dollar and Svensson (2000). ients’ public spending. 570. Svensson (2003). 547. Development Assistance Committee (DAC) of the Organi- 571. Adam and Gunning (2002). sation for Economic Co-operation and Development (OECD). 572. Adam and Gunning (2002). 548. World Bank (2001l). 573. World Bank (2002r). 549. For example, the Global Fund to Fight AIDS, Tuberculosis 574. Angrist and others (2002). and Malaria; the Global Alliance for Vaccinations and Immuniza- 575. Miguel and Kremer (2001). tions; the Global Vaccine Fund; and the Global Environment Facil- 576. van de Walle (2002) and van de Walle and Cratty (2003). ity. For details on health-related global funds see Kalter (2003). 577. Riddel (1999). 550. Ostrom and others (2001). 578. World Bank (2002r). 551. Boyce and Haddad (2001). 579. Harrold and Associates (1995). 552. World Bank (2001f). 580. There have been a number of reviews of the sectorwide 553. Bräutigam (2000). approach, including Brown (2000a), Conway (2000), Foster (2000), 554. World Bank (1998a). Foster, Brown, and Conway (2000), Jones (1997), Jones and Lawson 555. Cohen and Wheeler (1997). (2001), and ; World Bank (2001b). 556. Data for the Development Gateway are provided by the 581. Fozzard and Foster (2001) and Kanbur and Sandler (1999). OECD DAC and other donor sources over several decades. Unfor- 582. Adam and Gunning (2002). tunately, the database has no indication of the number of projects 583. Brown (2000b). ongoing at any given time. 584. Mackinnon (2003). 557. A Her�ndahl index of donor concentration is �rst calcu- 585. The perception—rather than the reality—of �duciary risk lated by summing the squared shares of aid over all donor agencies may reduce political support for foreign aid in the donor country. operating in the recipient country (O’Connell and Saludo (2001)). But that is a political issue in rich countries, not a service-delivery This index, which ranges from 0 to 1, is then subtracted from 1 to issue in poor countries. form an index of donor fragmentation, with high values indicating 586. World Bank (2002r). greater fragmentation (Knack and Rahman (2003)). 587. World Bank (2002r). 558. Index values do not necessarily rise with aid levels or num- 588. Most bilateral donors give more aid to countries that vote ber of projects: doubling each donor’s aid or number of projects similarly to them in the United Nations General Assembly, where but keeping the number of donors and their activity shares con- each nation regardless of size has one vote (Alesina and Dollar stant leaves the index values unchanged. (2000) and Wang (1999)). 559. Data are from the OECD DAC. The trend may overstate the 589. Halonen (2003). worsening of donor fragmentation to the extent pooling of donor 590. See, for example, Tarp and Hjertholm (2000). (c) The International Bank for Reconstruction and Development / The World Bank References The word processed describes informally reproduced works that may Health Sector in Uganda.� Abt. Associates Inc.; MEASURE Evalua- not be commonly available through libraries. tion Project Working Paper 01-35. Bethesda, Md. Abed, George T., and Sanjeev Gupta. 2002. Governance, Corruption, & Akin, John S., David K. Guilkey, and E. Hazel Denton. 1995. “Qual- Economic Performance. Washington, D.C.: International Monetary ity of Services and Demand for Health Care in Nigeria: A Multi- Fund. nomial Probit Estimation.� Social Science and Medicine 40(11): Acharya, Arnab, Fuzzo de Lima, and Mick Moore. 2003.“The Prolifer- 1527–37. ators: Transactions Costs and the Value of Aid.� Institute of Devel- Akin, John S., and Paul Hutchinson. 1999. “Health-Care Facility opment Studies: Brighton, U.K. Processed. Choice and the Phenomenon of Bypassing.� Health Policy and Action learning program on participatory processes for PRSP. 2003. Planning 14(2):135–151. “Reform Of Basic Education in El Salvador.�World Bank: Washing- Alatas, Vivi. 1999. “Analyzing Indonesia’s Poverty Pro�le.� PhD’s the- ton, D.C. Available on line at http://www.worldbank.org/participa- sis. Princeton University. tion/web/web�les/eseducation.htm. Processed. Alatas, Vivi, Lant Pritchett, and Anna Wetteberg. 2003.“Voice Lessons: Adam, Christopher S., and Jan Willem Gunning. 2002. “Redesigning Local Government Organizations, Social Organizations, and the the Aid Contract: Donors’ Use of Performance Indicators in Quality of Local Governance.� Kennedy School of Government, Uganda.� World Development 30(12):2045–56. Harvard University: Cambrige, Mass. Processed. Agarwal, Bina. 2001. “Participatory Exclusions, Community Forestry, Alderman, Harold, and Victor Lavy. 1996. “Household Responses to and Gender: An Analysis for South Asia and a Conceptual Frame- Public Health Services: Cost and Quality Tradeoffs.� World Bank work.� World Development 29(10):1623–48. Research Observer 11(1):3–22. Aghion, Phillipe, and Jean Tirole. 1997.“Formal and Real Authority in Alesina, Alberto, Reza Baqir, and William Easterly. 1999. “Public Organizations.� Journal of Political Economy 105(1):1–29. Goods and Ethnic Divisions.� Quarterly Journal of Economics Ahmad, Junaid. 1996. “The Structure of Urban Governance in South 114:1243–1284. African Cities.� International Tax and Public Finance 3(2):193–213. Alesina, Alberto, and David Dollar. 2000. “Who Gives Foreign Aid to ————. 1999. “Decentralizing Borrowing Powers.� World Bank Whom and Why?� Journal of Economic Growth 5(1):33–63. PREM Note 15. Washington, D.C. Allan, Kevin, Graeme Gotz, and Carmel Joseph. 2001. Johannesburg, ————. 2003. “Creating Incentives for Fiscal Discipline in the New An African City in Change. Johannesburg: Zebra Press. South Africa.� In Jonathan Rodden, Gunnar Eskeland, and Jennie Anand, Sudhir, and Martin Ravallion. 1993.“Human Development in Litback, eds., Decentralization and Hard Budget Constraints. Cam- Poor Countries: On the Role of Private Incomes and Public Ser- bridge: MIT Press. vices.� Journal of Economic Perspectives 7(1):133–150. Ahmad, Junaid, B. N. Goldar, M. Jakariya, and Smita Misra. 2002. Anderson, James, Daniel Kaufmann, and Francesca Recanatini. “Fighting Arsenic, Listening to Rural Communities: Findings from 2003. “Service Delivery, Poverty, and Corruption—Common a Study on Willingness to Pay for Arsenic-free, Safe Drinking Water Threads from Diagnostic Surveys.� Background paper for the in Rural Bangladesh.� World Bank, Water and Sanitation Program- WDR 2004. South Asia: Dhaka, Bangladesh. Available on line at Andrews, Matthew. 2001. “Adjusting External Audits to Facilitate http://www.wsp.org/pdfs/sa_arsenic_learning.pdf. Results-Oriented Government.� The International Journal of Gov- Ainsworth, Martha, Chris Beyrer, and Agnes Soucat. 2003. “AIDS and ernment Auditors 28(2):10–13. Public Policy: The Lessons and Challenges of ‘Success’ in Thai- Andrews, Matthew, and J. Edgardo Campos. 2003. “The Management land.� Health Policy 64:13–37. of Public Expenditures and Its Implications for Service Delivery.� Aitken, Jean-Marion. 1994. “Voices from the Inside: Managing Dis- Background paper for the WDR 2004. trict Health Services in Nepal.� International Journal of Health Plan- Andrews, Matthew, and Anwar Shah. 2003. “Towards Citizen- ning and Management 9(4):309–40. oriented Local-level Budgets in Developing Countries.� In Anwar Akin, John, Paul Hutchinson, and Koleman Strumpf. 2001. “Decen- Shah, eds., Ensuring Accountability When there is No Bottom Line. tralization and Government Provision of Public Goods: The Public Washington, D.C.: World Bank. 228 (c) The International Bank for Reconstruction and Development / The World Bank References 229 Angrist, J. D., and Victor Lavy. 1999.“Using Maimonides’ Rule to Esti- Batley, Richard. 1999. “The Role of Government in Adjusting Coun- mate the Effect of Class Size on Scholastic Achievement.� Quarterly tries: An Overview of Findings.� International Development Journal of Economics 114(2):533–75. Department, University of Birmingham. Monograph. Angrist, Joshua, Eric Bettinger, Erik Bloom, Elizabeth M. King, and Beatty, Jack. 1992. The Rascal King: The Life and Times of James Michael Kremer. 2002. “Vouchers for Private Schooling in Colom- Michael Curley, 1874-1958. Reading, Mass.: Addison-Wesley. bia: Evidence from a Randomized Natural Experiment.� American Bebbington, A. 1997. “New States, New NGOs? Crises and Transitions Economic Review 92(5):1535–58. among Rural Development NGOs in the Andean Region.� World Antia, N. H. 1994. Kerala Shows the Way to Health, International Con- Development 25(11):1755–65. gress on Kerala Studies, Abstracts, Volume 2. Thiruvananthapuram: Beck, Thorsten, George Clarke, Alberto Groff, Philip Keefer, and A.K.G. Centre for Research and Studies. Patrick Walsh. 2001. “New Tools in Comparative Political Econ- Apoyo Opinión y Mercado S.A. 2002. “Sentimientos Encontrados.� omy: The Database of Political Institutions.� World Bank Economic Opinión Data: Resumen de Encuestas a la Opinión Pública 21(2):1–1. Review 15(1):165–176. Appadurai, Arjun. 2001. “Deep Democracy: Urban Governmentality Becker, Gary. 1971. The Economics of Discrimination. Chicago: Uni- and the Horizon of Politics.� University of Chicago: Chicago. versity of Chicago Press. Processed. Behrman, Jere R., and John Hoddinott. 2001. “An Evaluation of the Appleton, Simon. 2001. “What Can We Expect from Universal Pri- Impact of PROGRESA on Preschool Child Height.� IFPRI FCND mary Education?� In Ritva Reinikka and Paul Collier, eds., Uganda’s Discussion Paper 104. Washington, D.C. Available on line at Recovery: The Role of Farms, Firms, and Government. Washington, www.ifpri.org. D.C.: World Bank. Bell, Clive, and Shantayanan Devarajan. 1987. “Intertemporally Con- Azfar, Omar. 2002. “Corruption and the Delivery of Health and Edu- sistent Shadow Prices in an Open Economy: Estimates for Cyprus.� cation Services.� University of Maryland, IRIS: College Park, Md. Journal of Public Economics 32(3):263–85. Processed. Bennet, S., and B. Mills A. McPake. 1997. Private Health Providers in Azfar, Omar, and Tugrul Gurgur. 2001. “Does Corruption Affect Developing Countries. Serving the Public Interest? Lodon and New Health and Education Outcomes in the Philippines?� University of Jersey: Zed Books. Maryland, IRIS: College Park, Md. Processed. Beschel, Robert P. Jr., and Nick Manning. 2000. “The ‘Brain’ of Gov- Bahl, Roy, and Johannes Linn. 1992. Urban Public Finance in Develop- ernment: Central Mechanisms for Policy Formulation and Coordi- ing Countries. New York: Oxford University Press for the World nation.� In Salvatore Schiavo-Campo and Pachampet Sundaram, Bank. eds., To Serve and to Preserve: Improving Public Administration in a Balakrishnan, Suresh. 2002. “Citizen Report Cards and Basic Services Competitive World. Manila: Asian Development Bank. for the Poor.� Public Affairs Centre: Bangalore, India. Processed. Besley, Timothy, and Robin Burgess. 2002. “The Political Economy of Bale, Malcolm, and Tony Dale. 1998. “Public Sector Reform in New Government Responsiveness: Theory and Evidence from India.� Zealand and Its Relevance to Developing Countries.� World Bank Quarterly Journal of Economics 117(4):1415–51. Research Observer 13(1):103–121. Betancourt, Roger, and Suzanne Gleason. 2000. “The Allocation of Banerjee, Abhijit, Shawn Cole, Esther Duflo, and Leigh Linden. 2003. Publicly-Provided Goods to Rural Households in India: On Some “Improving the Quality of Education in India: Evidence from Consequences of Caste, Religion and Democracy.� World Develop- Three Randomized Experiments.� Massachusetts Institute of Tech- ment 28(12):2169–82. nology, Department of Economics: Cambridge, Mass. Processed. Bhargava, Alok, Dean T. Jamison, L. J. Lau, and Christopher J. Murray. Bang, A. T., R. A. Bang, S. B. Baitule, M. H. Reddy, and M. D. Desh- 2001. “Modeling the Effects of Health on Economic Growth.� Jour- mukh. 1999. “Effect of Home-Based Neonatal Care and Manage- nal of Health Economics 20(3):423–440. ment Septics on Neonatal Mortality: Field Trial in Rural India.� Bhat, R. 1996.“Regulation of the Private Health Sector in India.� Inter- Lancet 354(9194):1955–61. national Journal of Health Planning and Management 11:253–74. Bardhan, Pranab. 2002. “Decentralization of Governance and Devel- Bhuiya, A., M. Chowdhury, F. Ahmed, and A. Adams. 2001. opment.� Journal of Economic Perspectives 16(4):185–205. “Bangladesh: An Intervention Study of Factors Underlying Bardhan, Pranab, and Dilip Mookherjee. 2002. “Decentralizing Anti- Increasing Equity in Child Survival.� In T. Evans, M. Whitehead, Poverty Program Delivery in Developing Countries.� University of F. Diderichsen, A. Bhuiya, and M. Wirth, eds., Challenging California, Berkeley, Institute of International Studies: Berkeley, Inequities in Health: From Ethics to Action. New York: Oxford Uni- C.A. Processed. versity Press. Baris, E. 2003. “Resource Allocation and Purchasing in East Asia and Bhuiya, A., U. Rob, and M. R. Quaderi. 1998. “Ensuring Community Paci�c Region.� World Bank: Washington, D.C. Processed. Participation in MCH-FP Activities in Rural Bangladesh Lessons Barnighausen, T., and R. Sauerborn. 2002. “One Hundred and Eigh- Learned from a Pilot Project.� International Centre for Diarrhoeal teen Years of the German Health Insurance System: Are there Any Disease Research and Population Council: Dhaka, Bangladesh. Lessons for Middle- and Low-Income Countries?� Social Science Bhushan, Indu. 2003. “Contracting Health Services in Cambodia.� and Medicine 54(10):1559–87. Background note for the WDR 2004. Barro, Robert J. 1991.“Economic Growth in a Cross-Section of Coun- Bhushan, Indu, Sheryl Keller, and Brad Schwartz. 2002.“Achieving the tries.� Quarterly Journal of Economics 106(2):407–443. Twin Objectives of Ef�ciency and Equity: Contracting for Health (c) The International Bank for Reconstruction and Development / The World Bank 230 WORLD DEVELOPMENT REPORT 2004 Services in Cambodia.� Asian Development Bank, Policy Brief Bräutigam, Deborah. 2000. “Aid Dependence and Governance.� Series 6. Manila. Expert Group on Development Issues Study (EDGI) 2000:1. Stock- Bidani, Benu, and Martin Ravallion. 1997.“Decomposing Social Indi- holm, Sweden. cators Using Distributional Data.� Journal of Econometrics Brazil Ministry of Health. 2001. “Evolution of Numbers of Commu- 77(1):125–139. nity Health Agents 1994-2001.� Brazil Ministry of Health: Bierschenk, T., J. P. Olivier de Sardan, and J. P. Chauveau. 1997. Brasilia. Courtiers Locaux en Développement. Les villages Africaines en Quête Brenzel, Logan, and Pierre Claquin. 1994. “Immunization Programs de Projets. Paris: Karthala. and Their Costs.� Social Science and Medicine 39(4):527–36. Bils, Mark, and Peter J. Klenow. 2000. “Does Schooling Cause Brinkerhoff, Derick, and Mark McEuen. 1999. “New NGO Partners Growth?� American Economic Review 90(5):1160–1183. for Health Sector Reform in Central Asia: Family Group Practice Birdsall, Nancy, and Francois Orivel. 1996. “Demand for Primary Associations in Kazakhstan and Kyrgyzstan.� Special Initiatives Schooling in Rural Mali: Should User Fees Be Increased?� Educa- Report 19. Bethesda, Md. tion Economics 4(3):279–96. Briscoe, John. 1997. “Lessons for the Water Sector.� World Bank: Block, Steven A. 2002. “Political Business Cycles, Democratization, Washington, D.C. Processed. and Economic Reform: The Case of Africa.� Journal of Development Brook, Penelope, and Alain Locussol. 2001. “Easing Tariff Increases, Economics 67(1):205–28. Financing the Transition to Cost-Recovering Water Tariffs in Blokland, M., O. Braadbaart, and K. Schwartz. eds. 1999. Private Busi- Guinea.� In Penelope Brook and Suzanne M. Smith, eds., Contract- ness, Public Owners, Government Share Holding in Water Enter- ing for Public Devices: Output-Based Aid and Its Applications. Wash- prises. The Hague: Ministry of Housing, Spatial Planning, and the ington, D.C.: World Bank. Environment, Government of the Netherlands and the Water Sup- Brooke, James. 1993. “Brazilian State Leads Way in Saving Children.� ply and Sanitation Collaborative Council. The New York Times, May 14. Page: A. Blomquist, William, and Elinor Ostrom. 1958. “Institutional Capacity Brown, Adrienne. 2000a. “Current Issues in Sector-Wide Approaches and the Resolutions of a Commons Dilemma.� Policy Studies for Health Development: Mozambique Case Study.� World Health Review 5(2):383–93. Organization WHO/ISS 4. Geneva. Blondel, Jean, and Nicholas Manning. 2002. “Do Ministers Do What ————. 2000b. “Current Issues in Sector-Wide Approaches for They Say? Ministerial Unreliability, Collegial and Hierarchical Gov- Health Development: Uganda Study.� World Health Organization ernments.� Political Studies 50:455–76. who/iss 3. Geneva. Bloom, G., and H. Standing. 2001. “Pluralism and Marketization in Bruns, Barbara, Alain Mingat, and Ramahatra Rakatomalala. 2003. “A the Health Sector: Meeting Health Needs in Contexts of Social Chance for Every Child: Achieving Universal Primary Education by Change in Low and Middle-Income Countries.� University of Sus- 2015.� World Bank: Washington, D.C. sex, Institute of Development Studies Working Paper 136. Burguess, Simon, Carol Propper, and Deborah Wilson. 2002. “Does Brighton, U.K. Performance Monitoring Work?: A Review of the Evidence from Boix, Carles, and Daniel N. Posner. 1998. “Social Capital: Explaining the UK Public Sector, Excluding Health Care.� University of Bristol, its Origins and Effects on Government Performance.� British Jour- Department of Economics, CMPO Working Paper Series 02/49. nal of Political Science 28, 4(October):686–693. Bristol, U.K. Bonilla-Chacin, Maria E., and Jeffrey Hammer. 2003.“Life and Deaths Burki, Shaid Javed, Guillermo Perry, and William Dillinger. 1999. among the Poorest.� World Bank: Washington, D.C. Processed. “Beyond the Center: Decentralizing the State.� World Bank: Wash- Bonilla-Chacin, Maria E., Edmundo Murrugarra, and Moukim ington, D.C. Temourov. 2003.“Health Care during Transition and Health Systems Burnside, Craig, and David Dollar. 2000a. “Aid, Growth, the Incentive Reforms: Evidence from the Poorest CIS Countries.� Paper presented Regime and Poverty Reduction.� In Christopher L. Gilbert and at the Initiative of the CIS-7 Conference. Lucerne. January 20. David Vines, eds., The World Bank Structure and Policies. Cam- Boston, Jonathan. 1996.“Separation of Policy and Operations in Gov- bridge, U.K.: Cambridge University Press. ernment: The New Zealand Experience.� In International Gover- ————. 2000b. “Aid, Policies and Growth.� American Economic nance Network, eds., Policy and Operations. Ottawa: Canadian Review 90(4):847–868. Centre for Management Development. Cai, W. W., J. S. Marks, C. H. Chen, Y. X. Zhuang, L. Morris, and J. R. Botchway, Karl. 2001. “Paradox of Empowerment: Reflections on a Harris. 1998. “Increased Caesarean Sections Rates and Emerging Case Study from Northern Ghana.� World Development Patterns of Health Insurance in Shanghai, China.� American Jour- 29(1):135–53. nal of Public Health 88:777–80. Bowen, James. 1981. A History of Western Education. The Modern Cambodia National Institute of Statistics, Ministry of Health, and West: Europe and the New World. London: Methuen. ORC Macro. 2001. “Cambodia Demographic and Health Survey Boyce, Daniel, and Afef Haddad. 2001. “Thematic Review on Project 2000.� Cambodia National Institute of Statistics, Ministry of Implementation Units: An Analysis of Ongoing and Completed Health: Phnom Penh, Cambodia. Projects in Latin America and the Caribbean.� World Bank, A study Cameron, Lisa. 2001. “The Impact of the Indonesian Financial Crisis carried out on behalf of the Latin American and Caribbean Region on Children, An Analysis Using the 100 Villages Data.� Bulletin of Quality Enhancement Team: Washington, D.C. Indonesian Economic Studies 37(1):7–41. (c) The International Bank for Reconstruction and Development / The World Bank References 231 Campos, J. Edgardo, and Sanjay Pradhan. 1997. “Evaluating Public tems Development Thematic Group (training material series): Expenditure Management Systems: An Experimental Methodology Washington, D.C. with an Application to the Australia and New Zealand Reforms.� ChileCompra. 2002. “Chile Public Procurement System: Strategic Journal of Policy Analysis and Management 16(3):423–45. Plan 2002-2004.� Gobierno de Chile, Ministerio de Hacienda: San- Carnoy, Martin. 1997. “Is Privatization through Education Vouchers tiago, Chile. Really the Answer? A Comment.� World Bank Research Observer Chomitz, Kenneth, Gunawan Setiadi, Azrul Azwar, and Nusye Ismail 12(1):105–16. Widiyarti. 1998. “What Do Doctors Want?: Developing Incentives Carothers, Thomas. 1999. Aiding Democracy Abroad. The Learning for Doctors to Serve in Indonesia’s Rural and Remote Areas.�World Curve. Washington, D.C.: Carnegie Endowment for International Bank Policy Research Working Paper 1888. Washington, D.C. Peace. Chowdhury, Sadia. 2001. “Educating Mothers for Health: Output- Carrin, G, A. Ron, Y. Hui, W. Hong, Z. Tuohong, Z. Licheng, Z. Shuo, Based Incentives for Teaching Oral Rehydration in Bangladesh.� In Y. Yide, Ch. Jiaying, J. Qicheng, Z. Zhaoyang, Y. Jun, and L. Xuesh- J. Penelope and Suzanne M. Smith, eds., Contracting for Public Ser- eng. 1999. “The Reform of the Rural Cooperative Medical System vices: Output Based Aid and Its Applications. Washington, D.C.: in the People’s Republic of China: Interim Experience in 14 Pilot World Bank. Counties.� Social Science and Medicine 48(7):961–72. Claeson, M., E. R. Bos, T. Mawji, and I. Pathmanathan. 2000. “Reduc- Case, A., and A. Deaton. 1999. “School Inputs and Educational Out- ing Child Mortality in India in the New Millenium.� Bulletin of the comes in South Africa.� Quarterly Journal of Economics World Health Organization 78(10):1192–99. 114(3):1047–84. Claeson, M., C. Grif�n, T. Johnston, A. Soucat, A. Wagstaff, and A. Case, Anne. 2001. “Does Money Protect Health Status? Evidence from Yazbec. 2003. “Health, Nutrition, and Population.� In Jeni Klug- South African Pensions.� National Bureau of Economic Research man, eds., A Sourcebook for Poverty Reduction Strategies Vol. II: Working Paper 8495. Cambridge, Mass. Macreconomics and Sectoral Approaches. Washington, D.C.: World Castañeda, Tarsicio. 1999. “The Design, Implementation and Impact Bank. of Food Stamp Programs in Developing Countries.� World Bank, Coady, David. 2003. “Alleviating Structural Poverty in Developing Human Development Network. Washington, D.C. Countries: The Approach of PROGRESA in Mexico.� Background Castro-Leal, Florencia, Julie Dayton, Lionel Demery, and Kalpana paper for the WDR 2004. Menhra. 2000. “Public Spending on Health Care in Africa: Do the Cochi, S. L., C. A. De Quadros, S. Dittman, S. O. Foster, J. Z. Galvez Poor Bene�t?� Bulletin of the World Health Organization Tan, F. C. Grant, J. M. Olivé, H. A. Pigman, C. E. Taylor, and K. 78(1):66–74. Wang. 1998. “Group Report: What Are the Societal and Political Cebu Study Team. 1991. “Underlying and Proximate Determinants of Criteria for Disease Eradication?� In W. R. Dowdle and D. R. Hop- Child Health: The Cebu Longitudinal Health and Nutrition Study.� kins, eds., The Eradication of Infectious Diseases. Berlin: John Wiley American Journal of Epidemiology 133(2):185–201. & Sons: West Sussex. Chabal, Patrick, and Jean-Pascal Daloz. 1999. Africa Works: Disorder as Cohen, John M., and John R. Wheeler. 1997. “Building Sustainable Political Instrument. Oxford, U.K.: James Currey. Professional Capacity in African Public Sectors: Retention Con- Chandran, Ashok. 1999. “Effectiveness of Public Action in Rural straints in Kenya.� Public Administration and Development Health Service Delivery: A Case Study from Kerala.� Masters’s the- 17:307–24. sis. London School of Economics, Development Studies Institute. Cohen, Wilbur J. 1957. Retirement Policies under Social Security: A Leg- Chang, Ha-Joon. 2002. Kicking away the Ladder: Development Strategy islative History of Retirement Ages, the Retirement Test, and Disabil- in a Historical Perspective. London: Anthem Press. ity Bene�ts. Berkeley: University of California Press. Chappell, Henry W., and William Keech. 1990. “Citizen Information, Collier, Paul. 1997. “The Failure of Conditionality.� In C Gwin and J. Rationality, and the Politics of Macroeconomic Policy.� In John M. Nelson, eds., Perspectives on Aid and Development. Washington, Ferejohn and James H. Kuklinsky, eds., Information and Democratic D.C.: Overseas Development Council. Processes. Urbana: University of Illinois Press. Collier, Paul, and David Dollar. 2002. “Aid Allocation and Poverty Chase, Robert. 2002. “Supporting Communities in Transition: The Reduction.� European Economic Review 46(8):1475–1500. Impact of Armenian Social Investment Fund.� World Bank Eco- Collier, Paul, and Catherine Pattillo. 2000. Reducing the Risk of Invest- nomic Review 16(2):219–40. ment in Africa. Basingstone: Macmillan. Chaudhury, Nazmul, and Jeffrey Hammer. 2003. “Ghost Doctors: Commission on Macroeconomics and Health. 2001. Macroeconomics Absenteeism in Bangladeshi Health Facilities.� Background paper and Health: Investing in Health for Economic Development. Geneva: for the WDR 2004. World Health Organization. Chaudhury, Nazmul, Jeffrey S. Hammer, Michael Kremer, Karthik ————. 2002. “Mobilizing Domestic Resources for Health.� World Muralidharan, and F. Halsey Rogers. 2003. “Teacher and Health Health Organization, report of working group 3 on Macro-Eco- Care Provider Absenteeism : A Multi-Country Study.� World Bank: nomics and Health chaired by Kesi Botchewey and Alan Tait: Washington, D.C. Geneva. Chawla, Mukesh. 2001. “How Well Does the Health Sector in Georgia Community Driven Development. 2002. “Sourcebook for Commu- Serve the Poor: An Examination of Public and Out-of-Pocket nity Driven Development in Sub-Saharan Africa.� Community Expenditures on Health.� World Bank, prepared for the Health Sys- Driven Development, Environment and Socially Sustainable (c) The International Bank for Reconstruction and Development / The World Bank 232 WORLD DEVELOPMENT REPORT 2004 Development Network: Washington, D.C. Available on line at Das Gupta, Monica. 1987. “Selective Discrimination against Female http://essd.worldbank.org/CDDWk2000.nsf/GWeb/cdd_concepts. Children in Rural Pubjab, India.� Population and Development Conning, Jonathan, and Michael Kevane. 2002. “Community-Based Review 13(1):77–100. Targeting Mechanisms for Social Safety Nets: A Critical Review.� Das Gupta, Monica, P. Khaleghian, and R. Sarwal. 2003. “Governance World Development 30(3):375–94. of Communicable Disease Control Services: A Case Study and Conway, Tim. 2000. “Current Issues in Sector-Wide Approaches for Lessons from India.� Background paper for the WDR 2004. Health Development: Cambodia Case Study.� World Health Orga- Das, Jishnu, Stefan Dercon, Promila Krishnan, and James Habyari- nization WHO/GPE/00.2. Geneva. mana. 2003. “Rules versus Discretion: Public and Private Funding Coombe, Carol. 2000. “Managing the Impact of HIV/AIDS on the in Zambian Education.� World Bank: Washington, D.C. Processed. Education Sector.� The Center for the Study of AIDS, University of Das, Jishnu, and Jeffrey Hammer. 2003. “Money for Nothing: The Pretoria: Pretoria, South Africa. Processed. Dire Straits of Indian Health Care.� World Bank: Washington, D.C. Cornell, Stephen, and Joseph Kalt. 1995. “Where Does Economic Processed. Development Really Come From? Constitutional Rule among Davey, Shelia. eds. 2000. Health a Key to Prosperity - Success Stories in the Contemporary Sioux and Apache.� Economic Inquiry Developing Countries. Geneva: World Health Organization. 33(3):402–426. Davies, Betty Ann. 2001. “Reforming Birth and the (Re) making Mid- ————. 1997. “Successful Economic Development and Hetero- wifery in North America.� In R. Devries, van Teijlingen Benoit C., geneity of Governmental Form on American Indian Reserva- and S. Wrede, eds., Birth by Design: Pregnancy, Maternity Care, and tions.� In Merilee S. Grindle, eds., Getting Good Government: Midwifery in North America and Europe. London: Routledge. Capacity Building in the Public Sectors of Developing Countries. Davies, Ian. 2000. “The Debate Over Popular Education in El Sal- Cambridge, Mass.: Harvard University Press. vador.� Edgewood College, Madison W.I.: LASA, Miami, Florida, ————. 2000. “Where’s the Glue? Institutional and Cultural Foun- March 2000. Available on line at http://forlang.edgewood.edu/ dations of American Indian Economic Development.� Journal of ian/Education.htm. Processed. Socio Economics 29(5):443–70. Davis, Shelton H., and Harry Anthony Patrinos. 2002. “Investing in Cotlear, D. 2000. “Peru: Reforming Health Care for the Poor.� World Latin America’s Indigenous Peoples: The Human and Social Capi- Bank, Human Development Department (LCSHD) Paper Serie 57. tal Dimensions.� World Bank: Washington, D.C. Processed. Washington DC. de Soto, Hernando. 2000. The Mystery of Capital: Why Capitalism Tri- Cotteril, P., and S. Chakaraborty. 2000. “Lebanon Provider Payment umphs in the West and Fails Everywhere Else. New York: Perseus Case Study.� World Bank, HNP network, background paper to the Book Group. Resource Allocation and Purchasing Research Project: World Deaton, Angus. 