30263 WBI Development Studies Building State Capacity in Africa New Approaches, Emerging Lessons Edited by Brian Levy and Sahr Kpundeh Jointly prepared by the World Bank Institute and the Africa Region Vice Presidency of the World Bank Copyright © 2004 The International Bank for Reconstruction and Development / THE WORLD BANK 1818 H Street, N.W. Washington, D.C. 20433, U.S.A. All rights reserved Manufactured in the United States of America First printing September 2004 The World Bank Institute was established by the World Bank in 1955 to train officials con- cerned with development planning, policymaking, investment analysis, and project implementation in member developing countries. At present the substance of WBI’s work emphasizes macroeconomic and sectoral policy analysis. Through a variety of courses, semi- nars, workshops, and other learning activities, most of which are given overseas in cooperation with local institutions, WBI seeks to sharpen analytical skills used in policy analy- sis and to broaden understanding of the experience of individual countries with economic and social development. Although WBI’s publications are designed to support its training activities, many are of interest to a much broader audience. This report has been prepared by the staff of the World Bank. The judgments expressed do not necessarily reflect the views of the Board of Executive Directors or of the governments they represent. The material in this publication is copyrighted. The World Bank encourages dissemination of its work and will normally grant permission promptly. Permission to photocopy items for internal or personal use, for the internal or personal use of specific clients, or for educational classroom use is granted by the World Bank, provided that the appropriate fee is paid directly to the Copyright Clearance Center, Inc., 222 Rosewood Drive, Danvers, MA 01923, U.S.A., telephone 978-750-8400, fax 978-750-4470. Please contact the Copyright Clearance Center before photocopying items. For permission to reprint individual articles or chapters, please fax your request with com- plete information to the Republication Department, Copyright Clearance Center, fax 978-750- 4470. All other queries on rights and licenses should be addressed to the World Bank at the address above or faxed to 202-522-2422. The backlist of publications by the World Bank is shown in the annual Index of Publications, which is available from the Office of the Publisher. ISBN 0-8213-6000-0 e-ISBN 0-8213-6001-9 Library of Congress Cataloging-in-Publication Data has been applied for. Contents Foreword v Frannie A. Léautier and Callisto Madavo Acknowledgments vii About the Authors ix 1. Governance and Economic Development in Africa: Meeting the Challenge of Capacity Building 1 Brian Levy 2. Comparative Experience with Public Service Reform in Ghana, Tanzania, and Zambia 43 Mike Stevens and Stefanie Teggemann 3. Building State Capacity in Africa: Learning from Performance and Results 87 Poul Engberg-Pedersen and Brian Levy 4. Reforming Pay Policy: Techniques, Sequencing, and Politics 109 Kithinji Kiragu, Rwekaza Mukandala, and Denyse Morin 5. Cabinets, Budgets, and Poverty: Political Commitment to Poverty Reduction 149 Harry Garnett and William Plowden 6. Public Expenditure Accountability in Africa: Progress, Lessons, and Challenges 179 Bill Dorotinsky and Rob Floyd 7. Emerging Legislatures: Institutions of Horizontal Accountability 211 Joel D. Barkan, Ladipo Adamolekun, and Yongmei Zhou with Mouftaou Laleye and Njuguna Ng’ethe 8. Process Interventions Versus Structural Reforms: Institutionalizing Anticorruption Reforms in Africa 257 Sahr Kpundeh iii iv Contents 9. The Politics of Decentralization in Africa: A Comparative Analysis 283 Stephen N. Ndegwa and Brian Levy 10. Education Decentralization in Africa: A Review of Recent Policy and Practice 323 Alec Ian Gershberg and Donald R. Winkler 11. Reflections 357 Dele Olowu, Joel D. Barkan, and Njuguna Ng’ethe Index 367 Foreword The World Bank and other donors are fully committed to modalities of development support that put countries in the driver’s seat, with the poverty reduction strategy process—prepared by national governments, on the basis of close consultation with civil society—providing the framework for that support. An effective poverty reduction strategy process and a pro- ductive partnership can be built only on a platform of strong public capac- ity: capacity to formulate policies; capacity to build consensus; capacity to implement reform; and capacity to monitor results, learn lessons, and adapt accordingly. Building the requisite capacities turns out to be a formidable challenge. For these reasons, enhancing the capacity of African states has risen to the top of the continent’s development agenda. The process of learning how capacity can be built effectively is a continu- ing one. The World Bank’s efforts to help build capacity in Africa date back at least to the World Bank’s 1989 study, Sustainable Growth with Equity: A Long-Term Perspective for Sub-Saharan Africa, which highlighted the centrality of institutions for Africa’s development performance. Between 1987 and 1997, the Bank implemented at least 70 civil service reform projects across the African continent, with mixed results. Many of the lessons from this ini- tial round of effort were highlighted in the World Development Report 1997: The State in a Changing World and in a 2000 follow-up study, Reforming Public Institutions and Strengthening Governance: A World Bank Strategy. In recent years, a number of African governments, sometimes working in partnership with the World Bank and other development partners, have moved forward with new-style programs to build public sector capacity. This volume aims to share some of the lessons for the design and implementation of public sector capacity building that are emerging from this new generation of operational practice. The volume also exemplifies an increasingly collaborative way of work- ing within the World Bank Group, in this case between the World Bank Institute and the Bank’s Africa Regional Vice Presidency. Some of the back- ground research was funded in part by the World Bank Institute. Other v vi Foreword research projects were collaborative efforts by staff of the two groups plus experts from outside the Bank. All were discussed in depth with our African development partners and with African scholars at a workshop sponsored by the World Bank Institute workshop in Johannesburg, South Africa, on June 10–11, 2003. We feel sure that this model of working together consti- tutes the most effective way for the World Bank Group to contribute, in its role as a knowledge Bank, to the challenge of building state capacity in Africa. Frannie A. Léautier Callisto Madavo Vice President Vice President World Bank Institute Africa Region, World Bank Acknowledgments Within the past decade, dating back to World Bank President James Wolfen- sohn’s 1996 speech decrying the “cancer of corruption” and the World Devel- opment Report 1997: The State in a Changing World, the World Bank has intensified its efforts to address the challenge of building state capacity. Within the Bank, the discourse has been especially rich between practition- ers with long experience of the on-the-ground challenges of strengthening governments and those with a more academic orientation, whose work reflects the renewed interest in institutions by economists and political sci- entists. Many people within the World Bank have contributed to this climate of intensified, collaborative engagement that made this volume possible; in particular, the editors would like to acknowledge the contributions of Cheryl Gray and Sanjay Pradhan. This volume draws on the experiences of various colleagues that have in one way or another been involved in thinking about or implementing pub- lic sector reforms in Africa. Many people inside and outside the World Bank provided helpful comments and participated in either the Johannesburg workshop or in some of the brown bag lunches organized in the World Bank at which earlier drafts of the chapters were presented. Many thanks to Paula Donovan, Roumeen Islam, and in particular Alan Gelb for all their support throughout the preparation of this volume. Thanks also to many other staff in the World Bank, notably including Daniel Kaufmann and Michael Sarris, for their continuing support and willingness to generously share their time and ideas. We are also grateful to the many government officials and other experts from various countries for taking time from their busy schedules to partici- pate in the Johannesburg workshop and for their valuable contributions. Some of them include Winnie Byanyima (Uganda), Onchari Kenani (Kenya), Moses Kondowe (Zambia), Appiah Koranteng (Ghana), Bradford Malumbe (Zambia), Apollinaire Ndorukwigira (Africa Capacity Building Foundation), Adetunji Olaopa (Nigeria), Sahr George Pessima (Sierra Leone), and Jotham Tumwesigwe (Uganda). We hope that the various vii viii Acknowledgments lessons on capacity building and public sector reforms shared in this vol- ume will help us rethink some of our current approaches. For specific chapters, the authors would like to acknowledge the helpful comments and contributions from Ed Campos, Poul Engberg-Pedersen, Alan Gelb, and Phil Keefer (chapter 1); the U.K. Department for Interna- tional Development (the major funder for the research project on which chapter 4 is based); Rick Stapenhurst (chapter 7); Phyllis Dininio, Roumeen Islam, Michael Johnston, Tilla Sewe McAnthony, and Joel Turkewitz (chap- ter 8); Navin Girishankar, Ezzeddine Moudoud, Dele Olowu, and Leonardo Villalon (chapter 9); and Ben Meade (chapter 10). The editors also thank the government of Italy, which funded part of this work through the World Bank Institute/Italy Governance Trust Funds, and Alice Faintich for ably overseeing the editing of the volume. Of course, the usual caveat applies: the views expressed in the individual chapters are those of the authors and should not be attributed to the World Bank Group. About the Authors Ladipo Adamolekun is lead public sector management specialist at the World Bank, where he works on governance and public sector management issues. Prior to joining the World Bank in 1987, he was professor of public administration at Obafemi Awolowo University, Ife, Nigeria. He has pub- lished many books and numerous articles on politics and administration in Africa. His latest book is an edited volume entitled Public Administration in Africa: Main Issues and Selected Country Studies (Boulder, Colorado: Westview Press, 1999). Joel D. Barkan is professor of political science at the University of Iowa and senior consultant on governance at the World Bank. He specializes in state- society relations and democratization in Africa, and has received fellow- ships from the Woodrow Wilson International Center for Scholars and the United States Institute of Peace. His most recent articles are “Democracy in Africa: What Future?” (Harvard International Review, Summer 2002), and “Kenya After Moi” (Foreign Affairs, January-February 2004). Bill Dorotinsky is lead public sector specialist working on public expendi- ture issues in the World Bank’s Public Sector Reform and Capacity Building Unit. Prior to joining the Bank, Dorotinsky worked with the U.S. Treasury Office of Technical Assistance advising the governments of Argentina, Croa- tia, and Hungary. He also spent 12 years with the U.S. Office of Manage- ment and Budget addressing such issues as the implementation of performance budgeting, the reform of financial management reform, and the reform and financing of health and two years seconded to the District of Columbia during its fiscal crisis in the mid-1990s, including serving as deputy chief financial officer. Poul Engberg-Pedersen is a senior public sector specialist in the Public Sec- tor Reform and Capacity Building Unit of the World Bank. During 1994-2002, he was director of the Centre for Development Research in Copenhagen. Before that, he was head of research and development for COWIconsult ix x About the Authors Denmark. Engberg-Pedersen specializes in the institutional dimensions of development cooperation. His publications include Aid: Understanding Inter- national Development Cooperation (London: Zed Books, 2003). Rob Floyd is an economist and public sector management specialist at the World Bank, where he has focused on expenditure management systems and civil service reform, particularly in Africa. Previously he served as an adviser to the governments of Hungary and Thailand on reform implementation and was director of research at a private consulting firm. He currently serves as an aide to James Wolfensohn, president of the World Bank, and is respon- sible for a portfolio that includes poverty reduction, public sector and eco- nomic management, capacity building, and development economics. Harry Garnett is lead public sector management specialist in the Public Sec- tor Reform and Capacity Building Unit of the Africa Region of the World Bank. He is responsible for supporting public sector reform projects in a number of African countries, including strengthening presidencies and cab- inet offices. He has worked on public sector management issues in more than 30 developing countries worldwide. Alec Ian Gershberg is associate professor at the Milano Graduate School of Management and Urban Policy, New School University, New York. He is also a research associate with the National Bureau of Economic Research in Cambridge, Massachusetts. He has researched and written extensively on education reform and school governance in developing countries. His most recent publication is Beyond “Bilingual” Education: New Immigrants and Public School Policies in California (Washington, D.C.: Urban Institute Press, 2004). In July 2004 he joins the World Bank as senior education economist in the Human Development Network Anchor. Kithinji Kiragu is a partner with Pricewaterhouse Coopers’ Public Sector Advisory Services Group for the Africa Central Region. In 1995-98, he worked for the Tanzanian government as chief technical adviser for the public service reform program. Since the late 1980s, he has focused on pub- lic sector management reforms in Africa. In addition to extensive consult- ing, he has published and presented papers on public sector reforms in Sub-Saharan Africa at many regional and international forums. Sahr Kpundeh is senior public sector specialist working with the Gover- nance Team at the World Bank Institute and with the Public Sector Reform and Capacity Building Unit in the World Bank’s Africa Region. He is cur- rently responsible for leading the Bank’s public sector reform program in Kenya, as well as for developing and implementing governance learning About the Authors xi programs for client countries. His most recent work is a chapter he contributed entitled “Corruption and Political Insurgency in Sierra Leone” for Democracy and Terror: The Sierra Leone Civil War (Pretoria, South Africa: UNISA Press, 2004). From 1991 to 1995, Kpundeh was program officer for the Panel on Issues in Democracy and States in Transition at the National Academy of Sciences in Washington, D.C. Mouftaou Laleye is director of a policy analysis unit for the National Assembly of Benin. He is also a management consultant and a former senior lecturer in public administration at Obafemi Awolowo University, Ife, Nige- ria. He has worked as a consultant for several international bodies, includ- ing the Economic Commission for Africa and the World Bank. He has written or contributed to a number of publications on politics and adminis- tration in Africa. Brian Levy worked in the World Bank’s Africa Vice Presidency from 1991 to 2003 on the challenges of strengthening the institutional underpinnings of African development, for the last four years as sector manager of the Africa Public Sector Reform and Capacity Building Unit. He was a member of the core team that produced the World Bank World Development Report 1997: The State in a Changing World. He has published numerous books and articles on the interactions between institutions and development in Africa, East Asia, and elsewhere. Prior to joining the Bank he was assistant professor in devel- opment economics at Williams College in Williamstown, Massachusetts. Denyse Morin is senior public sector specialist in the World Bank’s Africa Public Sector and Capacity Building Unit. She is currently leading World Bank programs in support of capacity building and public service and judi- cial reforms in Tanzania and Uganda. Prior to joining the Bank in 1994, Morin was a senior program officer with the Canadian International Devel- opment Agency, where she worked on development programs in developing countries, including long-term assignments in Pakistan, Tanzania, and Zaire. Rwekaza Mukandala is professor of political science and public adminis- tration at the University of Dar es Salaam and specializes in comparative politics and governance issues. He has researched, consulted, and advised on various governance and institutional issues for such bodies as the gov- ernment of Tanzania, the United Nations, the United Nations Development Programme, the U.S. Agency for International Development, the World Bank, and the International Monetary Fund. Stephen N. Ndegwa is a public sector governance specialist with the World Bank in the Public Sector Governance central unit. His main areas of special- xii About the Authors ization are decentralization, political economy, and institutional reform, and he has worked in both the East Asia and Africa regional vice presidencies. Prior to joining the Bank, he was associate professor of government and director of the Program on Civil Society and Governance at the College of William and Mary in Williamsburg, Virginia. He is the author or editor of several books and articles on civil society, political economy, and develop- ment in Africa. Njuguna Ng’ethe is an associate research professor at the Institute for Development Studies, University of Nairobi, Kenya. Ng’ethe is a political scientist specializing in decentralization and civil society issues, fields in which he has written and consulted extensively. Between 1988 and 1994, he was director of the Institute for Development Studies. He also served as the founding executive director of the Institute of Policy Analysis and Research, Kenya’s first independent think tank, between 1994 and 1997. Dele Olowu currently works as a principal governance expert at the African Development Bank. He also serves on the Board of the European Centre for Development Policy and Management in Maastricht, Nether- lands. Prior to taking up his position at the African Development Bank, he was professor of public administration at Obafemi Awolowo University, Ile- Ife, Nigeria, and worked for the United Nations Economic Commission for Africa and the Institute of Social Studies in the Netherlands. One of his most recent publications, with Jim Wunsch, is Local Governance in Africa: The Chal- lenge of Democratic Decentralization (Boulder, CO: Lynne Reinner, 2003). William Plowden is a former senior U.K. civil servant who worked in the Cabinet Office and the Department of Industry. He was director general of the Royal Institute of Public Administration in London. He has also taught government and public administration at the London School of Economics, which he is now a visiting professor. He has advised governments in East- ern Europe and the former Soviet Union, Africa, and Latin America and the Caribbean on public sector reform and on strengthening presidencies and cabinet offices. Mike Stevens is a consultant specializing in governance and public sector management. He retired from the World Bank in 2003 after a 20-year career spent working chiefly on public sector reform issues, and played a central role in the development of the Bank’s governance and anticorruption poli- cies. Prior to joining the Bank, he worked as an economist for various gov- ernments in Africa and Asia and in teaching and banking in the United Kingdom and the United States. About the Authors xiii Stefanie Teggemann works with client governments to improve public sec- tor performance in Guinea, Madagascar, and Rwanda for the Public Sector Reform and Capacity Building Unit of the World Bank’s Africa Region. Before that she specialized in governance and anticorruption at the World Bank Public Sector Group in Washington, D.C. and the Kennedy School of Government at Harvard University. An Asia expert, she also managed a research project on the effects of corruption on the urban poor for the World Bank’s office in Indonesia. Donald R. Winkler is a senior research economist with Research Triangle Institute, International. Prior to his current assignment, he was professor of public policy at the University of Southern California for 10 years and held several positions at the World Bank, including manager of the education team for the Latin America region. He has written widely for refereed jour- nals and other publications on education decentralization, intergovernmen- tal finance, and the implementation of public policy reforms. Yongmei Zhou is an economist with the Public Sector Reform and Capacity Building Unit of the World Bank’s Africa Region. She is particularly inter- ested in the evolution of economic and political institutions and how they shape individual incentives and behavior. She has been leading the design and implementation of governance and public sector reform programs in Ghana and Sierra Leone. 1 Governance and Economic Development in Africa: Meeting the Challenge of Capacity Building Brian Levy The evolution of Africa’s development strategies has mirrored broader changes in our understanding of how development proceeds. In the years immediately following World War II, most economic analysts believed that financial capital plus engineering capability were sufficient to secure develop- ment. By the late 1970s, however, analysts realized that not only the quantity but also the efficiency of investment were key to sustainable growth and, fur- thermore, that economic incentives that resulted from economic policy were crucial determinants of investment efficiency. Economic policy reform (that is, structural adjustment) became the central focus of Africa’s development dis- course. By the early 1990s the realization began to dawn that policies them- selves were built on an underpinning of social, political, and state institutions and that weaknesses in this institutional foundation could undercut the eco- nomic policy reform agenda in three ways: by short-circuiting efforts at policy change, by failing to provide a robust platform of credibility and conflict reso- lution for market activity, and by being unable to provide complementary physical and social infrastructures. A decade later most analysts agree about the kinds of state institutions that are capable of supporting sustainable, market-oriented development. An ability to describe the characteristics of effective states does not, however, conjure them into existence out of thin air. The process of building state capacity is a dynamic one; knowing the desired end point is only moderately helpful in discovering the path from institutional weakness to progressively stronger capability. 1 2 Governance and Economic Development in Africa: Meeting the Challenge of Capacity Building For more than 15 years the World Bank has been working in partnership with countries across Africa to help build the requisite state capacities. The challenge has been difficult, and the record is quite mixed. Consequently, reformers must continuously learn what does—and does not—work. This book aims to contribute to this learning process by sharing some recent experiences and evolving insights of practitioners and researchers from (principally, but not exclusively) the World Bank, who have focused their professional lives on the challenge of helping build state capacity in Africa.1 This introductory chapter has three principal goals. First, it summarizes the empirical record of Africa’s development performance over the past quar- ter of a century in a way that highlights the interactions between economic performance and the quality of governance. Second, it introduces each of the contributions to the book. Third, it distills from the individual contributions a succinct statement of how state capacity might most effectively be built. Governance and Economic Performance: Some Empirical Patterns This section lays out an analytical framework that highlights the links between governance and economic performance, noting some key trends in Africa’s economic and governance performance since independence. Read- ers familiar with Africa’s development experience can proceed directly to the next section (“Building State Capacity”), which introduces the chapters on capacity building included in this book. Readers interested in a more detailed empirical analysis of the interactions between governance and eco- nomic performance can turn to appendix 1A of this chapter. The Governance Diamond Figure 1.1 introduces the analytical framework as a “governance diamond,” which highlights seven sets of interactions among the following four broadly defined categories:2 1. Authors shared early drafts of the chapters in this book with the World Bank’s development partners from more than a dozen countries at a workshop held in South Africa in June 2003 and revised their work on the basis of those discussions. 