1997. The Analysis of Household Surveys: A Micro- Bank. econometric Approach to Development Policy. Baltimore and Lon- Criel, B. 1998. District-Based Health Insurance in Sub-Saharan Africa. don: Johns Hopkins University Press for The World Bank. Part I: From Theory to Practice. Antwerp: ITG Press. Dehn, Jan, Ritva Reinikka, and Jakob Svensson. Forthcoming.“Survey Crocker, Keith, and Scott E. Mastens. 2002. “Prospects for Private Tools for Assessing Service Delivery.� In Francois Bourguinon and Water Provision in Developing Countries: Lessons from 19th Cen- Luiz Pereira da Silva (eds.) Evaluating the Poverty and Distribu- tury America.� In Mary Shirley, eds., Thirsting for Ef�ciency: The tional Impact of Economic Policies. New York: Oxford University Economics and Politics of Urban Water System Reform. Amsterdam, Press. New York: World Bank, Pergamon. Deichmann, Uwe, and Somnik V. Lall. 2003. “Are you satis�ed? Citi- Crook, Richard. 2002. “Urban Service Partnerships: “Street Level zen Feedback and Delivery of Urban Services.� World Bank: Wash- Bureaucrats� and Environmental Sanitation in Kumasi and ington, D.C. Processed. Accra, Ghana: Coping with Organizational Change in the Public Department for International Development and Water and Environ- Bureaucracy.� Paper presented at the the WDR Conference on mental Health at London and Loughborough. 1998.“Guidance Man- Making Services Work for Poor People Conference. Eynsham Hall, ual on Water Supply and Sanitation Programmes.�WEIDC for DIFD: Oxfordshire. November. London. Available on line at http://www.lboro.ac.uk/well/resources/ Crouch, Luis, and Henry F. Healey. 1997.“Education Reform Support. books-and-manuals/guidance-manual/guidance-manual.htm. Volume One: Overview and Bibliography.� Of�ce of Sustainable Devarajan, Shantayanan. 2002. “Growth Is Not Enough.� Word Bank: Development, Bureau for Africa, United States Agency for Interna- Washington, D.C. Processed. tional Development (USAID), Publication Series, Paper 47. Wash- Devarajan, Shantayanan, David Dollar, and Torgny Holmgren. 2001. ington, D.C. Aid and Reform in Africa Lessons from Ten Case Studies. Washing- Cuellar, Carlos J., William Newbrander, and Barbara K. Timmons. ton, D.C.: World Bank. 2000. “The PROSALUD Model for Expanding Access to Health Devarajan, Shantayanan, Margaret J. Miller, and Eric L. Swanson. 2002. Services.� World Bank, HNP Network: Washington D.C. “Goals for Development: History, Prospects, and Costs.� World Daramola, A. G., Hawa A. Biu, S. Ogoh Alubo, Agi S. P. I. Shebu, Bank Policy Research Working Paper 2819. Washington, D.C. Uchenna M. Nzewi, and Paul Francis. 1998. Hard Lessons: Primary Devarajan, Shantayanan, Andrew Sunil Rajkumar, and Vinaya Swa- Schools, Community, and Social Capital in Nigeria. Washington, roop. 1999. “What Does Aid to Africa Finance?� World Bank Policy D.C.: World Bank. Research Working Paper 2092. Washington, D.C. (c) The International Bank for Reconstruction and Development / The World Bank References 233 Devarajan, Shantayanan, and Vinaya Swaroop. 1998. “The Implica- Easterly, William, and Ross Levine. 1997. “Africa’s Growth Tragedy: tions of Foreign Aid Fungibility for Development Assistance.� Policies and Ethnic Divisions.� Quarterly Journal of Economics World Bank Policy Research Working Paper Series 2022. Washing- 112(4):1203–50. ton, D.C. Eichler, Rena, Paul Auxilia, and John Pollock. 2001. “Output-Based Di Tella, Rafael, and William D. Savedoff. eds. 2001a. Diagnosis Cor- Health Care: Paying for Performance in Haiti.� World Bank, View- ruption: Fraud in Latin America’s Public Hospitals . Washington, point, Note 236. Washington, D.C. D.C.: The Inter-American Development Bank. El Salvador Evaluation Team. 1997. “El Salvador’s EDUCO Program: Di Tella, Raphael, and William Savedoff. 2001b. The Corruption That A First Report on Parents’ Participation in School-Based Manage- Compromises Health: Studies of Public Hospitals in Latin America. ment.� World Bank Working Paper Series on Impact Evaluation of Washington, D.C.: Inter-American Development Bank. Education Reforms 4. Washington, D.C. Available on line at Di Tella, Raphael, and Ernesto Schargrodsky. Forthcoming. “The Role http://www.worldbank.org/research/projects/ of Wages and Auditing during a Crackdown on Corruption in the impact.htm. City of Buenos Aires.� Journal of Law and Economics. Erikson, Dan, Annie Lord, and Peter Wolf. 2003. “Cuba’s Social Ser- Diaz-Cayeros, Alberto, and Beatriz Magaloni. 2002. “Public Services vices: A Review of Education, Health, and Sanitation.� Background Mediated by the Political Process.� Stanford University: Stanford, paper for the WDR 2004. C.A. Processed. Eriksson, Joh, Alcira Kreimer, and Margaret Arnold. 2000. “El Sal- ————. 2003. “The Politics of Public Spending: Part I - The Logic vador: Post-Conflict Reconstruction.� World Bank Operation Eval- of Vote Buying. Part II -The Programa Nacional de Solidaridad uation Department, Country Case Study Series: Washington, D.C. (PRONASOL) in Mexico.� Background paper for the WDR 2004. Eskeland, Gunnar S., and Deon Filmer. 2002. “Autonomy, Participa- Diop, F., A. Yazbeck, and R. Bitran. 1995. “The Impact of Alternative tion, and Learning in Argentine Schools.� World Bank Policy Cost Recovery Schemes on Access and Equity in Niger.� Health Pol- Research Working Paper 2766. Washington, D.C. icy and Planning 10(3):223–40. Estévez, Federico, Beatriz Magaloni, and Alberto Diaz-Cayeros. 2002. Dixit, Avinash. 2000. “Incentives and Organizations in the Public Sec- “The Erosion of One-party Rule: Clientelism, Portfolio Diversi�ca- tor: An Interpretive Review.� Journal of Human Resources tion, and Electoral Strategy.� Paper presented at the the 2002 37(4):696–772. Annual Meetings of the American Political Science Association Con- Dixon, Geoffrey. 2002. “Thailand’s ‘Hurdle’ Approach to Budget ference. Boston. August. Reform.� World Bank PREM Note 73. Washington, D.C. Evans, A. 2003. “Decentralization: A Review of Staf�ng Practices in Dollar, David, and Jakob Svensson. 2000. “What Explains the Success Eight Countries.� World Bank: Washington, D.C. Processed. or Failure of Structural Adjustment Programmes?� Economic Jour- Evans, Peter. 1996. “Government Action, Social Capital and Develop- nal 110(466):894–917. ment: Reviewing the Evidence on Synergy.� World Development Domenighetti, Gianfranco, Pierangelo Luraschi, Antoine Casabianca, 24(6):1119–32. Felix Gutzwiller, Alberto Spinelli, Ennio Pedrinis, and Francesca Evans, Peter B., and James E. Rauch. 1999. “Bureaucratic Structure Repetto. 1988.“Effect of Information Campaign by the Mass Media and Bureaucratic Performance in Less Developed Countries.� Jour- on Hysterectomy Rates.� The Lancet 2:1470–3. nal of Public Economics 75(1):49–71. Dreze, Jean, and Haris Gazdar. 1996. “Uttar Pradesh: The Burden of Evans, Peter B., and James Rausch. Forthcoming. “Bureaucracy and Inertia.� In Jean Dreze and Amartya Sen, eds., Indian Development: Growth: A Cross-National Analysis of the Effects of ‘Weberian’ State Selected Regional Perspectives. New York: Oxford University Press. Structures on Economic Growth.� Journal of Public Economics. Dreze, Jean, and Amartya Sen. 2002. India: Development and Partici- Evers, Pieter. 2003. “Village Governments and Their Communities.� pation. Oxford: Oxford University Press. World Bank/Bappenas Local Level Institutions Study: Jakarta. ————. eds. 1991. The Political Economy of Hunger: Famine Pre- Processed. vention. Oxford: Clarendon Press. Faguet, Jean-Paul. 2001. “Does Decentralization Increase Govern- Dreze, Jean, and Nicholas Stern. 1987. “The Theory of Cost-Bene�t ment Responsiveness to Local Needs? Evidence from Bolivia.� Analysis.� In A. J. Auerbach and M. Feldstein, eds., Handbook in World Bank Policy Research Working Paper 2516. Washington, Public Economics Vol. 2. Amsterdam, New York: North-Holland. D.C. Dror, D. M., and A. Preker. 2003. Social Re-insurance - A New Ferejohn, John. 1974. Pork Barrel Politics: Rivers and Harbors Legisla- Approach to Sustain Community Health Financing. Geneva: Inter- tion, 1947-1968. Standford C.A.: Stanford University Press. national Labor Organization. Ferejohn, John, and James H. Kuklinsky. eds. 1990. Information and Duflo, Esther. 2001. “Schooling and Labor Market Consequences of Democratic Processes. Urbana: University of Illinois Press. School Construction in Indonesia: Evidence from an Unusual Pol- Ferrinho, P., and W. Van Lerberghe. 2003. “Managing Health Profes- icy Experiment.� American Economic Review 91(4):795–813. sionals in the Context of Limited Resources: a Fine Line Between Dumol, Mark. 2000. The Manila Water Concession: A Key Government Corruption and the Need for Moonlighting.� Background paper Of�cial’s Diary of the World’s Largest Water Privatization. Washing- for the WDR 2004. ton, D.C.: World Bank. Ferriol, Angela, Victoria Perez, Didio Quintana, and Aida Atienza. Dutta, Antara. 2003. “Public Service Delivery: The PROBE and Its 2003. “Servicios de Educacion, Salud, Agua y Saneamiento en Impact.� Background paper for the WDR 2004. Cuba.� Background paper for the WDR 2004. (c) The International Bank for Reconstruction and Development / The World Bank 234 WORLD DEVELOPMENT REPORT 2004 Feyziogly, Tarhan, Vinaya Swaroop, and Min Zhu. 1998.“A Panel Data Wastewater Systems: Global Lessons for Improving Management. Analysis of the Fungibility of Foreign Aid.� World Bank Economic New York: John Wiley and Sons, Inc. Review 12(1):29–58. Fozzard, Adrian, and Mick Foster. 2001. “Changing Approaches to Filmer, Deon. 2000.“The Structure of Social Disparities in Education: Public Expenditure Management in Low-Income Aid Dependent Gender and Wealth.� World Bank Policy Research Working Paper Countries.� United Nations University, World Institute for Devel- 2268. Washington, D.C. opment Economic Research Discussion Paper 107. Helsinki. ————. 2002. “Teacher Pay in Indonesia.� World Bank: Washing- Frank, Jonas, Federico Starnfeld, and Janos Zimmerman. Forthcom- ton, D.C. Processed. ing. “Linea de Referencia: El Proceso de Descentralización en ————. 2003a. “Determinants of Health and Education Out- Argentina, Bolivia, Brasil, Chile, Colombia, Ecuador, Perú y comes.� Background note for the WDR 2004. Venezuela.� Quito: GTZ/CONAM. ————. 2003b. “The Incidence of Public Expenditures on Health Freedom House. 1999. “Democracy’ s Century: A Survey of Global and Education.� Background paper for the WDR 2004. Political Change in the 20th Century.� Freedom House: New York. Filmer, Deon, Jeffrey S. Hammer, and Lant H. Pritchett. 2000. “Weak ————. 2002. “Freedom in the World 2002.� Freedom House: New Links in the Chain: A Diagnosis of Health Policy in Poor Coun- York. tries.� World Bank Research Observer 15(2):199–224. Frey, Bruno S. 1997. “The Cost of Price Incentives: An Empirical ————. 2002. “Weak Links in the Chain II: A Prescription for Analysis of Motivation Crowding-Out.� American Economic Review Health Policy in Poor Countries.� World Bank Research Observer 87(4):746–55. 17(1):47–66. Fuentes, Patricio, and Reiko Niimi. 2002. “Motivating Municipal Filmer, Deon, Samuel Lieberman, and David Ariasingam. 2002. Action for Children: The Municipal Seal of Approval in Ceará, “Indonesia and Education for All.� World Bank, Development Brazil.� Environment and Urbanisation 14(2):123–133. Research Group and Human Development Unit of East Asia Gachuhi, D. 1999. “The Impact of HIV/AIDS on Education Systems Region: Washington, D.C. Processed. in the Eastern and Southern Africa Region and the Response of Filmer, Deon, and Lant Pritchett. 1999a. “The Effect of Household Education Systems to HIV/AIDS: Life Skills Programs.� Paper pre- Wealth on Educational Attainment: Evidence from 35 Countries.� sented at the Sub-Saharan Africa Conference on EFA 2000 Confer- Population and Development Review 25(1):85–120. ence. Johannesburg, South Africa. December 21. Filmer, Deon, and Lant H. Pritchett. 1999b. “The Impact of Public Galabawa, J. C. J., F. E. M. K. Senkoro, and A. F. Lwaitama. 2000. The Spending on Health: Does Money Matter?� Social Science and Med- Quality of Education in Tanzania. Dar-es-Salaam: Institute of icine 49(10):1309–23. Kiswahili Research. ————. 2001. “Estimating Wealth Effects without Expenditure Galiani, Sebastian, Paul Gertler, and Ernesto Schargrodsky. 2002. Data—or Tears: An Application to Educational Enrollments in “Water for Life: The Impact of the Privatization of Water Services States of India.� Demography 38(1):115–132. on Child Mortality.� Stanford University, Center for Research on Fiorina, Morris P. 1990. “Information and Rationality in Elec- Economic Development and Policy Reform Working Paper 154. tions.� In John Ferejohn and James H. Kuklinsky, eds., Informa- Stanford. tion and Democratic Processes. Urbana: University of Illinois Galiani, Sebastian, and Ernesto Schargrodsky. 2002. “School Decen- Press. tralization and Education Quality: The Role of Fiscal De�cits.� Uni- Fiorina, Morris P., and Kenneth Shepsle. 1990. “A Positive Theory of versidad de San Andres: Buenos Aires, Argentina. Processed. Negative Voting.� In John Ferejohn and James H. Kuklinsky, eds., Gauri, Varun. 2003. “A Rights-Based Approach to Health Care and Information and Democratic Processes. Urbana: University of Illi- Education.� Background note for the WDR 2004. nois Press. Gauri, Varun, and Ayesha Vawda. 2003.“Vouchers for Basic Education Fitz-Gibbon, Carol Taylor. 1996. Monitoring Education: Indicators, in Developing Countries: A Principal-Agent Perspective.� Back- Quality and Effectiveness. London: Cassell. ground paper for the WDR 2004. Foster, Mick. 2000. “New Approaches to Development Cooperation: Gazdar, Haris. 2000. “State, Community, and Public Schooling—A What Can We Learn from Experience with Implementing Sector- Political Economy of Universal Education in Pakistan.� London Wide Approaches?� Overseas Development Institute Working School of Economics: London, U.K. Processed. Paper 140. London. Available on line at http://www.odi.org.uk/ Gertler, Paul. 2000. “Final Report: The Impact of PROGRESA on pppg/publications/working_papers/140.html. Health.� International Food Policy Research Institute: Washington, Foster, Mick, Adrienne Brown, and Tim Conway. 2000.“Sector Health D.C. Available on line at http://www.ifpri.org/. Approaches for Health Development: A Review of Experience.� Gertler, Paul, and Simone Boyce. 2001. “An Experiment in Incentive World Health Organization WHO/GPE/00.1. Geneva. Based Welfare: The Impact of Mexico’s PROGRESA on Health.� Foster, S. D. 1990. “Improving the Supply and Use of Essential Drugs University of California at Berkeley: Berkeley, C.A. Available on line in Sub-Saharan Africa.� World Bank Policy Research Working at http://faculty.haas.berkeley.edu/gertler/. Processed. Paper 456. Washington, D.C. Gertler, Paul, and Jeffrey S. Hammer. 1997a. “Financing and Allocat- Foster, Vivien. 2002. “Ten Years of Water Service Reform in Latin ing Public Expenditures in the Health Sector.� In G. Scheiber and D. America: Towards an Anglo-French Model.� In Paul Seidenstat, Chollet, eds., Health Reform in Developing Countries. Washington, David Haarmeyer, and Simon Hakim, eds., Reinventing Water and D.C.: World Bank. (c) The International Bank for Reconstruction and Development / The World Bank References 235 ————. 1997b. “Strategies for Pricing Publicly Provided Health Good, Harry G., and James D. Teller. 1969. A History of Western Edu- Services.� In George J. Schieber, eds., Innovations in Health Care cation. London: Macmillan. Financing: Proceedings of a World Bank Conference. Washington, Gorter, A., P. Sandiford, Z. Segura, and C. Villabela. 1999. “Improved D.C.: World Bank. Health Care for Sex Workers: A Voucher Program for Female Sex Gertler, Paul, and John Molyneaux. 1995. “Experimental Evidence on Workers in Nicaragua.� Research for Sex Work 2 the Effect of Raising User Fees for Publicly Delivered Health Care Govindaraj, Ramesh, Christopher J. L. Murray, and Chellaraj Gna- Services: Utilization, Health Outcomes and Private Provider naraj. 1995. “Health Expenditures in Latin America and the Response.� RAND: Santa Monica, CA. Processed. Caribbean.� World Bank Technical Paper 274. Washington, D.C. GfK Praha—Institute for Market Research. 2001. “Corruption Cli- Gradstein, Mark, and Moshe Justman. 2002.“Education, Social Cohe- mate—Central and Eastern Europe: Results of and International sion, and Economic Growth.� American Economic Review Research Project on Corruption in 11 Central and Eastern Euro- 92(4):1192–1204. pean Countries.� GfK Praha: Prague. Greaney, Vincent, Shahidur R. Khandker, and Mahmudul Alam. 1999. Gibson, John. 2000.“Who’s Not in School? Economic Barriers to Uni- Bangladesh: Assessing Basic Learning Skills. Dhaka, Bangladesh: versal Primary Education in Papua New Guinea.� Paci�c Economic University Press. Bulletin 15(2):46–58. Green, Andy. 1990. Education and State Formation: The Rise of Educa- Gilson, L. 1997.“The Lessons of User Fee Experience in Africa.� Health tion Systems in England, France, and the USA. New York: St. Mar- Policy and Planning 12(4):273–85. tin’s Press. Gilson, L., D. Kalyalya, F. Kuchler, S. Lake, H. Oranga, and M. Ouendo. Grindle, Merilee S. Forthcoming. Despite the Odds: Contentious Poli- 2000. “The Equity Impacts of Community Financing Activities in tics and Education. Princeton N.J.: Princeton University Press. Three African Countries.� International Journal of Health Planning Grossman, Gene, and Elhanan Helpman. 1999. “Competing for and Management 15(4):291–317. Endorsements.� American Economic Review 89(3):501–24. ————. 2001. “Strategies for Promoting Equity: Experience with Guasch, J. Luis. 2003. “Concessions of Infrastructure Services: Inci- Community Financing in Three African Countries.� Health Policy dence and Determinants of Renegotiation—An Empirical Evalua- 58(1):37–67. tion and Guidelines for Optimal Concession Design.� World Bank Girishankar, Navin. 2002. “Institutional Underpinnings of Public and University of California, San Diego: Washington, D.C. and San Expenditure Management in Malawi: Current Practice, Change Diego, C.A. Processed. Prospects, and Critical Pathways.� World Bank: Washington, D.C. Gundlach, Erich, and Ludger Wößman. 2001. “The Fading Product- Processed. ivity of Schooling in East Asia.� Journal of Asian Economies Glaeser, Edward L., and Andrei Shleifer. 2002. “The Curley Effect.� 12(3):401–17. National Bureau of Economic Research Working Paper 8942. Cam- Gundlach, Erich, Ludger Wößman, and Jens Gmelin. 2001. “The bridge, Mass. Decline of Schooling Productivity in OECD Countries.� Economic Glewwe, Paul. 1999. “Why Does Mother’s Schooling Raise Child Journal 111(471):C135–C147. Health in Developing Countries? Evidence from Morocco.� Journal Gunning, Jan Willem. 2001. “Rethinking Aid.� In Boris Pleskovic and of Human Resources 34(1):124–59. Nicholas Stern, eds., Annual World Bank Conference on Develop- Glewwe, Paul, Nauman Ilias, and Michael Kremer. 2000. “Teacher ment Economics. Washington, D.C.: World Bank. Incentives.� University of Minesota, Competitions Economics Inc., Gupta, Sanjeev, Hamid R. Davoodi, and Erwin R. Tiongson. 2002. and Harvard University: Minesota, W.I. and Cambridge, Mass. “Corruption and the Provision of Health Care and Education Ser- Processed. vices.� In George T. Abed and Sanjeev Gupta, eds., Governance, Cor- Glewwe, Paul, M. Kremer, and Sylvie Moulin. 1997. “Textbooks and ruption and Economic Performance. Washington, D.C.: The Interna- Test Scores: Evidence from a Prospective Evaluation in Kenya.� Har- tional Monetary Fund. vard University: Cambridge, Mass. Processed. Gupta, Sanjeev, Marijn Verhoeven, and Erwin R. Tiongson. Forth- Glinskaya, Elena, and Jyotsna Jalan. 2003.“Improving Primary School coming. “Public Spending on Health Care and the Poor.� Health Education in India: An Impact Assesment of DPEP—Phase I.� Economics. World Bank: Washington, D.C. Processed. ————. 2002. “The Effectiveness of Government Spending on Goetz, Anne-Marie, and John Gaventa. 2001. “Bringing Citizen Voice Education and Health Care in Developing and Transition Econom- and Client Focus into Service Delivery.� Institute of Development ics.� European Journal of Political Economy 18(4):717–737. Studies Working Paper 138. Brighton, U.K. Gwatkin, D., and M. Guillot. 2000.“The Burden of Disease among the Goetz, Anne-Marie, and Rob Jenkins. 2002. “Voice, Accountability Global Poor: The Current Situation, Future Trends, and Implica- and Human Development: The Emergence of a New Agenda.� tions for Strategy.� World Bank: Washington DC. Background Paper for the HDR 2002, Occasional Paper 2002/4. Gwatkin, D., S. Rutstein, K. Johnson, R. Pande, and A. Wagstaff. 2000. New York. “Socioeconomic Differences in Health, Nutrition and Population.� Gómez-Lobos, Andrés, and Dante Contreras. 2000. “Subsidy Policies World Bank, Health, Nutrition and Population Discussion Paper. for the Utility Industries: A Comparison of the Chilean and Washington D.C. Colombian Water Subsidy Schemes.� Universidad de Chile, Depar- Gwatkin, Davidson, and Others. 2000. “Theme Section: Poverty and tamento de Economía: Santiago de Chile. Processed. Health.� Bulletin of the World Health Organization 78(1):2–96. (c) The International Bank for Reconstruction and Development / The World Bank 236 WORLD DEVELOPMENT REPORT 2004 Habyarimana, James, Jishnu Das, Stefan Dercon, and Pramila Krish- Hoxby, C. M. 2000.“The Effects of Class size on Student Achievement: nan. 2003. “Sense and Absence: Absenteeism and Learning in Zam- New Evidence from Population Variation.� Quarterly Journal of bian Schools.� World Bank: Washington, D.C. Processed. Economics 115(4):1239–85. Haddad, L., and S. Gillespie. 2001. “Attacking the Double Burden of Hoy, Michael, and Emmanuel Jimenez. 2003. “The Impact on the Malnutrition in Asia and the Paci�c.�Asian Development Bank and Urban Environment of Incomplete Property Rights.� World Bank: International Food Policy Research Institute: Manila and Washing- Washington, D.C. Processed. ton, D.C. Hsieh, Chang, and Miguel Urquiola. 2003. “When Schools Compete, Haddad, Salim, and Pierre Fournier. 1995. “Quality, Cost and Utiliza- How Do They Compete? An Assessment of Chile’s Nationwide tion of Health Services in Developing Countries. A Longitudinal School Voucher Program.� Princeton University and World Bank: Study in Zaire.� Social Science and Medicine 40(6):743–53. Princeton, N.J. and Washington, D.C. Processed. Hadi, A. 2001. “Promoting Health Knowledge through Micro-Credit Hughes, D. 1993. “General Practitioners and the New Contract: Pro- Programmes: Experience of BRAC in Bangladesh.� Health Promo- moting Better Health Through Financial Incentives.� Health Policy tion International 16(3):219–28. 25:39–50. Halonen, Maija. 2003. “Coordination Failure in Foreign Aid.� Back- Human Settlements Program, International Institute for Environ- ground paper for the WDR 2004. ment and Development. 2003. “Local Action for Global Water Hammer, Jeffrey S., and William G. Jack. 2001. “The Design of Incen- Goals: Addressing Inadequate Water and Sanitation in Urban tives for Health Care Providers in Developing Countries: Con- Areas, (or the State of the World’s Cities: Water and Sanitation).� tracts, Competition and Cost Control.� World Bank Policy International Institute for Environment and Development for UN- Research Working Paper 2547. Washington, D.C. HABITAT: London, U.K. Processed. Hammer, Jeffrey S., Ijaz Nabi, and James Cercone. 1995. “The Dis- Hunt, Paul. 2002. “The International Human Rights Treaty Obliga- tributional Impact of Social Sector Expenditures in Malaysia.� In tions of States Parties in the Context of Service Provision.� UN D. van de Walle and K. Nead, eds., Public Spending and the Poor: Committee on the Rights of the Child: Geneva. Theory and Evidence. Baltimore: Johns Hopkins University Press. Hutchinson, Paul. 2001. “Combating Illness.� In Ritva Reinikka and Hanson, K., J. Kamwanga, B. McPake, and F. Ssengooba. 2002. Paul Collier, eds., Uganda’s Recovery: The Role of Farms, Firms, and “Towards Improving Hospital Performance in Uganda and Zam- Government. Washington, D.C.: World Bank. bia: Reflections and Opportunities for Autonomy.� Health Policy ICM. 2003. “Draft Statement on the Philosophy and Model of Mid- 61(1):73–94. wifery Care.� ICM: Hague, Netherlands. Processed. Hanson, Kara. 2000. “Implementing Health Sector Reforms in Africa: IMF. 2002. “Actions to Strengthen the Tracking of Poverty-Reducing A Review of Eight Country Experiences.� UNICEF; Division of Public Spending in Heavily Indebted Poor Countries.� Interna- Evaluation, Policy and Planning: Paris. tional Monetary Fund: Washington, D.C. Hanushek, Eric A. 2002. “Evidence, Politics, and the Class Size IMF, and IDA. 2002. “Review of the Poverty Reduction Strategy Paper Debate.� In Mishel Lawrence and Richard Rothstein, eds., The Class (PRSP) Approach: Early Experience with Interim PRSPs and Full Size Debate. Washington, D.C.: Economic Policy Institute. PRSPs.� International Monetary Fund: Washington, D.C. Harrold, Peter, and Associates. 1995. “The Broad Sector Approach to Indonesia: Ministry of National Education. 2002. “Indonesia Edu- Investment Lending: Sector Investment Programs.� World Bank cation at a Glance�. Jakarta, Indonesia: Ministry of National Discussion Paper 302. Washington, D.C. Education. Available on line at http://www.depdiknas.go.id/ Hart, Julian. 1971. “The Inverse Care Law.� The Lancet 360(9328): statistik.htm. 405–12. Indonesia: Ministry of Religious Affairs. 2002.“Project Lensa�. Jakarta, Hays, Samuel P. 1959. Conservation and the Gospel of Ef�ciency: The Indonesia: Ministry of Religious Affairs. Available on line at Progressive Conservation Movement 1890-1920. Cambridge, Mass.: www.depag.net. Harvard University Press. Institute For Health Sciences, Kunming Medical College, and World Hino, T. 1993. “Community Participation in ‘Programme de Restruc- Bank. 2001. “Report on Maternal and Child Health Poverty Allevi- turation de L’Hydraulique Villageoise’ in Côte d’Ivoire.� Africa ation Fund Study in China.� Institute for Health Sciences, Kunming Country Department 11. World Bank: Washington, D.C. Medical College (HIS, KMC) and World Bank. Processed. Hirschman, Albert O. 1970. Exit, Voice, and Loyalty: Responses to International Budget Project. 2000. “A Taste of Success: Examples of Declines in Firms, Organizations, and States. Cambridge. Mass.: the Budget Work of NGOs.� International Budget Project of the Harvard University Press. Center on Budget and Policy Priorities: Washington, D.C. Holmes, Malcolm. 2002. “MTEF Synthesis Summary.� Overseas International Energy Agency. 2002. “Energy and Poverty.� In Interna- Development Institute: London, U.K. Processed. tional Energy Agency, eds., World Energy Outlook 2002. Paris: Inter- Holmstrom, Bengt, and Paul Milgrom. 1991.“Multitask Principal-Agent national Energy Agency. Analysis: Incentive Contracts, Asset Ownership and Job Design.� Jour- International Labor Organization (ILO). 2002. “Social Dialogue in nal of Law, Economics and Organization 7(special issue):24–52. the Health Services: Institutions, Capacity and Effectiveness.� Hossain, Naomi, and Mick Moore. 2002. “Arguing for the Poor: Elites International Labor Organization: Geneva. Available on line at and Poverty in Developing Countries.� Institute of Development http://www.ilo.org/public/english/dialogue/sector/techmeet/ Studies Working Paper 148. Brighton, U.K. jmhs02/jmhs-r.pdf. (c) The International Bank for Reconstruction and Development / The World Bank References 237 International Labour Organization. 2001. “The Impact of Decentral- Review of Country Experience and Lessons Learned.� Bureau of ization and Privatization on Municipal Services.� International Health and Planning, MOPH, Thailand: Bangkok. Processed. Labor Organization: Geneva. Jonsson, Bengt, and Phillip Musgrove. 1997. “Government Financing International Monetary Fund, and World Bank. 2003. “Progress of Health Care.� In G. Schieber, eds., Innovations in Health Care. Report and Critical Steps in Scaling Up: Education for All, Health, Washington, D.C.: World Bank Discussion Paper 365. HIV/AIDS, Water and Sanitation.� Development Committee (Joint Joshi, Anuradha, and Mick Moore. 2000. “The Mobilization Potential Ministerial Committee of the Boards of Governors of the Bank and of Anti-Poverty Programmes.� Institute of Development Studies the Fund on the Transfers of Real Resources to Developing Coun- Discussion Paper 374. Brighton, U.K. tries): Washington, D.C. Kalter, Henry. 2003.“Global Health Funds.� Background paper for the Irwin, Timothy. 2003. “Free-Entry Competition in Infrastructure as a WDR 2004. Response to Poor Governance.� World Bank: Washington, D.C. Kanbur, Ravi, and Todd Sandler. 1999. “The Future of Development Processed. Assistance: Common Pools and International Public Goods.� Over- Iyer, Parameswaran. 2002. “Willingness to Charge and Willingness to seas Development Council, Policy Essay 25. Baltimore. Pay: The World Bank-assisted China rural Water Supply and Sani- Kapur, Devesh, and Richard Webb. 2000. “Governance-Related Con- tation Program.� Water and Sanitation Program, Field Note ditionalities of the International Financial Institutions.� United Jacobson, Charles D., and Joel A. Tarr. 1996. “No Single Path: Owner- Nations and Center for International Development Harvard Uni- ship and Financing of Infrastructure in the 19th and 20th Cen- versity; G-24 Discussion Paper 6. New York and Boston. turies.� In Ashoka Mody, eds., Infrastructure Delivery: Private Initia- Kaufmann, Daniel, Sanjay Pradhan, and Randy S. Ryterman. 1998. tive and the Public Good. Washington, D.C.: World Bank. “New Frontiers in Diagnosing and Combating Corruption.� World Jaffré, Y., J.-P. Olivier, and Olivier de Sardan. eds. 2002. Les Dysfonc- Bank PREM Note 7. Washington, D.C. tionnements des Systèmes de Soins. Rapport du Volet Socio-anthro- Keefer, Philip. 2002. “Clientelism, Credibility, and Democracy.� World pologique. Enquêtes sur l’Accès aux Soins dans 5 Capitales d’Afrique Bank: Washington, D.C. Processed. de l’Ouest. Marseille: UNICEF-Coopération Francaise. ————. 2003. “Personal Communication.� World Bank: Washing- Jaffré, Y., and A. Prual. 1993.“Le Corps des Sages Femmes, Entre Iden- ton, D.C. Processed. tités Sociales et Professionnelles.� Sciences Sociales et Sante Keefer, Philip, and Stuti Khemani. 2003. “The Political Economy of XI(2):63–80. Social Service Provision.� Background paper for the WDR 2004. Jaffré, Yannick, and Alain Prual. 1994. “Midwives in Niger: An Kelly, Michael J. 1999. “The Impact of HIV/AIDS on Schooling in Uncomfortable Position between Social Behaviors and Health Care Zambia.� Paper presented at the XIth International Conference on Constraints.� Social Science and Medicine 38(8):1069–1073. AIDS and STDs in Africa Conference. Janovsky, K. 2002. “Review of Experiences in Health Services Decen- ————. 2000. “Planning for Education in the Context of tralization.� World Health Organization: Geneva. HIV/AIDS.� UNESCO, International Institute for Educational Jayasuriya, Ruwan, and Quentin Wodon. 2002. “Explaining Country Planning: Paris. Ef�ciency in Improving Health and Education Indicators: The Role Khemani, Stuti. Forthcoming. “Political Cycles in a Developing Econ- of Urbanization.� World Bank: Washington, D.C. Processed. omy: Effect of Elections in the Indian States.� Journal of Develop- Jenkins, Rob, and Anne-Marie Goetz. 1999. “Accounts and Account- ment Economics. ability: Theoretical Implications of the Right-to-Information ————. 2003. “Partisan politics and Intergovernmental Transfers Movement in India.� Third World Quarterly 20(3):603–22. in India.� World Bank: Washington, D.C. Processed. ————-. 2002. “Civil Society Engagement and India’s Public Dis- King, Elizabeth, Estelle James, and A. Suriyadi. 1996. “Finance, Man- tribution System: Lessons from the Rationing Kruti Samiti in agement and Costs of Public and Private Schools in Indonesia.� Mumbai.� Paper presented at the the WDR Conference on Making Economics of Education Review 15(4):387–98. Services Work for Poor People Conference. Eynsham Hall, Oxford- shire. November. King, Elizabeth, and Berk Ozler. 2002. “What’s Decentralization Got To Do with Learning? Endogenous School Quality and Student Jimenez, Emmanuel, and Yasuyuki Sawada. 1999. “Do Community- Performance in Nicaragua.� World Bank, Development Research Managed Schools Work? An Evaluation of El Salvador’s EDUCO Group: Washington, D.C. Available on line at http://www.world- Program.� World Bank Economic Review 13(3):415–441. bank.org/research/projects/impact.htm. Processed. ————. 2002. “Does Community Management Help Keep Kids in Kizilbash, Hamid H. 1998. “Teaching Teachers to Teach.� In Pervez Schools? Evidence using Panel Data from El Salvador’s EDUCO Hoodbhoy, eds., Education and the State: Fifty Years of Pakistan. Program.� World Bank: Washington, D.C. Processed. Karachi: Oxford University Press. Jones, Stephen. 1997.“Sector Investment Programs in Africa: Issues and Kleemeier, Elizabeth. 2000. “The Impact of Participation on Sustain- Experience.� World Bank Technical Paper 374. Washington, D.C. ability: An Analysis of the Malawi Rural Piped Scheme Program.� Jones, Stephen, and Andrew Lawson. 2001. “Moving from Projects to World Development 28(5):929–44. Programmatic Aid.� World Bank Operation Evaluation Depart- Klein, Michael, and Neil Roger. 1994. “Back to the Future: The Poten- ment, OED Working Paper Series 5. Washington, D.C. tial in Infrastructure Privatisation.� In Richard O’Brien, eds., Jongudomsuk, Pongpisut, Jadej Thammatuch-aree, and Prae Chitti- Finance and the International Economy. Oxford, U.K.: Oxford Uni- nanda. 2002. “Pro-Poor Health Financing Schemes in Thailand: A versity Press. (c) The International Bank for Reconstruction and Development / The World Bank 238 WORLD DEVELOPMENT REPORT 2004 Knack, Stephen. 2001. “Does Foreign Aid Promote Democracy?� Uni- Leonard, Kenneth L. 2002. “When Both States and Markets Fail: versity of Maryland, IRIS Center Working Paper 238. College Park, Asymmetric Information and the Role of NGOs in African Health Md. Care.� International Review of Law and Economics 22(1):61–80. Knack, Stephen, and Aminur Rahman. 2003. “Aid Intensity, Donor Leonard, Kenneth L., Gilbert R. Mliga, and Damen Haile Mariam. Fragmentation and the Quality of Governance.� Background paper 2002. “Bypassing Health Centers in Tanzania: Revealed Preferences for the WDR 2004. for Observable and Unobservable Quality.� Journal of African Knippenberg, Rudolf, Eusebe Alihonou, Agnes Soucat, Kayode Oyeg- Economies 11(4):441–71. bite, Maria Calivis, Ian Hopwood, Reiko Niimi, Mamadou Pathe Levy, Santiago, and Evelyne Rodríguez. 2002. “El Programa de Edu- Diallo, Mamadou Conde, and Samuel Ofosu-Amaah. 1997. cación, Salud y Alimentación de México, Progresa.� Instituto Mexi- “Implementation of the Bamako Initiative: strategies in Benin and cano del Seguro Social y Secretaría de Hacienda y Crédito Público: Guinea.� International Journal of Health Planning and Management Ciudad de Mexico, Mexico. Processed. 12 Suppl 1:S29–S47. Levy-Bruhl, Daniel, Agnes Soucat, Raimi Osseni, Jean-Michel Ndiaye, Koenen-Grant, Julie, and Harry Garnett. 1996. “Improving Policy Boubacar Dieng, Bethune Xavier De, Alpha Telli Diallo, Mamadou Formulation in Zambia.� Implementing Policy Change Project Conde, Mohamed Cisse, Yarou Moussa, Kandjoura Drame, and Case Study 2. Washington, D.C. Rudolf Knippenberg. 1997. “The Bamako Initiative in Benin and Koenig, Michael A., Gillian H. C. Foo, and Ketan Joshi. 2000. “Quality Guinea: Improving the Effectiveness of Primary Health Care.� of Care within the Indian Family Welfare Programme: A Review of International Journal of Health Planning and Management 12 Suppl Recent Evidence.� Studies in Family Planning 31(1):1–18. 1:S49–S79. Kremer, Michael. 2002. “Innovations and Donor Assistance.� Univer- Lewis, Bernard. 1961. The Emergence of Modern Turkey. Oxford, U.K.: sity of Maryland, IRIS: College Park, Md. Processed. Oxford University Press. Kremer, Michael, and Andrei Sarychev. 2000. “Why Do Governments Lewis, Maureen. 2000. Who is Paying for Health Care in Eastern Europe Operate Schools?� Harvard University: Cambridge, Mass. Available and Central Asia? Washington, D.C.: World Bank, Europe and Cen- on line at http://post.economics.harvard.edu/faculty/kremer/papers. tral Asia Region, Human Development Unit. html. Processed. Lewis, Maureen, Gunnar S. Eskeland, and Ximena Traa-Valerezo. 1999. Kunfaa, Ernest Y., and Tony Dogbe. 2002. “Empty Pockets.� In Deepa “Challenging El Salvador’s Rural Health Care Strategy.� World Bank Narayan and Patti Petesch, eds., Voices of the Poor: From Many Policy Research Working Paper 2164. Washington D.C. Lands. New York: World Bank and Oxford University Press. Lewis, Maureen A., Gerald M. La Forgia, and Margaret B. Sulvetta. Kusek, Jody, Ray Rist, and Elyzabeth White. 2003. “How Will We 1996. “Measuring Public Hospital Costs: Empirical Evidence from Know the Millennium Development Goal Results When We See the Dominican Republic.� Social Science and Medicine 43(2): Them?� World Bank: Washington, D.C. Processed. 221–34. Kutzin, Joseph. 2003. “Health Expenditures, Reforms and Policy Pri- Liang, Xiaoyan. 1999. “Teacher Pay in 12 Latin American Countries: orities for the Kyrgyz Republic.� MANAS Health Policy, Policy How Does Teacher Pay Compare to other Professions, What Deter- Research Paper 24. Bishkek,Kyrgyztan. mines Teacher Pay, and Who are the Teachers?� World Bank, Ladd, Helen F. 2002. “School Vouchers: A Critical View.� Journal of Human Development Department, LCSHD Paper Series 49. Wash- Economic Perspectives 16(4):3–24. ington, D.C. Lakshman, M, and Mark Nichter. 2001. “Contamination of Medicine Lieberman, Samuel. 2003. “The Role of Government in Influencing Injection Paraphernalia Used by Registered Medical Practitioners Child Health, Saving Lives.� Background paper for “Working in South India: An Ethnographic Study.� Social Science and Medi- Together to Accelerate Progress Towards the Health and Nutrition cine 51:11–28. Millennium Development Goals�. World Bank: Washington, D.C. Processed. Lamboray, Jean Louis. 2000. “HIV and Health Reform in Phayao: From Crisis to Opportunity.� UNAIDS: Geneva. Available on line at Liese, B., and and others. 2003. “Human Resources Crisis in Health http://www.unaids.org/publications/documents/care/general/ Services in Sub-Saharan Africa.� Background paper for the WDR una0004e.pdf. 2004. Langsten, Ray, and Kenneth Hill. 1995. “Treatment of Childhood Lindelow, M., P. Ward, and N. Zorzi. 2003.“Expenditure Tracking and Diarrhea in Rural Egypt.� Social Science and Medicine 40(7): Service Delivery Survey: The Health Sector in Mozambique. Report 989–1001. on Preliminary Findings.� World Bank: Washington, D.C. Processed. Lanjouw, Peter, and Martin Ravallion. 1999. “Bene�t Incidence, Pub- lic Spending Reforms and the Timing of Program Capture.� The Lisulo, Angela. 2003. “Costa Rica: Health Policies.� Background paper World Bank Economic Review 13(2):257–73. for the WDR 2004. Lazear, Edward P. 2000. “The Future of Personnel Economics.� Eco- Litvack, J. I., and C. Bodart. 1993. “User Fees Plus Quality Equals nomic Journal 110(467):F611–39. Improved Access to Health Care: Results of a Field Experiment in Cameroon.� Social Science and Medicine 37(3):369–83. Le Houerou, Philippe, and Robert Taliercio. 2002. “Medium Term Expenditure Frameworks: From Concept to Practice. Preliminary Litvack, Jennie, Junaid Ahmad, and Richard Bird. 1998. “Rethinking Lessons from Africa.� World Bank, Africa Region Working Paper Decentralization in Developing Countries.� World Bank Sector 28. Washington, D.C. Studies Series September 1998. Washington, D.C. (c) The International Bank for Reconstruction and Development / The World Bank References 239 Lloyd, Cynthia B., Sahar El Tawila, Wesley H. Clark, and Barbara S. Marchant, T. J., Armstrong Schellenberg J.R.M., T. Edgar, R. Nathan, S. Mensch. 2001. “Determinants of Educational Attainment among Abdulla, O. Musaka, H. Mponda, and C. Lengeler. 2002. “Socially Adolescents in Egypt: Does School Quality Make a Difference?� Marketed Insecticide-Treated Nets Improve Malaria and Anaemia Population Council Policy Research Division Working Papers 150. in Pregnancy in Southern Tanzania.� Tropical Medicine and Inter- New York. national Health 7(2):149–58. Lobo, Albert, and Suresh Balakrishnan. 2002.“Report Card on Service Marek, T., I. Diallo, B. Ndiaye, and J. Rakotosalama. 1999. “Success- of Bhoomi Kiosks.� Public Affairs Centre: Bangalore, India. ful Contracting of Preventions Services: Fighting Malnutrition Processed. in Senegal and Madagascar.� Health Policy and Planning 14(4): Lorrain, Dominique. 1992. “The French Model of Urban Services.� 382–89. West European Politics 15(2):77–92. Martens, Bertin, Uwe Mummert, Peter Murrell, and Paul Seabright. Lott, John R. Jr. 1999. “Public Schooling, Indoctrination, and Totali- 2002. The Institutional Economics of Foreign Aid. Cambridge, U.K.: tarianism.� Journal of Political Economy 107(6):S127–S157. Cambridge University Press. Lundberg, Paul. 2002. “A Comparison of Decentralization in Pakistan Masud, Mohammad O. 2002. “Co-producing Citizen Security: The and Nepal.� Paper presented at the the 2nd International Conference Citizen-Police Liaison Committee in Karachi.� Institute of Devel- on Decentralization Conference. Manila. June 25. opment Studies Working Paper 172. Brighton, U.K. Mackey, Jo Ann. 2002. “Using a Health Belief Model in Teaching Pre- Mauro, Paulo. 1998. “Corruption and the Composition of Govern- ventive Health Care Principles to Israeli RN’s.� Paper presented at ment Expenditures.� Journal of Public Economics 69:263–79. the CITA Conference Conference. University of Massachusetts. Mayer, Elizabeth Ann. 1997. “Only Very Few Girls Even Start to Go to November. School in Yemen.� Women’s International Network News 23(3): Mackinnon, John. 2003. “How Does Aid Affect the Quality of Public 52–52. Expenditure? What We Know and What We Do not Know.� Back- McPake, B. 1996. “Public Autonomous Hospitals in Sub-Saharan ground paper for the WDR 2004. Africa: Trends and Issues.� Health Policy 35(2):155–77. Macy, J., and R. Quick. 2002. “The Safe Water System - A Household- McPake, Barbara, Delius Asiimwe, Francis Mwesigye, Matthius Based Water Quality Intervention Program for the Developing Ofumbi, Pieter Streefland, and Asaph Turinde. 2000. “Coping World.� Water Conditioning and Puri�cation Magazine 44(4):-–-. Strategies of Health Workers in Uganda.� In Pauolo Ferrinho and Madaus, G. F., and V. Greaney. 1985.“The Irish Experience in Compe- Wim Van Lerberghe, eds., Providing Health Care Under Adverse tency Testing: Implications for American Education.� American Conditions: Health Personnel Performance and Individual Coping Journal of Education 93(2):268–94. Strategies (Studies in Health Services and Organization, 16). Maiga, Z., Traore Nafo F., and A. El Abassi. 1999. “La Reforme du Antwerp, Belgium: ITG Press. Secteur Sante au Mali.� Studies in Health Services Organisation & Medina, Luis Fernando, and Susan Stokes. 2002. “Clientelism as Polit- Policy 12:132–132. ical Monopoly.� Paper presented at the the 2002 Annual Meetings of Makinen, M., H. Waters, M. Rauch, N. Almagambetova, R. Bitran, L. the American Political Science Association Conference. Boston. Gilson, D. McIntyre, S. Pannarunothai, A. L. Prieto, G. Ubilla, and August 29. S. Ram. 2000. “Inequalities in Health Care Use and Expenditures: Mehrotra, S., and S. W. Jarrett. 2002. “Improving Basic Health Service Empirical Data from Eight Developing Countries and Countries in Delivery in Low-Income Countries: ‘Voice’ to the Poor.� Social Sci- Transition.� Bulletin of the World Health Organization 78(1):55–65. ence and Medicine 54(11):1685–90. Mamdani, Mahmood. 1996. Citizen and Subject: Contemporary Africa Mehryar, A. H., A. Aghajanian, and S. Ahmadnia. 2003. “Primary and the Legacy of Late Colonialism. Princeton, N.J.: Princeton Uni- Health Care and Rural Poverty Eradication in the Islamic Republic versity Press. of Iran.� Background paper for the WDR 2004. Manning, N. 1998. “Unbundling the State Autonomous agencies and Mencher, Joan. 1980. “The Lessons and Non-lessons of Kerala: Agri- Service Delivery.� World Bank: Washington, D.C. Processed. cultural Labourers and Poverty.� Economic and Political Weekly Manning, Nick. 2001. “The legacy of the New Public Management in 15(41-43):1784–802. Developing Countries.� International Review of Administrative Sci- Mesoamerica Nutrition Program Targeting Study Group. 2002. “Tar- ences 67:297–312. geting Performance of Three Large Scale, Nutrition Oriented Social Manning, Nick, and Neil Parison. 2003. “Building on Strengths: Programs in Central America and Mexico.� Food Nutrition Bulletin Lessons from Comparative Public Administration Reforms.� 23(2):162–74. Higher School of Economics and World Bank: Moscow. Processed. Meyer, C. A. 1995. “Opportunism and NGOs Entrepreneurship and Manor, James. 2002. “Partnerships between Governments and Civil Green North-South Transfers.� World Development 23(8):1277–89. Society for Service Delivery in Less Developed Countries: Cause for Miguel, Edward, and Michael Kremer. 2001. “Worms: Education and Concern.� Paper presented at the the WDR Conference on Making Health Externalities in Kenya.� National Bureau of Economic Services Work for Poor People Conference. Eynsham Hall, Oxford- Research Working Paper Series 8481. Cambridge, Mass. shire. November. Miguel, Szekely. 1998. The Economics of Poverty, Inequality, and Wealth Mansuri, Ghazala, and Vijayendra Rao. 2003. “Evaluating Commu- Accumulation in Mexico. New York: St. Martin’s Press. nity Driven Development: A Critical Review of the Research.� Millot, Benoit, and Julia Lane. 2002. “The Ef�cient Use of Time in World Bank: Washington, D.C. Processed. Education.� Education Economics 10(2):209–28. (c) The International Bank for Reconstruction and Development / The World Bank 240 WORLD DEVELOPMENT REPORT 2004 Mills, A., and J. Bromberg. 1998. “Experiences of Contracting: An Musgrave, Richard. 1959. The Theory of Public Finance: A Study in Overview of the Literature.� World Health Organization 33. Public Economy. New York: McGraw-Hill. Geneva. Nag, Moni. 1989. “Political Awareness as a Factor in Accessibility of Mills, A., J. Broomberg, and C. Hongoro. 1997. “Improving the Ef�- Health Services: A Case Study of Rural Kerala and West Bengal.� ciency of District Hospitals: Is Contracting an Option?� Tropical Economic and Political Weekly 24(8):417–26. Medicine and International Health 2:116–26. Nahar, S., and A. Costello. 1998. “The Hidden Cost of ‘Free’ Maternity Mills, A., R. Brugha, K. Hanson, and B. McPake. 2002.“Approaches for Care in Dhaka, Bangladesh.� Health Policy and Planning Improving Delivery in the Non-State Sector: What Is the Evidence 13(4):417–22. on What Works, Where and Why?� Paper presented at the Making Naimoli, Joseph, and Denise Vaillancourt. 2003. “Performance-Based Services Work for the Poor Conference. Eynsham Hall, Oxfordshire. Management in an Evolving Decentralized Public Health System in November 4. West Africa: The Case of Burkina Faso.� Background paper for the Mingat, Alain. 2003. “Management of Education Systems in Sub- WDR 2004. Saharan African Countries: A Diagnostic and Ways towards Narayan, Deepa. eds. 2002. Empowerment and Poverty Reduction: A Improvement in the Context of the EFA-FTI.� World Bank: Wash- Sourcebook. Washington, D.C.: World Bank. ington, D.C. Processed. Narayan, Deepa, Robert Chambers, Meera Kaul Shah, and Patti Ministère de la Santé Bénin. 2003. “Revue des Dépenses Publiques Petesch. 2000a. Crying out for Change: Voices of the Poor. New York: dans le Secteur de la Santé.� Ministère de la Santé Bénin: Porto- Oxford University Press for the World Bank. Novo. Narayan, Deepa, Raj Patel, Kai Schafft, Anne Rademacher, and Sarah Ministry of Education Ethiopia. 2000. “Alternate Routes to Basic Pri- Koch-Schulte. 2000b. Voices of the Poor: Can Anyone Hear Us? New mary Education.� Ministry of Education: Addis Ababa. York: Oxford University Press. Ministry of Health Guinea. 2002. “Guinea, A Country Status Report Narayan, Deepa, and Patti Pettesch. eds. 2002. From Many Lands: on Health and Poverty, Health, Nutrition, and Population Inputs Voices of the Poor. Washington, D.C.: World Bank and Oxford Uni- for the PRSP and HIPC Process.� Ministry of Health Guinea: versity Press. Conakry. National Democratic Institute for International Affairs, and World Mintz, Patricia, William D. Savedoff, and Jorge Pancorvo. 2000. Bank. 1998. “Promoviendo Transparencia y Gobernabilidad: Expe- “Bolivia: PROSALUD and the Insurance Program for Mothers and riencias en la República Argentina, Colombia y los Estados Children (SNMN).� World Bank: Washington, D.C. Unidos.� National Democratic Institute for International Affairs Mizala, Alejandra, and Pilar Romaguera. 2000. “School Performance and World Bank: Buenos Aires. and Choice.� Journal of Human Resources 35(2):392–417. Nauman, Huma. 1990. “Primary Teacher Certi�cate Program in ————. 2002. “Evaluacion del Desempeño e Incentivos en la Edu- NWFP: A Study.� Project BRIDGES, HIID: Cambridge, Mass. cación Chilena.� Cuadernos de Economía 39(118):353–94. Navarro, Juan Carlos, and Rafael de la Cruz. 1998. “Federal State and Moens, F. 1990.“Design, Implementation and Evaluation of Commu- Nonpro�t Schools in Venezuela.� In William D. Savedoff, eds., Orga- nity Financing Schemes for Hospital Care in Developing Coun- nization Matters: Agency Problems in Health and Education in Latin tries: A Prepaid Health Plan in Bwamanda Health Zone, Zire.� America. Washington, D.C.: Inter American Development Bank. Social Science and Medicine 30(12):1319–27. Nelson, Joan. 2000. The Politics of Social Sector Reforms. Washington, Molyneaux, Jack, and Paul Gertler. 1999. “Evaluating the Impact of D.C.: Overseas Development Council. IDT.� The RAND Corporation: Santa Monica, C.A. Processed. Newman, John, Menno Pradhan, Laura B. Rawlings, Geert Ridder, Moore, Mark H., David Thacher, Andrea Dodge, and Tobias Moore. Ramiro Coa, and Jose Luis Evia. 2002. “An Impact Evaluation of 2002.“Recognizing Public Value in Policing: The Challenge of Mea- Education, Health and Water Supply Investments by the Bolivian suring Police Performance.� Policy Executive Research Forum: Social Investment Fund.� World Bank Economic Review Washington, D.C. 16(2):241–74. Moore, Mick, and James Putzel. 2001. “Thinking Strategically about Nichols, Len M., Nicholas Prescott, and Kai Hong Phua. 1997. “Med- Politics and Poverty.� Institute of Development Studies Working ical Savings Accounts for Developing Countries.� In George J. Paper 101. Brighton, U.K. Schieber, eds., Innovations in Health Care Financing: Proceedings of Mtemeli, D. N. 1994. “An Investigation into Why Women in Mutasa a World Bank Conference. Washington, D.C.: World Bank. District Prefer to Deliver at Home and Not in Health Institu- Nickson, Andrew, and Claudia Vargas. 2002. “The Limitations of tions.� Provincial Medical Direction: Manicaland, Zimbabwe. Water Regulation: The Failure of the Cochabamba Concession in Processed. Bolivia.� Bulletin of Latin American Research 21(1):99–120. Murnane, Richard J., and David K. Cohen. 1986. “Merit Pay and the Nieves, I., Gerard M. La Forgia, and Jaume Ribera. 2000. “Guatemala: Evaluation Problem: Why Most Merit Pay Plans Fail and a Few Sur- Large-Scale Government Contracting of NGO’s to Extend Basic vive.� Harvard Educational Review 56(1):1–17. Health Services to Poor Populations in Guatemala.� World Bank. Murphy, Paud, Carla Bertoncino, and Lianqin Wang. 2002.“Achieving Washington DC. Available on line at www.worldbank.org. Universal Primary Education in Uganda: The ‘Big-Bang’ Nitayaramphong, S., S. Srivanichakom, and Y. Pongsupap. 2000. Approach.� World Bank, Human Development Network, Educa- “Strategies to Respond to Health Manpower Needs in Rural Thai- tion Notes: Washington, D.C. land.� Studies in Health Services Organisation & Policy 16:55–72. (c) The International Bank for Reconstruction and Development / The World Bank References 241 NRI and World Bank. 2003.“Public Expenditure and Service Delivery Improve Services for the Poor.� Background paper for the WDR in Papua New Guinea: Draft.� World Bank: Washington, D.C. 2004. O’Connell, Stephen A., and Charles C. Saludo. 2001. “Aid Intensity in Perry, H., N. Robinson, D. Chavez, O. Taja, C. Hilari, D. Shanklin, and Africa.� World Development 29(9):1527–52. J. Wyon. 1999. “Attaining Health for All through Community Part- O’Connor, Thomas H. 1995. The Boston Irish: A Political History. nerships: Principles of the Census-Based, Impact-Oriented (CBIO) Boston: Northeastern University Press. Approach to Primary Health Care Developed in Bolivia, South OECD. 2001. “Knowledge and Skills for Life: First Results from PISA America.� Social Science and Medicine 48(8):1053–67. 2000.� OECD: Paris. Persson, Torsten, and Guido Tabellini. 2000. Political Economics: Olson, Mancur. 1971. The Logic of Collective Action: Public Goods and Explaining Public Policy. Cambridge, Mass.: MIT Press. the Theory of Groups. Cambridge, Mass.: Harvard University Press. Peters, David H., Abdo S. Yazbeck, Adam Wagstaff, G. N. V. Ramana, Onyango-Ouma, Washington, Rose Laisser, Musiba Mbilima, Mar- Lant Pritchett, and Rashmi R. Sharma. 2003. Better Health Systems garet Araoye, Patricia Pittman, Irene Agyepong, Mairo Zakari, for India’s Poor: Findings, Analysis and Options. Washington, D.C.: Sharon Fonn, Marcel Tanner, and Carol Vlassof. 2001. “An Evalua- World Bank. tion of Health Workers for Change in Seven Settings: A Useful Picazo, Oscar F. 2002. “Better Health Outcomes from Limited Management and Health System Development Tool.� Health Policy Resources: Focusing on Priority Services in Malawi.� World Bank, and Planning 16(Sup. 1):24–32. Africa Region Human Development Working Paper Series. Wash- Organization for Economic Cooperation and Development. 2001. ington, D.C. “Knowledge and Skills for Life: First Results from PISA 2000.� Platteau, Jean-Philippe. 2003. “Rushing to Help the Poor Through Organisation for Economic Cooperation and Development: Paris. Participation May Be Self-Defeating.� Paper presented at the the Ostrom, Elinor. 1990. Governing the Commons. New York: Cambridge WDR Conference on Making Services Work for Poor People Confer- University Press. ence. Berlin, Germany. July. Ostrom, Elinor, Clark Gibson, Sujai Shivakumar, and Krister Ander- Platteau, Jean-Philippe, and Frederic Gaspart. 2003. “The ‘Elite Cap- son. 2001. “Aid Incentives and Sustainability: An Institutional ture’ Problem in Participatory Development.� Centre for Research Analysis of Development Cooperation.� SIDA Studies in Evalua- on the Economics of Development, University of Namur: Namur, tion 02/01. Stockholm, Sweden. Belgium. Processed. Ottaway, M, and T. Chung. 1999. “Debating Democracy Assistance: Plummer, Janelle. 2003. “Independent Providers in Water and Sanita- Toward a New Paradigm.� Journal of Democracy 10(4):99–128. tion.�World Bank, draft note prepared for the Water and Sanitation Program: Washington, D.C. Processed. Pakistan Institute for Environment Development Action, and Project Management Team. 1994. “The State of the Public Sector Primary Population Services International. 2003. “Pro�les.� Population Ser- Health Care Services, District Sheikhupura, Punjab, Pakistan.� vices International: Washington D.C. Available on line at Bamako Initiative technical report series, UNICEF 31. http://www.psi.org/resources/pro�les.html. Pannarunothai, S., and A. Mills. 1997. “The Poor Pay More: Health- Porignon, D., E. M. Soron’Gane, T. E. Lokombe, D. K. Isu, P. Hennart, Related Inequality in Thailand.� Social Science and Medicine and W. Van Lerberghe. 1998. “How Robust Are District Health Sys- 44(12):1781–90. tems? Coping with Crisis and Disasters in Rutshuru, Democratic Republic of Congo.� Tropical Medicine and International Health Pannarunothai, S., S. Srithamrongsawat, M. Kongpan, and P. Thum- 3(7):559–65. vanna. 2000. “Financing Reforms for the Thai Health Card Scheme.� Health Policy and Planning 15(3):303–11. Preker, A. S., G. Carrin, D. Dror, M. Jakab, W. Hsiao, and D. Arhin. 2001. “Role of Communities in Resource Mobilization and Risk Papua New Guinea Of�ce of National Planning. 1999. “Papua New Sharing: A Synthesis Report.� World Bank, HNP Discussion Paper. Guinea Human Development Report 1998.� Government of Papua Washington DC. New Guinea and United Nations Development Program: Port Moresby, Papua New Guinea. Prendergast, Carnice. 2001. “Selection and Oversight in the Public Sector, with the Los Angeles Police Department as an Example.� Parker, Ronald, and Tauno Skytta. 2000. “Rural Water Projects: National Bureau of Economic Research Working Paper 8664. Cam- Lessons from OED Evaluations.� World Bank Operations Evalua- bridge, Mass. tion Department, Working Paper Series 3. Washington, D.C. Price, N. 2001.“The Performance of Social Marketing in Reaching the Pathmanathan, Indra, Jerker Liljestrand, Jo M. Martins, Lalini Poor and Vulnerable in AIDS Control Programmes.� Health Policy Rajapaksa, Craig Lissner, Amala de Silva, Swarna Selvaraju, and and Planning 16(3):231–39. Prabha Joginder Singh. 2003. Investing Effectively in Maternal Health in Malaysia and Sri Lanka. Washington D.C.: World Bank. Pritchett, Lant. Forthcoming.“It Pays to Be Ignorant.� Journal of Policy Reform. Paul, Samuel. 2002. Holding the State to Account: Citizen Monitoring in Action. Bangalore: Books for Change. Pritchett, Lant, and Duriya Farooqui. 2003. “The Economics of Edu- cation Reeducated.� Kennedy School of Government and Center Paxson, Christina, and Norbert A. Schady. 2002. “The Allocation and for Global Development: Cambridge, Mass. Processed. Impact of Social Funds: Spending on School Infrastructure in Peru.� World Bank Economic Review 16(2):297–319. Pritchett, Lant, and Deon Filmer. 1999. “What Education Production Functions Really Show: A Positive Theory of Education Expendi- Pearson, Mark, and Paolo Carlo Belli. 2003. “The Impact of ture.� Economics of Education Review 18(2):223–39. Resource Allocation Mechanisms, and Priority Setting Policies to (c) The International Bank for Reconstruction and Development / The World Bank 242 WORLD DEVELOPMENT REPORT 2004 Pritchett, Lant, and Michael Woolcock. 