2. My intent is to be heuristic, rather than comprehensive. Thus I intentionally defined the categories broadly. Political interests, for example, can be viewed as broad enough to include cultural determinants of behavior. Similarly, the category political institutions can be viewed as broad enough to include the justice system. But even with this expansive interpretation, the framework remains incomplete. For example, international influences on African governance are not included and the quality of leadership also fits uncomfortably into the framework. Brian Levy 3 • Political interests. Defined here to comprise the social and class struc- ture of civil society. • Formal political institutions. Defined here to comprise both the consti- tutional structure of the state (legislative structure and oversight; judicial independence; structure of decentralized, intergovernmental relationships; rules governing political representation, and so forth) and the formal, representative political leadership that results from this constitutional structure.3 • Bureaucracy. Defined here to comprise public employees and their associated responsibility (under the direction and oversight of formal political leaders) for formulating policy, regulating economic activity, and delivering services. • Economy. Defined here to comprise a society’s productive factors and their associated levels of economic activity. Figure 1.1. Africa’s Governance Diamond Bureaucracy C5 C7 Formal political C6 Economy institutions C2 C4 C3 C1 Political interests Source: Author. 3. Note that the new institutional economics (see North 1990 for an important synthesis), which is an important source of inspiration for this chapter, emphasizes that the “rules of the game” comprise informal as well as formal institutions. Explor- ing the role of informal institutions is beyond the scope of this chapter, but the reader should remember that social norms, specifically the relative salience of hori- zontal as opposed to vertical networks of obligation and accountability, are likely to be an important determinant of how some key interactions in the governance dia- mond operate in practice. 4 Governance and Economic Development in Africa: Meeting the Challenge of Capacity Building As figure 1.1 highlights, these four categories make up an interdepen- dent system and together shape the trajectories over time of economic, pub- lic sector, and more broadly, state performance. Logically, 16 sets of causal interactions exist among the four categories. The focus here is on the follow- ing seven sets of interactions that appear to be the most dominant. • Causal relationship 1: economy ⇒ political interests. This relationship highlights how the structure and competitiveness of economic pro- duction help shape the patterns of prevailing political interests. • Causal relationships 2 and 3: political interests ⇔ formal political institu- tions. One facet of this interdependent relationship (depicted as C2 in figure 1.1) comprises the way in which formal institutions provide a “container” for political activity, by structuring the legitimate chan- nels of influence-seeking, including contestation for political leader- ship. The second facet of the relationship (depicted as C3 in figure 1.1) is how powerful political interests seek to influence formal insti- tutions, not only through legitimate influence seeking, but also through efforts to reshape to their advantage the structure of the for- mal institutions themselves. • Causal relationship 4: political interests ⇒ bureaucracy. This relationship highlights efforts on the part of political interests in civil society to influence bureaucratic behavior directly, unmediated by formal insti- tutional structures. Note that the stronger the container for politics provided by formal institutions is, the less important this causal channel is likely to be. • Causal relationship 5: formal political institutions ⇒ bureaucracy. High- lighted here is the agency relationship between political principals and their bureaucratic agents. Politicians have the authority both to set the policy, regulatory, and service delivery goals of bureaucratic actors and to monitor their performance in relation to these goals. • Causal relationship 6: formal political institutions ⇒ economy. This rela- tionship highlights the extent to which formal institutions provide stability and predictability to the economic environment and have associated influence on levels of private investment and economic activity. • Causal relationship 7: bureaucracy ⇒ economy. This relationship high- lights the effects on the economy of bureaucratic decisions and actions in relation to policies, regulation, and service delivery. Appendix 1A lays out in some detail how Africa’s capacity building chal- lenges and opportunities have been shaped by interactions among these Brian Levy 5 seven causal relationships. Indeed, attention to these interactions underlies much of the analysis both in the this chapter and throughout the book. Africa’s Neopatrimonial Downward Spiral Upon independence many African countries were bequeathed governance structures with a strong shell of formal institutions, both political and bureaucratic. Often, however, these seemingly richly articulated formal institutions turned out to be something of a shell, reflecting a combination of colonial legacy and high-minded aspirations of independence. In reality, the ability of many states to structure (beyond a narrow urban segment) political interests in a way that was supportive of a developmental project was weak. The short-run consequence (Migdal 1988) was that the mode of gover- nance shifted rapidly from a formal system of checks and balances to a de facto (and, in many countries, also in part de jure) system of rule that has been described by African scholars as “neopatrimonial” (Dia 1996; Lewis 1996). Through the 1970s and 1980s neopatrimonialism seemed to provide a stable, if not especially dynamic, form of rule. In practice, though, as the governance diamond implies—and as the subsequent unfolding of events in country after country confirmed—slow-moving, but inexorable, forces of decay underlay the neopatrimonial model. One area of decay was in the bureaucracy. At independence most bureaucracies were governed by formal rules and initially subject to rela- tively slight informal pressure from political interests. With the rise of neopatrimonial rule, the mode of governing bureaucracy shifted from the clarification, monitoring, and enforcement of formal rules to informal rules set without transparency, and sometimes increasingly capriciously, by a country’s political leadership. The consequence was a decline in bureau- cratic performance. The decline of bureaucracy influenced economic performance by affect- ing policymaking, regulation, and service delivery. Neopatrimonial rule gen- erally operated by conferring discretionary rents on favored allies, giving little attention to the impact of rentier policies on economic growth, the effi- ciency of public services, or the quality of business regulation. The classic consequences, evident in country after country, included the disruption of markets, rising costs of doing business, urban bias, and increased protection- ism. The intensification of a rentier relationship between business and gov- ernment progressively transformed the structure of the business class itself, bringing about a progressive rise of a politically dependent business class. 6 Governance and Economic Development in Africa: Meeting the Challenge of Capacity Building In some countries (Sierra Leone, for example) the neopatrimonial down- ward spiral proceeded to the point of state collapse. In others domestic political intervention preempted the cycle of decline. This book focuses on the capacity building experiences of the latter group—countries that have retained (or reconstructed) functioning states. Adjustment and Economic Performance Reversal of the neopatrimonial downward spiral began in a small number of African countries in the mid-1980s and accelerated into the 1990s. The two most familiar facets of this reversal are structural adjustment economic reforms and the wave of democratization that swept the continent in the early 1990s. The trajectories of adjustment and democratization have been documented extensively elsewhere (Bratton and Van Der Walle 1997; World Bank 1994). This section simply provides an initial empirical overview (with more detail in appendix 1A) of the economic impact of adjustment on a set of 21 African countries.4 The 21 countries can be grouped according to the degree to which the process of economic reform has been sustained and longstanding. Three groups are identified. • The first group comprises the sustained adjusters—eight countries that consistently remained on track in their programs of reform throughout the 1988–96 rounds of financial support provided through the multi- donor Special Program for Africa. These countries are Benin, Burkina Faso, Ghana, Malawi, Mali, Mozambique, Uganda, and Zambia. • The second group comprises eight later adjusters—countries that initi- ated strong policy reforms in the 1990s and continued to sustain their reforms through the end of 2001. These countries are Cameroon, Chad, Guinea, Madagascar, Mauritania, Niger, Senegal, and Tanzania. • The third group comprises five governance polarized countries—Côte d’Ivoire, Kenya, Nigeria, Togo, and Zimbabwe. Although all of these countries initiated some policy reforms during the 1990s, in all cases domestic political disruptions of one kind or another overshadowed efforts at economic policy reform. 4. Beginning with the full set of 48 Sub-Saharan African countries, three sets of outlier countries were excluded: very small countries, countries affected by pro- found internal conflict, and the current middle-income, consistently high perform- ing countries. Furthermore, to facilitate comparison with earlier analysis, the selection of countries was limited to those that were included in the 1994 World Bank study, Adjustment in Africa. Brian Levy 7 Table 1.1. Average Annual Growth in Real GDP, 21 African Countries, 1975–2000 (percent) Country 1975–84 1985–89 1991–96 1996–2000 Eight sustained adjusters 1.7 3.2 3.7 4.7 Eight later adjusters 2.4 2.7 1.9 4.2 Five governance polarized 2.3 4.1 2.3 1.4 Source: Appendix 1A; World Bank (2002, p. 15). Table 1.1 reports the average annual real growth rates in GDP for these three groups of countries over four distinct periods.5 Overall, the sustained adjusters’ commitment to policy reform since the late 1980s appears to have paid off, because the growth accelerated in each period. Among the later adjusters, the overall pattern is one of slow growth prior to policy reform and subsequent acceleration. As for the last group, the contrast with the oth- ers is clear—the absence of any sustained acceleration in growth, and possi- bly even a trend decline. In sum as explored further in appendix 1A, the data strongly suggest that, although important country-to-country variations are evident, adjustment contributes to an acceleration of aggregate economic growth and to productivity-improving changes in economic structure. Governance Trends and Challenges Table 1.2 summarizes the responses to a question included in a 1996 survey of more than 3,600 firms in each of 69 countries (including 22 in Africa, 18 of which are included in the sample used in this chapter), conducted for the World Development Report 1997: The State in a Changing World (World Bank 1997). The countries were asked the following, to be answered both as of the time of the survey and as of 10 years earlier: “Please rate your overall perception of the relation between government and/or bureaucracy and private firms on the following 6-point scale.” Each firm scored its perceptions of the relationship on a 1–6 scale. A score of 1 signaled that government was perceived as a helping hand, and 6 signaled that it was perceived as an opponent. The difference between the 5. The World Bank (2002) reports directly the growth rates for 1975–1984, 1985–1989, and 1990–2000, and annual data for 1991–2000. This chapter used the 1991, 1996, and 2000 data to calculate growth rates for the subperiods 1991–96 and 1996–2000. The gap in information between 1989 and 1991 is an artifact of the way the data were made available and has no implication for the patterns detailed in table 1.1 and the text (based on inspection of the 1990–2000 growth rates, not reported here). 8 Governance and Economic Development in Africa: Meeting the Challenge of Capacity Building Table 1.2. Trends in Relationships between Private Firms and Government, 18 African Countries, 1986 and 1996 Net helping hand (+) or opponent (–) Country 1996 1986 Net change Average for sustained adjusters +11 –43 +54 Benin –23 –43 +20 Ghana +39 –48 +87 Malawi +17 –29 +46 Mali +11 –29 +40 Mozambique –19 –55 +36 Uganda +37 –38 +75 Zambia +19 –56 +75 Average for later adjusters –27 –16 +11 Cameroon –24 +7 –31 Chad –31 0 –31 Guinea –10 –10 0 Madagascar –42 –37 –5 Senegal –20 –8 –12 Tanzania –36 –51 +15 Average for governance polarized –14 +6 –20 Côte d’Ivoire –16 –2 –14 Kenya –2 –13 –11 Nigeria +6 +61 –55 Togo –21 +10 –31 Zimbabwe –36 –24 –12 Source: Survey conducted for the World Bank; World Bank (1997). proportion of respondents that scored the relationship 1 or 2 and the propor- tion that scored the relationship 5 or 6 provides one simple way to summa- rize the extent to which government is perceived as a net helping hand or opponent. In Ghana, for example, 48 percent more firms reported that as of 1986 they perceived the government to be an opponent than a helping hand. By contrast, for 1996, 39 percent more firms perceived the government as a helping hand than as an opponent. Table 1.2 summarizes the results for 18 countries, organized into the three subgroups used in this chapter. The results point to a remarkable turnaround between 1986 and 1996 among the sustained adjusters concerning the impact of public sector gover- nance on the economy—an average gain in the net helping hand ratio of 54 percent. In 1986 the ratio was negative in all seven countries; by 1996 it had turned to positive in five of the seven (and, although still net negative, had Brian Levy 9 improved by 20 points or more in the remaining two). As for the other coun- tries, the ratio had worsened in four of the five governance polarized coun- tries, as well as in four of the six later adjusters for which data were available; only Tanzania enjoyed significant gains, although still below those of the sustained adjusters. As appendix 1A explores in more detail these governance gains appear to be founded on changes in expectations, rather than on actual improvements in the performance of public institu- tions. Thus the question arises, is this seemingly positive trend sustainable? The comparative evidence reported in table 1.3 on performance of the 21 African countries in controlling corruption highlights the concern. Data on Table 1.3. Control of Corruption: Comparative Global Rank, 21 African Countries, 1996 and 2002 Country 1996 Percentile ranking 2002 Percentile ranking Average for sustained adjusters 31 31 Benin — 34 Burkina Faso 44 58 Ghana 36 43 Malawi 12 20 Mali 44 46 Mozambique 33 15 Uganda 32 16 Zambia 16 17 Average for later adjusters 39 33 Cameroon 7 9 Chad — 13 Guinea 66 36 Madagascar 66 62 Mauritania — 63 Niger 44 8 Senegal 39 53 Tanzania 9 16 Average for governance polarized 31 15 Côte d’Ivoire 72 23 Kenya 9 11 Nigeria 5 3 Togo 13 33 Zimbabwe 55 6 — Data not available (averages are calculated using available data). Source: Kaufmann, Kraay, and Mastruzzi (2003). 10 Governance and Economic Development in Africa: Meeting the Challenge of Capacity Building corruption form a useful proxy for the quality of political institutions because they signal the extent to which, first, these institutions can constrain public actors from pursuing private ends and, second, can direct private action aimed at influencing public officialdom to formal, rather than infor- mal, channels. Although corruption is difficult to measure—and all mea- sures are recognized as having wide margins of error—worldwide data provided by the World Bank Institute are widely regarded as among the most comprehensive and reliable (and careful, insofar as standard errors and point estimates are reported) available.6 The data set is for over 150 countries for two-year periods from 1996 to 2002. Table 1.3 reports the percentile rank in controlling corruption for each of the 21 African countries, relative to the worldwide population, for 1996 and for 2002. (A country in the 5th percentile is better at controlling corruption than only 4 percent of the sample; by contrast a country in the 95th percentile is better than 94 percent of the sample.) The results give little comfort about the quality of Africa’s political institutions as of 1996, about their subse- quent direction of change, or about any positive impact of economic reform on corruption: • As of 2002 only 6 of the 21 countries ranked above the worst perform- ing 40 percent of countries worldwide in their control of corruption. • The period from 1996 to 2002 shows no evidence of systematic improvement: only three countries improved their percentile ranking by 10 or more points, and the ranking worsened by at least that amount for six countries. • As of 1996 sustained adjusters were no better at controlling corrup- tion than other groups; and over the subsequent six-year period, although one country in the sustained adjuster group (Burkina Faso) improved by more than 10 points, the performance of two others (Mozambique and Uganda) worsened by at least that amount. Overall, then, Africa’s governance landscape as of the first decade of the 21st century offers a mixed picture. In the decades immediately following independence, a majority of countries succumbed to a downward gover- nance spiral, characterized by perverse incentives on the part of political and economic actors and the corresponding decay of political and bureau- cratic institutions. Beginning in the mid-1980s a combination of external 6. Access to these data is available at http://www.Worldbank.org/wbi. For details of the data set and methodology, see Kaufman, Kraay, and Zoido-Lobaton (1999). Brian Levy 11 pressure and internal discontent set in motion in many countries processes of economic and political liberalization, which broke the iron grip of these perverse political incentives and their consequences for governance. Even so the transformation does not appear to have been dramatic enough to transform a vicious spiral into a virtuous one. The risks thus remain high that, absent continuing momentum of positive change, a down- ward spiral can readily reassert itself. Enhancing the capacity of political and bureaucratic institutions emerges as crucial for Africa’s development prospects. This book explores what we are learning about how capacity enhancement might best be achieved. Building State Capacity: Emerging Lessons Efforts to help build state capacity in independent Africa are not new. For example, between 1987 and 1997 the Bank implemented at least 70 civil ser- vice reform projects across the African continent (World Bank 1999). The results of these interventions were mixed at best: of those completed by 1997, only 29 percent were rated satisfactory in a review by the World Bank’s arms-length Operations Evaluations Department. Viewed through the lens of hindsight, a principal reason for the limited success of the first round of efforts to build state capacity was the implicit presumption that the weakness of public administration was managerial and could be remedied in a straightforward manner through a combination of organizational overhaul and financial support to procure the requisite specialist technical advice, training, and hardware. By contrast, a central les- son of experience—one that the governance diamond in figure 1.1 endeav- ors to capture heuristically—is that public administrations are embedded in a complex, interdependent system. This system incorporates not only the bureaucratic apparatus as a whole, but also political institutions and social, economic, and political interests more broadly. The approaches to building state capacity explored in this book take this interdependence as a starting point. They thus aim to complement a narrowly technocratic focus on the supply side of public management and give more careful attention to the demand-side incentives for performance. Two facets of state capacity building are highlighted in the chapters that follow. Chapters 2–6 explore recent experience with, and lessons from, efforts to build Africa’s administrative capacity, a principal focus of state reform efforts of the 1990s. Chapters 7–10 examine recent experiences and emerging lessons from efforts across Africa to foster institutions that strengthen gov- ernment accountability to citizens. This focus on accountability follows 12 Governance and Economic Development in Africa: Meeting the Challenge of Capacity Building directly from the priority in state capacity building initiatives given to demand-side incentives for performance. Other topics on which an explo- ration of state capacity building might usefully have focused, but which, given limitations both of space and of the professional experience of the con- tributors, are not addressed in this book include the following: judicial and legal reform, sector-specific capacity building challenges (although chapter 10 looks at the education sector in the context of decentralization), and the supply-side issue of how to build the effective training capacity needed for capacity building. Building Administrative Capacity Chapters 2 and 3 provide two distinct birds-eye views of the experiences of programs to strengthen governmental administrative capacity. In chapter 2 Stevens and Teggemann provide an overview of the challenges of compre- hensive second generation administrative reform using a review of imple- mentation experiences in Ghana, Tanzania, and Zambia. In chapter 3 Engberg-Pedersen and Levy seek general patterns from the results of 15 World Bank–supported state capacity building projects (implemented across 11 African countries). Stevens and Teggemann begin by distinguishing between first and sec- ond generation reforms. In contrast to first generation reforms, which were preoccupied overwhelmingly with controlling the wage bill, the focus of the second generation efforts was on enabling governments to be not only affordable, but also effective. Stevens and Teggemann examine in depth the countries’ experiences with five key components of comprehensive second generation programs: • Pay, including the selective accelerated salary enhancement program • State restructuring, with special emphasis on the introduction of exec- utive agencies • Performance management, including the introduction of a perfor- mance improvement management system and a funding mechanism • Strengthening of control systems, including financial management, audit, administrative controls, and civil society engagement • Training. They find that only in Tanzania do the results seem to be fulfilling the program’s initial promise. Ghana’s reforms barely went beyond the analytic phase before being put on hold following a change in government. Zambia made improvements in some limited areas and is intensifying its efforts toward a focused agenda for improving expenditure accountability. Brian Levy 13 To account for these divergent reform experiences, Stevens and Tegge- mann use the evocative metaphor of a play where the combination of stage, actors, and props define the quality of its ensemble: The political and economic context provides the stage on which reforms are conceived and implemented. The actors involved in the reform process as well as the implementation arrangements shape critically the depth at which change affects exist- ing institutional layers and at what speed. Questions to be asked include if leadership and commitment are sufficient, where leader- ship is situated and if it has sufficient authority; and how this config- uration translates into project structures. Project implementation needs to be carried out by well-qualified staff who are embedded in existing institutional structures or at a minimum have strong links into the political realm. These first two factors—the stage and the actors—must be favorably aligned for the program components to take hold and unleash impact. But the kinds of props the actors employ in the battle (the right reforms), their quality and fit (are reform components technically sound and fit the ensemble of reforms and country context?) as well as the timing of when to bring the props on stage (are the basics in place before more sophisticated reforms are started?) are equally important. Stevens and Teggemann conclude that only in Tanzania was the political environment sufficiently favorable to warrant the pursuit of a comprehen- sive program of administrative reform. In countries (including, in hind- sight, Ghana and Zambia) in which, according to Stevens and Teggemann, “the political environment is less favorable, reform components need to be carefully selected and tailored to start with basic and politically less con- tentious reforms while creating opportunities for more comprehensive reform and generating momentum.” The message that, in most settings, the way forward for administrative reform is likely to be an incremental one is reinforced by the results reported by Engberg-Pedersen and Levy in chapter 3. They find that comprehensive efforts at administrative reform achieved some success in only three of seven sample countries where they were attempted—Tanzania, plus two additional countries with a politically propitious environment (Cape Verde and Uganda)—and have achieved little in the remaining four countries— Ghana and Zambia, plus two additional countries (Guinea and Kenya) where political constraints short-circuited ambitious programs. Among the more narrowly focused reforms, the sample reveals a strik- ing difference in the performance of capacity building initiatives focused on 14 Governance and Economic Development in Africa: Meeting the Challenge of Capacity Building improving expenditure accountability (eight of their sample of nine compo- nents were rated successful) and those focused on human resource manage- ment (only two of eight components were rated successful). To account for this difference, Engberg-Pedersen and Levy suggest that “large groups with vested interests and legitimate claims in the bureaucracy may perceive the [expenditure accountability] activities as more technical and less threaten- ing than the more politically sensitive [administrative reforms], which may have more immediate redistributional consequences.” They speculate that another reason might be the more readily observable results for expenditure accountability initiatives, and they provide extensive suggestions about how a results-focus might more effectively and more broadly be incorpo- rated into administrative reform and state capacity building. Chapters 4, 5, and 6 examine in depth the experiences of selected African countries with specific administrative reforms. In chapter 4, Kiragu, Mukandala, and Morin focus in depth on pay policy, one of the five reforms highlighted by Stevens and Teggemann. Kiragu, Mukandala, and Morin distinguish between technically rational, politically rational, and politically reactive approaches to pay policy and use these distinctions to review empirically the evolution of public sector pay in eight countries. Only Botswana has consistently sustained pay policies and outcomes that approximate accepted notions of good practice in this area. And only Tanza- nia and Uganda have managed to reroute politically reactive policies in more technically and politically rational directions. The remaining five countries (Benin, Burkina Faso, Ghana, Senegal, and Zambia) have been stuck in politically rational or reactive modes, although with some cross- country variation, including a few short-lived episodes of attempted reform. Kiragu, Mukandala, and Morin link these differences in performance to differences across countries in the political constraints and opportunities for pay reform, specifically to cross-country variations in the characteristics of institutional actors, of political system and stability variables, and of civil society. They conclude that the capacity to adopt technically rational pay policies is greatest in settings that are characterized by high levels of politi- cal and administrative institutionalization, but that have relatively low lev- els of political competition (that is, Botswana, Tanzania, and Uganda). By contrast, settings that have high levels of political competition but low insti- tutionalization (Benin, Ghana, and Zambia) seem especially prone to politi- cally reactive approaches to pay reform. They underscore, though, that even [i]f a political leader temporarily has space to be technically rational . . . the crucial challenge is to sustain this space in the face of inevitably Brian Levy 15 rising political competition . . . This need points to the adoption of pay policies that nurture and build institutions and governing coali- tions, rather than those that risk being polarizing. Chapters 5 and 6 focus on capacities related to public expenditure man- agement. The quality of national budget formulation, execution, and report- ing has risen to the top of Africa’s development agenda for two reasons. The first reason is the move to democracy, and the associated expectations that citizens have the right to hold governments accountable for how public resources are used. The second reason is the increased recognition by donors of the centrality of “country ownership” to development effective- ness and a consequent desire to transfer donor resources to African coun- tries in ways that support, rather than circumvent, national processes of prioritization and oversight. As the chapters make clear, the realities on the ground fall far short of idealized, best practice approaches to budgeting. Garnett and Plowden’s point of departure in chapter 5 is the poverty reduction strategy paper (PRSP) process, which in many African countries has become an important basis for resource transfers by the World Bank, the International Monetary Fund, and increasingly, other donors as well. As Garnett and Plowden note, “PRSPs generally have been participatory . . . But the extent to which the policies and programs contained therein will be implemented depends on how the PRSP process relates to the ongoing poli- cymaking processes of government.” They highlight the budget as a key locus where general political promises are translated into concrete actions and hypothesize that “ongoing involvement and consensus-building among cabinet members . . . is key to both fiscal sustainability and sustain- able pursuit of a coherent, prioritized set of policies.” They examine this proposition empirically by exploring the processes of budget formulation in Mozambique, South Africa, Uganda, and Zambia. In all four countries, the finance ministry emerges as the leading player, but large variations are evident across countries both in its effectiveness and in the extent to which its decisions are embedded in a process that builds effective consensus among the leadership about the legitimacy of the result. Neither Zambia nor Mozambique have been able to implement effective processes for prioritization. In Zambia an ad hoc supplementary budget appropriations process undercuts the initial budget almost to the point of irrelevancy. In Mozambique large-scale funding from donors, which (at least until very recently) has been separated from the national budget, has given de facto fiscal independence to favored line ministries. Uganda, by contrast, has been highly effective in prioritizing public spending, but its process has been dominated almost entirely by a dominant and capable 16 Governance and Economic Development in Africa: Meeting the Challenge of Capacity Building Ministry of Finance, Planning, and Economic Development, which has enjoyed the firm backing of the powerful national president. Only in South Africa do Garnett and Plowden document a clear commitment to a more consensual prioritization process, although this commitment, too, is a con- tinuing evolution from the dominant-finance-ministry model, which pre- vailed in the immediate postapartheid period. In sum their case studies suggest that the road linking a participatory PRSP process to a genuinely inclusive and consensual process of budget formulation is, at best, a long and circuitous one. In chapter 6 Dorotinsky and Floyd report and reflect on the results of efforts to benchmark the quality of expenditure accountability systems in 20 African countries—not only budget formulation, but also budget execution and budget reporting. In 2001 staff of the World Bank, the International Monetary Fund, and 20 participating countries worked together to bench- mark 15 distinct technical features of budgeting. “The initial results,” Dorotinsky and Floyd conclude, “are sobering. The state of public expendi- ture management systems in many countries emerged . . . as poor, with weaknesses in some very basic elements of these systems. Moreover, the reform of expenditure accountability systems is turning out to be more diffi- cult than expected.” Budget formulation emerges as least problematic, relatively speaking. Two-thirds of the 20 countries meet at least half of the seven benchmarks in this area. By contrast, only 50 percent of the countries meet at least two of the four budget execution benchmarks. To improve budget execution, almost three-fourths of the benchmarked countries have begun to invest in computerized financial management information systems. Dorotinsky and Floyd review the interaction between computerized financial management information system investments and the quality of budget execution, and they conclude that technology can add value only in the context of an underlying commitment to disciplined decisionmaking and internal man- agement systems that are geared to monitoring compliance. Budget reporting is highlighted “as a core element of transparency, empowerment, and accountability,” but because only 40 percent of coun- tries meet at least two of the four benchmarks in this area, it emerges as the weakest of the three areas. As Dorotinsky and Floyd highlight, many coun- tries have put in place rules of the game that formally allow supreme audit institutions to scrutinize expenditures independently and report the results publicly. In practice, however, impact has been limited in the face of • Public accounts, which are often inaccurate, cover only part of public expenditures, and are made available only after long delays Brian Levy 17 • Supreme audit institutions that lack the skill base to do their job effectively • Extremely uneven parliamentary oversight. Yet for all of the shortcomings of the previous status quo, Dorotinsky and Floyd find reasons to be hopeful that the 2001 benchmarking exercise might have been an important wake-up call. For one thing a follow-up implemen- tation update conducted in 2003 documented intensified country-level activ- ity to fix the shortcomings, and significant progress was being made among the lowest rated countries. For another the preoccupation with national bud- get systems reflects a growing willingness in principle on the part of donors to provide aid in the more fungible form of budget support, and both the intensified efforts to strengthen national budget systems and the evident link between stronger national budget systems and development effectiveness seem to be helping accelerate this process. Finally, the intensified focus on expenditure accountability comes at a time when (as highlighted throughout this book) development practitioners are acknowledging the limitations of comprehensive public sector capacity building efforts. The risk is thus diminishing that efforts to improve expenditure accountability will become lost in the crowd of an overloaded reform agenda. This last point highlights a more general proposition. For each of the reform areas considered in chapters 2–6, a well-understood set of mecha- nisms has the potential to align the administrative system to support devel- opmental goals. Each of the chapters takes these normative best practices as the starting point and examines the degree to which they have been realized in practice. In all but a few cases the results so far have fallen far short of the normative model. What plausibly underlies this implementation shortfall is the politicians’ dilemma highlighted by Kiragu, Mukandala, and Morin in chapter 4 in the context of pay policy: the need to balance the technocratic logic that shapes many proposals for reform against the political impera- tives of building and sustaining alliances with powerful patrons, of avoid- ing conflict with powerful social groups, and of maintaining electoral support. Depending on the severity of the political constraints, even well- intentioned leaders are thus limited in how ambitious a reform agenda they can adopt. A key lesson from chapters 2–6 is that it can be counterproduc- tive to ignore these constraints and bulldoze along with an agenda of com- prehensive best practice solutions. In most countries the emphasis could more usefully be placed, first, on approaches to administrative reform that are more partial and more tailored to country-specific windows of opportu- nity and, second, on determined follow-through to achieve results in the few areas selected for focused effort. 18 Governance and Economic Development in Africa: Meeting the Challenge of Capacity Building Strengthening Outward Accountability Because it has become clearer that the challenges of building state capacity are at least as political as they are technical, the focus has intensified on the demand side of capacity. The intent is to alter the incentives of political lead- ers by reshaping state institutional arrangements in ways that require them increasingly to respond to a broad array of civic pressures for performance and not simply to the elites who benefited from the status quo. Broadly speaking these efforts to strengthen outward accountability can be divided into two subcategories: first, efforts to strengthen horizontal accountability by strengthening the direct links between the central government and the citizens and, second, efforts to strengthen downward accountability by transforming institutional arrangements to shift resources and responsibil- ity from central government to other subnational or community actors,7 who are closer to the service delivery frontline and hence have potentially stronger ties to users. Whatever the specific horizontal or vertical institu- tional transformation, new capacities will be needed across the board to support these new roles. The central bureaucracy, for example, will have a changed—but not necessarily a less important—role, a role that is more transparent and more participatory in its actions and is focused more on strategy, regulation, and monitoring than on the direct delivery of services. Strengthening Horizontal Accountability Chapters 7 and 8 examine the development of capacity with respect to two very different facets of horizontal accountability. Chapter 7 focuses on the evolution, role, and impact of national parliaments in four African democra- cies. As representatives of the “voice of the people,” as the source of legiti- mate legislative authority, and as one of the two branches of the state responsible for executive oversight (the other is, of course, the judiciary), national parliaments have, in principle, a fundamental, dominant role in assuring horizontal accountability. Chapter 8 focuses less on a specific insti- tution than a strategy for horizontal accountability, namely the efforts to engage citizens in high profile campaigns against corruption. The two chap- ters make clear that, as with efforts to strengthen administrative capacity, efforts to build horizontal capacity will not succeed if they are based on some idealized view of best practice systems. Rather, the starting point must be a realistic view of how horizontal accountability operates in practice. 7. Although some might include privatization in the menu, the focus in this book is on activities for which the domain of responsibility remains collective. Brian Levy 19 Barkan, Adamolekun, and Zhou’s empirical investigation identified large disparities in the effectiveness of parliaments across four countries, all of which are formally democratic, and also within countries over time. The Kenyan Parliament of the past five years emerged as unequivocally the most independently assertive. The Ghanaian and Beninese legislatures are described as semi-independent and certainly more independent as of 2002 than 10–15 years earlier. The Senegalese legislature, however, continued to be almost entirely subservient to the executive. These variations in indepen- dence translate into variations in how parliamentarians allocate their time between policy-related and constituency-support activities. The Kenyans are the most (and the Senegalese are the least) preoccupied with policy- related activities. But even in Kenya, real engagement with the budgeting process (as distinct from other aspects of policymaking) is limited, and even this engagement tends to focus narrowly on the implementation of spend- ing commitments within the districts of individual members. A combina- tion of constitutional rules and, as explored in chapter 5, executive control over the timing of the process precludes significant parliamentary influence on budget formulation. Furthermore multiyear delays in the presentation of audits have led some parliamentarians to refer disparagingly to audit com- mittees as the “postmortem committees.” To account for the observed variations in the effectiveness of parliaments, Barkan, Adamolekun, and Zhou focus on the impact of four distinct factors. First is the consistent pull, evident in all countries, toward patronage-driven, deliver-the-goods-to-constituents behavior by legislators—a result of the rural, local-geography-focused organization of political interests. Second are differences in formal rules, which Barkan, Adamolekun, and Zhou acknowledge explain part—although they insist only a modest part—of the cross-country variations. The third factor, which they highlight as a power- ful proximate source of variation across countries, comprises differences across countries in the pay and institutional resources available to parlia- mentarians. They note, though, that “the amount of resources available to legislators . . . is unlikely to rise unless forcefully demanded by the legisla- tors themselves.” They argue that the composition of legislatures has shifted in favor of “reformers” and opportunistic “patronage-seekers” at the expense of “incumbent authoritarians,’” and they link this shift to a combi- nation of a generational change (younger legislators are better educated and more globally connected), the rise of civic activism among urban elites, and the rise of multiparty democracy across Africa. Because they locate the changes in legislative performance in the assertiveness of legislators them- selves, and because they link this assertiveness to shifts in legislator compo- sition as a result of domestic social changes, Barkan, Adamolekun, and 20 Governance and Economic Development in Africa: Meeting the Challenge of Capacity Building Zhou conclude on the cautionary note that “the modernization of the African legislature . . . cannot be ‘orchestrated’ from outside.” In chapter 8, Kpundeh reviews recent initiatives across Africa to combat corruption; anticorruption emerged in the mid-1990s as an especially potent means of mobilizing civic energy for good government. Kpundeh distin- guishes between structural interventions aimed at strengthening the capacity, transparency, and accountability of state institutions (that is, interventions of the kind described throughout this book) and softer process interventions. The latter include efforts to mobilize civic pressure for change (for example, by documenting transparently the extent of the corruption problem); work- ing to reshape the attitudes of public officials; and engaging in high-profile, but often ad hoc and politically opportunistic, actions to crack down on cor- rupt political and bureaucratic leaders. Much of the energy in the late 1990s focused exclusively on these process interventions, and Kpundeh highlights the risk of this partial approach, namely, that [i]n the absence of the sustained results that can come only from structural reforms, citizens increasingly will become disenchanted with the emptiness of their leaders’ exhortations and withdraw again from the public arena, leaving the terrain free for a new round of mis- governance. Dedicated anticorruption agencies frequently are established as a quick- fix response to the problem. Kpundeh examines their impact through the lens of this structure–process dichotomy; he notes very different trajectories across countries. Botswana’s Directorate for Corruption and Economic Crimes operates against the backdrop of strong complementary state insti- tutions and is a useful part of the overall arsenal of tools to ensure that pub- lic resources are well used. By contrast, Malawi’s Anti-Corruption Bureau and Sierra Leone’s Anti-Corruption Commission are criticized as “phony populist” mechanisms—typical examples of process interventions meant to give the appearance of action while changing little. Tanzania’s Prevention of Corruption Bureau, Uganda’s Inspectorate of Government, and Zambia’s Anti-Corruption Commission (that is, since that country’s change of presi- dential leadership in 2002; earlier it had been a typical example of phony populism) occupy an intriguing intermediate space. They are effective mechanisms for combating corruption in the short term, en route to more far-reaching comprehensive reforms of the public sector. According to Kpundeh a combination of, first, strong top-level leader- ship (including the readiness and authority to override opposition from oth- erwise influential elites) and, second, a sustained partnership with civilian Brian Levy 21 actors committed to opposing corruption distinguishes the more effective short-term transitional mechanisms from their phony populist counter- parts. More broadly he concludes that the most effective way forward in the fight against corruption appears to be through a dynamic interaction between process and structural interventions, with the intent of initiating a virtuous cycle—a means to educate and move changes forward. Increased participation generates pressure for structural interven- tions that improve public performance, and the resultant structural interventions sustain change by encouraging more participation, education, and hence another round of calls for more refined struc- tural interventions. How might a virtuous cycle progressively transform the environment for horizontal accountability? Taken together, the conclusions in chapters 6–8 point to an intriguing opportunity: • The discussion on budget reporting in chapter 6 highlights the weak- nesses in many African countries of ex post audit systems, but notes that (given recent improvements in the legal framework) some straightforward ways of getting improvements exist and points also to rising interest among donors to provide the requisite support (as part of the broader interest in expenditure accountability). • Implicit in the finding in chapter 7 that parliamentarians disparage the “postmortem” audit committee is the prospect that, if audits were more timely and of higher quality, with greater likelihood of follow- up by the executive, the function of budget oversight would be taken more seriously. • Chapter 8 points to a growing maturity in anticorruption process interventions and an increasing focus