2002. “Solutions When the Two—Equity and Excellence in Education for Development. Lon- Solution is the Problem: Arraying the Disarray in Development.� don: Cassel. Center for Global Development Working Paper 10. Washington, Reinikka, Ritva, and Jakob Svensson. 2001. “Explaining Leakage in D.C. Public Funds.� World Bank, Policy Research Woking Paper 2709. Pritchett, Lant H. 2002. “When Will They Ever Learn? Why All Gov- Washington, D.C. ernments Produce Schooling.� John F. Kennedy School of Govern- ————. 2003a. “The Power of Information: Evidence from an ment, Harvard University: Cambridge, Mass. Processed. Information Campaign to Reduce Capture.�World Bank: Washing- Pritchett, Lant H., and Lawrence H. Summers. 1996. “Wealthier is ton, D.C. Processed. Healthier.� The Journal of Human Resources 31(4):841–68. ————. 2003b. “Working for God? Evaluating Service Delivery Probe Qualitative Research Team. 2002.“Healthcare Seeking Behavior of Religious Not-for-Pro�t Health Care Providers in Uganda.� in Andhra Pradesh.� Probe Qualitative Research: Bombay. Policy Research Working Paper Series 3058. Washington, D.C. PROBE Team in association with Centre for Development Econom- Riddel, Roger C. 1999. “The End of Foreign Aid to Africa? Concerns ics. 1999. Public Report on Basic Education in India. New Delhi: about Donor Policies.� African Affairs 98:309–35. Oxford University Press. Ritzen, Jo, Lianqin Wang, and Yael Duthilleul. 2002. “Education as an Prud’homme, Remy. 1995. “The Dangers of Decentralization.� World Agent of Social Cohesion.�World Bank, Human Development Net- Bank Research Observer 10(2):201–20. work: Washington, D.C. Processed. Psacharopoulos, Georges, Jorge Valenzuela, and Mary Arends. 1996. Roberts, John. 2002. “Results-Oriented Public Expenditure Manage- “Teacher Salaries in Latin America: A Review.� Economics of Educa- ment:Lessons from Country Case Studies.� Overseas Development tion Review 15(4):401–6. Institute: London, U.K. Processed. Public Services International, and Education International. 2000. Robinson, James, and Thierry Verdier. 2002. “The Political Economy “Improving the Effectiveness of the Public Sector—If Not Us, of Clientelism.� Centre for Economic Policy Research Discussion Who?� Public Services International and Education International: Paper Series 3205. London. Paris and Brussels. Available on line at http://www.world-psi.org. Rodden, Jonathan, Gunnar Eskeland, and Jennie Litvack. eds. 2003. Putnam, Robert, Robert Leonardi, and Raffaella Nanetti. 1992. Mak- Decentralization and Hard Budget Constraints. Cambridge, ing Democracy Work: Civic Traditions in Modern Italy. Princeton Mass.: MIT Press. N.J.: Princeton University Press. Rojanapithayakorn, W., and R. Hanenberg. 1996. “The 100% Con- Radoki, Carole. 2003. “What Are the Most Effective Strategies for dom Program in Thailand.� AIDS 10(1):1–7. Understanding and Channeling the Preferences of Service Users to Rosskam, Ellen. 2003.“No Pills, No Bandages, Nothing.� International Make Public Services More Responsive.� Background paper for the Labor Organization: Geneva. Processed. WDR 2004. Rowe, Alexander K., Faustin Onikpo, Marcel Lama, Francois. Cokou, Raghupati, Usha, and Vivien Foster. 2003. “A Scorecard for India, Tar- and Michael S. Deming. 2001. “Management of Childhood Illness iffs and Subsidies in South Asia.� World Bank, PPIAF and Water at Health Facilities in Benin: Problems and Their Causes.� American and Sanitation Program Paper 2. Washington, D.C. Journal of Public Health 91(10):1625–35. Rajkumar, Andrew Sunil, and Vinaya Swaroop. 2002. “Public Spend- Saadah, F., M. Pradhan, and R. Sparrow. 2001. “The Effectiveness of the ing and Outcomes: Does Governance Matter?� World Bank Policy Health Card as an Instrument to Ensure Access to Medical Care for Research Working Paper 2840. Washington, D.C. Available on line the Poor during the Crisis.�World Bank: Washington, D.C. Processed. at http://econ.worldbank.org/view.php?type=5&id=15176. Samrasinghe, Daya, and John Akin. 1994. “Health Strategy and Ramachandran, V. K. 1996. “On Kerala’s Development Achieve- Financing Study: Final Report.� Sri Lanka, Ministry of Health and ments.� In Jean Dreze and Amartya Sen, eds., Indian Development: Women’s Affairs: Columbo. Selected Regional Perspectives. New York: Oxford University Press. Save the Children. 2002. “New Products into Old Systems: The Ini- Rao, Vijayendra. 2003. “Experiments in Participatory Econometrics: tial Impact of the Global Alliance for Vaccines and Immuniza- Improving the Connection Between Economic Analysis and the tions (GAVI) at Country Level.� Save the Children and London Real World.� Economic and Political Weekly, May 18. School of Hygiene and Tropical Medicine: London. Available on Rao, Vijayendra, and Ana Maria Ibáñez. 2003. “The Social Impact of line at http://www.savethechildren.org.uk/development/latest/ Social Funds in Jamaica: A Mixed-Methods Analysis of Participa- Gavi_report_cover.pdf. tion, Targeting and Collective Action in Community Driven Devel- Savedoff, William, and Pablo Spiller. 1999. Spilled Water: Institutional opment.� World Bank Policy Research Working Paper 2970. Wash- Commitment in the Provision of Water Services. Washington, D.C.: ington, D.C. Inter-American Development Bank. Rao, Vijayendra, and Michael Walton. Forthcoming. “Culture and Savedoff, William D., and T. Paul Schultz. eds. 2000. Wealth from Public Action: An Introduction.� In Vijayendra Rao and Michael Health. Washington, D.C.: Inter-American Development Bank. Walton (eds.) Culture and Public Action. Stanford University Press. Sawada, Yasuyuki. 1999. “Community Participation, Teacher Effort, Reimers, Fernando. 1997. “The Role of the Community in Expand- and Educational Outcome: The Case of El Salvador’s EDUCO Pro- ing Educational Opportunities: The EDUCO Schools in El Sal- gram.� William Davidson Institute Working Papers 307. University vador.� In James Lynch, Celia Modgil, and Sohan Modgil, eds., of Michigan. Available on line at http://eres.bus.umich.edu/ Education and Development: Tradition and Innovation: Volume web/dwpauth.html. (c) The International Bank for Reconstruction and Development / The World Bank References 243 Schacter, Mark, Phillip Haid, and Julie Koenen-Grant. 1999. Cabinet Shah, Anwar. 2003a. “On Getting the Giant to Kneel: Approaches to a Decision-Making: Lessons from Canada; Lessons from Africa. Change in the Bureaucratic Culture: A Primer on Results-oriented Ottawa: Institute on Governance. Management and Evaluation.� In Anwar Shah, eds., Ensuring Schedler, Andreas. 1999.“Conceptualizing Accountability.� In Andreas Accountability When There is No Bottom Line. Washington, D.C.: Schedler, Larry Diamond, and Marc F. Plattner, eds., The Self- World Bank. Restraining State: Power and Accountability in New Democracies. Shah, Shekhar. 2003b. “Walk Before you Run: Governance and Public Boulder and London: Lynne Rienner. Sector Reforms in Poverty Reduction Strategies.� Background Schellenberg, J. R., S. Abdulla, R. Nathan, O. Musaka, T. J. Marchant, paper for the WDR 2004. N. Kikumbih, A. K. Mushi, H. Mponda, H. Minja, H. Mshinda, M. Shah, Shekhar, and Manju Rani. 2003. “Worlds Apart: Why are Kerala Tanner, and C. Engeler. 2001. “Effect of Large-Scale Social Market- and Uttar Pradesh so Different in Human Development?� Back- ing of Insecticide-treated Nets on Child Survival in Rural Tanza- ground paper for the WDR 2004. nia.� The Lancet 357(9264):1241–7. Shi, Min, and Jakob Svensson. 2003.“Political Budget Cycles in Devel- Schick, Allen. 1996. The Spirit of Reform: Managing the New Zealand oped and Developing Countries.� World Bank: Washington, D.C. State Sector in a Time of Change. Wellington: State Services Com- Processed. mission and the Treasury. Shirley, Mary. eds. 2002. Thirsting for Ef�ciency: The Economics and ————. 1998. “Why Most Developing Countries Should Not Try Politics of Urban Water System Reform. Oxford, U.K.: Elsevier Press. New Zealand’s Reforms.� World Bank Research Observer 13(1): Shleifer, Andrei, and Rober W. Vishny. 1993. “Corruption.� Quarterly 123–31. Journal of Economics 108(3):599–617. Schieber, G, and A. Maeda. 1997. “A Curmudgeon’s Guide to Financ- Shreenivasan, Jain. 2002. “Last Word: Development Scam.� The Week, ing Health Care in Developing Countries.� In George J. Schieber, February 10. eds., Innovations in Health Care Financing: Proceedings of a World Sillers, Son. 2002. “Strengthening Basic Education through Institu- Bank Conference. Washington, D.C.: World Bank. tional Reform: Linking Authority, Accountability, and Trans- Schiotz, Ania. 2002. “Hygiene in Nineteenth-Century Norway.� Insti- parency.� USAID Issues Paper 1. Washington, D.C. tut for Allmenn- og Samfunnsmedisin, University of Oslo: Oslo, Singerman, Diane. 1995. Avenues of Participation: Family, Politics and Norway. Processed. Networks in Urban Cairo. Princeton, N.J.: Princeton University Schleicher, Andreas, Maria Teresa Siniscalco, and Neville Postlewaite. Press. 1995. “The Conditions of Primary Schools: A Pilot Study in the Singh, Janmejay, and Parmesh Shah. 2003. “Making Services Work for Least Developed Countries.� UNESCO and UNICEF: Paris. Poor People: The Role of Participatory Public Expenditure Man- Processed. agement.� Background note for the WDR 2004. Schleifer, Andrei, and Robert W. Vishny. 1994.“Politicians and Firms.� Smith, Warrick. 1997a. “Utility Regulators: Decisionmaking Struc- Quaterly Journal of Economics 109(4):995–1025. tures, Resources and the Start-up Strategy.� World Bank, Finance, Schneider, Helen, and Natasha Palmer. 2002. “Getting to the Truth? Private Sector and Infrastructure Network Viewpoint 129. Wash- Researching User Views of Primary Health Care.� Health Policy and ington, D.C. Planning 17(1):32–41. ————. 1997b. “Utility Regulators: Roles and Responsibilities.� Schorr, Lisbeth B. 1997. Common Purpose: Strengthening Families and World Bank, Finance, Private Sector and Infrastructure Network Neighborhoods to Rebuild America. New York: Doubleday. Viewpoint 128. Washington, D.C. Schuknecht, Ludger. 1996. “Political Business Cycles and Fiscal Poli- ————. 2003. “Regulating Infrastructure for the Poor: Regulatory cies in Developing Countries.� Kyklos 49(2):155–70. System Design.� In Penelope Brook and Tymothy Irwin, eds., Infra- Schultz, T. Paul. 2001. “School Subsidies for the Poor: Evaluating the structure for Poor People—Public Policy For Private Provision. Wash- Mexican PROGRESA Program.� Yale University Economic Growth ington, D.C.: World Bank. Center Discussion Paper 834. New Haven, Conn. Available on line Soeters, Robert, and Fred Grif�ths. 2003. “Improving Government at http://www.econ.yale.edu/~egcenter/EGCdiscussion3.htm. Health Services through Contract Management: A Case from Cam- Schwartz, Klaas, and Blokland Maarten. 2002. “Government-Owned bodia.� Health Policy and Planning 18(1):74–83. Public Limited Companies in the Dutch Water Sector.� In Paul Sei- Soucat, Agnes, T. Gandaho, and D. Levy-Bruhl. 1997. “Health Seeking denstat, David Haarmeyer, and Simon Hakim, eds., Reinventing Behavior and Household Health Expenditures in Benin and Water and Wastewater Systems: Global Lessons for Improving Man- Guinea: the Equity Implications of the Bamako Initiative.� Interna- agement. New York: John Wiley and Sons, Inc. tional Journal of Health Planning and Management 12(Supp. Scott, James C. 1998. Seeing Like a State: How Certain Schemes to 1):S137–S164. Improve the Human Condition Have Failed. New Haven: Yale Uni- Soucat, Agnes, and Manju Rani. 2003a.“Making Health and Nutrition versity Press. Services Work for Poor People.� Background note for the WDR Seidenstat, Paul, David Haarmeyer, and Simon Hakim. eds. 2002. 2004. Reinventing Water and Wastewater Systems: Global Lessons for ————. 2003b. “Reaching the Poor with or without Quali�ed Pro- Improving Management. New York: John Wiley and Sons, Inc. fessionals: A Review of Obstacles to and Levers for Reaching the Sen, Amartya. 2002.“Why Half the Planet is Hungry.� Observer of Lon- Poor with Health and Nutrition Services.� Background paper for don, June 16. the WDR 2004. (c) The International Bank for Reconstruction and Development / The World Bank 244 WORLD DEVELOPMENT REPORT 2004 ————. 2003c. “Who Do the Poor Really Get Their Services From: Planning, and Economic Development: Kampala, Uganda. Myths and Reality of the Public-Private Mix.� Background paper Processed. for the WDR 2004. Tynan, Nicola. 2002. “London’s Private Water Supply, 1582-1902.� In Stasavage, David. 2003.“On the Role of Democracy in Uganda’s Move to Paul Seidenstat, David Haarmeyer, and Simon Hakim, eds., Rein- Universal Primary Education.� Background paper for the WDR 2004. venting Water and Wastewater Systems: Global Lessons for Improving Stewart, John. 1996. “A Dogma of Our Times - The Separation of Management. New York: John Wiley and Sons, Inc. Policy-Making and Implementation.� Public Money and Manage- U.K. Department of International Development. 2002. “Transport’s ment July–September: 33–40. Role in Achieving the Millenium Development Goals.� U.K. Stiglitz, Joseph E. 2000. Economics of the Public Sector. New York: W.W. Department of International Development, Transport Resource Norton. Center: London, U.K. Studdert, D. M., E. J. Thomas, H. R. Burstin, B. I. Zbar, E. J. Orav, and UNAIDS. 2000. “Report on the Global HIV/AIDS Epidemic.� T. A. Breannd. 2000. “Negligent Care and Malpractice Claiming UNAIDS: Geneva. Behavior in Utah and Colorado.� Medical Care 38(3):250–60. UNAIDS, and WHO. 2003. “Aids Epidemic Update.� Joint United Sundaram, Pachampet. 2001.“Transfers and Postings in India.� World Nations Programme on HIV/AIDS (UNAIDS) and World Health Bank: Washington, D.C. Processed. Organization (WHO): Switzerland. Suwal, J. V. 2001. “The Main Determinants of Infant Mortality in UNESCO. 2002. “EFA Global Monitoring Report, 2002: Is the World Nepal.� Social Science and Medicine 53(12):1667–81. on Track?� UNESCO: Paris. Available on line at http://www.unesco. Svensson, Jakob. 2003. “Why Conditional Aid Doesn’t Work and org/education/efa/monitoring/monitoring_2002.shtml. What Can Be Done about It.� Journal of Development Economics UNICEF. 2001. “Progress since the World Summit for Children: A 70(2):381–402. Statistical Review.� UNICEF: New York. Available on line at Sweeting, E. M. 2001. “Basic Education Projects in Indonesia since http://www.unicef.org/pubsgen/wethechildren-stats/. 1990: A Study in Sustainability.� World Bank: Washington, D.C. UNICEF. 2002. “Accelerating Early Child Survival and Development Processed. in High Under-Five Mortality Areas in the Context of Health Talero, Eduardo. 2001. “Electronic Government Procurement: Con- Reform and Poverty Reduction: A Results Based Approach, Pro- cepts and Country Experiences.� World Bank, Operational Policy posal for CIDA.� UNICEF: New York. and Services: Washington, D.C. Processed. UNICEF. 2002. “End of Decade Databases: Child Mortality�. Geneva, Tan, Jee-Peng, Agnes Soucat, and Alain Mingat. 2000. “Enhancing UNICEF. Available on line at http://www.childinfo.org/cmr/revis/ Human Development in the HIPC/PRSP Context: Progress in the db2.htm. Africa Region During 2000.� World Bank, Africa Region, Human UNICEF-World Bank 2003. Marginal Budgeting for Bottlenecks: Development Sector: Washington, D.C. How to reach the impact frontier of health and nutrition services Tangcharoensathien, V, and S Nittayaramphong. 1994.“Thailand: Pri- and accelerate progress towards the Millennium Development vate Health Care Out of Control?� Health Policy and Planning Goals. A budgeting model and an application to low-income 9(1):31–40. countries (forthcoming). Tarp, Finn, and Peter Hjertholm. eds. 2000. Foreign Aid and Develop- United Nations. 1961. “United Nations Demographic Yearbook�. The ment: Lessons Learnt and Directions for the Future. London, New United Nations. Monograph. York: Routledge. Uriarte, Miren. 2002. “Cuba Social Policy at the Crossroads:Main- Taylor, J. 2003. “Contracting for Health Services.� In A. Preker and A. taining Priorities, Transforming Practice.� Oxfam America: Harding, eds., Private Participation in Health Services. Washington Boston. D.C.: World Bank. Urquiola, Miguel. 2001. “Identifying Class Size Effects in Developing Tendler, Judith. 1998. Good Government in the Tropics. Baltimore: Countries: Evidence from Rural Schools in Bolivia.� World Bank John Hopkins University Press. Policy Research Working Paper 2711. Washington, D.C. The City of Johannesburg Council. 2001. Johannesburg: An African van de Walle, Dominique. 1998. “Targeting Revisted.� World Bank City in Change. Cape Town: Zebra Press. Research Observer 13(2, August):231–248. The Economist. 2002.“Kenya’s election: Waiting for the Newish Man.� ————. 2002. “Choosing Rural Road Investments to Help Reduce The Economist, December 19. Poverty.� World Development 30(4):575–89. Thomas, Duncan, Victor Lavy, and John Strauss. 1996. “Public Policy van de Walle, Dominique, and Dorothyjean Cratty. 2003. “Impact and Anthropometric Outcomes in Côte D’Ivoire.� Journal of Public Evaluation of a Rural Road Rehabilitation Project.� World Bank: Economics 61(3):155–192. Washington, D.C. Processed. Torres Parodi, C. 2003. “Health and Ethnicity.� Background note for van de Walle, Dominique, and Kimberly Nead. eds. 1995. Public the WDR 2004. Spending and the Poor: Theory and Evidence. Baltimore: Johns Hop- kins University Press. Tremolet, Sophie. 2002. “Rural Water Service.� World Bank Private Sector Finance Viewpoint 249. Washington, D.C. van der Berg, Servaas, and Ronelle Burger. 2003. “The Stories behind the Numbers: An Investigation of Efforts to Deliver Ser- Tumusiime-Mutebile, Emmanuel. 2003. “Uganda’s Experience with vices to South Africa’s Poor.� Background paper for the WDR the Medium-term Expenditure Framework.� Ministry of Finance, 2004. (c) The International Bank for Reconstruction and Development / The World Bank References 245 Van Domelen, Julie. 2002. “Social Funds: Evidence on Targeting, Poverty Working Group. 2002. “Child Health: Reaching the Poor.� Impacts and Sustainability.� Journal of International Development World Bank: Washington, D.C. Processed. 14(5):627–42. Wagstaff, Adam, and E. van Doorslaer. Forthcoming. “Catastrophe Van Lerberghe, W., W. Ammar, R. el Rashidi, M. Awar, A. Sales, and A. and Impoverishment in Paying for Health Care: With Applica- Mechbal. 1997. “Reform Follows Failure: II. Pressure for Change in tions to Vietnam: 1993–1998.� Health Economics. the Lebanese Health Sector.� Health Policy and Planning Wagstaff, Adam, Eddy van Doorslaer, and Naoko Watanabe. 2001. 12(4):312–19. “On Decomposing the Causes of Health Sector Inequalities with Van Lerberghe, W., and P. Ferrinho. 2003.“Civil Society Organizations an Application to Malnutrition Inequalities in Vietnam.� World and the Poor: The Unful�lled Expectations.� Background paper for Bank Policy Research Working Paper 2714. Washington, D.C. the WDR 2004. Waitzkin, Howard. B. 1991. The Politics of Medical Encounters: How Vasan, Sudha. 2002. “Ethnography of the Forest Guard: Contrasting Patients and Doctors Deal with Social Problems. New Haven, Conn.: Discourses, Conflicting Roles and Policy Implementation.� Eco- Yale University Press. nomic and Political Weekly, October 5. Walker, N., B. Schwarlander, and J. Bryce. 2002. “Meeting Interna- Vegas, Emiliana. 2002. “School Choice, Student Performance, and tional Goals in Child Survival and HIV/AIDS.� The Lancet Teacher and School Characteristics: The Chilean Case.� World 360(9329):284–89. Bank Policy Research Working Paper 2833. Washington, D.C. Wang, T. Y. 1999. “U.S. Foreign Aid and UN Voting: An Analysis of Vegas, Emiliana, Lant Pritchett, and William Experton. 1999. “Cómo Important Issues.� International Studies Quarterly 43:199–210. Atraer y Retener Docentes Cali�cados en la Argentina: Impacto del Wang, Youfa, C. Monteiro, and B. M. Popkin. 2002. “Trends of Over- Nivel y la Estructura de la Remuneración.� World Bank Depart- weight and Underweight in Children and Adolescents in the United ment of Human Development, LCSHD Paper Series 38. Washing- States, Brazil, China and Russia.� American Journal of Clinical ton, D.C. Nutrition 75:971–77. Victora, Cesar, J. Patrick Vaughan, Fernando C. Barros, Anamaria C. Warwick, Donald P., and Fernanda Reimers. 1995. Hope or Despair: Silva, and Elaine Tomasi. 2000a. “Explaining Trends in Inequities: Learning in Pakistan’s Primary Schools. Westport, Conn.: Praeger. Evidence from Brazilian Child Health Studies.� The Lancet Water and Sanitation Program. 2001. “Community Contracting in 356(9235):1093–98. Rural Water and Sanitation: The Swajal Project, Uttar Pradesh, Victora, Cesar G., Jennifer Bryce, Oliver Fontaine, and Roeland India.� Water and Sanitation Program: New Delhi. Monasch. 2000b. “Reducing Deaths from Diarrhoea through Oral Water and Sanitation Program (WSP-AF). 2003. “Building Partner- Rehydration Therapy.� Bulletin of the World Health Organization ships to Deliver Water supply and Sanitation Services to Low- 78(10):1246–55. income Communities, Good-Practice from Sub-Saharan Africa.� von Braun, Joachim, and Ulrike Grote. 2000. “Does Decentralization World Bank, Water and Sanitation Program (WSP-AF): Washing- Serve the Poor?� Paper presented at the IMF Conference on Fiscal ton, D.C. Processed. Decentralization Conference. Washington, D.C. November 21. Water Supply and Sanitation Collaborative Council. 2000. “Global von der Fehr, Nils-Henrik M., and Jose Jaime Millan. 2001. “Sus- Water Supply and Sanitation Assessment.� Water Supply and tainability of Power Sector Reform in Latin America: An Analyt- Sanitation Collaborative Council: Geneva. Available on line at ical Framework.� Inter American Development Bank Working www.wsscc.org. Paper. Washington, D.C. Available on line at www.iadb.org. Waters, Hatt, and Aselsson. 2002. “Working for the Private Sector for Wade, Robert. 1982. “System of Administrative and Political Corrup- Child Health.� The World Health Organization and The World tion: Canal Irrigation in South India.� Journal of Development Stud- Bank. Prepared for the USAID-sponsored SARA Project.: Washing- ies 18(April 1982):287–328. ton D.C. ————. 1985.“The Market for Public Of�ce: Why the Indian State Werner, David, Carol Thuman, and Jane Maxwell. 1992. Where There Is Not Better at Development?� World Development 13:467–97. is no Doctor: A Village Health Care Handbook. Palo Alto, Calif.: rev. ————. 1987. “Management of Common Property Resources: ed., Hesperian Foundation. Finding a Cooperative Solution.� World Bank Research Observer Wetterberg, Anna, and Scott Guggenheim. Forthcoming. “Capitaliz- 2(2):219–34. ing on Local Capacity: Institutional Change in the Kecamatan ————. 1988. Village Republics: Economic Conditions for Collec- Development Program.� In A. J. Bebbington, M. Woolcock, and S. tive Action in South India. Cambridge, U.K.: Cambridge Univer- Guggenheim (eds.) Practical Theory, Reflective Action: Social Capi- sity Press. tal and Development Projects at the World Bank. New York: Oxford Wagstaff, Adam. 2000. “Socioeconomic Inequalities in Child Mor- University Press. tality: Comparisons across Nine Developing Countries.� Bulletin Weyland, Kurt. 1997. “‘Growth with Equity’ in Chile’s New Democ- of the World Health Organization 78(1):19–29. racy?� Latin America Research Review 32(1):37–67. ————. 2002. “Child Health on a Dollar a Day: Some Tentative WHO (World Health Organization). 1999. “World Health Report�. Cross-Country Comparisons.� World Bank, Development The World Health Organization. Monograph. Research Group: Washington, D.C. Processed. ————. 2002. “25 Questions and Answers on Health and Human Wagstaff, Adam, Jennifer Bryce, Flavia Bustreo, Mariam Claeson, Rights.� Health and Human Rights Publication Series 1. Geneva, Niklas Danielsson, and WHO-World Bank Child Health and Switzerland. (c) The International Bank for Reconstruction and Development / The World Bank 246 WORLD DEVELOPMENT REPORT 2004 WHO (World Health Organization), and World Bank. 2002. “Dying ————. 2001b. Education and Health in Sub-Saharan Africa: A for Change: Poor People’s Experience of Health and Ill-Health.� Review of Sector-Wide Approaches. Washington, D.C.: World World Health Organization and World Bank: Geneva and Wash- Bank. ington, D.C. ————. 2001c. “Education and Training in Madagascar: Towards WHO (World Health Organization), Division of Child Health and a Policy Agenda for Economic Growth and Poverty Reduction.� Development. 1998.“CHD 1996-97 Report.�World Health Organi- World Bank, Human Development 4 (AFTH4): Washington, zation: Geneva. Available on line at http://www.who.int/ D.C. child-adolescent-health/publications/puboverview.htm#progress. ————. 2001d. “Evaluating the Impact of Health Care Reforms in WHO UNICEF Joint Monitoring Programme for Water Supply and Poland: The Challenges of Systemic Changes in a Transforming Sanitation. 2001. “Global Water Supply and Sanitation Assess- Environment.� World Bank ECA Region: Washington, D.C. ment 2000 Report.� World Health Organization and United ————. 2001e. “Honduras: Public Expenditure Management for Nations Children’s Fund: Geneva and New York. Available on line Poverty Reduction and Fiscal Sustainability.� World Bank: Wash- at http://www.who.int/water_sanitation_health/Globassessment/ ington, D.C. GlobalTOC.htm. ————. 2001f. “Implementation of World Bank-�nance Projects. Wilson, James Q. 1989. Bureaucracy: What Government Agencies Do A Note on Eastern Europe and Central Asia Experience with Pro- and Why they Do It. New York: Basic Books. jects Implementation Units.� World Bank: Washington, D.C. Wodon, Q. 1999. “Cost-Bene�t Analysis of Food for Education in ————. 2001g. “Indonesia Poverty Report.� World Bank: Washing- Bangladesh.� World Bank, Background paper for the 1998 ton, D.C. Bangladesh Poverty Assesment: Washington, D.C. Processed. ————. 2001h. “Nepal: Priorities and Strategies for Education Wolfensohn, James D. 1997. “The Challenge of Inclusion.� World Reform.� World Bank, Human Development Unit, South Asia Bank, Annual Meeting Address, Hong Kong SAR, China. Region: Washington, D.C. World Bank. 1990. World Development Report 1990: Poverty. New ————. 2001i. “Reform of the Electrical Power Sector in India. York: Oxford University Press. Report on Fact-Finding Mission (August 2001).� World Bank, ————. 1993. World Development Report 1993: Investing in Health. Energy and Infrastructure Unit, South Asia Regional Of�ce, Energy New York: Oxford University Press. and Water Department: Washington, D.C. ————. 1994a. “Better Health in Africa: Experience and Lessons ————. 2001j. “Vietnam, Growing Healthy: A Review of Viet- Learned.� World Bank: Washington, D.C. nam’s Health Sector.� World Bank, Human Development Sector ————. 1994b. “Indonesia’s Health Work Force: Issues and Unit, Vietnam Country Unit, East Asia and Paci�c Region: Wash- Options.� World Bank: Washington, D.C. ington, D.C. ————. 1994c. “World Development Report 1994: Infrastructure ————. 2001k. World Development Report 2000/2001: Attacking for Development.� Oxford University Press: New York. Poverty. New York: Oxford University Press. ————. 1996. “Argentina Cordoba - Public Sector Assessment: ————. 2001l.“Zambia Public Expenditure Review: Public Expen- Proposals for Reform Volume II Annexes.� World Bank: Washing- diture, Growth and Poverty: A Synthesis.� World Bank: Washing- ton, D.C. ton, D.C. ————. 1998a. Assessing Aid: What Works, What Doesn’t, and Why. ————. 2002a. A Case for Aid: Building a Consensus for Develop- New York: Oxford University Press. ment Assistance. Washington, D.C.: World Bank. ————. 1998b.“Basic Education in Indonesia: From Crisis to Recov- ————. 2002b. A Sourcebook for Poverty Reduction Strategies. Vol. 1. ery.� World Bank: Washington, D.C. Available on line at http:// Washington, D.C.: World Bank. www-wds.worldbank.org/servlet/WDS_IBank_Servlet?pcont ————. 2002c.“Achieving Education for All in Post-Conflict Cam- =details&eid=000094946_99031910544076. bodia.� Education Notes . Washington, D.C. Available on line at ————. 1998c. “Public Expenditure for Poverty Alleviation in www.worldbank.org/education. Northeast Brazil: Promoting Growth and Improving Services.� ————. 2002d. “Bangladesh: The Experience and Perceptions of World Bank: Washington, D.C. Public Of�cials.� World Bank: Washington, D.C. Processed. ————. 2000a. “Anticorruption in Transition: A Contribution to ————. 2002e. “Building on Strengths: Lessons from Comparative the Policy Debate, The World Bank.� World Bank: Washington, Public Administration Reforms.� World Bank: Washington, D.C. D.C. Processed. ————. 2000b. “Bolivia: From Patronage to a Professional State, ————. 2002f. “Decentralizing Indonesia: A Regional Public Institutional and Governance Review.� World Bank: Washington, Expenditure Review, Overview Report.� East Asia, The World Bank: D.C. Washington, D.C. Processed. ————. 2000c. “Making Transition Work for Everyone: Poverty ————. 2002g. “EdStats: Gender-disaggregated Education Pro- and Inequality in Europe and Central Asia.� World Bank: Washing- �les.� World Bank: Washington, D.C. Available on line at ton, D.C. http://sima/edstats/td.asp. ————. 2001a. “Decentralization in the Transition Economies: ————. 2002h. “Education and HIV/AIDS: A Window of Hope.� Challenges and the Road Ahead.� World Bank: Washington, D.C. World Bank: Washington, D.C. (c) The International Bank for Reconstruction and Development / The World Bank References 247 ————. 2002i. “Evaluation of the Comprehensive Development Clarke, Prema. “Education Reform in the Education Guarantee Framework (CDF): Synthesis Report.� World Bank: Washington, Scheme in Madhya Pradesh, India and the Fundescola Program in D.C. Processed. Brazil.� ————. 2002j.“Mexico: Second Basic Health Project: Implementa- Coady, David. “Alleviating Structural Poverty in Developing Coun- tion Completion Report.� World Bank: Washington, D.C. tries: The Approach of PROGRESA in Mexico.� ————. 2002k. “Pakistan Development Policy Review: A New Das Gupta, Monica, P. Khaleghian, and R. Sarwal. “Governance of Dawn?� World Bank: Washington, D.C. Communicable Disease Control Services: A Case Study and ————. 2002l. “Pakistan Poverty Assessment. Poverty in Pakistan: Lessons from India.� Vulnerabilities, Social Gaps, and Rural Dynamics.� World Bank: Diaz-Cayeros, Alberto, and Beatriz Magaloni . “The Politics of Public Washington, D.C. Spending: Part I - The Logic of Vote Buying. Part II -The Programa ————. 2002m. “Poverty in Pakistan: Vulnerabilities, Social Gaps, Nacional de Solidaridad (PRONASOL) in Mexico.� and Rural Dynamics.� World Bank: Washington, D.C. Dutta, Antara. “Public Service Delivery: The PROBE and Its Impact.� ————. 2002n. “Romania: Local Social Services Delivery Study.� Erikson, Dan, Annie Lord, and Peter Wolf. “Cuba’s Social Services: A World Bank: Washington, D.C. Review of Education, Health, and Sanitation.� ————. 2002o. “Rural Water Projects: Lessons Learned.� World Ferrinho, P., and W. Van Lerberghe. “Managing Health Professionals Bank Operations Evaluation Report 215. Washington, D.C. in the Context of Limited Resources: a Fine Line Between Corrup- ————. 2002p. Social Funds: Assessing Effectiveness. Washington, tion and the Need for Moonlighting.� D.C.: World Bank. Ferriol, Angela, Victoria Perez, Didio Quintana, and Aida Atienza. ————. 2002q. “The Right to Tell: The Role of Mass Media in Eco- “Servicios de Educacion, Salud, Agua y Saneamiento en Cuba.� nomic Development.� World Bank: Washington, D.C. Filmer, Deon. “The Incidence of Public Expenditures on Health and ————. 2002r. “World Bank Group Work in Low-Income Coun- Education.� tries Under Stress.� World Bank: Washington, D.C. Gauri, Varun, and Ayesha Vawda. “Vouchers for Basic Education in ————. 2002s. World Development Indicators. Washington, D.C.: Developing Countries: A Principal-Agent Perspective.� World Bank. Halonen, Maija. “Coordination Failure in Foreign Aid.� ————. 2003a. “Global Economic Prospects and the Developing Kalter, Henry. “Global Health Funds.� Countries: Investing to Unlock Global Opportunities.� World Keefer, Philip, and Stuti Khemani. “The Political Economy of Social Bank: Washington, D.C. Service Provision.� ————. 2003b. “Turkey-reforming the Health Sector for Improved Khaleghian, Peyvand.“Decentralization and Public Services: The Case Access and Ef�ciency.� World Bank, Human Development Sector of Immunization.� Unit: Washington, D.C. Knack, Stephen, and Aminur Rahman. “Aid Intensity, Donor Frag- World Health Organization. 2003. “Severe Acute Respiratory Syn- mentation and the Quality of Governance.� drome (SARS): Status of the Outbreak and Lessons for the Imme- Knippenberg, Rudolf, Fatoumata Traore Nafo, Raimi Osseni, Yero diate Future.� World Health Organization: Geneva. Boye Camara, Abdelwahid El Abassi, Agnes Soucat. “Increasing Wößmann, Ludger. 2003. “Schooling Resources, Educational Institu- client’s power to scale up health services for the poor: The Bamako tions, and Student Performance: The International Evidence.� Initiative in West Africa.� Oxford Bulletin of Economics and Statistics 65(2):117–170. Leonard, L. Kenneth and Melkiory Masatu. “Report on Basic Yip, Winnie, and Peter Berman. 2001. “Targeted Health Insurance in a Findings in Outpatient Facility Evaluations in Arusha Munici- Low Income Country and its Impact on Access and Equity in Access: pality, Arumer District and Monduli District, Arusha Region, Egypt’s School Health Insurance.� Health Economics 10(3):207–20. Tanzania.� Zhao, Feng, Agnes Soucat, and Niagale Traore. 2003. “The Poor in Liese, B., and and others. “Human Resources Crisis in Health Services Mali: Outcomes and Household Determinants of Child Health.� in Sub-Saharan Africa.� World Bank, Africa Region. Washington, D.C. Lisulo, Angela. “Costa Rica: Health Policies.� Mackinnon, John.“How Does Aid Affect the Quality of Public Expen- diture? What We Know and What We Do not Know.� Background papers for the WDR 2004 Mehryar, A. H., A. Aghajanian, and S. Ahmadnia. “Primary Health Anderson, James, Daniel Kaufmann, and Francesca Recanatini. “Ser- Care and Rural Poverty Eradication in the Islamic Republic of vice Delivery, Poverty, and Corruption - Common Threads from Iran.� Diagnostic Surveys.� Msuya, Joyce. “Horizontal and Vertical Delivery of Health Services: Andrews, Matthew, and J. Edgardo Campos. “The Management of What are The Trade Offs?� Public Expenditures and Its Implications for Service Delivery.� Naimoli, Joseph, and Denise Vaillancourt. “Performance-based Man- Bhalla, Surjit, Suraj Saigal, and Nabhojit Basu.“Girl’s Education is it— agement in an Evolving Decentralized Public Health System in Nothing else Matters.� West Africa: The Case of Burkina Faso.� Chaudhury, Nazmul, and Jeffrey Hammer. “Ghost Doctors: Absen- Patrinos, Harry. “A Review of Demand-Side Financing Initiatives in teeism in Bangladeshi Health Facilities.� Education.� (c) The International Bank for Reconstruction and Development / The World Bank 248 WORLD DEVELOPMENT REPORT 2004 Pearson, Mark, and Paolo Carlo Belli. 2003. “The Impact of agement.� Resource Allocation Mechanisms, and Priority Setting Policies to Soucat, Agnes, and Manju Rani. “Making Health and Nutrition Ser- Improve Services for the Poor.� vices Work for Poor People.� Radoki, Carole. “What Are the Most Effective Strategies for Under- Torres Parodi, C. “Health and Ethnicity.� standing and Channeling the Preferences of Service Users to Make Public Services More Responsive.� Consultation papers Roberts, John “Managing Public Expenditure for Development Caseley, Jonathan. “Understanding Service Delivery Reform and Results and Poverty Reduction.� Organizational Performance in an Indian State.� Robinson, James A. “Politician-Proof Policy.� Cotton, Andrew. “Sanitation: A Problem of Scale.� Saigal, Suraj. “Literature Review on Service Delivery in India.� Crook, Richard. Urban Service Partnerships: “Street Level Bureau- Shah, Shekhar, and Manju Rani. “Worlds Apart: Why are Kerala and crats� and Environmental Sanitation in Kumasi and Accra, Ghana: Uttar Pradesh Do Different in Human Development?� Coping with Organizational Change in the Public Bureaucracy.� Shah, Shekhar. “Walk Before you Run: Governance and Public Sector Grindle, Merilee S. “First in the Queue? Mainstreaming The Poor in Reforms in Poverty Reduction Strategies.� Service Delivery.� Soucat, Agnes, and Manju Rani. “Reaching the Poor with or with- Jenkins, Rob and Anne Marie Goetz. “Civil Society Engagement out Quali�ed Professionals: A Review of Obstacles to and Levers and India’s Public Distribution System: Lessons form The for Reaching the Poor with Health and Nutrition Services.� Rationing Kruti Samiti in Mumbai.� Soucat, Agnes, and Manju Rani. “Who Do the Poor Really Get Manor, James. “Partnerships between Governments and Civil Soci- Their Services From: Myths and Reality of the Public-Private ety for Service Delivery in Less Developed Countries: Cause for Mix.� Concern.� Stasavage, David. “On the Role of Democracy in Uganda’s Move to McPake, Barbara. “What Can Be Done About the Private Health Universal Primary Education.� Sector in Low-Income Countries.� StØre, Jonas Gahr. “The Provision of Social Services and the Moore, Mick. “Organisations that Reach the Poor: Why Coproduc- Emerging Welfare State. A Norwegian Perspective.� tion Matters.� van der Berg, Servaas, and Ronelle Burger. “The Stories behind the Nickson, Andrew. “The Role of The Non-State Sector in Urban Numbers: An Investigation of Efforts to Deliver Services to Water Supply.� South Africa’s Poor.� Platteau, Jean-Philippe. “Rushing to Help the Poor Through Partic- Van Lerberghe, W., and P. Ferrinho. “Civil Society and Health.� ipation May Be Self-Defeating.� Vegas, Emiliana “Do Changes in Teacher Contracts and Compensa- Plummer, Janelle. “Developing Inclusive Public-Private Partner- tion Affect Student Performance? Evidence from Togo.� ships: The Role of Small-Scale Independent Providers in the Delivery of Water and Sanitation Services.� Background notes for the WDR 2004 Powis, Benjamin. “Who Uses User Groups?� Alderman, Harold H. “Impact of Nutrition On Other Health Out- Rakodi, Carole. “What are the most Effective Strategies for Undere- comes;� standing and Channelling the Preferences of Service Users to Bhushan, Indu. “Contracting Health Services in Cambodia.� Make Public Services more Responsive?.� Filmer, Deon. “Determinants of Health and Education Outcomes.� Rose, Pauline. “Is The Non-State Education Sector Serving The Gauri, Varun. “A Rights-Based Approach to Health Care and Edu- Needs of The Poor?: Evidence From East and Southern Africa� cation.� Sibbons, Maureen.“Making Education Services Work for Poor People.� Hallman, Kelly. “Health and Nutrition Impacts of Arsenic Contam- Standing, Hilary and Gerry Bloom. “Beyond Public and Private? ination in Bangladesh.� Unorganized Markets in Health Care Delivery.� Hesselbarth, Susanne. Federal Ministry for Economic Cooperation and Development. “Program of Action 2015: Alliances against Poverty Chances and Challenges for the Water Sector.� Singh, Janmejay, and Parmesh Shah. “Making Services Work for Poor People: The Role of Participatory Public Expenditure Man- (c) The International Bank for Reconstruction and Development / The World Bank Introduction to Selected World Development Indicators T his year’s edition presents comparative socioeco- the United Nations and its specialized agencies, the Interna- nomic data for more than 134 economies in six tional Monetary Fund (IMF), and the Organisation for Eco- tables. An additional table provides data on basic nomic Co-operation and Development (OECD). Although indicators for 74 economies with sparse data or with popula- international standards of coverage, de�nition, and classi�- tions of less than 1.5 million. Data are for the most recent cation apply to most statistics reported by countries and year available and, for some indicators, an earlier year is pro- international agencies, there are inevitably differences in vided for comparison purposes. timeliness and reliability arising from differences in the The �rst two tables present data on the size of economies capabilities and resources devoted to basic data collection and several indicators on non-income poverty that are and compilation. For some topics, competing sources of data included in the Millennium Development Goals. Four addi- require review by World Bank staff to ensure that the most tional tables cover data on special topics related to the main reliable data available are presented. In some instances, WDR themes on health, education, service delivery, and for- where available data are deemed too weak to provide reliable eign aid. measures of levels and trends or do not adequately adhere to The indicators presented here are a selection from more international standards, the data are not shown. than 800 included in World Development Indicators 2003. Pub- The data presented are generally consistent with those in lished annually, World Development Indicators reflects a com- World Development Indicators 2003. However, data have been prehensive view of the development process. Its opening revised and updated wherever new information has become chapter reports on the Millennium Development Goals which available. Differences may also reflect revisions to historical grew out of agreements and resolutions of world conferences series and changes in methodology. Thus data of different organized by the United Nations (UN) in the past decade, and vintages may be published in different editions of World reaf�rmed at the Millennium Summit in September 2000 by Bank publications. Readers are advised not to compile data member countries of the UN. The other �ve main sections series from different publications or different editions of the recognize the contribution of a wide range of factors: human same publication. Consistent time-series data are available on capital development, environmental sustainability, macroeco- World Development Indicators 2003 CD-ROM and through nomic performance, private sector development, and the WDI Online. global links that influence the external environment for devel- All dollar �gures are in current U.S. dollars unless other- opment. World Development Indicators is complemented by a wise stated. The various methods used to convert from separately published database that gives access to over 1,000 national currency �gures are described in the Technical data tables and 800 time-series indicators for 225 economies Notes. and regions. This database is available through an electronic Because the World Bank’s primary business is providing subscription (WDI Online) or as a CD-ROM. lending and policy advice to its low- and middle-income members, the issues covered in these tables focus mainly on Data sources and methodology these economies. Where available, information on the high- Socioeconomic and environmental data presented here are income economies is also provided for comparison. Readers drawn from several sources: primary data collected by the may wish to refer to national statistical publications and pub- World Bank, member country statistical publications, lications of the OECD and the European Union for more research institutes, and international organizations such as information on the high-income economies. 249 (c) The International Bank for Reconstruction and Development / The World Bank 250 WORLD DEVELOPMENT REPORT 2004 Classi�cation of economies From time to time an economy’s classi�cation is revised and summary measures because of changes in the above cutoff values or in the econ- omy’s measured level of GNI per capita. When such changes The summary measures at the bottom of each table occur, aggregates based on those classi�cations are recalcu- include economies classi�ed by income per capita and by lated for the past period so that a consistent time series is region. GNI per capita is used to determine the following maintained. income classi�cations: low-income, $735 or less in 2002; middle-income, $736 to $9,075; and high-income, $9,076 and above. A further division at GNI per capita $2,935 is Terminology and country coverage made between lower-middle-income and upper-middle- The term country does not imply political independence but income economies. See the table on classi�cation of may refer to any territory for which authorities report sepa- economies in this volume for a list of economies in each rate social or economic statistics. Data are shown for group (including those with populations of less than 1.5 economies as they were constituted in 2002, and historical million). data are revised to reflect current political arrangements. Summary measures are either totals (indicated by t if Throughout the tables, exceptions are noted. the aggregates include estimates for missing data and non- reporting countries, or by an s for simple sums of the data Technical notes available), weighted averages (w), or median values (m) Because data quality and intercountry comparisons are often calculated for groups of economies. Data for the countries problematic, readers are encouraged to consult the Technical excluded from the main tables (those presented in Table 7) notes, the table on Classi�cation of economies by region and have been included in the summary measures, where data income, and the footnotes to the tables. For more extensive are available, or by assuming that they follow the trend of documentation see World Development Indicators 2003. reporting countries. This gives a more consistent aggre- Readers may �nd more information on the WDI 2003, gated measure by standardizing country coverage for each and orders can be made online, by phone, or fax as follows: period shown. Where missing information accounts for a third or more of the overall estimate, however, the group For more information and to order online: http://www. measure is reported as not available. The Technical Notes worldbank.org/data/wdi2003/index.htm To order by phone provides further information on aggregation methods. or fax: 1-800-645-7247 or 703-661-1580; Fax 703-661-1501 Weights used to construct the aggregates are listed in the technical notes for each table. To order by mail: The World Bank, P.O. Box 960, Herndon, VA 20172-0960, U.S.A. (c) The International Bank for Reconstruction and Development / The World Bank Selected World Development Indicators 251 Classi�cation of economies by region and income, FY2004 East Asia and Paci�c Latin America and Caribbean Sub-Saharan Africa High income OECD American Samoa UMC Argentina UMC Angola LIC Australia Cambodia LIC Belize UMC Benin LIC Austria China LMC Bolivia LMC Botswana UMC Belgium Fiji LMC Brazil LMC Burkina Faso LIC Canada Indonesia LIC Chile UMC Burundi LIC Denmark Kiribati LMC Colombia LMC Cameroon LIC Finland Korea, Dem. Rep. LIC Costa Rica UMC Cape Verde LMC France Lao PDR LIC Cuba LMC Central African Rep. LIC Germany Malaysia UMC Dominica UMC Chad LIC Greece Marshall Islands LMC Dominican Rep. LMC Comoros LIC Iceland Micronesia, Fed. Sts. LMC Ecuador LMC Congo, Dem. Rep. LIC Ireland Mongolia LIC El Salvador LMC Congo, Rep. LIC Italy Myanmar LIC Grenada UMC Côte d’Ivoire LIC Japan N. Mariana Islands UMC Guatemala LMC Equatorial Guinea LIC Korea, Rep. Palau UMC Guyana LMC Eritrea LIC Luxembourg Papua New Guinea LIC Haiti LIC Ethiopia LIC Netherlands Philippines LMC Honduras LMC Gabon UMC New Zealand Samoa LMC Jamaica LMC Gambia, The LIC Norway Solomon Islands LIC Mexico UMC Ghana LIC Portugal Thailand LMC Nicaragua LIC Guinea LIC Spain Timor-Leste LIC Panama UMC Guinea-Bissau LIC Sweden Tonga LMC Paraguay LMC Kenya LIC Switzerland Vanuatu LMC Peru LMC Lesotho LIC United Kingdom Vietnam LIC St. Kitts & Nevis UMC Liberia LIC United States St. Lucia UMC Madagascar LIC Europe and Central Asia St. Vincent & Grenadines LMC Malawi LIC Other high income Albania LMC Suriname LMC Mali LIC Andorra Armenia LMC Trinidad & Tobago UMC Mauritania LIC Antigua & Barbuda Azerbaijan LIC Uruguay UMC Mauritius UMC Aruba Belarus LMC Venezuela, RB UMC Mayotte UMC Bahamas, The Bosnia & Herzegovina LMC Mozambique LIC Bahrain Bulgaria LMC Middle East and North Africa Namibia LMC Barbados Croatia UMC Algeria LMC Niger LIC Bermuda Czech Rep. UMC Djibouti LMC Nigeria LIC Brunei Estonia UMC Egypt, Arab Rep. LMC Rwanda LIC Cayman Islands Georgia LIC Iran, Islamic Rep. LMC São Tomé & Principe LIC Channel Islands Hungary UMC Iraq LMC Senegal LIC Cyprus Kazakhstan LMC Jordan LMC Seychelles UMC Faeroe Islands Kyrgyz Rep. LIC Lebanon UMC Sierra Leone LIC French Polynesia Latvia UMC Libya UMC Somalia LIC Greenland Lithuania UMC Morocco LMC South Africa LMC Guam Macedonia, FYR LMC Oman UMC Sudan LIC Hong Kong, China Moldova LIC Saudi Arabia UMC Swaziland LMC Isle of Man Poland UMC Syrian Arab Rep. LMC Tanzania LIC Israel Romania LMC Tunisia LMC Togo LIC Kuwait Russian Fed. LMC West Bank & Gaza LMC Uganda LIC Liechtenstein Serbia & Montenegro LMC Yemen, Rep. LIC Zambia LIC Macao, China Slovak Rep. UMC Zimbabwe LIC Malta Tajikistan LIC South Asia Monaco Turkey LMC Afghanistan LIC Netherlands Antilles Turkmenistan LMC Bangladesh LIC New Caledonia Ukraine LMC Bhutan LIC Puerto Rico Uzbekistan LIC India LIC Qatar Maldives LMC San Marino Nepal LIC Singapore Pakistan LIC Slovenia Sri Lanka LMC Taiwan, China United Arab Emirates Virgin Islands (U.S.) This table classi�es all World Bank member economies, and all other economies with populations of more than 30,000. Economies are divided among income groups according to 2002 GNI per capita, calculated using the World Bank Atlas method. The groups are: low income (LIC), $735 or less; lower middle income (LMC), $736–2,935; upper middle income (UMC), $2,936–9,075; and high income, $9,076 or more. Source: World Bank data. (c) The International Bank for Reconstruction and Development / The World Bank Table 1 Size of the economy Population Surface Population Gross national PPP gross Gross domestic area density income a national income b product millions thousand people per sq. km $ billions Per $ billions Per % growth Per capita sq. km of land area capita $ capita $ % growth 2002 2002 2002 2002 2002 2002 2002 2001–2002 2001–2002 Albania 3 29 117 4.4 1,380 13 4,040 4.7 3.7 Algeria 31 2,382 13 53.8 1,720 167 c 5,330 c 4.1 2.5 Angola 14 1,247 11 9.2 660 24 c 1,730 c 17.1 13.8 Argentina 38 2,780 14 154.1 4,060 377 9,930 –10.9 –12.0 Armenia 3 30 109 2.4 790 9 3,060 12.9 13.5 Australia 20 7,741 3 386.6 19,740 528 26,960 3.5 2.5 Austria 8 84 98 190.4 23,390 230 28,240 1.0 0.9 Azerbaijan 8 87 95 5.8 710 24 2,920 10.6 9.7 Bangladesh 136 144 1,042 48.5 360 234 1,720 4.4 2.6 Belarus 10 208 48 13.5 1,360 53 5,330 4.7 5.1 Belgium 10 33 314 239.9 23,250 282 27,350 0.7 0.4 Benin 7 113 60 2.5 380 7 1,020 5.3 2.6 Bolivia 9 1,099 8 7.9 900 20 2,300 2.5 0.4 Bosnia & Herzegovina 4 51 81 5.2 1,270 24 5,800 3.9 2.4 Botswana 2 582 3 5.1 2,980 13 7,770 3.5 2.5 Brazil 174 8,547 21 497.4 2,850 1,266 7,250 1.5 0.3 Bulgaria 8 111 71 14.1 1,790 54 6,840 4.3 4.9 Burkina Faso 12 274 43 2.6 220 12 c 1,010 c 5.6 3.1 Burundi 7 28 275 0.7 100 4c 610 c 3.6 1.7 Cambodia 12 181 71 3.5 280 20 1,590 4.5 2.6 Cameroon 16 475 33 8.7 560 25 1,640 4.4 2.2 Canada 31 9,971 3 700.5 22,300 882 c 28,070 c 3.3 2.2 Central African Rep. 4 623 6 1.0 260 5c 1,190 c 4.2 2.6 Chad 8 1,284 6 1.8 220 8 1,000 10.9 7.8 Chile 16 757 21 66.3 4,260 143 9,180 2.1 0.9 China 1,281 9,598 d 137 1,209.5 940 5,625 e 4,390 e 8.0 7.2 Hong Kong, China 7 .. .. 167.6 24,750 182 26,810 2.3 1.5 Colombia 44 1,139 42 80.1 1,830 257 5,870 1.5 –0.1 Congo, Dem. Rep. 54 2,345 24 5.0 90 31 580 3.0 0.2 Congo, Rep. 3 342 9 2.2 700 2 700 3.5 0.7 Costa Rica 4 51 77 16.2 4,100 33 8,260 2.8 1.0 Côte d’Ivoire 17 322 53 10.3 610 24 1,430 –0.9 –3.0 Croatia 4 57 78 20.3 4,640 43 9,760 5.2 5.3 Czech Rep. 10 79 132 56.7 5,560 148 14,500 2.0 2.1 Denmark 5 43 127 162.7 30,290 158 29,450 1.6 1.3 Dominican Rep. 9 49 178 20.0 2,320 51 5,870 4.1 2.5 Ecuador 13 284 47 19.0 1,450 41 3,130 3.0 1.2 Egypt, Arab Rep. 66 1,001 67 97.6 1,470 246 3,710 3.0 1.1 El Salvador 7 21 315 13.5 2,080 30 4,570 2.3 0.4 Eritrea 4 118 43 0.7 160 4 950 9.2 6.5 Estonia 1 45 32 5.6 4,130 15 11,120 5.8 6.2 Ethiopia 67 1,104 67 6.4 100 48 720 5.0 2.7 Finland 5 338 17 122.2 23,510 132 25,440 1.6 1.4 France 59 552 108 1,342.7 f 22,010 f 1,556 26,180 1.0 0.6 Georgia 5 70 74 3.3 650 11 2,210 5.4 6.4 Germany 82 357 231 1,870.4 22,670 2,163 26,220 0.2 0.0 Ghana 20 239 88 5.4 270 40 c 2,000 c 4.5 2.6 Greece 11 132 82 123.9 11,660 194 18,240 4.0 3.6 Guatemala 12 109 111 20.9 1,750 47 3,880 2.0 –0.6 Guinea 8 246 32 3.1 410 15 1,990 4.3 2.1 Haiti 8 28 301 3.7 440 13 c 1,580 c –0.9 –2.7 Honduras 7 112 60 6.2 920 17 2,450 2.0 –0.6 Hungary 10 93 110 53.7 5,280 130 12,810 3.3 3.5 India 1,048 3,287 353 501.5 480 2,691 2,570 4.4 2.8 Indonesia 212 1,905 117 149.9 710 632 2,990 3.7 2.3 Iran, Islamic Rep. 66 1,648 40 112.1 1,710 415 6,340 5.9 4.2 Ireland 4 70 56 92.6 23,870 109 28,040 3.6 2.6 Israel 6 21 315 .. .. g .. .. .. .. Italy 58 301 197 1,097.9 18,960 1,467 25,320 0.4 0.4 Jamaica 3 11 241 7.4 2,820 9 3,550 1.0 0.1 Japan 127 378 349 4,265.6 33,550 3,315 26,070 –0.7 –0.8 Jordan 5 89 58 9.1 1,760 21 4,070 4.9 2.0 Kazakhstan 15 2,725 5 22.3 1,510 81 5,480 9.5 10.2 Kenya 31 580 55 11.3 360 31 990 1.8 –0.2 Korea, Rep. 48 99 483 473.0 9,930 785 16,480 6.3 5.7 Kuwait 2 18 118 .. .. g .. .. .. .. Kyrgyz Rep. 5 200 26 1.5 290 8 1,520 –0.5 –1.5 Lao PDR 6 237 24 1.7 310 9c 1,610 c 5.0 2.6 Latvia 2 65 38 8.1 3,480 21 8,940 6.1 7.2 Lebanon 4 10 434 17.7 3,990 20 4,470 1.0 –0.3 Lesotho 2 30 69 1.0 470 6c 2,710 c 3.8 2.6 Lithuania 3 65 54 12.7 3,660 34 9,880 6.7 6.9 Macedonia, FYR 2 26 80 3.5 1,700 13 6,210 0.3 0.1 Madagascar 16 587 28 3.9 240 12 720 –11.9 –14.4 Malawi 11 118 114 1.7 160 6 570 1.8 –0.3 Note: For data comparability and coverage, see the technical notes. Figures in italics are for years other than those speci�ed. 252 (c) The International Bank for Reconstruction and Development / The World Bank Table 1 Size of the economy—continued Population Surface Population Gross national PPP gross Gross domestic area density income a national income b product millions thousand people per sq. km $ billions Per $ billions Per % growth Per capita sq. km of land area capita $ capita $ % growth 2002 2002 2002 2002 2002 2002 2002 2001–2002 2001–2002 Malaysia 24 330 74 86.0 3,540 201 8,280 4.2 2.1 Mali 11 1,240 9 2.8 240 10 840 9.6 7.1 Mauritania 3 1,026 3 1.0 340 .. .. 5.1 2.2 Mexico 101 1,958 53 596.7 5,910 862 8,540 0.7 –0.8 Moldova 4 34 129 1.7 h 460 h 7 1,560 7.2 7.6 Mongolia 2 1,567 2 1.1 440 4 1,650 3.7 2.6 Morocco 30 447 66 35.4 1,190 109 3,690 4.5 2.9 Mozambique 18 802 24 3.9 210 .. .. 9.9 7.7 Myanmar 49 677 74 .. .. i .. .. .. .. Namibia 2 824 2 3.3 1,780 12 c 6,650 c 3.0 1.2 Nepal 24 147 169 5.6 230 33 1,350 –0.6 –2.8 Netherlands 16 42 477 386.8 23,960 443 27,470 0.1 –0.6 New Zealand 4 271 14 53.1 13,710 77 20,020 3.8 3.2 Nicaragua 5 130 44 .. .. i .. .. .. .. Niger 12 1,267 9 2.0 170 9c 770 c 3.0 –0.2 Nigeria 133 924 146 38.7 290 103 780 –0.9 –3.1 Norway 5 324 15 171.8 37,850 163 35,840 2.0 1.4 Pakistan 145 796 188 59.2 410 281 1,940 4.4 1.9 Panama 3 76 40 11.8 4,020 17 c 5,870 c 0.8 –0.7 Papua New Guinea 5 463 12 2.8 530 11 c 2,080 c –2.5 –4.7 Paraguay 6 407 14 6.4 1,170 25 c 4,450 c –2.2 –4.3 Peru 27 1,285 21 54.7 2,050 128 4,800 5.2 3.7 Philippines 80 300 268 81.5 1,020 342 4,280 4.6 2.4 Poland 39 323 127 176.6 4,570 391 10,130 1.2 1.2 Portugal 10 92 110 108.7 10,840 174 17,350 0.4 0.3 Romania 22 238 97 41.3 1,850 141 6,290 4.3 4.5 Russian Fed. 144 17,075 9 307.9 2,140 1,127 7,820 4.3 4.8 Rwanda 8 26 331 1.9 230 10 1,210 9.4 6.3 Saudi Arabia 22 2,150 10 .. .. j .. .. .. .. Senegal 10 197 52 4.7 470 15 1,510 2.4 0.0 Serbia & Montenegro 11 102 108 11.6 k 1,400 k .. .. .. .. Sierra Leone 5 72 73 0.7 140 3 490 6.3 4.2 Singapore 4 1 6,826 86.1 20,690 96 23,090 2.2 1.4 Slovak Rep. 5 49 112 21.4 3,950 66 12,190 4.4 4.3 Slovenia 2 20 99 19.6 9,810 35 17,690 2.9 2.9 South Africa 44 1,221 36 113.5 2,600 430 c 9,870 c 3.0 2.2 Spain 41 506 82 594.1 14,430 842 20,460 1.8 1.6 Sri Lanka 19 66 293 15.9 840 64 3,390 3.0 1.7 Sweden 9 450 22 221.5 24,820 224 25,080 1.9 1.5 Switzerland 7 41 183 274.2 37,930 226 31,250 –0.2 –0.2 Syrian Arab Rep. 17 185 93 19.2 1,130 55 3,250 3.1 0.6 Tajikistan 6 143 45 1.1 180 6 900 9.1 7.9 Tanzania 35 945 40 9.6l 280 l 19 550 5.8 3.6 Thailand 62 513 121 122.2 1,980 411 6,680 5.2 4.5 Togo 5 57 88 1.3 270 7 1,430 3.0 0.5 Tunisia 10 164 63 19.6 2,000 61 6,280 1.9 0.7 Turkey 70 775 90 174.0 2,500 426 6,120 7.8 6.1 Turkmenistan 6 488 12 6.7 1,200 25 4,570 14.9 12.6 Uganda 23 241 119 5.9 250 31 c 1,320 c 6.3 3.6 Ukraine 49 604 84 37.7 770 226 4,650 4.5 5.3 United Kingdom 59 243 244 1,486.2 25,250 1,523 25,870 1.5 1.4 United States 288 9,629 31 10,110.1 35,060 10,110 35,060 2.3 1.2 Uruguay 3 176 19 14.8 4,370 41 12,010 –10.8 –11.3 Uzbekistan 25 447 61 11.5 450 40 1,590 4.2 2.9 Venezuela, RB 25 912 28 102.6 4,090 127 5,080 –8.9 –10.6 Vietnam 81 332 247 34.9 430 180 2,240 7.1 5.8 Yemen, Rep. 19 528 35 9.4 490 14 750 4.2 1.1 Zambia 10 753 14 3.5 330 8 770 3.0 1.3 Zimbabwe 13 391 34 .. i .. 28 2,120 –5.6 –6.6 World 6,201 s 133,875 s 48 w 31,483.9 t 5,080 w 46,952 t 7,570 w 1.7 w 0.5 w Low income 2,495 33,612 77 1,071.7 430 5,092 2,040 4.1 2.3 Middle income 2,742 67,898 41 5,033.3 1,840 15,431 5,630 3.2 2.2 Lower middle income 2,411 54,970 45 3,352.4 1,390 12,378 5,130 4.8 3.9 Upper middle income 331 12,928 26 1,667.9 5,040 3,050 9,220 –1.5 –2.7 Low & middle income 5,237 101,510 53 6,101.7 1,170 20,474 3,910 3.3 2.0 East Asia & Paci�c 1,838 16,302 116 1,740.5 950 7,640 4,160 6.7 5.8 Europe & Central Asia 476 24,217 20 1,030.2 2,160 3,188 6,690 4.7 4.6 Latin America & Carib. 