on leveraging constructive changes in transparency and accountability institutions, rather than on quick-gratification statements of intent accompanied by politically ambiguous efforts to fry a big fish, so to speak. It follows that actions to strengthen supreme audit institutions, to inten- sify civic interest and engagement in how public resources actually are used, and to support parliamentary efforts to strengthen their expenditure oversight functions potentially are mutually reinforcing. More reliable, timely, and transparent information and the resulting intensified scrutiny on public spending can provide momentum for investment in systems to improve expenditure accountability. The resulting improvements in trans- parency can, in turn, make civic activism more effective. The capacities 22 Governance and Economic Development in Africa: Meeting the Challenge of Capacity Building of parliaments are likewise mutually reinforcing in relation to civic activists. Effective activists press parliamentarians to perform better, and well-functioning parliaments provide valuable points of leverage for activists. Parliaments are also mutually reinforcing in relation to systems of expenditure accountability. Better systems to account for how public monies are used enable parliamentarians to do their job better, and empowered parliamentarians are better positioned to push for improved information from the bureaucracy. If a virtuous spiral can be initiated across these different dimensions of expenditure accountability, the cumu- lative consequence even of some relatively limited initial actions could be far-reaching improvements in the horizontal accountability of the public sector and thereby progressively in public performance. Strengthening Downward Accountability Ordinarily downward accountability is viewed as synonymous with decen- tralization—that is, transforming intergovernmental systems in ways that empower local governments, which are accountable to their subnational electorates. This pattern is indeed the principal focus of chapters 9 and 10, although as both chapters make clear, the range of downward accountabil- ity options is broader. The political and economic virtues of democratic decentralization are perceived by its advocates to be far-reaching. The political virtues include, first, restraining arbitrary action by national leadership by expanding the number of check-and-balance points built into the constitutional framework and, second, deepening democracy and citizen empowerment by providing focal points at all levels of society for civic contestation. The economic virtues include, first, improving allocative efficiency by shifting decisions about how to prioritize down to the lowest feasible level and, second, improving the productivity of public resources by creating more direct mechanisms for citizen-users to pressure public officials for improved per- formance. The extent to which achievement of this ambitious agenda is con- sistent with the political realities on the ground is a key empirical question. In chapter 9 Ndegwa and Levy use case studies of Malawi, Senegal, and Uganda to explore in depth the political dynamics of decentralization. As they show, for a rapid and sustained decentralization to proceed effectively, a formidable set of enabling political conditions needs to be in place. Uganda came closest to meeting these, but even there the process proceeded in a way that left local governments potentially subordinate to central authorities (even in areas in which responsibility formally was assigned to Brian Levy 23 the local level). In Malawi and Senegal, the political conditions were more constraining and had corresponding limitations in the extent to which, at least so far, local governments have been empowered. Malawi’s high-profile Malawi Social Action Fund (MASAF) illustrates the ambiguities nicely.8 Since its inception in 1995, MASAF has made available well over US$100 million for local infrastructure investments. By all accounts, the resources have been used effectively, cost efficiently, and in ways that have been empowering to local communities. What remains unclear, though, is if MASAF will turn out to be an important stepping stone to genuinely accountable local government or if it will ossify as a populist centrally con- trolled vehicle for winning support for the incumbent political leadership.9 The Malawi and Senegal case studies lead Ndegwa and Levy to a variety of propositions about how to proceed with reforms to foster downward accountability in politically constrained settings. First, they suggest that at least as much attention should be given to building the requisite political coalitions for change as to refining the technical details of optimal decentral- ization. The case studies suggest that Senegal, and perhaps Malawi, may have fallen into the trap of analysis as a substitute for action. Second, in politically constrained settings, they advocate greater openness to hetero- dox, institutionally messy initiatives that strengthen downward account- ability and that can spur forward the process of political change. MASAF offers one example; community schools, highlighted in chapter 10, another. In chapter 10, Gershberg and Winkler assess three distinct approaches to education decentralization in Africa. The first two approaches comprise the assignment of responsibility for education to elected subnational govern- ments, either to an intermediate, regional tier (described in some countries as a provincial or state tier) or to smaller, more compact local authorities (for exam- ple, urban municipalities). Gershberg and Winkler note that, the world over, 8. Social funds along MASAF lines have been used in more than a dozen African countries, including on a large scale, Eritrea, Ethiopia, Ghana, Guinea, Madagascar, Nigeria, Senegal, Tanzania, and Zambia. 9. Although none of the other African countries that have social funds yet pro- vide guidance in this regard, the contrasting experience of two Latin American countries, Bolivia and Peru, is revealing. Peru’s social fund became progressively more closely tied to the office of President Fujimori, part of a populist machine, but with no long-term gains for institution building. By contrast, Bolivia successfully used the empowerment structures nurtured through its social fund to build unusu- ally robust mechanisms of downward accountability when it promulgated its Popu- lar Participation Law, which shifted control of the resources to newly empowered local governments. 24 Governance and Economic Development in Africa: Meeting the Challenge of Capacity Building shifting responsibility for education to subnational authorities generally is dri- ven not so much by sector-specific objectives as by a more comprehensive agenda of state transformation, with the aim of strengthening more broadly the relationship of accountability and legitimacy between government and cit- izens. Although opposing this agenda is counterproductive for the education sector, they suggest that no evidence shows that this agenda leads to direct improvements in education service delivery—in Africa or elsewhere. They illustrate this last point with the examples of Ethiopia, Nigeria, and South Africa, three leading African examples of decentralizing education responsibilities to regional authorities. Ethiopia’s regional authorities simply replicated at their level the top-down control structures, which until now had operated nationwide, with no gains in the operation of the system. South Africa’s central government retained a shared responsibility for edu- cation, and exercised its authority through regulatory and budget control that was so tight it tied the hands of regional authorities almost completely. In Nigeria responsibility for education seems to have been a political and patronage football, tossed back and forth between central, regional, and local tiers in ways that radically obscured who was actually accountable for sys- tem performance. Gershberg and Winkler find decentralization to local (rather than regional) governments to be rare in Africa. They examine two purported examples, Tanzania and Uganda, but find that, especially in Tan- zania, the reality is more one of continued control by the central education ministries through their deconcentrated local offices.10 Uganda’s experience is noteworthy, though, for its highly successful initiative to empower parent- teacher school management committees, including by direct monthly, grant payments to these committees; the amounts of these grants are posted on public notice boards. Community schools make up the third approach considered by Gersh- berg and Winkler. They describe community control of schools as “the most common form of educational decentralization in Sub-Saharan Africa,” and identify over a dozen countries in which these schools play a significant role in the education system (they make up, for example, 32 percent of primary schools in Mali and 50 percent in Tanzania). Interestingly despite the fact that the mushrooming of community schools was in part a response to state failure, it is an institutional form that is especially well-aligned with one of 10. Indeed deconcentration might be a candidate to be a fourth approach to edu- cation decentralization. It falls short of such a categorization, though, because it is purely an administrative reform and provides no necessary change in the account- ability relationship. Brian Levy 25 the most important positive lessons of international experience highlighted by Gershberg and Winkler. The lesson, for which they report some evidence from Africa, is that empowering parents and communities in school gover- nance reduces absenteeism of teachers and students, improves learning out- comes, and helps more broadly to strengthen civic engagement. The evident benefits of community schools underscore Ndegwa and Levy’s conclusion about the importance of openness to a variety of ways of fostering downward accountability. As with other arenas for state reform and capacity building considered in this book, a preoccupation with techni- cal best practice approaches (in this case to decentralization) can lead to paralysis, and to the neglect of politically feasible options (in this case for both improving downward accountability and achieving short-term devel- opmental gains at the service delivery frontline). But the opposite also holds true: an exclusive preoccupation with quick-fixes at the frontline, without attention to overall system coherence also is likely eventually to prove unsustainable—both because of the ad hoc nature of the technical arrange- ments for resource flows and, more fundamentally, because without more broadly institutionalized checks and balances, changes in political winds too readily overwhelm ad hoc innovations. As Ndegwa and Levy underscore, by virtue of both its institutional coherence and the way in which it locks-in a transformation of how power and resources flow through the polity, decen- tralization via intergovernmental systems reform should remain the guiding North Star to which more pragmatic and diverse strategies adopted in the short and medium term should converge over time. What seems to be called for, then, is a creative response to opportunities as they emerge for enhancing downward accountability, never losing sight of, but not being paralyzed by, the long-run imperatives of overall institutional coherence. An Emerging New Approach Taken together, the chapters in this book point to a paradigm shift in the analysis and operational approach to building state capacity—from a nar- row focus on organizational and public management approaches to a broader perspective that incorporates both the institutional rules of the game within which public organizations operate and the political dynamics. Four propositions summarize the emerging new approach. Moving pro- gressively more upstream from the bureaucratic frontline, these proposi- tions are the following: • Bureaucracies are agents of political principals; political principals set objectives, which bureaucracies are charged with implementing. 26 Governance and Economic Development in Africa: Meeting the Challenge of Capacity Building It follows that the behavior of public bureaucracies cannot be under- stood without attention to their politically derived objectives, and that the capacities of public bureaucracies cannot effectively be built in isolation from broader systemic changes, including strengthening the mechanisms for clarifying political purposes and communicating them to bureaucracies. • Improvements in systems of accountability strengthen pressure for performance on public actors. Thus an important focus of efforts to improve public performance should be building the capacities of accountability systems. These capacities include (a) hierarchical con- trol structures within bureaucracies, (b) downwardly accountable governance structures through which elected political leaders moni- tor bureaucratic behavior and through which citizens provide feed- back to politicians, and (c) downward accountability mechanisms that link citizen users more directly with service providers. Note that enriching the array of accountability systems also helps strengthen the checks and balances that restrain arbitrariness, focus government action on public purposes, and underpin political stability. • Political institutions and the structure of political interests that underpin systems of accountability are country specific, so efforts to build state capacity need to fit well with country-specific realities. Therefore the approach to building state capacity cannot be an undif- ferentiated, best practice, cookie-cutter approach. Rather, the central issue is understanding country-specific constitutional structures and patterns of political, social, and economic interests and to aim for a good fit between efforts to strengthen administrative and account- ability systems and these country-specific realities. • Change processes can be cumulative. This proposition follows directly from interdependence. Change in any one part of the system potentially induces change in other parts as well. It directs attention toward the identification of entry points for reform and capacity building, which, even though modest in themselves, have the poten- tial to catalyze further changes down the road, and thereby to pro- gressively expand the opportunities for building state capacity. Table 1.4 contrasts a traditional, technocratic, and managerial approach to building state capacity, depicted in column I, with the modifications, depicted in column II, that come from incorporating into the framework the institutional and political insights highlighted by the four propositions. The role of the two-way arrows warrants emphasizing up front: as the chapters in this book make clear, the emerging challenge is to harness the necessary, Brian Levy 27 Table 1.4. Building State Capacity—The New Approach Areas of contrast I II Analytical orientation Technical ⇔ Political Supply ⇔ Demand Focus areas Bureaucracy ⇔ Oversight mechanisms Central ⇔ Local Design principles Best practice ⇔ Next steps, good fit Comprehensive ⇔ Focused, results oriented Structure ⇔ Process Note: Column I depicts a traditional, technocratic, and managerial approach to building state capacity. Column II depicts the modifications that come from incorporating into the frame- work the institutional and political insights highlighted by the four propositions listed in text. Source: Author. specialist public management expertise highlighted in column I to processes of change management, which are embedded in a clear-sighted view of the underlying drivers of political and institutional change highlighted in col- umn II. Put differently, the new approach represents an extension and refinement of earlier efforts—not their abandonment. Three sets of contrasts between the traditional and newer approach are highlighted in table 1.4. The first contrast is in analytical orientation. The tra- ditional approach bestows privileges to specialist knowledge of the details of how public administrative systems function and has a corresponding analyt- ical focus on how to enhance the supply and use of human resources. The new approach requires also a skill in analyzing country-specific political institutions and dynamics and, based on that analysis, identifying potential entry points for enhancing demand for improved public performance. The second contrast is in the areas of focus. The new approach expands the focus areas beyond central bureaucracies to include the oversight mech- anisms for holding bureaucracies accountable and also, because of the enhanced opportunities afforded at community level for direct citizen par- ticipation and oversight, the institutional arrangements for public action at subnational and local levels. The third contrast is in the design principles underlying efforts to build state capacity. A typical traditional program attempted to emulate best prac- tice structures in industrialized countries (commonly, in the 1990s, the New Zealand model). Often, it also attempted comprehensive reform, on the pre- sumption that a chain is only as strong as its weakest link. The new approach works to modify best practice to achieve a good fit with country-specific 28 Governance and Economic Development in Africa: Meeting the Challenge of Capacity Building realities. Rather than remaining preoccupied with politically unachievable comprehensive reforms, the focus is on more modest, viable initiatives, espe- cially those for which results are observable. Careful attention is given to orchestrating the reform process in a way that not only helps facilitate suc- cess with the immediate agenda, but more broadly opens up opportunities for subsequent, cumulative change. Sequencing emerges in the new approach as central. Given country- specific realities, what is likely to be the most effective strategy for building state capacity? How should that strategy prioritize and sequence across the various arenas for possible engagement? Clearly, no one-size-fits-all answer to these questions exists. Even so, the patterns and lessons explored in this book offer some pointers about how to align a capacity building strategy to country-specific starting points. At one end of the spectrum are countries whose starting points for fur- ther enhancement of state capacity are strong. Political leadership has a clear developmentally oriented vision and a mandate for acting on that vision. A workable baseline of bureaucratic capabilities exists. Such settings provide the opportunity for embarking on a comprehensive program of state capacity building, as exemplified in this book by Tanzania’s program of administrative reform and capacity building (chapter 2) and the far- reaching program of decentralized capacity building in Uganda (chapter 9). At the other end of the spectrum considered in this book (which is con- siderably narrower than the actual, Africa-wide spectrum, since conflict- affected countries have not been included in the underlying empirical work) are countries that have been part of Africa’s democratic opening and that enjoy political stability, but that continue to operate largely along neopatrimonial lines. A consistent message of the studies in this book is that, in such countries, comprehensive programs of state capacity building are certain to fail. What is called for instead are some modest initial steps to identify, and build on, entry points for state capacity building that, although modest in themselves, have the potential to set in motion far-reaching cumulative changes. The expenditure accountability initiatives explored in chapters 6 and 8 seem to have such potential. The bulk of countries have an uneven landscape of reformers and neopatrimonials and lie more to the middle of the spectrum between bureaucratic capability and bureaucratic dysfunction. For this group, the opportunities for reform and capacity building will be spread idiosyncrati- cally across the institutional landscape, driven by where the champions happen to be active. Promising examples highlighted in this book include the strong positive impact on policy prioritization of Uganda’s investment Brian Levy 29 in the capacity of its Ministry of Finance, Planning, and Economic Develop- ment (chapter 5); Malawi’s social fund (chapter 9); community schools in Benin, Burkina Faso, and elsewhere in francophone Africa (chapter 10); Kenya’s Parliament (chapter 7); and the promising array of World Bank–supported expenditure accountability operations (summarized in chapter 6). As all of these examples attest, the key to moving forward is a readiness to live with messiness and with imperfection in relation to ideal technical designs, but at the same time not to lose sight of the overall, more coherent, long-run constellation of state capacities and institutional interre- lationships toward which the individual efforts are headed. In sum, the central message of this book is that when approaching the challenge of state capacity building, reformers must find a middle ground between the bipolar moods that have for decades plagued developmental theory and practice: exuberant optimism that some magic formula for development has been found, followed inevitably by deep disappointment over its limitations. By contrast, the way forward outlined here might best be described as hopeful realism. Realism, in the recognition that only in a few African countries (and, even there, perhaps only for a limited period of time) is the way forward likely to yield major gains in the short term. The world over, the work of building states that are both effective and account- able to their citizens is a centuries-long process; Africa’s independence from colonial rule came less than half a century ago. As the many country exam- ples presented throughout this book signal, hope in the conviction that, irre- spective of a country’s initial circumstance, some way forward for building state capacity is there to be tapped. Indeed, the process of cumulative causa- tion points to the possibility that small beginnings can set in motion pro- gressively more profound consequences. The Chinese, with two millennia of experience in state building capture the point best in the proverb by Lao- tzu: the journey of a thousand miles begins with a single step. Appendix 1A. Governance, Adjustment, and Economic Performance in Africa This appendix extends the empirical analysis in the “Governance and Eco- nomic Performance: Some Empirical Patterns” section in three ways. First, it uses the governance diamond of figure 1.1 to clarify more precisely both the key causal relationships that led to Africa’s neopatrimonial downward spi- ral and the extent and limitations of the subsequent governance turn- around. Second, it provides more detailed country-specific data on the relationship between adjustment and subsequent performance. Third, it 30 Governance and Economic Development in Africa: Meeting the Challenge of Capacity Building provides data that help disaggregate country-by-country governance trends into some specific subcategories. Cause-Effect Relationships in the Governance Diamond Figures 1A.1–1A.5 use the governance diamond to depict schematically the evolution since independence of governance in Africa. Figure 1A.1 maps each of the seven cause-and-effect relationships from figure 1.1 into a 4×4 matrix. The causal relationships in the figures should be read as flowing from row to column. The second cell in the political interests row, for example, signifies the impact of political interests on formal political institutions (corresponding to C3 in the governance diamond), and the first cell in the formal political institutions row signifies the impact of formal political institutions on political interests (corresponding to C2 in the governance diamond). Figure 1A.2 highlights how these relationships generally play out in well-functioning, mature democracies. Each individual relationship can be positive (indicated with asterisks) and help strengthen and sustain the developmental orientation of the governance diamond as a whole, fostering political openness, strong public sector performance, and a productive pri- vate economy, or negative (signified using x’s) and put pressure on the sys- tem, causing developmental dysfunction. See the legends with figures 1A.2–1A.5 for an explanation of the degrees of positive and negative rela- tionships. Two features of figure 1A.2 are noteworthy for the subsequent analysis. First, note the presumption, which is reflected in the negative impact of political interests on both formal political institutions and the bureaucracy—that is the x’s in the political interests row of figure 1A.2— and is presumed to apply universally, that political interests engage in self- interested lobbying, potentially to the point of bypassing both the formal institutional arrangements that are intended to structure this lobbying, and the agency relationships that are intended to insulate bureaucracies from informal lobbying. But, second, note that in well-functioning, mature democracies the five remaining causal relationships work to counteract the potentially corrosive effects of this lobbying: • Formal institutions are strong and thereby check extralegal activities by political interests, govern bureaucracies in ways that hold them accountable for achieving public ends, and provide credible signals of stability to private investors. • Accountable bureaucracies make policy and they regulate and deliver services in ways that support private economic activity. Brian Levy 31 • A diversified economy provides a variety of different business activi- ties so that putative lobbying efforts by these dispersed interests are likely to offset each other. The net result is a virtuous spiral (or, alternatively, a stable equilibrium) that supports openness, strong public sector performance, and a productive pri- vate economy. As will be argued in the remainder of this chapter, Africa’s development dilemma is precisely that such a virtuous spiral cannot be taken for granted. Figure 1A.1. The Key Relationships Formal Political political interests institutions Bureaucracy Economy Political C3 C4 interests Formal political C2 C5 C6 institutions Bureaucracy C7 Economy C1 Note: See figure 1.1 for an explanation of C1–C7. Source: Author. 