527 20,450 26 1,726.5 3,280 3,556 6,750 –0.5 –1.9 Middle East & N. Africa 306 11,135 28 670.0 2,230 1,657 5,410 .. .. South Asia 1,401 5,140 293 640.5 460 3,352 2,390 4.3 2.6 Sub-Saharan Africa 688 24,267 29 306.5 450 1,116 1,620 3.2 0.9 High income 965 32,365 31 25,383.7 26,310 26,622 27,590 1.3 0.8 a. Calculated using the World Bank Atlas method. b. PPP is purchasing power parity; see the technical notes. c.The estimate is based on regression; others are extrapolated from the latest International Comparison Programme benchmark estimates. d. Includes Taiwan, China; Macao, China; and Hong Kong, China. e. Estimate based on bilateral comparison between China, and USA (Ruoen and Kai, 1995). f. GNI and GNI per capita estimates include the French overseas departments of French Guiana, Guadeloupe, Martinique, and Reunion. g. Estimated to be high income (9,076 or more). h. Data excludes Transnistria. i. Estimated to be low income ($735 or less). j. Estimated to be upper middle income ($2,935–9,075). k. Data excludes Kosovo. l. Data refer to mainland Tanzania only. (c) The International Bank for Reconstruction and Development / The World Bank Table 2 Millennium Development Goals: eradicating poverty and improving lives Eradicate Achieve universal Promote Reduce Improve maternal health extreme poverty primary education gender child and hunger equality mortality Share of poorest Prevalence Primary Ratio of female to Under-�ve Maternal Births quintile in of child completion male enrollments mortality mortality ratio attended national income malnutrition rate (%) in primary and rate per per 100,000 by skilled or consumption % % of children secondary 1,000 live births health staff under 5 school (%) a modeled % of total estimates 1987–2001 b 1990 2001 1990 2001 1990 2000 1990 2001 1995 1990 2000 Albania .. .. 14 101 .. 90 102 42 25 31 .. 99 Algeria 7.0 c 9 6 82 .. 80 98 69 49 150 .. 92 Angola .. 20 .. .. 28 .. 84 260 260 1,300 .. .. Argentina .. .. .. .. 96 .. 103 28 19 85 .. 98 Armenia 6.7 c .. 3 .. .. .. 106 58 35 29 .. 97 Australia 5.9 d .. .. .. .. 96 100 10 6 6 100 100 Austria 7.0 d .. .. .. .. 90 97 9 5 11 .. .. Azerbaijan 7.4 c .. 17 47 100 94 101 106 96 37 .. 88 Bangladesh 9.0 c 66 48 50 70 72 103 144 77 600 7 12 Belarus 8.4 c .. .. 97 .. .. 101 21 20 33 .. 100 Belgium 8.3 d .. .. .. .. 97 106 9 6 8 .. .. Benin .. .. 23 23 39 .. 62 185 158 880 38 .. Bolivia 4.0 c 11 8 55 72 89 97 122 77 550 43 59 Bosnia & Herzegovina .. .. 4 .. 88 .. .. 22 18 15 .. 100 Botswana 2.2 c .. 13 114 .. 107 102 58 110 480 79 99 Brazil 2.2 d 7 .. 48 71 .. 103 60 36 260 .. .. Bulgaria 6.7 d .. .. 90 .. 94 97 19 16 23 .. 99 Burkina Faso 4.5 c .. 34 19 25 61 70 210 197 1,400 30 27 Burundi 5.1 c .. 45 46 43 82 79 190 190 1,900 20 25 Cambodia 6.9 c .. 45 71 70 .. 83 115 138 590 47 34 Cameroon 4.6 c 15 22 57 43 82 81 139 155 720 58 56 Canada 7.3 d .. .. .. .. 94 101 8 7 6 .. .. Central African Rep. 2.0 c .. .. 28 19 61 .. 180 180 1,200 66 44 Chad .. .. 28 19 19 .. 56 203 200 1,500 15 16 Chile 3.2 d .. 1 94 99 98 88 19 12 33 .. .. China 5.9 d 17 10 99 .. 81 98 49 39 60 .. 70 Hong Kong, China 5.3 .. .. .. .. .. .. .. .. .. 100 100 Colombia 3.0 d 10 7 72 85 104 104 36 23 120 94 86 Congo, Dem. Rep. .. .. .. 48 40 69 80 205 205 940 .. 70 Congo, Rep. .. .. .. 61 44 88 89 110 108 1,100 .. .. Costa Rica 4.4 d 3 .. 73 89 96 101 17 11 35 .. 98 Côte d’Ivoire 7.1 c .. 21 44 40 .. 71 155 175 1,200 50 47 Croatia 8.3 c .. .. 86 .. 97 .. 13 8 18 .. 100 Czech Rep. 10.3 d 1 .. 89 .. 94 101 12 5 14 .. .. Denmark 8.3 d .. .. .. .. 96 103 9 4 15 .. .. Dominican Rep. 5.1 d 10 5 .. 82 .. 106 65 47 110 92 .. Ecuador 5.4 c .. 14 99 96 97 100 57 30 210 56 69 Egypt, Arab Rep. 8.6 c 10 4 77 .. 78 94 104 41 170 37 61 El Salvador 3.3 d 15 12 61 80 100 98 60 39 180 90 90 Eritrea .. .. .. 22 35 82 77 155 111 1,100 .. .. Estonia 7.0 d .. .. 93 .. 99 99 17 12 80 .. .. Ethiopia 2.4 d 48 47 22 24 68 68 193 172 1,800 8 10 Finland 10.1 d .. .. .. .. 105 106 7 5 6 .. .. France 7.2 d .. .. .. .. 98 100 10 6 20 .. .. Georgia 6.0 c .. 3 .. 90 94 102 29 29 22 .. 96 Germany 5.7 d .. .. .. .. 94 99 9 5 12 .. .. Ghana 5.6 c 30 25 63 64 .. 88 126 100 590 55 44 Greece 7.1 d .. .. .. .. 93 101 11 5 2 .. .. Guatemala 3.8 d .. 24 43 52 .. 92 82 58 270 30 41 Guinea 6.4 c .. 33 16 34 43 57 240 169 1,200 .. 35 Haiti .. 27 17 28 70 .. .. 150 123 1,100 78 24 Honduras 2.0 d 18 17 66 67 103 .. 61 38 220 .. .. Hungary 10.0 c 2 .. 93 .. 96 100 17 9 23 .. .. India 8.1 c 64 .. 70 76 68 78 123 93 440 44 42 Indonesia 8.4 c .. 25 92 91 91 98 91 45 470 47 56 Iran, Islamic Rep. 5.1 c .. 11 94 .. 80 95 72 42 130 78 .. Ireland 6.7 d .. .. .. .. 99 .. 9 6 9 .. .. Israel 6.9 d .. .. .. .. 99 100 12 6 8 .. .. Italy 6.0 d .. .. .. .. 95 98 10 6 11 .. .. Jamaica 6.7 c 5 4 90 94 97 101 20 20 120 92 95 Japan 10.6 d .. .. .. .. 96 101 6 5 12 100 .. Jordan 7.6 c 6 .. 102 104 93 101 43 33 41 87 .. Kazakhstan 8.2 c .. 4 .. .. .. 98 52 99 80 .. 98 Kenya 5.6 c .. 22 87 63 .. 97 97 122 1,300 50 44 Korea, Rep. 7.9 d .. .. 96 96 93 100 9 5 20 98 .. Kuwait .. .. .. 56 .. 97 101 16 10 25 .. .. Kyrgyz Rep. 9.1 c .. .. .. 100 100 99 81 61 80 .. 98 Lao PDR 7.6 c .. 40 44 69 75 82 163 100 650 .. 21 Latvia 7.6 d .. .. 76 .. 96 101 18 21 70 .. .. Lebanon .. .. .. .. .. .. 102 37 32 130 95 95 Lesotho 1.4 c 16 18 75 68 124 107 148 132 530 40 60 Lithuania 7.9 c .. .. 88 .. 93 99 14 9 27 .. .. Macedonia, FYR 8.4 c .. 6 89 .. 94 98 33 26 17 88 97 Madagascar 6.4 c 41 .. 34 26 .. 97 168 136 580 .. 46 Malawi 4.9 c 28 25 33 64 79 94 241 183 580 50 56 Note: For data comparability and coverage, see the technical notes. Figures in italics are for years other than those speci�ed. 254 (c) The International Bank for Reconstruction and Development / The World Bank Table 2 Millennium Development Goals: eradicating poverty and improving lives—continued Eradicate Achieve universal Promote Reduce Improve maternal health extreme poverty primary education gender child and hunger equality mortality Share of poorest Prevalence Primary Ratio of female to Under-�ve Maternal Births quintile in of child completion male enrollments mortality mortality ratio attended national income malnutrition rate (%) in primary and rate per per 100,000 by skilled or consumption % % of children secondary 1,000 live births health staff under 5 school (%) a modeled % of total estimates 1987–2001 b 1990 2001 1990 2001 1990 2000 1990 2001 1995 1990 2000 d Malaysia 4.4 25 .. 91 .. 98 105 21 8 39 .. 96 Mali 4.6 c .. .. 11 23 57 66 254 231 630 .. .. Mauritania 6.4 c 48 32 34 46 67 93 183 183 870 40 57 Mexico 3.4 d 17 8 89 100 96 101 46 29 65 .. .. Moldova 7.1 c .. .. 67 79 103 102 37 32 65 .. .. Mongolia 5.6 c 12 13 .. 82 107 112 107 76 65 100 97 Morocco 6.5 c 10 .. 47 .. 67 83 85 44 390 .. .. Mozambique 6.5 c .. .. 30 36 73 75 235 197 980 .. .. Myanmar .. 32 .. .. .. 95 98 130 109 170 94 .. Namibia 1.4 d 26 .. 70 .. 111 104 84 67 370 .. 76 Nepal 7.6 c .. 48 51 65 53 82 145 91 830 .. 12 Netherlands 7.3 d .. .. .. .. 93 97 8 6 10 100 100 New Zealand 6.4 d .. .. .. .. 96 103 11 6 15 .. .. Nicaragua 2.3 c .. 12 45 65 .. 105 66 43 250 .. 61 Niger 2.6 c 43 40 18 20 54 67 320 265 920 .. 16 Nigeria 4.4 c 35 31 72 67 76 .. 190 183 1,100 31 42 Norway 9.7 d .. .. .. .. 97 101 9 4 9 100 .. Pakistan 8.8 c 40 .. 44 59 47 61 128 109 200 40 20 Panama 3.6 c 6 .. 87 94 96 100 34 25 100 .. 90 Papua New Guinea 4.5 c .. .. 53 .. 77 90 101 94 390 40 .. Paraguay 1.9 d 4 .. 65 78 95 99 37 30 170 71 71 Peru 4.4 d 11 7 85 98 93 97 75 39 240 78 .. Philippines 5.4 c 34 32 89 .. .. 103 66 38 240 .. 56 Poland 7.8 c .. .. 100 .. 96 98 22 9 12 .. .. Portugal 5.8 d .. .. .. .. 99 102 15 6 12 98 100 Romania 8.2 c 6 .. 96 .. 95 100 36 21 60 .. 98 Russian Fed. 4.9 c .. .. .. 96 .. .. 21 21 75 .. 99 Rwanda .. c 29 24 34 28 98 97 178 183 2,300 22 31 Saudi Arabia .. .. .. 60 .. 82 94 44 28 23 88 91 Senegal 6.4 c 22 18 45 41 69 84 148 138 1,200 42 51 Serbia & Montenegro .. .. 2 72 96 96 .. 26 19 15 .. 93 Sierra Leone 1.1 c 29 27 .. 32 67 77 323 316 2,100 .. 42 Singapore 5.0 d .. .. .. .. 89 .. 8 4 9 .. 100 Slovak Rep. 8.8 d .. .. 96 .. 98 101 14 9 14 .. .. Slovenia 9.1 d .. .. 99 .. 97 .. 10 5 17 100 .. South Africa 2.0 c .. .. 76 .. 103 100 60 71 340 .. 84 Spain 7.5 d .. .. .. .. 99 103 9 6 8 .. .. Sri Lanka 8.0 c .. 33 100 111 99 102 23 19 60 85 .. Sweden 9.1 d .. .. .. .. 97 115 7 3 8 .. .. Switzerland 6.9 d .. .. .. .. 92 96 8 6 8 .. .. Syrian Arab Rep. .. .. .. 98 .. 82 92 44 28 200 64 .. Tajikistan 8.0 c .. .. .. 95 .. 87 127 116 120 .. 77 Tanzania 6.8 c 29 29 65 60 97 99 163 165 1,100 44 35 Thailand 6.1 c .. .. 93 90 94 95 40 28 44 71 .. Togo .. 25 25 41 63 59 70 152 141 980 32 51 Tunisia 5.7 c 10 4 75 .. 82 100 52 27 70 80 90 Turkey 6.1 c .. 8 90 .. 77 84 74 43 55 77 81 Turkmenistan 6.1 c .. 12 .. .. .. .. 98 87 65 .. 97 Uganda 7.1 c 23 23 49 65 .. 89 165 124 1,100 38 .. Ukraine 8.8 c .. 3 58 .. .. 92 22 20 45 .. 99 United Kingdom 6.1 d .. .. .. .. 97 111 9 7 10 100 99 United States 5.2 d .. .. .. .. 95 100 11 8 12 99 99 Uruguay 4.5 d,e 6 .. 95 98 .. 105 24 16 50 .. 100 Uzbekistan 9.2 c .. .. .. 100 .. .. 65 68 60 .. 96 Venezuela, RB 3.0 d 8 4 91 78 101 105 27 22 43 97 95 Vietnam 8.0 c 45 34 .. 101 .. .. 50 38 95 95 70 Yemen, Rep. 7.4 c 30 .. .. 58 .. 50 142 107 850 .. 22 Zambia 3.3 c 25 .. 91 73 .. 92 192 202 870 41 .. Zimbabwe 4.6 c 12 13 97 .. 96 94 80 123 610 62 84 World .. w .. w .. w .. w 84 w 92 w 93 w 81 w .. w .. w Low income .. .. 68 .. 74 78 141 121 43 .. Middle income .. .. 94 .. 84 98 51 38 .. .. Lower middle income 18 10 95 .. 82 97 54 41 .. .. Upper middle income .. .. 90 .. 96 100 34 23 .. .. Low & middle income .. .. 83 .. 80 90 101 88 .. .. East Asia & Paci�c 19 15 98 .. 83 97 59 44 .. 70 Europe & Central Asia .. .. .. .. .. .. 44 38 .. .. Latin America & Carib. .. .. .. .. .. 102 53 34 .. .. Middle East & N. Africa .. .. 81 .. 79 95 77 54 .. .. South Asia 64 .. 70 74 68 79 129 99 39 42 Sub-Saharan Africa .. .. 57 .. 79 82 178 171 .. .. High income .. .. .. .. 96 101 10 7 .. .. a. Break in series between 1997 and 1998 due to change from International Standard Classi�cation of Education 1976 (ISCED76) to ISCED97. b. Data are for the most recent year available. c. Refers to expenditure shares by percentiles of population; ranked by per capita expenditure. d. Refers to income shares by percentiles of population; ranked by per capita income. e. Data refer to urban only. 255 (c) The International Bank for Reconstruction and Development / The World Bank Table 3 Expenditures on education and health Public expenditure per student a Recurrent Incidence of Health expenditure Incidence spending education of health on primary expenditure expenditure teacher salaries b Primary Secondary Tertiary % of total lowest highest Public Private Total lowest highest % of % of % of recurrent quintile quintile % of % of per quintile quintile GDP GDP GDP spending on GDP GDP capita per per per primary $ capita capita capita education 2000 2000 2000 2000 1991–2001 c 1991–2001 c 2000 2000 1997–2000 1991–2001 c 1991–2001 c Albania .. .. .. 82.5 .. .. 2.1 1.3 41 .. .. Algeria .. .. .. .. .. .. 3.0 0.6 64 .. .. Angola .. .. .. 81.0 .. .. 2.0 1.6 24 .. .. Argentina 12.5 16.4 17.7 .. .. .. 4.7 3.9 658 33 6 Armenia 4.0 22.2 17.9 47.1 7 29 3.2 4.3 38 13 39 Australia 15.9 13.9 24.9 .. .. .. 6.0 2.3 1,698 .. .. Austria 25.1 30.5 51.0 .. .. .. 5.6 2.4 1,872 .. .. Azerbaijan 24.8 0.9 13.1 84.2 18 22 0.6 0.2 8 .. .. Bangladesh 7.3 14.1 38.9 75.0 12 32 1.4 2.4 14 16 26 Belarus .. .. .. .. .. .. 4.7 1.0 57 .. .. Belgium 17.0 .. .. .. .. .. 6.2 2.5 1,936 .. .. Benin 10.3 12.1 108.2 73.6 .. .. 1.6 1.6 11 .. .. Bolivia 13.3 11.0 45.2 80.6 .. .. 4.9 1.8 67 .. .. Bosnia & Herzegovina .. .. .. .. .. .. 3.1 1.4 50 .. .. Botswana .. .. .. .. .. .. 3.8 2.2 191 .. .. Brazil 12.5 12.6 72.8 .. 18 d 25 d 3.4 4.9 267 .. .. Bulgaria 15.2 17.1 14.5 .. .. .. 3.0 0.9 59 13 25 Burkina Faso .. .. .. 69.3 .. .. 3.0 1.2 8 .. .. Burundi 10.9 66.6 923.6 77.9 .. .. 1.6 1.5 3 .. .. Cambodia 3.2 15.0 48.6 80.0 15 29 2.0 6.1 19 .. .. Cameroon 8.3 24.6 69.6 67.5 .. .. 1.1 3.2 24 .. .. Canada .. .. 46.1 .. .. .. 6.6 2.5 2,058 .. .. Central African Rep. .. .. .. 71.5 .. .. 1.4 1.5 8 .. .. Chad 9.5 28.5 423.7 65.8 .. .. 2.5 0.6 6 .. .. Chile 13.9 15.2 21.9 .. .. .. 3.1 4.1 336 .. .. China 6.1 12.1 85.8 .. .. .. 1.9 3.4 45 .. .. Hong Kong, China .. .. .. .. .. .. .. .. .. .. .. Colombia .. .. .. .. 23 14 5.4 4.2 186 27 13 Congo, Dem. Rep. .. .. .. 89.7 .. .. 1.1 0.4 9 .. .. Congo, Rep. 9.9 .. .. 79.7 .. .. 1.5 0.7 22 .. .. Costa Rica 14.9 19.4 55.7 .. 21 20 4.4 2.0 273 27 13 Côte d’Ivoire 14.7 35.7 139.6 77.5 13 35 1.0 1.7 16 11 32 Croatia .. .. .. .. .. .. 8.0 2.0 434 .. .. Czech Rep. 12.5 23.2 33.9 .. .. .. 6.6 0.6 358 .. .. Denmark 23.4 37.2 65.1 .. .. .. 6.8 1.5 2,512 .. .. Dominican Rep ... .. .. .. .. .. 1.8 4.5 151 .. .. Ecuador 4.3 8.9 .. .. 12 25 1.2 1.2 26 8 38 Egypt, Arab Rep. .. .. 39.4 .. .. .. 1.8 2.0 51 .. .. El Salvador 2.0 26.4 10.4 .. .. .. 3.8 5.0 184 .. .. Eritrea .. .. .. 70.4 .. .. 2.8 1.5 9 .. .. Estonia 24.5 30.8 33.0 .. .. .. 4.7 1.4 218 .. .. Ethiopia .. .. .. 79.5 .. .. 1.8 2.8 5 .. .. Finland 17.3 25.5 39.7 .. .. .. 5.0 1.6 1,559 .. .. France 18.0 29.3 30.3 .. .. .. 7.2 2.3 2,057 .. .. Georgia .. .. .. 84.0 .. .. 0.7 6.4 41 .. .. Germany 17.8 20.5 42.5 .. .. .. 8.0 2.6 2,422 .. .. Ghana .. .. .. 82.3 16 21 2.2 2.0 11 12 33 Greece 16.0 17.9 26.7 .. .. .. 4.6 3.7 884 .. .. Guatemala 4.9 12.1 .. .. .. .. 2.3 2.4 79 .. .. Guinea 9.5 .. .. 65.3 5 44 1.9 1.5 13 4 48 Haiti .. .. .. 90.0 .. .. 2.4 2.5 21 .. .. Honduras .. .. .. 88.0 .. .. 4.3 2.5 62 21 12 Hungary 17.7 18.7 30.5 .. .. .. 5.1 1.7 315 .. .. India 7.2 23.1 .. 76.8 .. .. 0.9 4.0 23 10 32 Indonesia 3.2 8.7 .. 80.1 15 29 0.6 2.1 19 12 29 Iran, Islamic Rep. 10.3 11.8 81.6 .. .. .. 2.5 3.0 258 .. .. Ireland 13.3 15.2 27.8 .. .. .. 5.1 1.6 1,692 .. .. Israel 21.2 22.5 31.6 .. .. .. 8.3 2.6 2,021 .. .. Italy 21.2 27.1 26.0 .. .. .. 6.0 2.1 1,498 .. .. Jamaica 16.2 26.8 80.0 .. 22 15 2.6 2.9 165 .. .. Japan 21.3 .. .. .. .. .. 6.0 1.8 2,908 .. .. Jordan 13.7 16.1 31.1 .. .. .. 4.2 3.9 137 .. .. Kazakhstan .. .. .. .. 8 26 2.7 1.0 44 .. .. Kenya 0.4 1.2 496.9 95.8 17 21 1.8 6.5 28 14e 24e Korea, Rep. 18.3 16.8 8.0 .. .. .. 2.6 3.4 584 .. .. Kuwait .. .. .. .. .. .. 2.6 0.4 586 .. .. Kyrgyz Rep. .. 18.3 32.2 78.2 14 27 2.2 2.2 12 .. .. Lao PDR 6.5 8.7 145.3 80.4 12 34 1.3 2.1 11 .. .. Latvia 23.6 25.2 22.5 .. .. .. 3.5 2.4 174 .. .. Lebanon 10.5 .. 9.3 .. .. .. 2.5 9.9 499 .. .. Lesotho 27.0 76.3 962.7 70.1 .. .. 5.2 1.1 28 .. .. Lithuania 61.4 .. 40.4 .. .. .. 4.3 1.7 185 .. .. Macedonia, FYR .. 30.6 44.8 .. 9 40 5.1 0.9 106 .. .. Madagascar 3.9 .. 76.2 57.6 8 41 2.5 1.0 9 12 30 Malawi .. .. .. 86.0 16 25 3.6 4.0 11 .. .. Note: For data comparability and coverage, see the technical notes. Figures in italics are for years other than those speci�ed. 256 (c) The International Bank for Reconstruction and Development / The World Bank Table 3 Expenditures on education and health—continued Public expenditure per student a Recurrent Incidence of Health expenditure Incidence spending education of health on primary expenditure expenditure teacher salaries b Primary Secondary Tertiary % of total lowest highest Public Private Total lowest highest % of % of % of recurrent quintile quintile % of % of per quintile quintile GDP GDP GDP spending on GDP GDP capita per per per primary $ capita capita capita education 2000 2000 2000 2000 1991–2001 c 1991–2001 c 2000 2000 1997–2000 1991–2001 c 1991–2001 c Malaysia 11.2 19.9 86.1 .. .. .. 1.5 1.0 101 .. .. Mali 13.7 .. 241.4 68.9 .. .. 2.2 2.7 10 .. .. Mauritania 11.7 36.4 .. 81.8 .. .. 3.4 0.9 14 .. .. Mexico 11.7 13.8 45.2 .. 19 21 2.5 2.9 311 .. .. Moldova 1.3 28.7 19.3 32.2 .. .. 2.9 0.6 11 .. .. Mongolia .. 40.6 26.8 85.0 .. .. 4.6 2.0 23 .. .. Morocco 20.5 49.9 102.7 .. 12 24 1.3 3.2 50 .. .. Mozambique .. .. .. 73.9 .. .. 2.7 1.6 9 .. .. Myanmar 1.6 1.9 19.4 .. .. .. 0.4 1.8 153 .. .. Namibia 20.7 34.0 147.1 .. .. .. 4.2 2.9 136 .. .. Nepal 14.2 15.6 98.7 80.0 11 46 0.9 .. .. .. .. Netherlands 15.4 21.8 43.0 .. .. .. 5.5 2.6 1,900 .. .. New Zealand 19.9 22.3 25.5 .. .. .. 6.2 1.8 1,062 .. .. Nicaragua 20.5 .. .. 67.3 11 35 2.3 2.1 43 18 18 Niger 22.3 81.0 441.0 74.1 .. .. 1.8 2.1 5 .. .. Nigeria .. .. .. 90.9 .. .. 0.5 1.7 8 .. .. Norway 29.2 .. 46.5 .. .. .. 6.6 1.2 2,832 .. .. Pakistan .. .. .. 80.7 14 29 0.9 3.2 18 .. .. Panama 15.8 24.4 47.7 .. 12 21 5.3 2.3 268 .. .. Papua New Guinea 11.1 18.0 40.4 .. .. .. 3.6 0.5 31 .. .. Paraguay .. 18.1 .. .. .. .. 3.0 4.9 112 .. .. Peru 8.0 10.6 22.0 .. 15 22 2.8 2.0 100 .. .. Philippines 14.3 12.5 23.2 .. .. .. 1.6 1.8 33 .. .. Poland 26.5 12.0 20.2 .. .. .. 4.2 1.8 246 .. .. Portugal 20.5 29.4 28.2 .. .. .. 5.8 2.4 862 .. .. Romania .. .. .. .. 22 17 1.9 1.0 48 .. .. Russian Fed. .. 20.5 15.8 .. .. .. 3.8 1.5 92 .. .. Rwanda 6.9 .. 571.6 91.4 .. .. 2.7 2.5 12 .. .. Saudi Arabia .. .. 86.9 .. .. .. 4.2 1.1 448 .. .. Senegal 13.6 33.1 244.6 63.4 .. .. 2.6 2.0 22 .. .. Serbia & Montenegro .. .. .. .. .. .. 2.9 2.7 50 .. .. Sierra Leone .. .. .. 66.9 .. .. 2.6 1.7 6 .. .. Singapore .. .. .. .. .. .. 1.2 2.3 814 .. .. Slovak Rep. 10.8 19.2 30.8 .. .. .. 5.3 0.6 210 .. .. Slovenia .. .. .. .. .. .. 6.8 1.8 788 .. .. South Africa 14.0 17.9 61.3 .. 14 35 3.7 5.1 255 16 17 Spain 18.8 25.5 19.8 .. .. .. 5.4 2.3 1,073 .. .. Sri Lanka .. .. .. .. .. .. 1.8 1.8 31 20 20 Sweden 23.5 28.3 53.5 .. .. .. 6.5 1.9 2,179 .. .. Switzerland 23.2 28.2 55.8 .. .. .. 5.9 4.8 3,573 .. .. Syrian Arab Rep. 12.9 23.3 .. .. .. .. 1.6 0.9 30 .. .. Tajikistan .. .. 9.9 .. .. .. 0.9 2.3 6 .. .. Tanzania .. .. .. 88.8 14 37 2.8 3.1 12 17 29 Thailand 12.5 12.8 38.2 .. .. .. 2.1 1.6 71 .. .. Togo 11.6 23.1 295.3 74.8 .. .. 1.5 1.3 8 .. .. Tunisia 16.2 28.4 89.8 .. .. .. 2.9 2.6 110 .. .. Turkey 17.6 11.8 72.1 .. .. .. 3.6 1.4 150 .. .. Turkmenistan .. .. .. .. .. .. 4.6 0.8 52 .. .. Uganda .. .. .. 73.8 13 32 1.5 2.4 10 .. .. Ukraine .. 21.2 28.2 .. .. .. 2.9 1.2 26 .. .. United Kingdom 14.0 14.9 26.3 .. .. .. 5.9 1.4 1,747 .. .. United States 17.9 22.4 .. .. .. .. 5.8 7.2 4,499 .. .. Uruguay 8.2 12.0 21.3 .. .. .. 5.1 5.8 653 .. .. Uzbekistan .. .. .. 73.0 .. .. 2.6 2.6 29 .. .. Venezuela, RB .. .. .. .. .. .. 2.7 2.0 233 .. .. Vietnam .. .. .. 55.0 18 21 1.3 3.9 21 12 29 Yemen, Rep. .. .. .. 73.3 19 22 2.1 2.8 20 .. .. Zambia .. .. .. 78.3 .. .. 3.5 2.1 18 .. .. Zimbabwe 13.2 20.1 200.9 75.0 .. .. 3.1 4.2 43.0 .. .. World .. m .. m .. m 5.4 w 3.9 w 482 w Low income .. .. .. 1.1 3.2 21 Middle income .. .. .. 3.0 2.9 115 Lower middle income .. .. .. 2.7 3.1 85 Upper middle income 12.4 .. .. 3.5 2.5 330 Low & middle income .. .. .. 2.7 2.9 71 East Asia & Paci�c 7.6 .. 40.1 1.8 2.9 44 Europe & Central Asia .. .. .. 4.0 1.5 108 Latin America & Carib. .. .. .. 3.3 3.7 262 Middle East & N. Africa .. .. .. 2.9 1.7 170 South Asia 7.3 .. .. 1.0 3.7 21 Sub-Saharan Africa .. .. .. 2.5 3.4 29 High income .. .. 6.0 4.2 2,735 a. Break in series between 1997 and 1998 due to change from ISCED76 to ISCED97. b. Source: Bruns, Barbara, Alain Mingat and Ramahatra Rakotomalala, 2003, “Achieving Universal Primary Education by 2015: A Chance for Every Child� (2003). Washington D.C., The World Bank, Table A.2. c. Data are for the most recent year available. d. Includes northeast and southeast Brazil only. e. Data refer to rural only. 257 (c) The International Bank for Reconstruction and Development / The World Bank Table 4 Service indicators Primary Primary Trained Health Child Tuberculosis Physicians Hospital Inpatient Access Access to teacher pupil- teachers personnel immunization treatment beds admission to an improved absence teacher in absence rate success rate improved sanitation rate ratio primary rate rate water facilities education source % of pupils % of % of total % of children % of per 1,000 per 1,000 % of % of % of total per total under age one registered people people population population population teacher cases Measles DPT 2002–2003 2000 2000 2001 2001 1999 1995–2000 a 1995–2000 a 1995–2000 a 2000 2000 Albania .. 22 .. .. 95 97 .. 1.3 3.2 .. 97 91 Algeria .. 28 93.7 .. 83 89 87 1.0 2.1 .. 89 92 Angola .. 35 .. .. 72 41 .. 0.1 .. .. 38 44 Argentina .. 22 .. .. 94 82 59 2.7 3.3 .. .. .. Armenia .. .. .. .. 93 94 88 3.2 0.7 8 .. .. Australia .. .. .. .. 93 92 84 2.5 7.9 16 100 100 Austria .. 13 .. .. 79 84 77 3.1 8.6 30 100 100 Azerbaijan .. 19 99.9 .. 99 98 88 3.6 9.7 6 78 81 Bangladesh .. 57 65.0 35 76 83 81 0.2 .. .. 97 48 Belarus .. 17 100.0 .. 99 99 .. 4.4 12.2 26 100 .. Belgium .. 12 .. .. 83 96 .. 3.9 7.3 20 .. .. Benin .. 54 65.0 .. 65 76 77 0.1 .. .. 63 23 Bolivia .. 24 74.2 .. 79 81 74 1.3 1.7 .. 83 70 Bosnia & Herzegovina .. .. .. .. 92 91 90 1.4 1.8 .. .. .. Botswana .. 27 89.2 .. 83 87 71 .. .. .. 95 66 Brazil .. 26 .. .. 99 97 11 1.3 3.1 0 87 76 Bulgaria .. 18 .. .. 96 96 .. 3.4 7.4 .. 100 100 Burkina Faso .. 47 80.4 .. 46 41 61 0.0 1.4 2 42 29 Burundi .. 50 .. .. 75 74 .. .. .. .. 78 88 Cambodia .. 53 95.9 .. 59 60 93 0.3 .. .. 30 17 Cameroon .. 63 .. .. 62 43 75 0.1 .. .. 58 79 Canada .. 15 .. .. 96 97 .. 2.1 3.9 10 100 100 Central African Rep. .. 74 .. .. 29 23 .. 0.0 .. .. 70 25 Chad .. 71 37.2 .. 36 27 .. .. .. .. 27 29 Chile .. 25 .. .. 97 97 83 1.1 2.7 .. 93 96 China .. 20 .. .. 79 79 96 1.7 2.4 4 75 38 Hong Kong, China .. .. .. .. .. .. 78 1.3 .. .. .. .. Colombia .. 26 .. .. 75 74 82 1.2 1.5 .. 91 86 Congo, Dem. Rep. .. 26 .. .. 46 40 69 0.1 .. .. 45 21 Congo, Rep. .. 51 64.6 .. 35 31 61 0.3 .. .. 51 14 Costa Rica .. 25 .. .. 82 88 81 0.9 1.7 9 95 93 Côte d’Ivoire .. 48 99.1 .. 61 57 63 0.1 .. .. 81 52 Croatia .. 18 .. .. 94 94 .. 2.3 .. .. .. .. Czech Rep. .. 18 .. .. 97 98 78 3.1 8.8 21 .. .. Denmark .. 10 .. .. 94 97 .. 3.4 4.5 20 100 .. Dominican Rep. .. 40 .. .. 98 62 81 2.2 1.5 .. 86 67 Ecuador 16 23 .. .. 99 90 75 1.7 1.6 .. 85 86 Egypt, Arab Rep. .. 22 .. .. 97 99 87 1.6 2.1 3 97 98 El Salvador .. 26 .. .. 99 99 78 1.1 1.6 .. 77 82 Eritrea .. 45 70.5 .. 88 93 44 0.0 .. .. 46 13 Estonia .. 14 .. .. 95 94 63 3.0 7.4 18 .. .. Ethiopia .. 55 70.4 .. 52 56 76 .. .. .. 24 12 Finland .. 16 .. .. 96 99 .. 3.1 7.5 27 100 100 France .. 19 .. .. 84 98 .. 3.0 8.2 23 .. .. Georgia .. 16 .. .. 73 86 61 4.4 4.8 5 79 100 Germany .. 15 .. .. 89 97 .. 3.6 9.1 24 .. .. Ghana .. 33 68.6 .. 81 80 55 0.1 .. .. 73 72 Greece .. 13 .. .. 88 88 .. 4.4 4.9 15 .. .. Guatemala .. 33 .. .. 90 82 81 0.9 1.0 .. 92 81 Guinea .. 44 .. .. 52 43 .. 0.1 .. .. 48 58 Haiti .. .. .. .. 53 43 70 0.2 0.7 .. 46 28 Honduras .. 34 .. .. 95 95 88 0.8 1.1 .. 88 75 Hungary .. 11 .. .. 99 99 .. 3.2 8.2 24 99 99 India 23 b 40 .. 43 56 64 82 .. .. .. 84 28 Indonesia 18 22 .. 42 59 60 50 .. .. .. 78 55 Iran, Islamic Rep. .. 25 96.5 .. 96 95 82 0.9 1.6 .. 92 83 Ireland .. 22 .. .. 73 84 .. 2.3 9.7 15 .. .. Israel .. 12 .. .. 94 95 .. 3.8 6.0 .. .. .. Italy .. 11 .. .. 70 95 71 6.0 4.9 18 .. .. Jamaica .. 36 .. .. 85 90 74 1.4 2.1 .. 92 99 Japan .. 20 .. .. 96 85 76 1.9 16.5 10 .. .. Jordan .. .. .. .. 99 99 88 1.7 1.8 11 96 99 Kazakhstan .. 19 .. .. 96 96 79 3.5 8.5 15 91 99 Kenya .. 30 96.6 .. 76 76 78 0.1 .. .. 57 87 Korea, Rep. .. 32 .. .. 97 99 .. 1.3 6.1 6 92 63 Kuwait .. 14 100.0 .. 99 98 .. 1.9 2.8 .. .. .. Kyrgyz Rep. .. 24 48.4 .. 99 99 83 3.0 9.5 21 77 100 Lao PDR .. 30 76.2 .. 50 40 84 0.2 .. .. 37 30 Latvia .. 15 .. .. 98 97 74 2.8 10.3 21 .. .. Lebanon .. 17 .. .. 94 93 96 2.1 2.7 17 100 99 Lesotho .. 48 74.2 .. 77 85 69 0.1 .. .. 78 49 Lithuania .. 16 .. .. 97 95 84 4.0 9.2 24 67 67 Macedonia, FYR .. 22 .. .. 92 90 .. 2.2 4.9 9 .. .. Madagascar .. 50 .. .. 55 55 .. 0.1 .. .. 47 42 Malawi .. 56 51.2 .. 82 90 71 .. 1.3 .. 57 76 Note: For data comparability and coverage, see the technical notes. Figures in italics are for years other than those speci�ed. 258 (c) The International Bank for Reconstruction and Development / The World Bank Table 4 Service indicators—continued Primary Primary Trained Health Child Tuberculosis Physicians Hospital Inpatient Access Access to teacher pupil- teachers personnel immunization treatment beds admission to an improved absence teacher in absence rate success rate improved sanitation rate ratio primary rate rate water facilities education source % of pupils % of % of total % of children % of per 1,000 per 1,000 % of % of % of total per total under age one registered people people population population population teacher cases Measles DPT 2002–2003 2000 2000 2001 2001 1999 1995–2000 a 1995–2000 a 1995–2000 a 2000 2000 Malaysia .. 18 .. .. 92 97 90 0.7 2.0 .. .. .. Mali .. 63 .. .. 37 51 68 0.1 0.2 1 65 69 Mauritania .. 42 .. .. 58 61 .. 0.1 .. .. 37 33 Mexico .. 27 .. .. 97 97 80 1.8 1.1 6 88 74 Moldova .. 20 .. .. 81 90 .. 3.5 12.1 19 92 99 Mongolia .. 32 92.9 .. 95 95 86 2.4 .. .. 60 30 Morocco .. 28 .. .. 96 96 88 0.5 1.0 3 80 68 Mozambique .. 64 61.8 .. 92 80 71 .. .. .. 57 43 Myanmar .. 32 85.4 .. 73 72 81 0.3 .. .. 72 64 Namibia .. 32 36.0 .. 58 63 50 0.3 .. .. 77 41 Nepal .. 37 44.5 .. 71 72 87 0.0 0.2 .. 88 28 Netherlands .. 10 .. .. 96 97 79 3.2 10.8 10 100 100 New Zealand .. 16 .. .. 85 90 .. 2.2 6.2 13 .. .. Nicaragua .. 36 .. .. 99 92 81 0.9 1.5 .. 77 85 Niger .. 42 84.1 .. 51 31 60 0.0 0.1 28 59 20 Nigeria .. .. .. .. 40 26 75 .. .. .. 62 54 Norway .. .. .. .. 93 95 77 2.9 14.6 17 100 .. Pakistan .. .. .. .. 54 56 70 0.6 .. .. 90 62 Panama .. 25 79.0 .. 97 98 80 1.7 2.2 .. 90 92 Papua New Guinea 15 36 .. 19 58 56 66 0.1 .. .. 42 82 Paraguay .. 20 .. .. 77 66 .. 1.1 1.3 .. 78 94 Peru 13 25 .. 26 97 85 93 0.9 1.5 1 80 71 Philippines .. 35 .. .. 75 70 87 1.2 .. .. 86 83 Poland .. 11 .. .. 97 98 69 2.2 4.9 16 .. .. Portugal .. 13 .. .. 87 96 85 3.2 4.0 12 .. .. Romania .. 20 .. .. 98 99 78 1.8 7.6 18 58 53 Russian Fed. .. 17 .. .. 98 96 65 4.2 12.1 22 99 .. Rwanda .. 51 .. .. 78 86 67 .. .. .. 41 8 Saudi Arabia .. 12 .. .. 94 97 66 1.7 2.3 11 95 100 Senegal .. 51 100.0 .. 48 52 .. 0.1 0.4 .. 78 70 Serbia & Montenegro .. 20 100.0 .. 90 93 .. 2.0 5.3 .. 98 100 Sierra Leone .. 44 78.9 .. 37 44 75 0.1 .. .. 57 66 Singapore .. .. .. .. 89 92 95 1.6 .. .. 100 100 Slovak Rep. .. 19 .. .. 99 99 79 3.5 7.1 20 100 100 Slovenia .. 14 .. .. 98 92 88 2.3 5.7 .. 100 .. South Africa .. 33 67.9 .. 72 81 60 0.6 .. .. 86 87 Spain .. 14 .. .. 94 95 .. 3.3 4.1 12 .. .. Sri Lanka .. .. .. .. 99 99 84 0.4 .. .. 77 94 Sweden .. 11 .. .. 94 99 .. 2.9 3.6 18 100 100 Switzerland .. 14 .. .. 81 95 .. 3.5 17.9 15 100 100 Syrian Arab Rep. .. 24 92.2 .. 93 92 84 1.3 1.4 .. 80 90 Tajikistan .. 22 .. .. 86 83 .. 2.0 .. .. 60 90 Tanzania .. 40 44.1 .. 83 85 78 0.0 .. .. 68 90 Thailand .. 21 .. .. 94 96 77 0.4 2.0 .. 84 96 Togo .. 34 80.0 .. 58 64 76 0.1 .. .. 54 34 Tunisia .. 23 .. .. 92 96 91 0.7 1.7 .. 80 84 Turkey .. .. .. .. 90 88 .. 1.3 2.6 8 82 90 Turkmenistan .. .. .. .. 98 95 .. 3.0 .. .. .. .. Uganda 26 59 45.0 35 61 60 61 .. .. .. 52 79 Ukraine .. 20 .. .. 99 99 .. 3.0 11.8 20 98 99 United Kingdom .. 18 .. .. 85 94 .. 1.8 4.1 15 100 100 United States .. 15 .. .. 91 94 76 2.8 3.6 12 100 100 Uruguay .. 21 .. .. 94 94 83 3.7 4.4 .. 98 94 Uzbekistan .. .. .. .. 99 97 79 3.1 8.3 .. 85 89 Venezuela, RB .. .. .. .. 49 70 82 2.4 1.5 .. 83 68 Vietnam .. 28 84.9 .. 97 98 92 0.5 1.7 8 77 47 Yemen, Rep. .. 30 .. .. 79 76 83 0.2 0.6 .. 69 38 Zambia 17 45 100.0 .. 85 78 .. 0.1 .. .. 64 78 Zimbabwe .. 37 .. .. 68 75 73 0.1 .. .. 83 62 World 27 m .. m 72 w 73 w .. w .. w 9w 81 w 55 w Low income 39 78.9 59 61 .. .. 76 43 Middle income 21 .. 86 85 1.9 3.3 6 82 60 Lower middle income 21 .. 85 84 1.9 3.3 6 81 58 Upper middle income 21 .. 91 92 1.8 3.3 11 .. .. Low & middle income 29 .. 71 71 .. .. 79 51 East Asia & Paci�c 21 .. 76 77 1.7 2.4 4 76 46 Europe & Central Asia .. .. 95 94 3.1 8.9 18 91 .. Latin America & Carib. 26 .. 91 89 1.5 2.2 2 86 77 Middle East & N. Africa 24 .. 92 92 .. .. 88 85 South Asia 42 66.5 58 65 .. .. 84 34 Sub-Saharan Africa 47 78.9 58 53 .. .. 58 53 High income 17 .. 90 94 3.0 7.4 15 .. .. a. Data are for the most recent year available. b. Average for 14 states. 259 (c) The International Bank for Reconstruction and Development / The World Bank Table 5 Foreign aid recipient indicators Net of�cial development Aid per Aid dependency ratios Donor assistance or of�cial aid capita fragmentation index $ millions $ Aid as Aid as % Aid as % Aid as % % of GNI of gross of imports of central capital of goods government formation and services expenditure 1996 2001 1996 2001 1996 2001 1996 2001 1996 2001 1996 2001 Albania 228 269 72 85 8.3 6.3 54.7 33.6 20.3 15.0 28.5 .. 0.9 Algeria 304 182 11 6 0.7 0.3 2.6 1.3 2.5 1.4 2.2 1.1 0.7 Angola 473 268 40 20 8.1 3.4 18.1 8.3 7.9 3.2 .. .. 0.9 Argentina 135 151 4 4 0.1 0.1 0.3 0.4 0.3 0.4 0.3 0.3 0.9 Armenia 293 212 90 69 18.3 9.7 91.8 53.8 31.8 20.9 .. .. 0.7 Australia .. .. .. .. .. .. .. .. .. .. .. .. .. Austria .. .. .. .. .. .. .. .. .. .. .. .. .. Azerbaijan 96 226 12 28 3.1 4.2 10.5 18.9 5.3 8.9 18.1 16.4 0.8 Bangladesh 1,236 1,024 10 8 3.0 2.1 15.2 9.4 15.8 9.8 .. 21.4 0.9 Belarus 77 39 8 4 0.5 0.3 2.2 1.4 1.0 0.4 1.6 1.1 0.8 Belgium .. .. .. .. .. .. .. .. .. .. .. .. .. Benin 288 273 51 42 13.3 11.6 76.3 60.1 36.1 36.0 .. .. 0.9 Bolivia 832 729 110 86 11.6 9.4 69.2 70.5 42.3 31.3 48.9 34.2 0.9 Bosnia & Herzegovina 845 639 239 157 33.5 12.7 73.6 111.1 33.8 23.8 .. .. 0.9 Botswana 75 29 48 17 1.6 0.6 6.2 2.5 2.9 1.0 4.3 .. 0.9 Brazil 288 349 2 2 0.0 0.1 0.2 0.3 0.3 0.4 .. .. 0.8 Bulgaria 182 346 22 44 1.9 2.6 22.6 12.5 2.8 3.7 3.8 7.4 0.7 Burkina Faso 420 389 41 34 16.9 15.7 61.8 61.7 55.0 57.4 .. .. 0.8 Burundi 111 131 18 19 12.5 19.3 102.3 274.3 69.9 80.7 44.6 39.8 0.9 Cambodia 422 409 38 33 13.6 12.4 51.8 66.9 30.5 20.1 .. .. 0.9 Cameroon 412 398 30 26 4.8 4.9 29.5 26.0 16.7 13.3 .. 31.1 0.8 Canada .. .. .. .. .. .. .. .. .. .. .. .. .. Central African Rep. 170 76 49 20 16.2 7.9 369.9 56.0 70.7 49.5 .. .. 0.8 Chad 296 179 43 23 18.8 11.3 123.7 26.9 57.1 18.1 .. .. 0.9 Chile 196 58 14 4 0.3 0.1 1.1 0.4 0.8 0.2 1.4 0.4 0.8 China 2,646 1,460 2 1 0.3 0.1 0.8 0.3 1.5 0.5 4.1 2.2 0.7 Hong Kong, China 13 4 2 1 0.0 0.0 0.0 0.0 0.0 0.0 .. .. 0.6 Colombia 189 380 5 9 0.2 0.5 0.9 3.1 1.0 2.0 1.3 1.9 0.7 Congo, Dem. Rep. 166 251 4 5 3.1 5.3 10.3 95.1 9.0 18.1 .. .. 