32 Governance and Economic Development in Africa: Meeting the Challenge of Capacity Building Figure 1A.2. Well-Functioning, Mature Democracies Formal Political political interests institutions Bureaucracy Economy Political x x interests Formal political *** ** ** institutions Bureaucracy ** Economy * *** Strongly positive. ** Positive. * Weakly positive. 0 Neutral. x Weakly negative. xx Negative. xxx Strongly negative. Source: Author. Figures 1A.3 and 1A.4 contrast the condition of African governance at inde- pendence with its condition once the neopatrimonial downward spiral, intro- duced in “Governance and Economic Performance: Some Empirical Patterns” had taken hold. Consider, first, the interactions between political interests and formal political institutions. The strong buttress provided by political Brian Levy 33 institutions in a mature democracy (C2 is scored with three stars in figure 1A.2) contrasts with Africa’s more fragile institutional legacy at independence (C2 is scored with one star in figure 1A.3). Insofar as formal institutions failed to constrain politics, power increasingly became exercised through informal channels that were built around relationships of patronage, not through Figure 1A.3. Africa, around 1965: The Institutional Legacy at Independence Formal Political political interests institutions Bureaucracy Economy Political xx xx interests Formal political * ** ** institutions Bureaucracy * Economy 0 *** Strongly positive. ** Positive. * Weakly positive. 0 Neutral. x Weakly negative. xx Negative. xxx Strongly negative. Source: Author. 34 Governance and Economic Development in Africa: Meeting the Challenge of Capacity Building Figure 1A.4. Africa, around 1980: The Impact of Neopatrimonialism Formal Political political interests institutions Bureaucracy Economy Political xxx xxx interests Formal political 0 x xx institutions Bureaucracy xx Economy xx *** Strongly positive. ** Positive. * Weakly positive. 0 Neutral. x Weakly negative. xx Negative. xxx Strongly negative. Source: Author. governance systems that are both formal and credible. As figure 1A.4 depicts, this fragility led to a progressive deterioration in governance in the two decades after independence. One set of pressures evident in the contrast between figures 1A.3 and 1A.4 was on the bureaucracy, causal relationships C4 and C5 in the governance Brian Levy 35 diamond. As noted in the section on “Governance and Economic Perfor- mance: Some Empirical Patterns,” with the rise of neopatrimonial rule, the mode of governing bureaucracy shifted from the clarification, monitoring, and enforcement of formal rules to informal rules set without transparency, and sometimes capriciously, by a country’s political leadership. The conse- quence, depicted by the negative scores for C4 and C5 in figure 1A.4, was a decline in bureaucratic performance. Now consider changes between figures 1A.3 and 1A.4 in the impact of formal political institutions and the bureaucracy on the economy (causal relationships C6 and C7). To begin with C7, the decline of bureaucracy as a result of neopatrimonialism led, in country after country, to the disruption of markets, rising costs of doing business, urban bias, and rising protection- ism. Perhaps as damaging to growth are the consequences through C6: with the rise in actual and perceived arbitrariness in economic policy, investor confidence and subsequently rates of investment declined. The conse- quence was progressive economic slowdown. Completing the circle, the intensification of a rentier relationship led to the emergence of an increasingly politically dependent business class. Although Africa’s relatively undiversified economy implied that the business class never was especially pluralistic (that is, never generated much by way of internal checks and balances), over time causal relationship C1 shifted to become a source of further intensification of the dysfunctional political dynamic. As summarized in the “Governance and Economic Performance: Some Empirical Patterns” section of this chapter and as detailed further later, adjustment and democratization appear to have halted this downward spi- ral of dysfunctional governance and worsening economic performance. Interpreting this seeming reversal through the lens of the governance dia- mond suggests, however, that the gains remain fragile, too fragile for confi- dence that a sustainable virtuous spiral is underway within Africa’s governance diamond. Consider, first, the status of formal political institutions. Consistent with the data in table 1.2, figure 1A.5 depicts some turnaround in the quality of formal political institutions, to the point that they no longer are a drag on the economy. But these institutions still appear much weaker than at inde- pendence, either as a platform of credibility for private investors or as a vehicle for directing the bureaucracy in a developmental direction. Further- more, the ability of political institutions to contain politics remains weak, although in the wake of some bitter lessons about how terribly wrong things can go, the press of political interests on formal political institutions and the bureaucracy arguably is weaker today than at independence. 36 Governance and Economic Development in Africa: Meeting the Challenge of Capacity Building Figure 1A.5. African Governance around 2000—A Nascent Turnaround? Formal Political political interests institutions Bureaucracy Economy Political x x interests Formal political * x * institutions Bureaucracy x Economy 0 *** Strongly positive. ** Positive. * Weakly positive. 0 Neutral. x Weakly negative. xx Negative. xxx Strongly negative. Source: Author. Nor do we have any reason to believe that economic reforms, and the subsequent private sector response, have generated sufficient structural change to immunize (through C1) the governance diamond against a rever- sal. For one thing, the economies (and the absolute magnitudes of recent changes) remain quite small, typically too small to support the kind of Brian Levy 37 diversified economic base on which workable pluralism might comfortably rest. For another thing, inclusive economic activities of a kind that have been demonstrated to buttress pluralism (Moore 1966)—notably including labor-intensive export-oriented manufacturers, dynamic small and medium enterprises, and smallholder cash crop agriculture—have not so far emerged as major drivers of economic expansion (Wood 2002). A prudent conclusion (reflected in figure 1A.5) might thus be that across many African countries, although the private sector no longer is part of the problem of dysfunctional governance, it also is not yet part of the solution. Adjustment and Economic Performance—A Disaggregation Table 1A.1 provides country-specific data on average annual growth rates over four distinct periods between 1975 and 2000. As noted in the chapter ’s text, within each of the three subcategories are some noteworthy country- by-country variations. Consider first the sustained adjusters. Alone among these eight countries, Zambia’s economy failed to rebound: over the 1990s its GDP growth averaged less than 1 percent per annum. For another, the disaggregated data suggest a possible leveling off of growth rates as economies stabilize to the new policy equilibrium: in Ghana, Malawi, and Uganda growth was somewhat slower for 1996–2000 than for 1991–96. Also (although perhaps reflecting global stimuli) six of the eight countries grew more slowly in 2000 than they did in 1999. Even so, the overall trend is clear. Turning to the later adjusters, again the overall pattern—slow growth prior to policy reform, with subsequent acceleration—is fairly consistent across countries, although with some qualifiers. Note first that seven of the eight countries in the group are francophone, that the 1994 Franc Zone devaluation was a signal event of the 1990s, and that the acceleration of growth in 1996–2000 is consistent with the impact of that devaluation. Sec- ond, note that three countries—Guinea, Mauritania, and Senegal—although not included in the eight sustained-adjuster countries (as defined by their continually remaining on track within the Special Program for Africa), have had relatively stable, and relatively reform-oriented economic policy regimes since the 1980s. Their growth trajectories resemble those of the sus- tained adjusters—steady since the late 1980s, with some acceleration in the 1990s. (Tanzania, the eighth country in the group, also has seen a steady acceleration of growth over the 1990s.) As for the governance polarized third group, the overall pattern—the absence of any evidence of accelerating growth and possibly even a trend decline—disguises especially sharp period-by-period discontinuities in 38 Governance and Economic Development in Africa: Meeting the Challenge of Capacity Building Table 1A.1. Average Annual GDP Growth, 21 African Countries, 1975–2000 (percent) Country 1975–84 1985–89 1991–96 1996–2000 1999 2000 Sustained adjusters 1.7 3.2 3.7 4.7 — — Benin 3.8 1.5 4.4 5.2 4.7 5.8 Burkina Faso 3.6 4.4 4.1 4.9 6.3 2.2 Ghana –1.1 5.2 4.1 4.2 4.4 3.7 Malawi 3.2 1.9 2.7 3.2 4.0 1.7 Mali 2.3 0.8 3.2 5.2 6.1 4.5 Mozambique — 6.0 3.7 8.1 7.5 1.6 Uganda — 3.4 7.7 5.3 7.5 3.5 Zambia 0.2 2.3 –0.1 1.7 2.0 3.5 Later adjusters 2.4 2.7 1.9 4.2 — — Cameroon 8.5 –0.1 –0.2 4.7 4.4 4.2 Chad 0.4 0.7 0.7 3.3 1.0 0.6 Guinea — 4.7 4.5 3.7 3.6 2.0 Madagascar –0.2 2.3 1.4 4.3 4.7 4.8 Mauritania 1.6 3.3 3.2 5.2 4.1 5.2 Niger 2 4.2 0.9 3.1 0.1 2.0 Senegal 2.1 3.5 2.6 5.4 5.1 5.6 Tanzania — — 2.3 4.0 3.6 5.1 Governance polarized 2.3 4.1 2.3 1.4 — — Côte d’Ivoire 2.2 2.0 3.0 2.7 1.6 –2.3 Kenya 4.7 5.9 2.1 1.2 1.3 –0.2 Nigeria –0.7 5.0 2.4 2.4 1.1 3.8 Togo 2.1 3.4 1.9 0.9 2.4 0.7 Zimbabwe 3.0 4.2 2.1 –0.1 –0.7 –4.9 —No consistent data available. Source: World Bank (2002, p.15). individual countries, reflecting in part11 the governance-related instabilities to which these countries have been subject. As tables 1A.2 and 1A.3 highlight, adjustment also has contributed to a productivity-enhancing restructuring of the economies of the more success- ful reformers. As has long been understood, the pre-reform combination of overvalued exchange rates, import controls, and price controls on basic commodities imparted a powerful urban bias to the economic incentives 11. Oil price shocks, combined with economic policies that accentuated rather than muted their impact, are of course an important additional part of the Nigerian story. Brian Levy 39 and structure of many African economies. Thus, trends in agricultural growth provide a useful measure of the extent to which adjustment reforms are having their intended effect of reversing this urban bias. Table 1A.2 summarizes the trends in agricultural value added, organized into the three subgroups of countries. The observed pattern is remarkably con- sistent across all groups—a shift in the mid-1980s in the trend rate of growth from under 1.5 percent to above 3.0 percent. Note that this shift occurred at a similar period for all three groups of countries, reflecting a broad, early recog- nition of the need to correct egregious anti-agricultural policy biases and indi- cating that at least this part of the reform agenda was readily viable domestically. Note also that the gains in agricultural growth were on average somewhat larger among the earlier, more sustained adjusters. Although more disaggregated data (not presented here) reveal some within-group variation across countries, the overall pattern is a broadly consistent one. Table 1A.3 summarizes the trends in industrial value added. Industrial value added is defined broadly, following United Nations Standard Interna- tional Trade Classification categories, to include mining and natural resource processing, not simply narrow manufacturing. Again, the aggregate patterns support the proposition that adjustment contributed both to an acceleration of growth and (given the sharp declines in protectionism that accompanied industrial growth) a productivity enhancing restructuring. Note, though, that Table 1A.2. Average Annual Growth in Agricultural Value Added, 21 African Countries, 1975–2000 (percent) Subgroup 1975–84 1985–89 1991–96 1996–2000 Sustained adjusters 1.4 4.7 4.1 4.1 Later adjusters 1.4 3.1 3.2 3.3 Governance polarized 1.3 3.4 3.0 3.1 Source: World Bank (2002, p. 16). Table 1A.3. Average Annual Growth in Industrial Value Added by Subgroup, 21 African Countries, 1975–2000 (percent) Subgroup 1975–84 1985–89 1991–96 1996–2000 Sustained adjusters 0.6 2.7 3.6 7.4 Later adjusters 4.7 2.3 1.2 5.0 Governance polarized 2.3 5.1 1.7 0.8 Source: World Bank (2002, p. 17). 40 Governance and Economic Development in Africa: Meeting the Challenge of Capacity Building in the industrial subcategory, the cross-country variations within each sub- group are very large. Only seven countries (Burkina Faso, Côte d’Ivoire prior to its current crisis, Guinea, Mali, Mauritania, Senegal, and Uganda) have sus- tained positive and substantial growth in industrial value added since at least the mid-1980s. An additional eight (Benin, Cameroon, Chad, Ghana, Mada- gascar, Mozambique, Niger, and Tanzania) passed through cycles of boom and bust in the 1970s and 1980s, before settling into a path of seemingly sus- tainable recovery in the 1990s. For six countries (Kenya, Malawi, Nigeria, Togo, Zambia, and Zimbabwe), the experience of industrial development over the last quarter century has been less happy, although only in Zambia was the level of industrial value added lower in 2000 than it was in 1980. Sources of Change in Governance Although the results in table 1.2 of the text signaled some positive gover- nance changes, they did not shed much light on which causal relationships from the governance diamond most accounted for these changes. A second set of questions from the 1996 survey (for which responses are summarized in table 1A.2) helps clarify the causal links, by distinguishing sharply between the credibility and the bureaucratic quality dimensions of gover- nance. The survey questions were the following: • In the last 10 years, has predictability of laws and policies increased? Remained the same? Decreased? • In the last 10 years, have difficulties in dealing with government offi- cials increased? Remained the same? Decreased? The former question probes credibility, the latter probes bureaucratic quality. Table 1A.4 reports the net changes over the past 10 years (the pro- portion reporting increased minus the proportion reporting decreased) for the 18 countries for which data are available In nine countries, the proportion of firms reporting an increase in at least one of the predictability and bureaucratic quality dimensions of governance exceeded the proportion of firms reporting a decrease by 20 percentage points or more. A comparison of the relative magnitude of the changes across the two indicators reveals that the governance gains came dispropor- tionately through gains in predictability: the magnitude of change in the predictability scores exceeded that of the bureaucratic quality scores for six countries, with the reverse in only two. Across the nine countries, the aver- age gains in predictability (32 points per country) were double the 16 point average gains in bureaucratic quality (16 points) for those same countries. Brian Levy 41 Table 1A.4. Changes in Credibility and Bureaucratic Quality in 18 African Countries, between 1986 and 1996 (net change, percent) Change in Changes in difficulties predictability with officials Country (+ = improvement) (+ = improvement) Average for sustained adjusters +34 +16 Benin +28 +9 Ghana +35 +37 Malawi +21 –4 Mali +45 +25 Mozambique +68 –4 Uganda +33 +20 Zambia +9 +31 Average for later adjusters –3 –29 Cameroon –7 –79 Chad –8 –38 Guinea +25 +25 Madagascar –26 –26 Senegal +12 –12 Tanzania –11 –43 Average for governance polarized –5 –12 Côte d’Ivoire +22 +2 Kenya +12 –3 Nigeria –19 –21 Togo –6 –28 Zimbabwe –32 –8 Source: Survey conducted for the World Bank; World Bank (1997). The disproportionate gain in predictability relative to bureaucratic qual- ity suggests that governance gains associated with economic reform may be rooted in large part in changes in expectations, rather than actual improve- ments in the performance of public institutions. Although changes in expec- tations can be sufficient to set in motion accelerated private investment and growth, they will not suffice to secure improvements in the provision of public services in the short and medium run. Nor does a largely perceptual shift give confidence that two critical roles of political institutions high- lighted in the governance diamond, the container of politics (C2) and the disciplining principal in relation to the bureaucratic agent (C5), indeed are 42 Governance and Economic Development in Africa: Meeting the Challenge of Capacity Building the source of the observed turnaround. But if political institutions remain fragile, broader questions arise about the sustainability of reform. References Bratton, Michael, and Dominique Van Der Walle. 1997. Democratic Experiments in Africa. Dia, Mamadou. 1996. Africa’s Management in the 1990s and Beyond: Reconciling Indige- nous and Transplanted Institutions. Directions in Development Series. Washington, DC: World Bank. Kaufmann, Daniel, Aart Kraay, and Massimo Mastruzzi. 2003. “Governance Matters III: Governance Indicators for 1996–2002.” Policy Research Working Paper no. 3106. World Bank, Washington, D.C. Kaufmann, Daniel, Aart Kraay, and Pablo Zoido-Lobaton. 1999. “Governance Mat- ters.” Policy Research Department Working Paper no. 2196. World Bank, Wash- ington, D.C. Lewis, Peter. 1996. “Economic Reform and Political Transition in Africa: The Quest for a Politics of Development.” World Politics 49(1): 92–129. Migdal, Joel S. 1988. Strong Societies and Weak States: State-society Relations and State Capabilities in the Third World. Princeton, NJ: Princeton University Press. Moore, Barrington, Jr. 1966. Social Origins of Dictatorship and Democracy. Boston: Bea- con Press. North, Douglass. 1990. Institutions, Institutional Change, and Economic Performance. New York: Cambridge University Press. Wood, Adrian. 2002. “Could Africa Be Like America”? Paper presented at 14th Annual Bank Conference on Development Economics, April, Washington, D.C. World Bank. 1994. Adjustment in Africa. Washington, D.C. _____. 1997. World Development Report 1997: The State in a Changing World. New York: Oxford University Press. _____. 1999. Civil Service Reform: A Review of World Bank Assistance. Operations Evalua- tion Department. Report no. 19211. Washington, D.C. _____. 2002. African Development Indicators. Washington, D.C. 2 Comparative Experience with Public Service Reform in Ghana, Tanzania, and Zambia Mike Stevens and Stefanie Teggemann This chapter examines the public service reform experience of three anglo- phone African countries—Ghana, Tanzania, and Zambia—which have embarked on what might be termed second-generation public service reform programs. Each one is distinctive, yet all three have common threads running through them. Performance has varied with country, and although they have attempted to implement many of the same changes, the results have been quite different. The reforms in these countries are ongoing, however, and lessons are still being learned. This chapter is an initial effort to draw out some of the most important features of these country programs, hold them up to examination, and identify factors that contribute to their success or failure. The assessment finds that the single most important success factor is a favorable political environment that provides the power base, incentives, and commitment to implement even difficult reforms. Furthermore imple- mentation arrangements and staffing need to be sufficiently embedded in existing administrative and political structures. Finally program compo- nents must be appropriately tailored to the country context, and its readi- ness for reform must be technically sound and strategically sequenced. This chapter starts with a brief recapitulation of the story of the public service in anglophone Africa, sketching out characteristics of what came to be known as first- and second- reforms. It then turns to Ghana, Tanzania, and Zambia as examples of second-generation reform programs, identifies and explores key features of these countries’ programs, and makes compar- isons. The chapter concludes with observations on drivers of success and potential pitfalls to be avoided. 43 44 Comparative Experience with Public Service Reform in Ghana, Tanzania, and Zambia Setting the Scene The present wave of what have come to be called second-generation civil ser- vice reforms needs to be set in its historical context, because the approaches used in these reforms are in part a reflection of how governments attempted earlier civil service reforms and how those efforts turned out. Early Reforms The first wave of civil service reform in anglophone Sub-Saharan Africa (SSA) came soon after independence, when departing expatriates had to be replaced with indigenous staff. Most countries developed training and local- ization (or Africanization) programs, and many of the present-day public service training institutions, now sadly run down, date from this period. This process was effectively completed for most countries in the first decade of independence. The turnaround of expatriate and local staff was fairly quick and promotion for the first generation of graduates came rapidly. Although some transition efforts failed, on the whole the transformation was completed quite effectively. In a number of countries, the Africanization also overlapped, or was closely followed by, a process of adaptation of civil ser- vice processes, pay scales, and structures to the change of paradigm, from law and order administration to the development state, in which the public sector played the leading role in spearheading development. Pay scales were adjusted to local conditions—at least to a degree, because pay of technical, professional, and managerial staff remained high, and many of the departing expatriate allowances were retained. Class structures were abolished in the interest of internal mobility within a unified public service, incomes policies and manpower planning were introduced, cadres with schemes of service and new public service ministries were created, and the powers of public ser- vice commissions (PSCs) were reduced (abolished in Tanzania). In making these changes many countries in East and Southern Africa, and also Nigeria (although the reforms were never fully implemented there) were strongly influenced by the round of public management modernization reforms in old commonwealth countries sparked off by the Fulton Commission in the United Kingdom. Senior African civil servants, now retired, look back at this period as a golden age, when things worked, pay befitted the status of a pub- lic officer, and the public service was often the dominant partner in the rela- tionship with ministers. In almost all countries an erosion of pay (see figure 2.1), skills, morale, and values then followed. This erosion had many causes, among them the overextension of the state; the pursuit of inward looking and often socialist Mike Stevens and Stefanie Teggemann 45 development policies, which shrank the tax base; periods of arbitrary rule by the military; and the spread of patronage and corruption. By the early 1980s most countries had experienced a serious loss of capacity with the departure overseas of professional and managerial staff and a lowering of skill level among those who remained through the decline of training policies and institutions. The formal rules of budgeting and personnel management remained in place, but in practice a dual system emerged with informal rules of how staff were hired, promoted, or paid and how budgets were actually implemented. Low levels of formal pay spawned a range of pay-augmenting behaviors ranging from moonlighting (even “day-lighting”), abuse of travel, seeking project work with donor salary supplements, parastatal board sit- ting allowances, and director’s advances all the way to outright fraud and corruption. In the process productivity and the public interest suffered, and many civil services were more concerned about the absorption and redistrib- ution of budget inputs than the production of outputs in the form of key state functions and service delivery. Figure 2.1. Index of Real Average Pay in the Tanzanian Civil Service, 1969–2000 140 Series 1 120 100 Index (1976 = 100) 80 60 40 20 0 70 72 74 76 78 80 82 84 86 88 90 92 94 96 98 00 19 19 19 19 19 19 19 19 19 19 19 19 19 19 19 20 Year Note: Data for 1995–2000 are provisional. Source: Valentine (2002, p. 4). Based on Valentine (1999); data provided by the Civil Service Department. 46 Comparative Experience with Public Service Reform in Ghana, Tanzania, and Zambia First-Generation Reforms The first efforts at tackling the public service crisis in SSA came in the wake of the structural adjustment programs launched during the 1980s with the support of the International Monetary Fund and the World Bank. The Gam- bia, Ghana, and Guinea were the first countries to undergo what have been termed first-generation civil service reform programs. These first-generation reforms had a heavy emphasis on controlling the wage bill, through retrenchment and the identification of ghost workers and their removal from payrolls. Although the countries attempted to tackle a central problem of the public service, namely, bloated payrolls, they were only partly successful. In Ghana many of those retrenched from employment in central government ministries and departments found jobs again in the country’s large parastatal sector, principally in subvented agencies. Even when downsizing was suc- cessful, numbers tended to creep up again, leaving little room to improve pay and conditions. Many of the pathologies of an eroded public service remained. Second-Generation Reforms The shortcomings of the first-generation reforms led to a further wave, second-generation reforms, which began in the 1990s. These reforms were much broader than the first-generation reforms, which were centered on the wage bill. First, a more differentiated approach to pay was taken. First-generation reforms for the most part assumed that salaries could be decompressed and improved across the board if the downsizing of staff was pursued res- olutely. Second-generation reforms acknowledge that this approach was not practical and that alternative strategies needed to be devised, based on more selective improvement and the development of a medium-term pay reform strategy. At the heart of the approach was a shift away from pay scales determined by job evaluation and internal relativities, toward align- ment with relevant labor market rates within the country. Another thrust was consolidating the array of donor practices to supplement salaries, which made sense in a narrow project context, but were seen as harmful to sustainable capacity building. Second, a renewed effort was made to focus government on its core functions, in an effort to reverse the relentless expansion of program com- mitments that governments accumulated in the era of the development state. The emphasis was on the classic public goods functions of the gov- ernment, such as law and order, regulation of the private sector, economic Mike Stevens and Stefanie Teggemann 47 management, and the provision of social services in areas that have large externalities, such as primary education and health care. Third, the governance framework for particular policy functions was reassessed. Inspired in part by the executive agencies (EAs) of several Organisation of Economic Co-operation and Development countries, gov- ernments created autonomous agencies, subjecting these agencies to a gov- ernance framework that gave managers greater freedom from central government rules in return for achieving specific performance standards and targets. Fourth, second-generation reforms also seek to reverse the centralization of government, which occurred under the development state paradigm when the mandates of local governments were sharply curtailed in many countries. Guided by the principle of subsidiarity, the thrust is to push ser- vice delivery down to the local level, where more information is available, and, it is hoped, the greater scope for community action will hold local gov- ernment service providers accountable. Another element is a growing emphasis on performance. Ministries, departments, and agencies (MDAs) increasingly use three-year strategic plans and annual business plans, identifying output and outcome targets against which performance can be measured and publicized. In several countries (for example, Tanzania and Zambia) performance improvement funds have been created to encourage MDAs to reengineer existing processes to make them faster and more efficient (so-called “quick wins” programs). Personnel practices are also being overhauled to reward individ- ual performance more transparently. At the same time many programs seek to reinforce the merit principle in public service, through the updating of ethical codes and conflict of interest rules and the rehabilitation of tradi- tional institutions for protecting the merit principle, such as the PSC.1 1. Generally, this rehabilitation is a good thing because PSCs can protect the public service from unwarranted political interference and ensure that the merit principle is respected in appointments and promotions. Many countries are emerg- ing from a period during which the neutrality of the public service was not respected, and interference in public service management was arbitrary. PSCs typi- cally ensure a healthy public service by setting standards for appointment and pro- motions and overseeing how these processes are followed. Commissioners appointed to PSCs tend to be former civil servants or persons distinguished in other fields. The conservatism inherent in many PSCs could, however, become a barrier to reform, for example, by resisting the appointment of outsiders to head key agencies or setting aside the seniority principle for the same purposes. They might also seek to block experiments with differentiated pay. 48 Comparative Experience with Public Service Reform in Ghana, Tanzania, and Zambia Efforts are also made to make training more relevant by refurbishing train- ing institutions, modernizing curriculum, and creating a better balance between supply-driven and demand-led modes of training provision. Second-generation public service reforms also recognize that comple- mentary action on other fronts is necessary. Thus they are often accompa- nied by parallel efforts to modernize budgeting and financial management and to strengthen audit institutions. All three countries are attempting to introduce budgeting based upon medium-term expenditure frameworks. Because countries undergoing reform are often newly reestablished democ- racies, an emphasis is also put on building in mechanisms for making gov- ernment more transparent and participatory. This emphasis has led to efforts to modernize records management across government. Second-Generation Reforms in Ghana, Tanzania, and Zambia: A Comparison Although they are not the only countries in Africa seeking to implement second-generation reforms (Kenya, Mozambique, and Uganda are also practitioners), Ghana, Tanzania, and Zambia are leading examples, and each has a rich experience from which to draw lessons: • Tanzania’s Public Service Reform Project was launched in October 1999. It has proven to be the most progressive reformer and has made promising progress in tackling a comprehensive set of reforms. • Zambia’s Public Sector Capacity Building Project (PSCAP) was approved in February 2000. Its record to date is a mixed picture: although it has some improvements to show, overall, its administra- tive reform agenda is lagging behind. PSCAP II is being recast to focus predominantly on expenditure accountability in the hope of laying the groundwork for administrative reform. • Ghana did not sustain its promise of successful economic and admin- istrative reforms demonstrated during the 1980s. The Public Sector Management Reform Project (PSMRP) was approved in April 1999, but it barely went beyond the analytic phase and had little adminis- trative change to show. Reforms came to a complete halt with the election of President John Kufuor ’s government, which is in the process of redefining its approach to public service reform. How can the different outcomes of these programs be explained? This chapter argues that three factors are critical to successful public sector reform. First, the political context must be conducive to reform. Second, the Mike Stevens and Stefanie Teggemann 49 reforms must be driven by committed leaders in the public service, and implementation arrangements need to be firmly embedded into administra- tive and political domains. Third, the design must be right and contain rele- vant program components in a sequence that works. Implementing reform can be likened to producing a play, in which the combination of stage, actors, and props define the quality of its ensemble. The political and economic context provides the stage on which reforms are conceived and implemented. The actors involved in the reform process as well as the implementation arrangements shape critically the depth at which change affects existing institutional layers and at what speed. Ques- tions to be asked include if leadership and commitment are sufficient, where leadership is situated and if it has sufficient authority; and how this configuration translates into project structures. Project implementation needs to be carried out by well-qualified staff who are embedded in existing institutional structures or at a minimum have strong links into the political realm. These first two factors—the stage and the actors—must be favorably aligned for the program components to take hold and unleash their impact. But the kinds of props the actors employ in the battle (the right reforms), their quality and fit (are reform components technically sound and fit the ensemble of reforms and country context?) as well as the timing of when to bring the props on stage (are the basics in place before more sophisticated reforms are started?) are equally important. Ghana, Tanzania, and Zambia have all fared differently on these three levels. The following section will contrast their respective endeavors in staging reform, beginning with an overview of the political and economic context, extending to an analysis of the degree of leadership and implemen- tation arrangements, and concluding with an evaluation of their experience to date in implementing five key reform components. The Enabling Environment: Political and Economic Context This section sketches the key features of the respective political and eco- nomic environments that shaped the success or failure of public sector reform. It will show that reforms are necessarily situated in a specific coun- try context, to which they need to respond if they are to be successful POLITICAL CONTEXT. Political will does not just happen. It is embedded in a political system and culture that provide incentives for reform at best or, at a minimum, do not discourage reform initiatives on the part of civil ser- vice leaders. These systems, in turn, are run by elites who may beholden to 50 Comparative Experience with Public Service Reform in Ghana, Tanzania, and Zambia interest groups that have a vested interest in the status quo or that support only a particular part of the reform agenda. Thus political systems and cul- tures, as well as the existence of powerful interest groups, make up the political landscape, which critically shapes the reform options available. Understanding the political context is thus key to crafting viable reform programs. As chapter 4 shows in a system with a high degree of political stress, a technocratic approach to pay reform is difficult to implement because the government will be preoccupied with securing political sup- port. This chapter argues that this principle applies more broadly to public service reforms and that politically contentious reform areas rely more heavily on the existence of a stable political system. Tanzania’s political environment stands out as one that is characterized by relative civic peace, unity, and stability and the dominance of a single party (Kiragu and Mukandala 2004). Under the socialist system, power was concentrated in the ruling party, originally the Tanganyika African National Union, which later became the Chama Cha Mapinduzi (CCM), and down- ward accountability to voters through intraparty competition was limited. CCM’s policies of equity, social justice, and electoral competition enjoyed broad popular support, which outlasted the socialist era. Even after the introduction of multiparty elections, the CCM continued to win the major- ity of votes and has even increased its share from 1995 to 2000 in both parlia- mentary and presidential elections. Trade unions, after their resurgence, remained acquiescent, constrained by the belief that they would threaten the existing peace and tranquility if they made radical demands. Tanzania’s political environment of uncompetitive pluralism as well as political stabil- ity resulted in low political stress, which allowed the administration to push through a rational model of pay reform quite successfully. Progress was not linear, however. The first half of the 1990s saw the revival of a multiparty system (1992), political party campaigns, and the reemergence of unions, including a teachers union. Increasing political competition led to the politi- cization of pay reform. This development encouraged a proliferation of allowances for all levels that stealthily favored senior civil servants. And today, as recent events are showing, achievements are still fragile, requiring technocrats to manage carefully the political environment. In Zambia President Frederick Chiluba and his ruling Movement for Multi-party Democracy (MMD) party was elected on an anti–Kaunda plat- form. The loose coalition of church groups, intellectuals, union members, and human rights advocates, all of whom opposed President Kenneth Kaunda, was united not by ideology or party program, but by the aim of unseating Kaunda. Once this goal was achieved and supporters realized Mike Stevens and Stefanie Teggemann 51 that campaign promises of democracy, good governance, and economic growth could not be fulfilled, frictions within the ruling party became apparent, epitomized by the formation of various splinter groups, for exam- ple, the National Party in 1993 and three other new parties in 1998.2 In the absence of a stable coalition united by a common goal, the MMD in turn resorted increasingly to patronage and clientelism to maintain its political base, thereby repeating some of Kaunda’s pitfalls. Another complicating factor for public service reform has been the opposition by trade unions, tra- ditionally a powerful actor on the political scene. Under one-party rule trade unions had been the only organized opposition force, maintaining a great deal of independence from the ruling party. President Chiluba himself was a trade unionist and did not confront them. Unions have used their prominence and strength repeatedly in negotiations with the government to secure generous concessions, as the recent agreement on severance pack- ages for retrenched public officials illustrates (see box 2.1). Ghana witnessed substantial economic and administrative reforms under authoritarian populist rule of Jerry Rawlings in the 1980s. Against a backdrop of economic crisis, he had managed to harness support for techni- cally sound yet difficult reforms through “incorporation, cooption, or out- right suppression” (Kiragu and Mukandala 2004, p. 132). Although Ghana did not have the same legacy of an accountable and popular political party as Tanzania, Rawlings was able to win internal support for reform with the state house, cabinet, technocrats, and government-supported populist- revolutionary organizations. With the introduction of multiparty politics in 1992, however, this coalition no longer held together, prompting the reform process to stall and reverse. The story of Ghana in the 1990s resembles that of Zambia in the sense that the ruling National Democratic Congress party was no longer united by a programmatic vision and much energy was diverted toward sustaining patronage networks as a basis for political sur- vival. Under the latter Rawlings regime, the National Institutional Renewal Program (NIRP) was chaired by Vice President Atta Mills, who positioned himself as a reformer. The electoral loss of Mills left the administrative reform process entirely orphaned, and the new government distanced itself from the reform agenda of the past. 2. The government was quite successful, however, at economic liberalization and retrenchment of public servants during its first four or five years in power. An achievement that the MMD turned into electoral success—it won 131 of 150 seats during the 1996 legislative elections—but did not use to push the envelope on public sector reform. 52 Comparative Experience with Public Service Reform in Ghana, Tanzania, and Zambia Box 2.1. Retrenchment Costs in Zambia: A Flexible Donor Response to Political Realities Although Zambia encounters similar challenges to Tanzania, it has taken much longer to begin reform steps in the right direction. As pointed out earlier, this delay results from the high level of political stress and the subsequent politi- cization of pay reform. At the same time, however, we must acknowledge that donor support has not always taken these specificities of Zambia’s political landscape into account. An illustrative example is donors’ hesitancy to finance severance pay. Because of strong union pressure, retrenchment costs stipulated by the 1997 Employment Act were equivalent to about 12 years’ pay per retrenched staff. The government considered the political costs of budging from that agreement to be too high, despite pressure by the International Monetary Fund and the World Bank to go ahead and the simultaneous refusal by donors to finance retrenchment. A salary decompression that increased top salaries only added to the separation cost, making the implementation of retrenchment simply unaffordable. Consequently, public officials identified for retrenchment were sent home while they kept getting paid. Only in this year ’s budget has the government agreed to cover two-thirds of retrenchment costs and the credit agreement of PSCAP has been amended to contribute the remaining $10 mil- lion. The latter constitutes a successful donor response, which builds on the recognition of the government’s political realities. Although Zambia is an extreme case of extraordinarily high retrenchment costs, other SSA countries are not far behind. Typically, statutorily mandated retrenchment costs equal one month’s pay for each year of service plus the acceleration of pension benefits for staff age 45 and above. Severance payments often postpone the onset of wage bill savings, even if downsizing is achieved, and actually raise, in most cases, the wage bill in the very short run, and that increase makes severance payments an obstacle to reductions in public service size in the first place. THE ECONOMIC REALITY. A country’s economic situation and public ser- vice reform are inextricably intertwined. Public service reform becomes nec- essary when uncontrolled growth of the establishment, combined with unsustainable levels of pay and often aggravated by a deterioration of the economy, force the wage bill to surpass its economically sustainable level. The economic crisis of the 1970s and 1980s in Tanzania, for instance, forced the government to backtrack on its previous policy of adequate public ser- vice pay, mostly inherited from colonial times. At the same time the ability to move forward with public service reform is also affected by the state of the economy. Economic growth generates rev- enue, which enables a country to accommodate new policy priorities; meet the costs, such as severance payments or competitive pay for key posts; or rebalance the staff mix of departments. Furthermore if the economy is buoyant, Mike Stevens and Stefanie Teggemann 53 scope for employment outside the public sector will be greater, thus reducing resistance to downsizing. Conversely if the economy is in crisis, budgets will remain unpredictable, little leeway will exist for new policies, and the continuation of cash budgets will perpetuate poor service delivery and undermine any efforts to improve performance. Governments will be unable to fund separation schemes, and staff will hang on to their jobs in the absence of alternative employment. Reform Leadership and Structures Political commitment, leadership, and implementation arrangements are three key enabling factors for successful reforms. The following paragraphs describe how these factors played out in Ghana, Tanzania, and Zambia. POLITICAL COMMITMENT. That political commitment is essential for pub- lic service reform to be successful is a basic truth. However well designed technically, no reform will succeed if either collective determination in the key central management agencies to press forward or strong belief on the part of the president in the importance of a well-functioning public service is lacking. Those undertaking the reforms must be certain they always have the president’s backing, and because public service reform is likely to be politically challenging, it has to be one of a small handful of policy objec- tives at the top of the president’s agenda. Only in Tanzania does this resolve seem to be clearly the case. In Zambia commitment at the senior official level seems strong, but those at the highest level have other preoccupations. The program in Ghana had been affected by the change in government in January 2001 and the distrust with which the new government viewed the reform programs of the old. LEADERSHIP FOR REFORM. In addition to political commitment at the top, strong leadership is needed to implement reform successfully. This leader- ship must function at and link various levels: from the president down to political and administrative tiers of government. First, the existing leadership structure for the public service must be sup- portive of reform. The countries we are examining typically have a head of the civil or public service who has the power to approve pay (with the agree- ment of the finance ministry), staffing structures, and grading; personnel rules; training policy; and many civil service appointments. These responsi- bilities are in part shared with a civil or public service commission, headed by either former civil servants or otherwise distinguished persons. Estab- lished to protect the public service from unwarranted political interference, 54 Comparative Experience with Public Service Reform in Ghana, Tanzania, and Zambia PSCs have a statutory mandate to protect the merit principle in the public service and may perform a significant role in appointments and promotions in the public