0.9 Congo, Rep. 429 75 160 24 26.4 3.8 62.7 10.0 17.6 3.4 56.8 10.5 0.7 Costa Rica –10 2 –3 1 –0.1 0.0 –0.5 0.1 –0.2 0.0 –0.4 0.1 0.9 Côte d’Ivoire 965 187 67 11 8.6 1.9 65.6 18.2 19.3 4.3 35.6 10.6 0.7 Croatia 133 113 29 26 0.7 0.6 3.1 2.3 1.3 1.0 1.5 1.3 0.9 Czech Rep. 129 314 12 31 0.2 0.6 0.6 1.8 0.4 0.7 0.6 1.4 0.8 Denmark .. .. .. .. .. .. .. .. .. .. .. .. .. Dominican Rep. 100 105 13 12 0.8 0.5 3.9 2.1 1.3 0.9 4.8 2.0 0.8 Ecuador 253 171 22 13 0.2 0.9 0.8 3.2 4.1 2.1 .. .. 0.9 Egypt, Arab Rep. 2,199 1,255 37 19 3.2 1.3 19.6 8.2 11.6 5.6 10.0 .. 0.7 El Salvador 302 234 52 37 2.9 1.7 19.3 10.7 8.2 3.7 .. 66.9 0.8 Eritrea 159 280 43 67 24.6 40.9 71.6 115.2 27.3 52.3 .. .. 0.9 Estonia 59 69 42 50 1.4 1.3 4.9 4.5 1.7 1.2 4.0 4.1 0.8 Ethiopia 818 1,080 14 16 13.7 17.5 80.6 95.9 55.9 53.6 .. 39.3 0.9 Finland .. .. .. .. .. .. .. .. .. .. .. .. .. France .. .. .. .. .. .. .. .. .. .. .. .. .. Georgia 310 290 58 55 10.3 9.0 93.3 48.9 .. 21.5 .. 82.7 0.7 Germany .. .. .. .. .. .. .. .. .. .. .. .. .. Ghana 651 652 37 33 9.6 12.6 32.2 51.2 25.5 19.2 .. .. 0.9 Greece .. .. .. .. .. .. .. .. .. .. .. .. .. Guatemala 194 225 19 19 1.2 1.1 9.7 7.1 5.1 3.5 .. .. 0.8 Guinea 299 272 44 36 7.9 9.2 44.6 41.3 28.4 27.4 .. 32.6 0.9 Haiti 370 166 50 20 12.8 4.4 45.2 14.4 46.8 13.2 140.8 54.2 0.8 Honduras 359 678 62 103 9.4 10.8 28.2 34.7 14.1 18.3 .. .. 0.9 Hungary 204 418 20 41 0.5 0.8 1.7 3.0 0.9 1.1 1.0 1.9 0.8 India 1,897 1,705 2 2 0.5 0.4 2.3 1.6 3.2 2.2 3.3 2.0 0.8 Indonesia 1,123 1,501 6 7 0.5 1.1 1.6 4.9 1.7 2.5 3.4 4.3 0.7 Iran, Islamic Rep. 169 115 3 2 0.2 0.1 0.8 0.3 0.9 0.5 0.5 0.2 0.7 Ireland .. .. .. .. .. .. .. .. .. .. .. .. .. Israel 2,217 172 389 27 2.3 0.8 9.4 3.7 5.2 0.3 4.7 0.3 0.1 Italy .. .. .. .. .. .. .. .. .. .. .. .. .. Jamaica 58 54 23 21 0.9 0.7 3.1 2.3 1.4 1.0 2.2 1.8 0.9 Japan .. .. .. .. .. .. .. .. .. .. .. .. .. Jordan 507 432 117 86 7.6 4.9 24.0 18.9 8.7 6.7 21.6 15.1 0.8 Kazakhstan 125 148 8 10 0.6 0.7 3.7 2.6 1.6 1.3 .. 4.6 0.7 Kenya 597 453 22 15 6.6 4.0 38.4 31.1 16.1 10.8 22.3 .. 0.9 Korea, Rep. –149 –111 –3 –2 0.0 0.0 –0.1 –0.1 –0.1 –0.1 –0.2 .. 0.5 Kuwait 3 4 1 2 0.0 0.0 0.1 0.1 0.0 0.0 0.0 0.1 0.8 Kyrgyz Rep. 231 188 50 38 12.9 12.9 50.1 68.5 21.4 29.2 56.5 69.5 0.8 Lao PDR 332 243 69 45 17.8 14.6 61.2 62.9 42.5 40.6 .. .. 0.8 Latvia 72 106 29 45 1.4 1.4 7.5 5.1 2.3 2.4 4.5 4.8 0.9 Lebanon 232 241 57 55 1.7 1.4 6.0 8.6 2.9 .. 4.7 3.3 0.8 Lesotho 104 54 55 26 8.2 5.5 18.9 18.4 8.9 6.9 21.9 .. 0.9 Lithuania 91 130 25 37 1.2 1.1 4.7 5.0 1.8 1.8 4.6 4.1 0.8 Macedonia, FYR 106 248 53 122 2.4 7.3 11.9 39.4 5.7 12.4 .. .. 0.9 Madagascar 357 354 26 22 9.3 7.8 76.7 49.6 30.5 188.8 51.4 48.7 0.9 Malawi 492 402 52 38 20.5 23.4 174.9 210.2 42.8 38.3 .. .. 0.9 Note: For data comparability and coverage, see the technical notes. Figures in italics are for years other than those speci�ed. 260 (c) The International Bank for Reconstruction and Development / The World Bank Table 5 Foreign aid recipient indicators—continued Net of�cial development Aid per Aid dependency ratios Donor assistance or of�cial aid capita fragmentation index $ millions $ Aid as Aid as % Aid as % Aid as % % of GNI of gross of imports of central capital of goods government formation and services expenditure 1996 2001 1996 2001 1996 2001 1996 2001 1996 2001 1996 2001 Malaysia –457 27 –22 1 –0.5 0.0 –1.1 0.1 –0.5 0.0 –2.1 .. 0.3 Mali 491 350 50 32 19.1 13.9 81.9 62.7 49.1 28.4 .. .. 0.8 Mauritania 272 262 116 95 25.7 26.6 131.3 97.4 43.7 56.6 .. .. 0.8 Mexico 287 75 3 1 0.1 0.0 0.4 0.1 0.2 0.0 0.6 –0.1 0.8 Moldova 36 119 8 28 2.1 7.5 8.9 40.2 2.8 9.7 7.6 35.4 0.8 Mongolia 201 212 87 88 19.4 20.5 71.8 67.5 33.5 28.7 90.8 65.9 0.7 Morocco 650 517 24 18 1.8 1.6 9.1 6.1 5.3 3.8 .. 5.9 0.8 Mozambique 888 935 55 52 33.2 28.2 149.7 70.8 74.9 20.3 .. .. 0.9 Myanmar 43 127 1 3 .. .. .. .. 1.8 4.1 0.3 0.3 0.7 Namibia 188 109 116 61 5.3 3.4 23.3 14.4 8.0 5.1 14.8 12.3 0.9 Nepal 391 388 19 16 8.6 6.7 31.8 28.8 23.8 19.1 51.0 39.4 0.9 Netherlands .. .. .. .. .. .. .. .. .. .. .. .. .. New Zealand .. .. .. .. .. .. .. .. .. .. .. .. .. Nicaragua 934 928 205 178 58.4 .. 180.0 .. 57.2 41.3 137.9 84.7 0.9 Niger 255 249 27 22 13.0 12.9 132.7 111.0 51.2 47.9 .. .. 0.8 Nigeria 190 185 2 1 0.6 0.5 3.8 2.2 1.3 1.1 .. .. 0.9 Norway .. .. .. .. .. .. .. .. .. .. .. .. .. Pakistan 884 1,938 7 14 1.4 3.4 7.3 20.7 5.1 13.1 6.2 16.2 0.8 Panama 49 28 18 10 0.6 0.2 1.8 0.8 0.5 0.3 2.2 0.6 0.8 Papua New Guinea 381 203 82 39 7.6 7.2 32.2 33.8 13.9 11.0 27.1 20.2 0.5 Paraguay 89 61 18 11 0.9 0.9 3.9 3.6 1.7 2.0 5.9 4.8 0.6 Peru 329 451 14 17 0.6 0.9 2.6 4.5 2.6 4.0 3.3 4.6 0.7 Philippines 901 577 13 7 1.0 0.8 4.5 4.6 2.0 1.5 5.9 4.2 0.5 Poland 1,167 966 30 25 0.9 0.5 4.1 2.5 2.7 1.6 2.1 1.5 0.8 Portugal .. .. .. .. .. .. .. .. .. .. .. .. .. Romania 233 648 10 29 0.7 1.6 2.6 7.4 1.8 3.8 2.1 5.3 0.8 Russian Fed. 1,282 1,110 9 8 0.3 0.4 1.2 1.6 1.3 1.3 .. 1.5 0.6 Rwanda 467 291 82 37 34.1 17.3 234.9 92.7 120.6 62.0 .. .. 0.9 Saudi Arabia 23 27 1 1 0.0 0.0 0.1 0.1 0.0 0.1 .. .. 0.7 Senegal 580 419 68 43 12.7 9.2 67.5 45.0 32.0 21.7 58.8 41.6 0.8 Serbia & Montenegro a 70 1,306 7 123 .. 11.3 .. 89.2 1.6 25.0 .. .. 0.9 Sierra Leone 184 334 40 65 20.0 45.8 195.2 563.9 51.2 110.1 132.3 52.5 0.8 Singapore 15 1 4 0 0.0 0.0 0.0 0.0 0.0 0.0 0.1 0.0 0.7 Slovak Rep. 98 164 18 30 0.5 0.8 1.3 2.5 0.7 0.9 1.1 2.1 0.9 Slovenia 82 126 41 63 0.4 0.7 1.9 2.6 0.8 1.1 1.1 1.7 0.8 South Africa 364 428 9 10 0.3 0.4 1.5 2.5 1.0 1.2 0.8 1.3 0.9 Spain .. .. .. .. .. .. .. .. .. .. .. .. .. Sri Lanka 487 330 28 18 3.6 2.1 14.4 9.6 7.5 4.4 12.6 8.0 0.7 Sweden .. .. .. .. .. .. .. .. .. .. .. .. .. Switzerland .. .. .. .. .. .. .. .. .. .. .. .. .. Syrian Arab Rep. 219 153 15 9 1.6 0.8 6.6 3.7 3.1 2.1 1.6 1.3 0.7 Tajikistan 103 159 17 25 10.5 15.5 44.0 124.0 11.7 18.4 .. 129.1 0.8 Tanzania 877 1,233 29 36 13.8 13.3 81.2 77.7 38.6 53.5 .. .. 0.9 Thailand 830 281 14 5 0.5 0.2 1.1 1.0 0.9 0.4 2.8 1.2 0.2 Togo 157 47 39 10 10.9 3.8 57.1 17.9 18.7 6.8 .. .. 0.8 Tunisia 124 378 14 39 0.7 2.0 2.5 6.9 1.3 3.3 1.9 3.6 0.7 Turkey 238 167 4 2 0.1 0.1 0.5 0.7 0.4 0.3 0.5 0.2 0.7 Turkmenistan 24 72 5 13 1.0 1.2 .. 3.3 1.2 2.3 .. .. 0.6 Uganda 676 783 34 34 11.3 14.1 69.6 68.9 40.5 48.3 .. 64.6 0.9 Ukraine 398 519 8 11 0.9 1.4 3.9 6.4 1.8 2.4 .. 4.7 0.7 United Kingdom .. .. .. .. .. .. .. .. .. .. .. .. .. United States .. .. .. .. .. .. .. .. .. .. .. .. .. Uruguay 35 15 11 5 0.1 0.1 1.1 0.6 0.8 0.3 0.6 0.3 0.8 Uzbekistan 88 153 4 6 0.6 1.4 2.2 6.6 1.8 4.5 .. .. 0.7 Venezuela, RB 38 45 2 2 0.1 0.0 0.3 0.2 0.2 0.2 0.3 0.1 0.8 Vietnam 939 1,435 13 18 3.9 4.4 13.6 14.1 7.3 7.7 16.5 18.0 0.7 Yemen, Rep. 247 426 16 24 4.8 5.0 18.6 25.9 7.0 9.1 10.8 23.0 0.9 Zambia 610 374 66 36 19.9 10.7 145.1 51.2 35.8 20.9 .. .. 0.9 Zimbabwe 371 159 31.8 12 4.5 1.8 23.4 22.5 10.5 7.3 12.5 .. 0.9 World 62,264 s 58,244 s 11 w 10 w 0.2 w 0.2 w 0.9 w 0.9 w 0.8 w 0.6 w .. w .. w Low income 24,618 24,611 11 10 2.6 2.5 10.4 10.9 9.4 8.4 .. .. Middle income 22,401 21,006 9 8 0.4 0.4 1.7 1.7 1.6 1.3 .. .. Lower middle income 18,557 17,145 8 7 0.5 0.5 1.9 1.9 2.3 1.8 .. .. Upper middle income 3,175 3,336 10 10 0.3 0.2 1.0 0.9 0.6 0.5 .. .. Low & middle income 58,925 57,208 12 11 1.0 0.9 3.9 3.9 3.6 2.9 .. .. East Asia & Paci�c 8,039 7,394 5 4 0.6 0.5 1.4 1.3 1.6 1.2 .. .. Europe & Central Asia 8,670 9,783 18 21 0.8 1.0 3.3 4.4 2.4 2.3 .. .. Latin America & Carib. 7,430 5,985 15 12 0.4 0.3 1.8 1.6 1.9 1.2 .. .. Middle East & N. Africa 5,884 4,836 22 16 1.0 0.7 5.0 3.3 3.6 2.7 .. .. South Asia 5,169 5,871 4 4 1.0 1.0 4.7 4.3 5.5 5.1 .. .. Sub-Saharan Africa 16,552 13,933 28 21 5.2 4.6 27.3 24.6 14.2 11.0 .. .. High income 3,339 1,036 4 1 0.0 0.0 0.1 0.0 0.1 0.0 .. .. Note: Regional aggregates include data for economies not speci�ed elsewhere. World and income group totals include aid not allocated by country or region. The 2001 data exclude aid from the World Food Programme. a. Aid to the states of the former Socialist Federal Republic of Yugoslavia that is not otherwise speci�ed is included in regional and income group aggregates. 261 (c) The International Bank for Reconstruction and Development / The World Bank Table 6 Aid flows from Development Assistance Committee members Net flows to part I countries Net of�cial development assistance Untied aid a $ millions % of GNI annual average Per capita % of general % of % change in of donor government bilateral ODA volume b country b disbursements commitments 1995–96 to $ $ 1996 2001 1996 2001 2000–2001 1996 2001 1996 2001 1996 2001 Australia 1,074 873 0.27 0.25 0.6 46 49 0.76 0.74 78.1 59.3 Austria 557 533 0.24 0.29 0.2 51 66 0.46 0.57 .. .. Belgium 913 867 0.34 0.37 3.5 67 85 0.68 0.82 .. 89.8 Canada 1,795 1,533 0.32 0.22 –2.6 59 51 0.68 0.57 31.5 31.7 Denmark 1,772 1,634 1.04 1.03 4.4 265 306 1.72 2.00 61.3 93.3 Finland 408 389 0.33 0.32 5.0 61 75 0.59 0.72 60.2 87.5 France 7,451 4,198 0.48 0.32 –6.6 95 72 0.93 0.66 38.7 66.6 Germany 7,601 4,990 0.32 0.27 –1.2 67 62 0.67 0.59 60.0 84.6 Greece 184 202 0.15 0.17 24.3 14 19 0.33 0.40 .. 17.3 Ireland 179 287 0.31 0.33 11.9 43 74 0.67 0.92 .. 100.0 Italy 2,416 1,627 0.20 0.15 –2.3 34 28 0.38 0.32 .. 7.8 Japan 9,439 9,847 0.20 0.23 3.0 73 89 0.58 0.64 98.9 81.1 Luxembourg 82 141 0.44 0.82 18.1 156 325 1.05 1.89 94.4 .. Netherlands 3,246 3,172 0.81 0.82 5.0 161 195 1.73 1.97 82.2 91.2 New Zealand 122 112 0.21 0.25 5.6 22 30 0.49 0.61 .. .. Norway 1,311 1,346 0.84 0.83 1.7 278 299 1.82 1.95 88.4 98.9 Portugal 218 268 0.21 0.25 6.7 18 26 0.47 0.58 100.0 57.7 Spain 1,251 1,737 0.22 0.30 7.3 25 43 0.50 0.79 0.0 68.9 Sweden 1,999 1,666 0.84 0.81 4.4 173 207 1.27 1.52 78.9 86.5 Switzerland 1,026 908 0.34 0.34 3.0 108 123 .. .. 92.9 96.1 United Kingdom 3,199 4,579 0.27 0.32 5.8 58 80 0.66 0.84 86.1 93.9 United States 9,377 11,429 0.12 0.11 3.2 38 39 0.37 0.36 28.4 .. Total or average 55,622 52,336 0.25 0.22 1.8 59 63 0.63 0.61 71.3 79.1 Net flows to part II countries Net of�cial development aid $ millions % of GNI annual average Per capita % change in of donor volume b country b 1995–96 to $ $ 1996 2001 1996 2001 2000–2001 1996 2001 Australia 10 5 0.00 0.00 2.8 0 0 Austria 226 212 0.10 0.11 0.7 21 26 Belgium 70 88 0.03 0.04 7.0 5 9 Canada 181 152 0.03 0.02 –5.4 6 5 Denmark 120 181 0.07 0.11 10.3 18 34 Finland 57 61 0.05 0.05 3.6 9 12 France 711 1,334 0.05 0.10 22.4 9 23 Germany 1,329 687 0.06 0.04 –20.0 12 8 Greece 2 9 0.00 0.01 66.2 0 1 Ireland 1 0 0.00 0.00 –61.8 0 0 Italy 294 281 0.02 0.03 7.0 4 5 Japan 184 84 0.00 0.00 –35.9 1 1 Luxembourg 2 9 0.01 0.05 12.5 4 20 Netherlands 13 214 0.00 0.06 16.8 1 13 New Zealand 0 0 0.00 0.00 –1.4 0 0 Norway 50 32 0.03 0.02 –11.0 11 7 Portugal 18 28 0.02 0.03 10.8 1 3 Spain 98 14 0.02 0.00 –31.5 2 0 Sweden 178 119 0.07 0.06 –0.5 15 15 Switzerland 97 63 0.03 0.02 –3.7 10 9 United Kingdom 362 461 0.03 0.03 1.8 7 8 United States 1,694 1,542 0.02 0.02 4.6 7 5 Total or average 5,696 5,574 0.03 0.02 0.2 6 7 a. Excluding administrative costs and technical cooperation. b. At 2000 exchange rates and prices. 262 (c) The International Bank for Reconstruction and Development / The World Bank Table 7 Key indicators for other economies Population Surface Gross national income Gross domestic product Life Reduce Education area expectancy child at birth mortality Thousands Thousands $ Per $ PPP % growth Per years Under-�ve Primary Adult sq. km millions capita millions Per capita mortality completion illiteracy $ capita % growth rate rate rate $ per 1,000 % of % ages relevant 15 and age group above 2002 2002 2002 a 2002 a 2002 b 2002 b 2001–2002 2001–2002 2001 2001 1995–2001 c 2001 d e Afghanistan 27,963 652 .. .. .. .. .. .. 43 257 8 .. American Samoa 70 0.2 .. .. f .. .. .. .. .. .. .. .. Andorra 70 0.5 .. .. g .. .. .. .. .. 7 .. .. Antigua & Barbuda 69 0.4 647 9,390 686 9,960 2.7 2.1 .. 14 .. .. Aruba 90 0.2 .. .. g .. .. .. .. .. .. .. .. Bahamas, The 314 13.9 4,533 14,860 4,867 15,960 .. .. 70 16 .. 5 Bahrain 672 0.7 7,246 11,130 10,350 15,900 .. .. 73 16 91 12 Barbados 269 0.4 2,614 9,750 4,173 15,560 .. .. 75 14 .. 0h Belize 253 23.0 750 2,960 1,352 5,340 3.7 1.2 74 40 82 7 Bermuda 60 0.1 .. .. g .. .. .. .. .. .. .. .. Bhutan 851 47.0 505 590 .. .. 7.7 4.8 63 95 59 .. Brunei 351 5.8 .. .. g .. .. .. .. 76 6 .. 8 Cape Verde 458 4.0 590 1,290 2,164 i 4,720 i 4.0 1.4 69 38 117 25 Cayman Islands 35 0.3 .. .. g .. .. .. .. .. .. .. .. Channel Islands 149 0.2 .. .. g .. .. .. .. 79 .. .. .. Comoros 586 2.2 228 390 959 1,640 3.0 0.5 61 79 .. 44 Cuba 11,263 110.9 .. .. j .. .. .. .. 77 9 .. 3 Cyprus 765 9.3 9,372 12,320 13,798 i 18,040 i 2.0 1.4 78 6 .. 3 Djibouti 657 23.2 590 900 1,361 2,070 1.6 –0.3 45 143 30 35 Dominica 72 0.8 228 3,180 348 4,840 –2.8 –2.7 76 15 103 .. Equatorial Guinea 481 28.1 327 700 2,689 5,590 0.2 –2.4 51 153 .. 16 Faeroe Islands 50 1.4 .. .. g .. .. .. .. .. .. .. .. Fiji 823 18.3 1,775 2,160 4,371 5,310 4.4 3.6 69 21 .. 7 French Polynesia 240 4.0 3,794 16,150 5,725 24,360 .. .. 73 12 .. .. Gabon 1,291 267.7 4,028 3,120 6,870 5,320 3.0 0.6 53 90 80 .. Gambia, The 1,376 11.3 392 280 2,316 i 1,680 i –0.6 –3.1 53 126 70 62 Greenland 60 341.7 .. .. g .. .. .. .. .. .. .. .. Grenada 102 0.3 356 3,500 644 6,330 –0.5 –1.8 73 25 106 .. Guam 159 0.6 .. .. g .. .. .. .. 78 9 .. .. Guinea-Bissau 1,253 36.1 193 150 935 750 –4.2 –6.3 45 211 31 60 Guyana 772 215.0 651 840 2,919 3,780 0.3 –0.4 63 72 89 1 Iceland 284 103.0 7,944 27,970 8,118 28,590 0.0 –0.7 80 4 .. .. Iraq 24,256 438.3 .. .. j .. .. .. .. 62 133 .. 60 Isle of Man 80 0.6 .. .. f .. .. .. .. .. .. .. .. Kiribati 95 0.7 77 810 .. .. 2.8 0.7 62 69 .. .. Korea, Dem. Rep. 22,519 120.5 .. .. e .. .. .. .. 61 55 .. .. Liberia 3,295 111.4 489 150 .. .. 4.2 1.6 47 235 .. 45 Libya 5,534 1,759.5 .. .. f .. .. .. .. 72 19 .. 19 Liechtenstein 30 0.2 .. .. g .. .. .. .. .. 11 .. .. Luxembourg 444 2.6 17,221 38,830 22,644 51,060 0.8 0.2 77 5 .. .. k Macao, China 443 .. 6,329 14,380 k 8,349 i 18,970 i .. .. 79 .. .. 6 Maldives 287 0.3 598 2,090 .. .. 2.3 0.0 69 77 .. 3 Malta 397 0.3 3,632 9,200 6,634 16,790 .. .. 78 5 .. 8 Marshall Islands 53 0.2 125 2,350 .. .. 4.0 .. 65 66 .. .. Mauritius 1,212 2.0 4,669 3,850 12,764 10,530 4.4 3.3 72 19 111 15 Mayotte 145 0.4 .. .. f .. .. .. .. .. .. .. .. Micronesia, Fed. Sts. 122 0.7 242 1,980 .. .. 2.0 0.2 68 24 .. .. Monaco 30 0.0 .. .. g .. .. .. .. .. 5 .. .. Netherlands Antilles 220 0.8 .. .. g .. .. .. .. .. .. .. 3 New Caledonia 220 18.6 2,989 14,050 4,670 21,960 .. .. 73 10 .. .. N. Mariana Islands 80 0.5 .. .. g .. .. .. .. .. .. .. .. Oman 2,539 309.5 19,137 7,720 32,788 12,910 2.2 –0.3 74 13 76 27 Palau 20 0.5 142 7,140 .. .. 3.0 .. .. 29 .. .. Puerto Rico 3,869 9.0 42,052 10,950 60,679 15,800 .. .. 76 .. .. 6 Qatar 610 11.0 .. .. g .. .. .. .. 75 16 44 18 Samoa 176 2.8 250 1,420 942 5,350 1.3 0.0 69 25 99 1 San Marino 30 0.1 .. .. g .. .. .. .. .. 6 .. .. São Tomé & Principe 154 1.0 45 290 .. .. 3.0 0.9 65 74 84 .. Seychelles 84 0.5 538 6,530 .. .. –2.4 –3.8 73 17 .. .. Solomon Islands 443 28.9 254 570 672 i 1,520 i –4.0 –6.7 69 24 .. .. Somalia 9,391 637.7 .. .. e .. .. .. .. 47 225 .. .. St. Kitts & Nevis 46 0.4 293 6,370 450 9,780 –4.3 –6.3 71 24 110 .. St. Lucia 159 0.6 609 3,840 792 5,000 –0.5 –1.6 72 19 106 .. St. Vincent & Grenadines 117 0.4 329 2,820 595 5,100 0.7 0.0 73 25 84 .. Sudan 32,365 2,505.8 11,471 350 54,561 1,690 10.6 8.3 58 107 46 41 Suriname 423 163.3 828 1,960 .. .. 2.7 2.0 70 32 .. .. Swaziland 1,088 17.4 1,285 1,180 4,928 4,530 1.8 –0.1 45 149 81 20 Timor-Leste .. 14.9 402 .. e .. .. .. .. .. 124 54 .. Tonga 101 0.8 143 1,410 641 6,340 1.6 1.1 71 20 .. .. Trinidad & Tobago 1,318 5.1 8,553 6,490 11,446 8,680 2.7 2.0 72 20 81 2 United Arab Emirates 3,049 83.6 .. .. g .. .. .. .. 75 9 80 23 Vanuatu 206 12.2 221 1,080 569 2,770 –0.3 –2.4 68 42 .. .. Virgin Islands (U.S.) 110 0.3 .. .. g .. .. .. .. 78 11 .. .. West Bank & Gaza 3,212 .. 2,982 930 .. .. –19.1 –22.2 72 25 .. .. a. Preliminary World Bank estimates calculated using the World Bank Atlas method. b. Purchasing power parity; see the technical notes. c. Data are for the most recent year available. d. Esti- mate does not account for recent refugee flows. e. Estimated to be low income ($735 or less). f. Estimated to be upper middle income ($2,936 to $9,075). g. Estimated to be high income ($9,076 or more). h. Less than 0.5. i. The estimate is based on regression; others are extrapolated from the latest International Comparison Programme benchmark estimates. j. Estimated to be lower mid- dle income ($736 to $2,935). k. Refers to GDP and GDP per capita. 263 (c) The International Bank for Reconstruction and Development / The World Bank 264 WORLD DEVELOPMENT REPORT 2004 Technical notes World Bank publications. Readers are advised not to com- These technical notes discuss the sources and methods used pile such data from different editions. Consistent time series to compile the indicators included in this edition of Selected are available from the World Development Indicators 2003 World Development Indicators. The notes follow the order CD-ROM. in which the indicators appear in the tables. Ratios and growth rates Sources For ease of reference, the tables usually show ratios and rates The data published in the Selected World Development of growth rather than the simple underlying values. Values Indicators are taken from World Development Indicators in their original form are available from the World Develop- 2003. Where possible, however, revisions reported since the ment Indicators 2003 CD-ROM. Unless otherwise noted, closing date of that edition have been incorporated. In addi- growth rates are computed using the least-squares regression tion, newly released estimates of population and gross method (see Statistical methods below). Because this method national income (GNI) per capita for 2002 are included in takes into account all available observations during a period, table 1. the resulting growth rates reflect general trends that are not The World Bank draws on a variety of sources for the sta- unduly influenced by exceptional values. To exclude the tistics published in the World Development Indicators. Data on effects of inflation, constant price economic indicators are external debt are reported directly to the World Bank by used in calculating growth rates. Data in italics are for a year developing member countries through the Debtor Reporting or period other than that speci�ed in the column heading— System. Other data are drawn mainly from the United Nations up to two years before or after for economic indicators and and its specialized agencies, from the International Monetary up to three years for social indicators, because the latter tend Fund (IMF), and from country reports to the World Bank. to be collected less regularly and change less dramatically Bank staff estimates are also used to improve currentness or over short periods. consistency. For most countries, national accounts estimates are obtained from member governments through World Bank Constant price series economic missions. In some instances these are adjusted by An economy’s growth is measured by the increase in value staff to ensure conformity with international de�nitions and added produced by the individuals and enterprises operating concepts. Most social data from national sources are drawn in that economy. Thus, measuring real growth requires esti- from regular administrative �les, special surveys, or periodic mates of GDP and its components valued in constant prices. censuses. The World Bank collects constant price national accounts For more detailed notes about the data, please refer to the series in national currencies and recorded in the country’s World Bank’s World Development Indicators 2003. original base year. To obtain comparable series of constant price data, it rescales GDP and value added by industrial ori- Data consistency and reliability gin to a common reference year, currently 1995. This process Considerable effort has been made to standardize the data, gives rise to a discrepancy between the rescaled GDP and the but full comparability cannot be assured, and care must be sum of the rescaled components. Because allocating the dis- taken in interpreting the indicators. Many factors affect data crepancy would give rise to distortions in the growth rate, it availability, comparability, and reliability: statistical systems is left unallocated. in many developing economies are still weak; statistical methods, coverage, practices, and de�nitions differ widely; Summary measures and cross-country and intertemporal comparisons involve The summary measures for regions and income groups, pre- complex technical and conceptual problems that cannot be sented at the end of most tables, are calculated by simple unequivocally resolved. Data coverage may not be complete addition when they are expressed in levels. Aggregate growth for economies experiencing problems, such as those deriving rates and ratios are usually computed as weighted averages. from internal or external conflicts, affecting the collecting The summary measures for social indicators are weighted by and reporting of data. For these reasons, although the data population or subgroups of population, except for infant are drawn from the sources thought to be most authorita- mortality, which is weighted by the number of births. See the tive, they should be construed only as indicating trends and notes on speci�c indicators for more information. characterizing major differences among economies rather For summary measures that cover many years, calcula- than offering precise quantitative measures of those differ- tions are based on a uniform group of economies so that the ences. Also, national statistical agencies tend to revise their composition of the aggregate does not change over time. historical data, particularly for recent years. Thus, data of Group measures are compiled only if the data available for a different vintages may be published in different editions of given year account for at least two-thirds of the full group, as (c) The International Bank for Reconstruction and Development / The World Bank Selected World Development Indicators 265 de�ned for the 1995 benchmark year. As long as this crite- same purchasing power over domestic GNI that the U.S. dol- rion is met, economies for which data are missing are lar has over U.S. GNI. PPP rates allow a standard compari- assumed to behave like those that provide estimates. Readers son of real price levels between countries, just as conven- should keep in mind that the summary measures are esti- tional price indexes allow comparison of real values over mates of representative aggregates for each topic and that time. The PPP conversion factors used here are derived from nothing meaningful can be deduced about behavior at the price surveys covering 118 countries conducted by the Inter- country level by working back from group indicators. In national Comparison Programme. For Organisation for addition, the estimation process may result in discrepancies Economic Co-operation and Development countries data between subgroup and overall totals. come from the most recent round of surveys, completed in 2000; the rest are either from the 1996 survey, or data from Table 1. Size of the economy the 1993 or earlier round, which have been extrapolated to Population is based on the de facto de�nition of population, the 1996 benchmark. Estimates for countries not included in which counts all residents regardless of legal status or citi- the surveys are derived from statistical models using avail- zenship—except for refugees not permanently settled in the able data. country of asylum, who are generally considered part of the PPP GNI per capita is PPP GNI divided by midyear pop- population of their country of origin. The values shown are ulation. midyear estimates for 2002. Population estimates are usually Gross domestic product (GDP) per capita growth is based on national censuses, but the frequency and quality of based on GDP measured in constant prices. GDP is the sum these vary by country. Errors and undercounting occur even of value added by all resident producers plus any product in high-income countries; in developing countries such taxes (less subsidies) not included in the valuation of output. errors may be substantial because of limits in the transport, Growth in GDP is considered a broad measure of growth of communications, and other resources required to conduct a an economy. GDP in constant prices can be estimated by full census. Intercensal estimates are usually interpolation or measuring the total quantity of goods and services produced extrapolations based on demographic models. in a period, valuing them at an agreed set of base year prices, Surface area is a country’s total area, including areas and subtracting the cost of intermediate inputs, also in con- under inland bodies of water and some coastal waterways. stant prices. Growth is calculated from constant price GDP Population density is midyear population divided by data in local currency. land area in square kilometers. Land area is a country’s total area excluding areas under inland bodies of water and Table 2. Millennium Development Goals: coastal waterways. eradicating poverty and improving lives Gross national income (GNI—formerly gross national Share of the poorest quintile in national consumption is product or GNP), the broadest measure of national income, the share of consumption (or, in some cases, income) that is the sum of value added by all resident producers plus any accrues to the poorest 20 percent of the population. Data on product taxes (less subsidies) not included in the valuation personal or household income or consumption come from of output plus net receipts of primary income (compensa- nationally representative household surveys. The data in the tion of employees and property income) from abroad. Data table refer to different years between 1987 and 2001. Foot- are converted from national currency to current U.S. dollars notes to the data indicate whether the ranking are based on using the World Bank Atlas method. This involves using a per capital income or consumption. Each distribution is three-year average of exchange rates to smooth the effects of based on percentiles of population-rather than of house- transitory exchange rate fluctuations. (See the section on sta- holds-with households ranked by income or expenditure per tistical methods below for further discussion of the Atlas person. method). Prevalence of child malnutrition is the percentage of GNI per capita is gross national income divided by children under �ve whose weight for age is less than minus midyear population. GNI per capita in U.S. dollars is con- two standard deviations from the median for the interna- verted using the World Bank Atlas method. The World Bank tional reference population ages 0–59 months. The reference uses GNI per capita in U.S. dollars to classify economies for population, adopted by the World Health Organization in analytical purposes and to determine borrowing eligibility. 