service, as well as acting as an appeals body in disciplinary cases. They will also have a voice setting personnel rules and related actions affecting the management of the public service. The effectiveness of PSCs varies from country to country, however. Although Tanzania’s PSC has greatly facilitated the reinstatement of a meritocracy, Ghana’s PSC has been largely ineffective. Second, as already mentioned, the support of the president is vital. The president, who must take an active interest in the reform process and assume responsibility for the political consequences, plays a key role in resolving ministerial level conflicts and ensuring that key agencies are headed by people supportive of reform. Within MDAs, the permanent sec- retary or chief executive and senior managers must act as change agents. Finally, reform also entails staff working to perform such tasks as draft- ing new laws and regulations, writing cabinet papers, training and support- ing training agents, drawing up terms of reference, and selecting and supervising consultants. IMPLEMENTATION ARRANGEMENTS. Even if political commitment and reform leadership are in place, much depends on the quality of the imple- mentation arrangements. Issues include the location of the implementing agency (whether within an existing MDA or as a separate organization) and its staffing (whether with civil servants, consultants, or expatriates) and remuneration. In Ghana, the NIRP Secretariat was charged with coordinating a number of reform initiatives (civil service reform, financial management, decentral- ization, governance) financed by external donors. The secretariat comprised a team of donor-funded local consultants reporting to an oversight commit- tee headed by the senior minister (similar in standing to a prime minister), who was in turn charged by the president with coordinating all government reform programs. Change teams in MDAs headed by the director of admin- istration or chief director and staffed with representatives of all levels were responsible for managing the change process. They commissioned local and external consultants to undertake the studies required to underpin the reform process. Change teams reported to the NIRP Secretariat, which was not integrated into the regular departmental structures of government, was staffed by consultants, and lacked the authority to fulfill its guiding role. Toward the end of 2003, following a critical assessment of all reform pro- grams, the NIRP Secretariat was wound down. Mike Stevens and Stefanie Teggemann 55 A similar model works in Zambia, where an implementation team of Zambian consultants, some former civil servants, under contract with and paid for by the Bank, manages the reform process. This team, which reports to the head of civil service, is embedded in the management development division under the cabinet office, which also spearheads the Public Service Reform Program launched in 1993. It has, however, generally proved more effective in getting the required decisions out of government than the NIRP Secretariat has been in Ghana. By contrast, in Tanzania the Public Service Reform Project is led from the Civil Service Department (CSD), with the permanent secretary pushing the reforms forward in a determined way. In turn, the permanent secretary is assisted by departmental directors. Although the Public Service Reform Project began with a separate project team inside the CSD led by a chief advisor, consultants—mostly expatriate and funded by donors—are now embedded in the departmental structure of the CSD. The permanent secre- tary of the CSD, furthermore, enjoys the full confidence of the minister for the public service and the president. On the face of it Tanzania would appear to have the most effective model, principally because it was able to make the transition from a project team of advisers to leadership fully integrated into the department directly responsi- ble for public service matters. By contrast in Ghana a disconnect appears to have arisen between the NIRP Secretariat and key ministers, on the one hand, and their chief directors in the central management agencies, on the other. So great is this disconnect that the current government has terminated all programs and is beginning afresh. This looming failure is a reflection of the fragmentation of the public service and the lack of unified leadership: the civil service department covers only 20 percent of the public service (teach- ers, doctors, and subvented agencies are not part of the civil service), and no single authority is mandated with overseeing the whole of the public service. The NIRP Secretariat, created to oversee reform programs pertaining to all of the public service, consequently did not have a single political home and itself lacked the necessary political links and authority to push through reforms. Zambia, where the unit enjoys the full confidence of the head of ser- vice but is not embedded in departmental structures, falls somewhere in between the experience of Ghana and Tanzania. These analyses suggest that a variety of implementation arrangements can work, but all of them share some common features. First, senior most leadership in government (in effect the president and the minister responsible for the public service) must be politically committed to the goals for the reform program, support its strategies, show a constant 56 Comparative Experience with Public Service Reform in Ghana, Tanzania, and Zambia interest in developments, and demand results. They also must be prepared to intervene occasionally to manage political risks and remove bureaucratic bottlenecks. Second, because reforms affect government operations, agents for reform must be placed in government structures or, if placed externally, agents must be given strong institutional and personal links to the bureaucracy. A link is also needed between those managing public service reform and other com- plementary reforms, such as in budgeting and financial management, not least because these should reinforce each other, but also because some areas of public service reform, such as pay, directly concern other central manage- ment agencies, such as the country’s ministry of finance. Third, a central unit can draft legislation, issue administrative instruc- tions, and commission studies, but change at the MDA level, if it is to come about, will happen only through the actions of senior management change agents who are prepared to overcome staff resistance and drive through the reform agenda. Externally funded consultants cannot effect such change. Rather the political leadership needs to ensure that the best managers (in exceptional cases, from outside the public service) are put in positions of authority in MDAs. Appointing the best managers will likely require the president’s direct intervention in public service appointments, setting aside the seniority principle, as well as the PSC’s support. By contrast, the role of foreign experts can be to advise, encourage, and build capacity, cognizant that their role is that of a temporary advisor and enabler rather than an implementer. Fourth, because working in the engine room of a reform program requires individuals to work long hours, staff of the central unit need to be rewarded accordingly. These rewards, however, may cause resentment from other staff in the public service ministry who manage the systems that the reforms both rely upon and seek to change. At the end of the day, individu- als make reforms happen, not committees or consultants. The role of a cen- tral reform secretariat is to show the way and to empower reform-minded departmental managers to act. Five Reform Components in Comparison In a short chapter doing justice to the many reform components pursued under second-generation programs is impossible. The assessment thus focuses on what we believe are five key components of public service reform: (a) pay, including staffing numbers, control of the wage bill, and the selected accelerated salary enhancement (SASE) scheme; (b) restructuring of the state; (c) performance management; (d) the strengthening of control Mike Stevens and Stefanie Teggemann 57 systems, including budgeting, financial management, auditing, administrative controls, and civil society engagement; and (e) training. As the discussion of first-generation reforms showed, none of these com- ponents can stand alone. The wage bill depends not only on levels of pay but also on the size of the public service, which is inextricably linked to issues of state restructuring; a lean and well-paid public service with no performance management is unlikely to deliver good services. Because of these interde- pendencies, second-generation reform programs introduced schemes that link increases in pay to better performance (SASE) and downsizing to better results- and client-orientation (EA model). PAY. Just as first-generation reforms tackled but did not fully resolve the pay, numbers, and wage bill “iron triangle,” so the reality of excess staffing, insufficient pay, and wage bill pressure remains the starting point for pay reform under second-generation reform programs. The data here are mixed. Although some countries (for example, Tanzania) were able to show some improvements in officials’ pay in real terms, others (for example, Zambia) experienced further decline in the 1990s. In all three countries, however, thinking on pay has shifted, from across- the-board increases to selective increases, epitomized in the SASE schemes. These schemes have figured prominently in the medium-term pay reform strategies that governments have been encouraged to develop to sharpen policy and provide a time frame for addressing pay issues. Success in implementing these reforms, however, has varied greatly from country to country. As detailed by Kiragu, Mukandala, and Morin in chapter 4, pay reform is a politically contentious issue that is easily sacri- ficed for reasons of generating and sustaining political support. Only gov- ernments sufficiently blessed with political space for reform, such as Tanzania, can afford to pursue a technically rational approach and do so quite successfully. In this light, the findings displayed in table 2.1 can be explained. The pay reform components of the projects in Ghana and Zam- bia, by contrast, have borne little fruit, because the ruling parties in both countries were not sufficiently prepared to attack these reforms (for a more detailed description see Kiragu and Mukandala 2004). Reform approaches and donor responses hence must be flexibly tailored to fit the country con- text, as illustrated by the case of severance pay in Zambia (box 2.1). In the longer run, however, the unanswered question looms concerning the structure of pay governments should adopt for their staff as resources gradually become available. The traditional model, which was in its heyday 30 years ago, paid civil servants well and was designed to maintain internal relativities across the whole of the then-unified public service, so that it Table 2.1. Pay—Intent and Results, by Country Country Reform intent Results so far Assessment Ghana • Dissolve hidden reward structures • Conducted three studies • No reforms under the Public Sector • Introduce SASE scheme • Backtracked from SASE scheme Management Reform Program • Decreased real pay (except for 1998 • Deteriorating reform record: went and 1999) from technical-rational model in 1985–90 to politically reactive model of pay reform in 1990–2000 (Kiragu and Mukandala 2004) Tanzania Implement medium-term pay reform • SASE scheme applied to 900 posts in • Overall progress, although sustained strategy, including the following: three MDAs commitment to pay increases essential • Decompress salaries • Government lags behind in • Pursued technical-rational model of • Introduce SASE scheme implementing medium-term pay pay reform in 1995–2000 (Kiragu and reform strategies Mukandala 2004) • Average real pay increased more 58 than threefold since 1993 • Wage bill remained stable at slightly above 4 percent Zambia Implement medium-term pay reform • Some allowances eliminated, some • Some progress in relation to starting strategy including the following: decompression conditions of declining real pay, • Consolidate allowances into salaries • Medium-term pay reform strategy increasingly compressed pay scales • Decompress salaries adopted but knocked off course (until 2000), preponderance of • Introduce SASE scheme • Wage bill out of control (8 percent of allowances (2001), but problems gross domestic product) reemerge • Civil service employment up at • Pursued politically reactive model of 119,000. pay reform 1985–2000 (Kiragu and • Retrenchment planned Mukandala 2004) • Introduction of SASE scheme being reconsidered Source: Authors. Mike Stevens and Stefanie Teggemann 59 could form the basis of both a national income policy and the interchange of staff between different branches. The consensus view is that the new model should take its cue from domestic labor market comparators, which tend to assign much higher values to technical, professional, and managerial staff than to unskilled drivers, messengers, and clerks, who currently may be better off working for government. The implication of this model is a much more decompressed salary structure, the more so as economies integrate into the global economy. The political consequences of this change have yet to be discussed in any of the countries now embarking on a SASE scheme bridge to an unknown destination. In all three countries, SASE schemes were envisaged as a donor-funded bridging mechanism that would help the government advance the benefits of their medium-term pay reform policies. Under SASE schemes, salaries of key technical and professional staff are raised to compete with those of the private sector. Donors provide funding on a sliding scale over a five-year period, with the government increasingly picking up the cost using savings from parallel downsizing efforts. The SASE scheme thus allows the govern- ment to retain key staff in the short term while it is building a lasting com- petitive salary scheme. It also rewards staff performance (public officials can lose their top-ups in case of underperformance) and encourages perfor- mance management at the agency level because participating MDAs must have a strategic plan in place, backed up by detailed implementation and annual plans. As a positive side effect, the SASE scheme provides an alter- native to the widespread donor practice of salary top-ups with its well- known distortion of the labor market and draining of talent from mainstream government functions. In reality only Tanzania has so far implemented a SASE scheme. Ghana has abandoned its earlier reform approach altogether, and Zambia is pon- dering the feasibility of its SASE scheme given that record management and financial management systems are weak and the government has deviated substantially from the medium-term pay reform strategy since its adoption, resulting in serious budget overruns. In Tanzania, the Medium-Term Public Service Pay Policy of 1999 set the stage for a SASE scheme, targeted at key professional, technical, and man- agerial staff who play crucial roles in service delivery, reform implementa- tion, and the production of strategic government outputs. It also sought to reduce the reliance on expatriate experts. Introduced in 2000 together with the job evaluation and job regrading exercise, the SASE scheme requires the adoption of an open staff performance appraisal system. To date, it has been implemented in three MDAs out of the 10 planned and covers about 900 posts. It provides salary increases of 25 percent, a figure that analysts 60 Comparative Experience with Public Service Reform in Ghana, Tanzania, and Zambia believe the government will be able to sustain after donor funding fades out. Although it is still too early to judge the success of the SASE scheme in Tanzania,3 a few issues became apparent that are of importance to other countries that consider introducing a SASE scheme. First, because of its lim- ited scope, the SASE scheme has not managed to prevent salary top-ups by donors, and donors have been reluctant to pool resources. Close donor cooperation (and sustained donor commitment) as well as donor–govern- ment cooperation is necessary to make a SASE scheme work. Second, on the government side, salary increases have not materialized as planned, pro- longing the reliance on donors. Third, the SASE scheme pay level is still below market for some classes of professionals, such as information tech- nology specialists, suggesting the need for additional measures to prevent these employees from leaving for the private sector. Fourth, initially, an extrabudgetary fund was established to fund the SASE scheme, thus cir- cumventing national budget procedures. Implementation of a SASE scheme in Tanzania benefited from a number of factors. First, a proper records management system enabled a smooth introduction. In addition, economic growth has been stable and predictable, which has made the SASE scheme in theory affordable. Last but not least, donor–government coordination has been relatively successful, although room for improvement remains. None of these factors is sufficiently in place in Zambia, hence its reservations. For any country embarking on a SASE scheme, risks that could severely compromise its successful implementation are numerous, including the risk of politicization because the selection of the MDAs and posts might be dri- ven by politics and patronage rather than merit (see table 2.2.). More impor- tant, if retrenchment savings are insufficient or consumed by legally mandated severance packages, the government will not be able to sustain the SASE scheme once donors phase out their funding. Economic afford- ability and the political will to sustain a highly decompressed salary scale are factors that in fact equate to a litmus test for all SASE-scheme countries. Unwavering commitment is also required from donors. A loss of interest because of changing donor policies or priorities as well as temporary freezes of aid can severely shake the bridge that the SASE scheme is sup- posed to build and jeopardize its success. Whether governments can sustain competitive wages, led by national labor markets—the assumption driving the medium-term pay reform strategies—is also unclear. 3. For a detailed discussion of Tanzania’s SASE scheme experience to date see Valentine (2002). Mike Stevens and Stefanie Teggemann 61 Table 2.2. Addressing Risks in Implementing a SASE Scheme Risks Remedy Politicization of and patronage in Set clear criteria and transparent selecting posts compromises credibility selection mechanism Politicization of and patronage Create a transparent and well- in selecting participating MDAs publicized selection process compromises service delivery and Require strategic annual plans reform Top-ups seen as entitlement and Withdraw top-ups in case of under- preclude improvement in staff performance performance Verify through open performance appraisal systems MDAs do not deliver Hold MDA accountable for imple- mentation of annual plans through independent oversight Link SASE to performance improve- ment fund to build capacity and increase incentive Wavering donor commitment Diversify donor coalition sufficiently Ensure close donor cooperation Wavering government commitment Start SASE only when political will supports a medium-term pay reform strategy Ensure close donor-government relations Slow economic growth and insufficient Pick salary enhancement targets based revenue base threaten economic on realistic revenue projections affordability Maintain flexibility to readjust Source: Authors. Finally, despite early signs of success, Tanzania’s experience is not suffi- ciently advanced to draw conclusions about whether the SASE scheme will eventually deliver on its promise of retaining qualified staff or, put differently, whether staff that otherwise would have left (or not accepted a public sector job) decided to stay because of better pay. This conclusion ties into the ques- tion of whether the basic assumption underlying the SASE scheme, namely 62 Comparative Experience with Public Service Reform in Ghana, Tanzania, and Zambia that pay is the key incentive, is in fact true. Similarly, it remains to be seen whether a SASE scheme in combination with open performance reviews does in fact promote better performance. STATE RESTRUCTURING One of the characteristics of the development state is the overextension that this paradigm entails.4 The ensuing fiscal strain that an enlarged public service exerts on the national budget has, in many coun- tries, become fiscally unsustainable and, coupled with inefficient service delivery and inadequate performance, has led governments to rethink their roles. For each policy objective, governments must decide whether to inter- vene, and if they choose to intervene, whether to regulate a private sector response or provide a service directly. If the latter, they must decide, first, whether to take over service provision at the national or a subnational level and, next, whether responsibility should be given to central government departments, semiautonomous agencies, or state enterprises. From this deci- sion tree follow several reform elements, all of which entail substantial insti- tutional change and affect large parts of the public service. Understandably, this process makes restructuring a politically contentious reform to carry through because the potential losers of reform are, with a few exceptions, located within the central public sector and the winners are in the private sector or at the decentralized level, both of which represent constituencies who have less political capital. In addition restructuring inflicts immediate pain and pays off only in the medium to long term through cost savings and improved service delivery. To increase the chances of successful reform, the hardship inflicted upon retrenchees needs to be cushioned. The reform experiences in functional reviews, retrenchment, privatiza- tion, decentralization, and EA programs in our three case countries reveal a familiar pattern in which Tanzania has good progress to report, Zambia reports some achievements, and Ghana has barely moved beyond the ana- lytic stage (see table 2.3.). The following will discuss some of the challenges and differences in implementing reform. 4. The present machinery of government and the respective mandates of MDAs and the state-owned enterprises under their control exist because of decisions made over three decades on policy and how to implement it. Most SSA countries acquired a large number of state-owned enterprises in the first 30 years of indepen- dence, and increasingly these enterprises encountered severe performance prob- lems, resulting in serious fiscal drain. Under the first programs of structural adjustment, many of these enterprises have either been restructured or privatized, and stronger governance mechanisms have been put in place for the state-owned enterprises remaining under public ownership. State-owned enterprise employees, however, lie outside the public service. Table 2.3. State Restructuring—Intent and Results, by Country Country Reform intent Results so far Assessment Ghana • Perform functional analysis of MDAs • Functional reviews commissioned Reform barely went beyond studies • Reform SAs for 22 ministries despite a large public sector (around • Retrench 28,000 public servants • SA staff on public payroll. No coher- 600,000 since 1998–99) and about • Decentralize ent SA strategy, rarely independent. 300 SAs • Nurture public–private cooperation Restructuring plans not implemented • No real reduction of public sector because staff went from civil service to SAs • No progress on decentralization • The responsibility for providing public–private cooperation was transferred to Ministry of Private Sector Development 63 Tanzania • Reduce ministerial structures • Restructured 13 ministries and 8 Promising progress except for • Establish EA program independent departments; reduced decentralization • Decentralize size of civil service to 257,000 by 2001 • Ensure public–private participation from 270,000 in 1998 • Fewer than 20 EAs established • Signs of backtracking on decentralization • Some services contracted out to pri- vate sector Zambia • Restructure • Divested four agencies • Total of about 40 agencies but no • Establish right-sizing • Public service reduced overarching legal framework • Ensure private sector participation • No progress on private sector • Delays in retrenchment have held participation up restructuring efforts SA Subvented agency. Source: Authors. 