1983, is based on children from the United States, who are PPP Gross national income, which is GNI converted to assumed to be well nourished. Estimates of child malnutri- international dollars using purchasing power parity (PPP) tion are from national survey data. The proportion of chil- conversion factors, is included because nominal exchange dren who are underweight is the most common indicator of rates do not always reflect international differences in rela- malnutrition. Being underweight, even mildly, increases the tive prices. At the PPP rate, one international dollar has the risk of death and inhibits cognitive development in children. (c) The International Bank for Reconstruction and Development / The World Bank 266 WORLD DEVELOPMENT REPORT 2004 Moreover, it perpetuates the problem from one generation Maternal mortality ratio is the number of women who to the next, as malnourished women are more likely to have die from pregnancy-related causes during pregnancy and low-birth-weight babies. childbirth, per 100,000 live births. The data shown here have Primary completion rate is the total number of students been collected in various years and adjusted to a common successfully completing (or graduating from) the last year of 1995 base year. The values are modeled estimates based on primary school in a given year, divided by the total number of an exercise carried out by the World Health Organization children of of�cial graduation age in the population. The pri- (WHO) and United Nations Children’s Fund(UNICEF). In mary completion rate reflects the primary cycle as nationally this exercise maternal mortality was estimated with a regres- de�ned, ranging from three to four years of primary educa- sion model using information on fertility, birth attendants, tion (in a very small number of countries) to �ve or six years and HIV prevalence. This cannot be assumed to provide an (in most countries) and seven or eight years (in a relatively accurate estimate of maternal mortality in any country in small number of countries). For any country it is therefore the table. consistent with the gross and net enrollment ratios. The Births attended by skilled health staff are the percentage numerator may include coverage children who have repeated of deliveries attended by personnel trained to give the neces- one or more grades of primary school but are now graduat- sary supervision, care, and advice to women during preg- ing successfully as well as who entered school early. The nancy, labor, and the postpartum period, to conduct deliver- denominator is the number of children of of�cial graduation ies on their own, and to care for newborns. The share of age, which could cause the primary completion rate to exceed births attended by skilled health staff is an indicator of a 100 percent. There are other limitations that contribute to health system’s ability to provide adequate care for a preg- completion rates exceeding 100 percent, such as the use of nant women. Good antenatal and postnatal care improves estimates for population, different times of the year that the maternal health and reduces maternal and infant mortality. school and population surveys are conducted, and other dis- But data may not reflect such improvements because health crepancies in the numbers used in the calculation information system are often weak, material deaths are Ratio of female to male enrollments in primary and sec- underreported, and rates of maternal mortality are dif�cult ondary school is the ratio of the number of female students to measure. enrolled in primary and secondary school to the number of male students. Eliminating gender disparities in education Table 3. Expenditures on education and health would help to increase the status and capabilities of women. Public expenditure per student is the public current spend- This indicator is an imperfect measure of the relative accessi- ing on education divided by the number of students by level, bility of schooling for girls. With a target date of 2005, this is as a percentage of gross domestic product (GDP) per capita. the �rst of the targets to fall due. School enrollment data are Data on education are compiled by the UNESCO Institute reported to the UNESCO Institute for Statistics by national for Statistics from of�cial responses to surveys and from education authorities. Primary education provides children reports provided by education authorities in each country. with basic reading, writing, and mathematics skills along The data on education spending in the table refer solely to with an elementary understanding of such subjects as his- public spending—government spending on public educa- tory, geography, natural science, social science, art, and tion plus subsidies for private education. The data generally music. Secondary education completes the provision of exclude foreign aid for education. They may also exclude basic education that began at the primary level, and aims at spending by religious schools, which play a signi�cant role in laying foundations for lifelong learning and human develop- many developing countries. Data for some countries and for ment, by offering more subject-or skill-oriented instruction some years refer to spending by the ministry of education using more specialized teachers. only (excluding education expenditures by other ministries Under-�ve mortality rate is the probability that a new- and departments and local authorities). born baby will die before reaching age �ve, if subject to cur- Recurrent spending on primary teacher salaries is the rent age-speci�c mortality rates. The probability is expressed total amount spent on primary as a percent of total recur- as a rate per 1,000. The main sources of mortality date are rent spending on primary education (the latter including vital registration systems and direct or indirect estimates spending on personnel other than teachers). The data refer based on sample surveys or censuses. To produce harmo- to the primary education level of the education system only. nized estimates of under-�ve mortality rates that make use For countries with a �ve or six year primary system, the data of all available information in a transparent way, a method- are for the of�cial primary cycle. For countries with primary ology that �ts a regression line to the relationship between systems either longer than 6 years, or shorter than 5 years, mortality rates and their reference dates using weighted least the data are an estimate of a hypothetical 6-year equivalent squares was developed and adopted by both UNICEF and system (although based on actual enrollment, teacher, the World Bank. spending data, etc. through grade 6 in that country). The (c) The International Bank for Reconstruction and Development / The World Bank Selected World Development Indicators 267 data are estimates for 2000 based on the latest years for mation on health expenditures from national and local gov- which data are available. The data are derived from Bruns, ernment budgets, national accounts, household surveys, Mingat, and Rakatomalala (2003). Incidence of education insurance publications, international donors, and existing expenditures (lowest and highest quintiles). tabulations. Incidence of education and health expenditures (lowest Total health expenditure is the sum of public and private and highest quintiles). Average expenditure incidence stud- health expenditure. It covers the provision of health services ies relate household data on the use of public services by dif- (preventive and curative), family planning activities, nutri- ferent quintiles of the population to average spending on tion activities, and emergency aid designated for health but those services by the government. Results from these studies does not include provision of water and sanitation. The data provide a cross-sectional snapshot of who bene�ts from in the table are the product of an effort by the World Health public spending on services. Note that this is not necessarily Organization (WHO), the Organization for Economic Co- the same as who would bene�t from the marginal resources operation and Development (OECD), and the World Bank devoted to the sector. The data are accompanied by several to collect all available information on health expenditures caveats. First, while the data are often based on the best from national and local government budgets, national sources available, they are often limited when it comes to accounts, household surveys, insurance publications, inter- assessing the unit costs of services. Second, cross-country national donors, and existing tabulations. comparability is hampered by the fact that studies differ in the detail to which they differentiate average spending: for Table 4. Service indicators example some use a uniform estimate, some estimate sepa- Primary teacher absence rate is the percentage of primary rate unit costs for urban and rural areas, some for different school teachers who were absent from a random sample of provinces, and so on. Third, since the value of spending schools during surprise visits. might differ for different populations (for example spending Absenteeism of public servants from their jobs has long on urban dwellers might go much further towards providing been discussed as an impediment to effective public service quality services than an equal amount spent on people in delivery in developing countries, yet there has been relatively remote rural areas) the label “expenditure incidence� is dis- little systematic empirical evidence on this issue. As back- tinguished from “bene�t incidence�. Fourth, the results do ground research for this World Development Report, several not include the incidence of raising funds. A fairly regressive country studies were conducted. A multi-county study pattern of spending might still be pro-poor if it is �nanced Bangladesh, Ecuador, India (20 States), Indonesia, Peru, and through a very progressive tax system. Fifth, it is hard to Uganda (Chaudhury and others 2003). Additional studies know what a “good� allocation is without comparing it to with virtually identical methodologies were conduncted in other types of social spending. Details on the sources for Papua New Guinea (NRI and World Bank 2003) and Zambia these results, as well as a disaggregation by types of expendi- (Habyarimana and others 2003). tures, are available in Filmer (2003) WDR Background Note. The common survey methodology was built around Public health expenditure consists of recurrent and cap- unannounced visits to a nationally representative random ital spending from government (central and local) budgets, sample of primary schools and primary health care centers. external borrowings and grants (including donations from The study used clustered random sampling: after stratifying international agencies and nongovernmental organizations), each country (or Indian state) geographically, districts were and social (or compulsory) health insurance funds. The data randomly selected on a population-weighted basis, and then in the table are the product of an effort by the World Health facilities were randomly selected in each district. Enumera- Organization (WHO), the Organization for Economic Co- tors visited each facility and, after verifying workers’ sched- operation and Development (OECD), and the World Bank ules, recorded which of them were absent. to collect all available information on health expenditures The �gures in the table are preliminary calculations, from national and local government budgets, national based on data from surveys conducted mostly in late 2002 accounts, household surveys, insurance publications, inter- and early 2003. Further research will re�ne the calculations, national donors, and existing tabulations. in some cases drawing on data from additional visits to each Private health expenditure includes direct household facility. In addition, these facility surveys have collected a (out-of-pocket) spending, private insurance, spending by wealth of information now being used to probe the causes of non-pro�t institutions serving households (other than teacher and health personnel absence in the different coun- social insurance), and direct service payments by private tries. corporations. The data in the table are the product of an Note that these studies did not measure “absenteeism,� effort by the World Health Organization (WHO), the Orga- which is a term that is usually used to imply unjusti�able or nization for Economic Co-operation and Development unexplained absence, but instead reported on rates of (OECD), and the World Bank to collect all available infor- “absence.� That is, they reported the number of staff who (c) The International Bank for Reconstruction and Development / The World Bank 268 WORLD DEVELOPMENT REPORT 2004 were supposed to be on duty but were in fact absent from the Tuberculosis treatment success rate is the percentage of facility - without regard to the reasons for absence. Many new, registered smear-positive (infectious) cases that were personnel were doubtless absent for valid reasons, such as cured or in which a full course of treatment was completed. authorized leave or of�cial duties. Nevertheless, we report Data on the success rate of tuberculosis treatment are pro- the absence rates for two reasons: �rst, because the reasons vided for countries that have implemented the recom- for absence given by facility directors were typically not veri- mended control strategy: directly observed treatment, short �able; and second, because even authorized absences reduce course (DOTS). Countries that have not adopted DOTS or the quantity and quality of public services in these primary have only recently done so are omitted because of lack of schools and primary health centers. data or poor comparability or reliability of reported results. Primary pupil-teacher ratio is the number of pupils Physicians are graduates of any faculty or school of med- enrolled in primary school divided by the number of pri- icine who are working in the country in any medical �eld mary school teachers (regardless of their teaching assign- (practice, teaching, research). Data are from the WHO and ment). The comparability of pupil-teacher ratios across OECD, supplemented by country data. countries is affected by the de�nition of teachers and by dif- Hospital beds include inpatient beds available in public, ferences in class size by grade and in the number of hours private, general, and specialized hospitals and rehabilitation taught. Moreover, the underlying enrollment levels are sub- centers. In most cases beds for both acute and chronic care ject to a variety of reporting errors. They are based on data are included. Data are from the WHO and OECD, supple- collected during annual school surveys, which are typically mented by country data. conducted at the beginning of the school year. They do not Inpatient admission rate is the percentage of the popula- reflect actual number of attendance. And school administra- tion admitted to hospitals during a year. Data are from the tors may report exaggerated enrollments, especially if there WHO and OECD, supplemented by country data. is a �nancial incentive to do so. While the pupil-teacher ratio Access to an improved water source refers to the per- is often used to compare the quality of schooling across centage of the population with reasonable access to an ade- countries, it is often weakly related to the value added of quate amount of water from an improved source, such as a schooling systems (Behrman and Rosenzweig 1994). The household connection, public standpipe, borehole, pro- data are from the UNESCO Institute for Statistics, which tected well or spring, or rainwater collection. Unimproved compiles international data on education in cooperation sources include vendors, tanker trucks, and unprotected with national commissions and national statistical services. wells and springs. Reasonable access is de�ned as the avail- Trained teachers in primary school: are the percentage ability of at least 20 liters a person a day from a source within of primary school teachers who have received the minimum one kilometer of the dwelling. The data are based on surveys organized teacher training (preservice or in service) required and estimates provided by governments to the Joint Moni- for teaching. The share of trained teachers in primary toring Programme of the WHO and United Nations Chil- schools measures the quality of the teaching staff. It does not dren’s Fund (UNICEF). The coverage rates for water are take account of competencies acquired by teachers through based on information from service users on the facilities their professional experience or self-instruction, or of such their households actually use rather than on information factors as work experience, teaching methods and materials, from service providers, who may include nonfunctioning or classroom conditions, all of which may affect the quality systems. Access to drinking water from an improved source of teaching. Since the training teachers receive varies greatly, does not ensure that the water is safe or adequate, as these care should be taken in comparing across countries. The characteristics are not tested at the time of the surveys. data are from the UNESCO Institute for Statistics, which Access to improved sanitation facilities refers to the per- compiles international data on education in cooperation centage of the population with at least adequate access to with national commissions and national statistical services. excreta disposal facilities (private or shared but not public) Health personnel absence rate is the percentage of med- that can effectively prevent human, animal, and insect con- ical personnel at primary health clinics who were absent tact with excreta. Improved facilities range from simple but from a random sample of schools during surprise visits. (See protected pit latrines to flush toilets with a sewerage connec- the technical notes on the primary teacher absence rate for tion. To be effective, facilities must be correctly constructed further information). and properly maintained. The data are based on surveys and Child immunization rate is the percentage of children estimates provided by governments to the Joint Monitoring under one year of age receiving vaccination coverage for four Programme of the WHO and United Nations Children’s diseases—measles and diphtheria, pertussis (whooping Fund (UNICEF). The coverage rates for sanitation are based cough), and tetanus (DPT). A child is considered adequately on information from service users on the facilities their immunized against measles after receiving one dose of vac- households actually use rather than on information from cine, and against DPT after receiving three doses. service providers, who may include nonfunctioning systems. (c) The International Bank for Reconstruction and Development / The World Bank Selected World Development Indicators 269 Table 5. Foreign aid recipient indicators developing countries and territories. Flows to these recipi- Net of�cial development assistance or of�cial aid cover net ents that meet the criteria for ODA are termed of�cial aid. concessional flows to developing countries, transition Measures of aid flows from the perspective of donors dif- economies of Eastern Europe and the former Soviet Union fer from recipients’ perceived aid receipts for two main rea- and to certain advanced developing countries and territories sons. First, aid flows include expenditure items about which as determined by the Development Assistance Committee recipients may have no precise information, such as develop- (DAC) of the OECD. The flows are from members of the ment-oriented research, stipends and tuition costs for aid- DAC, multilateral development agencies, and certain Arab �nanced students in donor countries, or payment of experts countries. Data on aid are compiled by DAC and published hired by donor countries. Second, donors record their con- in its annual statistical report, Geographical Distribution of cessional funding (usually grants) to multilateral agencies Financial Flows to Aid Recipients, and in the DAC chairman’s when they make payments, while the agencies make funds annual report, Development Co-operation. The 2001 data available to recipients with a time lag and in many caes in the exclude aid from the World Food Programme because the form of soft loans where donors’ grants have been used to organization implemented an annual program budget in reduce the interest burden over the life of the loan. All data 2002, and the 2001 data are not yet consistent with the DAC in this table—including GNI, population, general govern- reporting system. ment disbursement—come from and are calculated by the Aid dependency ratios Net of�cial aid or of�cial devel- OECD. opment assistance as a percentage of GNI, gross capital for- Data are shown at current prices and dollar exchange mation and central government expenditure and aid per rates. capita provide a measure of the recipient country’s depen- Aid as a percentage of GNI shows the donor’s contribu- dency on aid. They are calculated using values in U.S. dollars tions of ODA or of�cial aid as a share of its gross national converted at of�cial exchange rates. Gross capital formation income. consists of outlays on additions to the �xed assets of the Average annual percentage change in volume and aid economy, net changes in the level of inventories, and net per capita of donor country are calculated using 2000 acquisitions of valuables. Central government expenditure exchange rates and prices. includes both current and capital (development) expendi- Aid as a percentage of general government disburse- tures and excludes lending minus repayments. For de�ni- ment shows the donor’s contributions of ODA as a share of tions of population and GNI, please see table 1. public spending. Donor fragmentation index A Her�ndahl index of Untied aid is the share of ODA that is not subject to donor concentration is calculated by summing the squared restrictions by donors on procurement sources. shares of aid over all donor agencies with positive gross dis- bursements of of�cial development assistance (ODA/OA) in Table 7. Key indicators for other economies the recipient country during the year. This index, which Population is based on the de facto de�nition of population, ranges from 0 to 1, is then subtracted from 1, to form an which counts all residents regardless of legal status or citi- index of donor fragmentation, with high values indicating zenship—except for refugees not permanently settled in the greater fragmentation. Data, and list of donor agencies, are country of asylum, who are generally considered part of the from the OECD DAC’s Geographical Distribution of Finan- population of their country of origin. The values shown are cial Flows to Aid Recipients. midyear estimates for 2002. Surface area is a country’s total area, including areas under inland bodies of water and some coastal waterways. Table 6. Aid flows from Development Assistance Gross national income (GNI) is the sum of value added Committee members by all resident producers plus any product taxes (less subsi- Net of�cial development assistance and net of�cial aid dies) not included in the valuation of output plus net record the actual international transfer by the donor of receipts of primary income (compensation of employees �nancial resources or of goods or services valued at the cost and property income) from abroad. Data are in current U.S. to the donor, less any repayments of loan principal during dollars converted using the World Bank Atlas method (see the same period. the technical notes for Table 1 and the section on statistical DAC maintains a list of countries and territories that are methods). aid recipients. Part I of the list comprises developing coun- GNI per capita is gross national income divided by tries and territories considered by DAC members to be eligi- midyear population. GNI per capita in U.S. dollars is con- ble for ODA. Part II comprises economies in transition: verted using the World Bank Atlas method. more advanced countries of Central and Eastern Europe, the PPP gross national income (GNI) is gross national countries of the former Soviet Union, and certain advanced income converted to international dollars using purchasing (c) The International Bank for Reconstruction and Development / The World Bank 270 WORLD DEVELOPMENT REPORT 2004 power parity rates. An international dollar has the same pur- It does not necessarily match the actual growth rate between chasing power over GNI as a U.S. dollar has in the United any two periods. States. (See the technical notes for Table 1). Gross domestic product (GDP) per capita growth is Exponential growth rate. The growth rate between two based on GDP measured in constant prices. GDP is the sum points in time for certain demographic data, notably labor of value added by all resident producers plus any product force and population, is calculated from the equation taxes (less subsidies) not included in the valuation of output. Growth is calculated from constant price GDP data in local r = ln (pn /p1 )/n, currency. (See the technical notes for Table 1). where pn and p1 are the last and �rst observations in the Life expectancy at birth is the number of years a new- period, n is the number of years in the period, and ln is the born infant would live if prevailing patterns of mortality at natural logarithm operator. This growth rate is based on a the time of its birth were to stay the same throughout its life. model of continuous, exponential growth between two Reduce child mortality—under-�ve mortality rate is points in time. It does not take into account the intermediate the probability that a newborn baby will die before reaching values of the series. Note also that the exponential growth age �ve, if subject to current age-speci�c mortality rates. The rate does not correspond to the annual rate of change mea- probability is expressed as a rate per 1,000. sured at a one-year interval which is given by Primary completion rate is the number of students suc- (pn – pn – 1)/pn – 1. cessfully completing the last year of (or graduating from) primary school in a given year, divided by the number of children of of�cial graduation age in the population. World Bank Atlas method. In calculating GNI and GNI per Adult illiteracy rate is the percentage of adults ages 15 capita in U.S. dollars for certain operational purposes, the and above who cannot, with understanding, read and write a World Bank uses the Atlas conversion factor. The purpose of short, simple statement about their everyday life. the Atlas conversion factor is to reduce the impact of exchange rate fluctuations in the cross-country comparison Statistical methods of national incomes. This section describes the calculation of the least-squares The Atlas conversion factor for any year is the average of a growth rate, the exponential (endpoint) growth rate, and the country’s exchange rate (or alternative conversion factor) for World Bank’s Atlas methodology for calculating the conver- that year and its exchange rates for the two preceding years, sion factor used to estimate GNI and GNI per capita in U.S. adjusted for the difference between the rate of inflation in dollars. the country, and through 2000, that in the G-5 countries (France, Germany, Japan, the United Kingdom, and the Least-squares growth rate. Least-squares growth rates are United States). For 2001 onwards, these countries include used wherever there is a suf�ciently long time series to per- the Euro Zone, Japan, the United Kingdom, and the United mit a reliable calculation. No growth rate is calculated if States. A country’s inflation rate is measured by the change more than half the observations in a period are missing. in its GDP deflator. The least-squares growth rate, r, is estimated by �tting a The inflation rate for G-5 countries (through 2000), or linear regression trendline to the logarithmic annual values the Euro Zone, Japan, the United Kingdom, and the United of the variable in the relevant period. The regression equa- States (for 2001 onwards), representing international infla- tion takes the form tion, is measured by the change in the SDR deflator. (Special drawing rights, or SDRs, are the IMF’s unit of account.) The ln Xt = a + bt, SDR deflator is calculated as a weighted average of the G-5 which is equivalent to the logarithmic transformation of the countries’ (through 2000, and the Euro Zone, Japan, the compound growth equation, United Kingdom, and the United States for 2001 onwards) GDP deflators in SDR terms, the weights being the amount Xt = Xo (1 + r)t . of each country’s currency in one SDR unit. Weights vary In this equation, X is the variable, t is time, and a = log Xo over time because both the composition of the SDR and the and b = ln (1 + r ) are the parameters to be estimated. If b* is relative exchange rates for each currency change. The SDR the least-squares estimate of b, the average annual growth deflator is calculated in SDR terms �rst and then converted rate, r, is obtained as [exp(b* )–1] and is multiplied by 100 to to U.S. dollars using the SDR to dollar Atlas conversion fac- express it as a percentage. tor. The Atlas conversion factor is then applied to a country’s The calculated growth rate is an average rate that is repre- GNI. The resulting GNI in U.S. dollars is divided by the sentative of the available observations over the entire period. midyear population to derive GNI per capita. (c) The International Bank for Reconstruction and Development / The World Bank Selected World Development Indicators 271 When of�cial exchange rates are deemed to be unreliable GDP deflator for year t, pt S$ is the SDR deflator in U.S. dol- or unrepresentative of the effective exchange rate during a lar terms for year t, Yt $ is the Atlas GNI per capita in U.S. period, an alternative estimate of the exchange rate is used in dollars in year t, Yt is current GNI (local currency) for year t, the Atlas formula (see below). and Nt is the midyear population for year t. The following formulas describe the calculation of the Atlas conversion factor for year t: Alternative conversion factors The World Bank systematically assesses the appropriateness 1  p ptS $   p ptS $   of of�cial exchange rates as conversion factors. An alterna- et* = e t − 2  t S$  + et −1  t S$  + et  3   pt − 2 pt − 2   pt −1 pt −1    tive conversion factor is used when the of�cial exchange rate is judged to diverge by an exceptionally large margin from and the calculation of GNI per capita in U.S. dollars for the rate effectively applied to domestic transactions of for- year t: eign currencies and traded products. This applies to only a small number of countries, as shown in Primary Data Docu- Y$ * t = (Yt /Nt)/e t mentation table in World Development Indicators 2003. Alter- native conversion factors are used in the Atlas methodology where et* is the Atlas conversion factor (national currency to and elsewhere in the Selected World Development Indicators the U.S. dollar) for year t, et is the average annual exchange as single-year conversion factors. rate (national currency to the U.S. dollar) for year t, pt is the (c) The International Bank for Reconstruction and Development / The World Bank (c) The International Bank for Reconstruction and Development / The World Bank