64 Comparative Experience with Public Service Reform in Ghana, Tanzania, and Zambia Functional reviews are undertaken to assess the functions of MDAs, eliminate overlaps and inefficiencies, and outsource those functions that more appropriately can be carried out by decentralized or nonstate entities. The elimination of superfluous personnel and, in some cases, the hiring of qualified staff follows suit, but these steps are often met with fierce resis- tance in case of the former (for example, Zambia where retrenchment of 4,000 staff was not be implemented for years because the government could not afford to pay the severance packages) or constrained by fiscal means for the latter (see the discussion on SASE schemes). Privatization of elements of service delivery or public–private partner- ships has borne fruit in Tanzania, both in terms of cost savings and better service delivery. A number of MDAs have contracted out grounds mainte- nance, cleaning, and security and reception, and demand from MDAs for private sector participation is growing as progress demonstrates success. In Ghana, responsibility for promoting private sector development under the public–private partnership component was transferred to the Ministry of Private Sector Development, which to date has not gone beyond studying the issue. No progress has been made on privatization. Similarly, no progress has been made in Zambia. Another dimension is reversing the centralization that characterized governance in all three countries (and many others). In Tanzania local gov- ernment was completely abolished for a decade before being reinstated in the mid-1980s. In Zambia local government structures continued to exist, but they were subordinated to parallel mechanisms of the one-party state. In Ghana local governments exist, but they remain under quite close control of central government. All governments, though, have declared policies of shifting more resources, responsibilities, and discretion to subnational tiers of government. Progress on decentralization is only partly supported by the three public sector reform programs under scrutiny in this chapter. In Tan- zania for instance, decentralization is one of the weakest components of the Public Service Reform Project and the main support to decentralization is channeled through the Local Government Reform Program launched in 1999. In Zambia, the PSCAP supported the formulation of government’s decentralization policy, but it will discontinue this agenda in the second phase of the program. Ghana has made no progress on decentralization during the lifetime of the PSMRP, although its political decentralization is fairly advanced in SSA comparison (Ndegwa 2002). One type of organizational change prominent in at least one of the three countries is the development of the EA model, drawing upon the experience of the United Kingdom (see box 2.2) and other similar countries. The rationale Mike Stevens and Stefanie Teggemann 65 Box 2.2. Executive Agency Model: The United Kingdom’s Experience EAs were first introduced in 1988 to enhance the responsiveness and effective- ness of government service provision. EAs are separate business units that have well-defined outputs, which are limited primarily to service delivery to internal and external customers of the government and rarely include policymaking. EAs are in charge of functions of vital interest to government such as prisons, civic administration, pensions, and job centers. The chief executive heading the EA has full managerial responsibil- ity and is evaluated and rewarded based on achieving defined outputs. EAs are accountable downward to citizens and outward to their departments and min- isters, who set annual targets and allocate resources accordingly. The United Kingdom experience has shown that EAs make government leaner and more effective. Since 1988 the civil service has been reduced from 570,000 to fewer than 480,000 while performance and functions have increased. EAs are generally held to be a success. Within a decade of the program’s inaugu- ration, 150 agencies had been established. Staff in EA and other executive organi- zations presently account for 373,000 or 78 percent of the central civil service. The keys to success are a clearly defined mandate and outputs and an effec- tive accountability framework and result-oriented performance management, documented by regular annual reports. Critically, the United Kingdom adopted the agency model from the starting point of an ethical, motivated, and well- paid civil service that operates in accordance with the formal rules. Source: Office of Public Service Reform (2002). for creating agencies is that some functions that entail implementation rather than policymaking can be carried out more efficiently by agencies that are one stage removed from departments. When an agency is created, a framework document details its roles and responsibilities as well as its operational autonomy. In return for budget resources, agencies commit themselves to performance targets and report on these targets in an annual report. Although cost recovery is an objective, most agencies will require a measure of subvention from the central government budget. Employees remain public officers and agencies normally have no separate corporate status—in other words, they and their staffs remain part of government. Yet sometimes agencies are given the freedom to recruit staff on terms and con- ditions more favorable than the civil service and closer to private sector lev- els for key technical, professional, and managerial staff. EAs carry some obvious risks: loss of political control, capture by private sector clients, fiscal problems arising from poor management, and loss of control over staffing as the government increasingly relies on the adequacy 66 Comparative Experience with Public Service Reform in Ghana, Tanzania, and Zambia of the internal controls of the agency itself rather than governmentwide financial and personnel rules and control systems. These risks are less daunting if the country already has a civil service that performs well, is ethi- cal and well-paid, and has a long tradition of acting in the public interest. These basics have been eroded in the past two to three decades in develop- ing countries in Africa and replaced by a high degree of informality. The country that has the most formally developed EA creation program is Tanzania, whose approach is explicitly modeled on the United Kingdom’s “next steps” EAs. Governed by a framework document, EAs are obliged to produce three-year strategic plans and an annual business plan, including key performance measures, which are reviewed by the permanent secretary of the parent ministry. The agency chief executive must produce an annual report on quality and quantity of service, with a set of accounts and finan- cial statements, for audit by the controller and accountant general and pre- sentation to parliament. The chief executive is also responsible for staff development and training. As a class of organization, EAs in Tanzania enjoy flexibility to deviate from civil service pay scales and reward staff with pay closer to private sector comparators. Most EAs are expected to break even on operating expenses, although they may require capital grants from the central government budget. Accounting standards are laid down by the accountant general, and EAs must establish a satisfactory system of internal audit. EAs have the flexibility to shift resources between different items of recurrent costs, and they may carry forward both capital and recurrent unspent budget funds. They are responsible for observing principles of open, competitive procurement, and as public entities they must observe the Public Procurement Act.5 Above all, they must achieve value for money. The chief executive is accounting officer6 for the agency, the permanent sec- retary of the parent ministry retains responsibility for overall strategic direc- tion, and the minister is politically responsible. The EA program in Tanzania is expanding, with some 19 agencies thus far created and nine more under preparation. They are covering areas such 5. Passed in 2000, the Public Procurement Act allows MDAs to carry out pro- curement of goods, works, and services following procedures set by government and subject to oversight by a central procurement monitoring and rulemaking body. 6. In anglophone systems of financial management, for each MDA an accounting/ controlling officer is appointed by the minister (or secretary) of finance to be responsi- ble for the proper management of all resources handled by that body. The accounting officer answers to the public accounts committee for any matters raised by the auditor general and taken up by the public accounts committee. Mike Stevens and Stefanie Teggemann 67 as business licensing, national archives, civil service training, and aviation regulation. The model appears to be working well, in terms of freeing the agency from the straitjacket of an unpredictable budget and empowering staff and management. Some analysts have expressed concerns, however, that the government may not live up to its promise of capital budget sup- port, which some agencies believe they were promised. This failure could affect their long-run ability to sustain activities and meet changing demand. Managers thus far appointed appear to be of high quality. Less clear is, first, how far the EA model can be extended to other areas, such as those in which the scope for cost recovery is much less and the agency will be more depen- dent on central government transfers, and second, whether the public sector will prove to have a sufficient supply of good managers to head the new agencies. Also as yet unknown is whether the parent ministries will be able to discharge their oversight responsibilities in a balanced way. Two decades ago, Tanzania had one of Africa’s largest parastatal sectors, and it failed to implement a sound governance framework for them. Ghana has a large (about 300) subvented agencies (SAs) sector, and although efforts are ongoing to reform the sector by closures, privatization, and better managed retention, none of the restructuring plans developed under the first phase of the PSMRP were ever implemented. Ghana’s SAs differ from the EA model in significant ways. SAs, by definition, receive transfers from the central government budget, and the class is much broader than in Tanzania. In Ghana, for example, the Ghana Education Ser- vice, which employs all the teachers, is a subvented agency. Nor is there yet an overall law governing the status of agencies. Rather, SAs have been established over the years through a mix of instruments, such as individual statutes, companies law, and executive order. Consequently, the account- ability arrangements are less transparent and generally weaker. Staff on agency payrolls are public servants, although in practice public service rules are applied less stringently, and some staff receive additional allowances. Many of Ghana’s 80,000 retrenchees reentered the public service through SAs after having received their retrenchment packages. Some agencies (such as the Ghana Institute of Public Administration, the civil service train- ing institution) are operating as autonomous agencies without subvention. Others (for example, the Ghana Education Service) are wholly budget dependent, and they have remained affected by the unpredictability of the cash-release system of budget implementation. All agencies have individual governing boards appointed by the president, who reserves the right to remove them, thus raising the cost of oversight and undermining SAs’ accountability to their sector ministries. 68 Comparative Experience with Public Service Reform in Ghana, Tanzania, and Zambia Zambia has about 40 agencies, 4 of which were created under PSCAP, spun off from their parent departments, namely, the National Institute of Public Administration; the civil service training institution; the Zambia Wildlife Authority, which manages wildlife parks; and the Zambia Voca- tional Training Institute. Agencies are established by act of Parliament, and work under way to adopt an overarching legal framework governing the operations of agencies has been abandoned following the suspension of the reform program. Current government policy is to reduce progressively the subvention to these agencies to zero, forcing them to stand on their own feet. In a different category are health boards, created some years ago in each district to run hospitals and clinics. Originally they were given respon- sibility for managing their own payrolls. After abuse, payroll autonomy has been taken back by the Ministry of Health. Still to be devised is a clear pol- icy for subvented agency governance, accompanied by a law defining their status and oversight arrangements. The foregoing discussion suggests that EA programs, such as the one Tanzania is pursuing, do have a future in SSA, but the category must be drawn carefully and not all presently designated agencies are necessarily good candidates. Indeed, as in the United Kingdom, Tanzania’s EA pro- gram has laid down stringent preconditions for agency creation. Regulatory agencies that can be largely self-financing seem obvious candidates, as long as capture can be avoided. Existing agencies that require subventions for the bulk of their financing can be put on a contractual basis, but this step will always be vulnerable to the risks of budget default on the part of central government and weak parent ministry oversight. Ineffective management is also a risk, such as the Zambia health board experience suggests. If EAs are sought as a means to escape dysfunctional central government ministries, they will not work. PERFORMANCE. Second-generation public service reform programs increasingly include performance management elements, drawing on many of the innovations introduced in Organisation of Economic Co-operation and Development countries. Introducing a performance dimension to the public service means making changes to the traditional system at both the personnel management level and the MDA level. The former includes merit principles in hiring, promotion, and firing; ethics; and staff appraisal sys- tems. Once again Tanzania is the country where these reforms have borne fruit (see table 2.4). Performance at the agency level is largely promoted through the devel- opment of strategic plans detailing performance targets and activities for Table 2.4. Performance—Intent and Results, by Country Country Reform intent Results so far Assessment Ghana • Develop sanctions and reward None No progress systems • Establish performance-based management system Tanzania • Establish performance improvement • Performance improvement model Substantive progress model supported by a PIF successfully introduced. PIF soon to be • Establish PSC fully operational • Reinstate public service ethics • PSC established • Prevention of Corruption Bureau, Ethics 69 Secretariat, Good Governance Coordination Unit, and Human Rights Commission all established and National Anti-Corruption Strategy and Action Plan produced, but most complaints of corruption remain uninvestigated Zambia • Strengthen policy formulation and • Eight ministries have developed Some progress but few results performance management strategic plans but little implementation • Establish a PIF • Fourteen ministries are piloting work plans for staff • PIF failed and is currently under review Source: Authors. 70 Comparative Experience with Public Service Reform in Ghana, Tanzania, and Zambia each agency. An innovative way of encouraging strategic planning is through performance improvement schemes, supported by performance improvement funds (PIFs), which all three countries initiated with varying degrees of success. Ghana was one of the first countries to venture down the performance improvement road in its Civil Service Performance Improvement Program. Launched in 1995 the Civil Service Performance Improvement Program deployed capacity development teams across a large number of MDAs. Using self-appraisal instruments, beneficiary surveys, and diagnostic workshops, they developed performance improvement plans , the heart of which was the development of a strategy to overcome constraints to per- formance and of an action plan to achieve this goal. To meet the transitional costs of implementing the performance improvement plans (such as the equipment purchases, consulting fees, and training expenses), a donor- funded performance improvement facility was created, under which a joint public–private sector committee reviewed applications and approved funding. Although many organizations duly prepared their performance improve- ment plans and were rewarded with small grants, in a larger sense little overall improvement in performance resulted, because the main constraints to better performance were beyond the control of either the agencies them- selves or the performance improvement plan process and were run out of the Office of the Head of Service. These constraints included the inability of the government budget to fund MDAs’ operating budgets reliably or to pay public servants salaries that were sufficient enough to motivate their work. With no basics in place, the performance improvement plan program was a building without foundations. Much the same approach has been adopted in Tanzania, where the PIF is set up to support the application of the performance improvement model. This model guides managers through a cycle of strategic planning, annual planning and performance budgeting, evaluation of plans, execu- tion, and monitoring and evaluation, the results of which ultimately feed back into the next round of strategic planning. Once plans are completed and approved, agencies receive resources from the PIF for implementation of their improvement plans or capacity building such as technical assis- tance, training in strategic planning, operational planning, and performance appraisal. Although the PIF, which is donor-funded and administered by the CSD, is currently outside the national budget, it is envisaged that it will be transformed into as a ring-fenced fund within the Consolidated Mike Stevens and Stefanie Teggemann 71 Fund and that flows into and out of the fund would be recorded in the budget.7 To ensure sufficient capacity to implement performance improvement models, a performance management unit in the CSD has been set up to pro- vide ministries with training and technical support. Managers will be trained in strategic planning, performance budgeting, and appraisal by technical experts. In addition, the CSD provides technical leadership. Finally MDAs will be held accountable for the implementation of their plans within budget through beneficiary surveys on the quality of service delivery and self-assessments by senior management teams. Citizens’ char- ters define and publicize minimum service standards for each MDA. All personnel will be held accountable for achieving their individual perfor- mance contracts, starting with the permanent secretary downward. The introduction of performance improvement models has been met with great enthusiasm, which has prompted the Tanzanian government to extend them beyond the three pilot MDAs plus one EA to all MDAs. To date, strategic plans have been finalized in 15 MDAs, and annual opera- tional and performance budgeting and action plans have been completed in 12 MDAs; 15 MDAs have launched client service charters; and open perfor- mance review and appraisal systems have been implemented in 3 MDAs while 7 others have been trained out of the planned 10. Some new challenges are emerging in the course of implementation. Ini- tial observations indicate that the strategic planning exercise must be better integrated into budget preparation, a requirement that calls for closer coop- eration between the Ministry of Finance and MDAs and the further devel- opment of the Medium-Term Expenditure Framework (MTEF) as the 7. All three countries have provisions under their organic finance laws for the creation by either the minister of finance or the accountant general of special funds. These funds are extrabudgetary: monies in the fund do not lapse at the end of the year, and withdrawals from the fund are not voted by the legislature. Flows into the fund from government or donors, however, still need to be shown in the budget as voted spending and also as spending by departments financed by the fund. At the end of the year, the accountant general must render a financial statement for every trust or special fund he or she administers. Properly speaking spending by a depart- ment financed from any source, whether a donor-managed PIF or a PIF that is part of the chart of accounts, still is required to be shown in the budget estimates and voted and thus is subject to medium-term expenditure framework discipline. The PIF could equally be established as a budget line in the CSD, but it then would be vulnerable to midyear cuts, and funds would lapse at the end of the financial year. 72 Comparative Experience with Public Service Reform in Ghana, Tanzania, and Zambia primary mechanism for coordinating resource allocation decisions. An issue of concern is that the selective support of MDAs potentially generates unin- tentional consequences: What if preferential treatment for a reform-minded ministry diverts resources away from ministries that are not performing as well but nevertheless have an important mandate? Zambia introduced a PIF that was intended to provide an incentive for and financing of costs of implementing performance improvement plans introduced in late 1990s (Irish Aid–funded), which had lacked funding and capacity building. Specifically the PIF would fund what were termed “quick wins,” that is, innovations to simplify procedures for better service delivery (intra-agency and to citizens) identified in MDAs strategic plans. Implementation to date, however, has not lived up to expectations. The logic of PIFs and quick wins was not sufficiently embraced by the MDAs. PIFs were seen as supplementary financing to government allocations. As a consequence, most applications for PIF funding were inappropriate (for example, cars and computers), lacked both innovation and a focus on per- formance improvement, and were not linked to MDAs strategic plans. CONTROL SYSTEMS AND ACCOUNTABILITY. Not only are control and accountability mechanisms key to keeping the reform process on track, they are also about getting the basics right—they are the foundations upon which public administrations rest and which enable them to conduct their core functions of policymaking, budget allocation, and delivery of goods and services. Controls can be achieved through three means, each of which will be discussed in the paragraphs that follow (see table 2.5): • Expenditure controls through budgeting and financial management including procurement • Administrative controls through personnel and performance man- agement systems • Mechanisms for democratic accountability to the legislature and for greater civil society engagement, both to improve policy and to strengthen accountability. Turning first to expenditure controls, public service reform is unlikely to be sustainable if not supported by improvements in the way budget sys- tems are operated. Indeed, budget reform, if it is done well, can powerfully reinforce the performance improvements sought by public service reforms. By any measure, budget systems are no longer working effectively in most SSA countries. As part of the state-led development paradigm, many SSA countries early on adopted a dual budget system linked to a five-year Table 2.5. Controls—Intent and Results, by Country Country Reform intent Results so far Assessment Ghana • Link MTEF to performance • MTEF operational since 1999 but not linked • Early progress on expenditure and management • Integrated payroll and personnel database administrative controls stalled • Expand coverage of existing inte- covers only part of the public service leading • Some stakeholder participation but grated personnel and payroll to little control over staffing levels and fraud not systematic throughout project database • Social assessments conducted in PSC, two • Perform stakeholder analysis and ministries, and five subvented agencies; consultation (social and social impact assessment on retrenchment beneficiary assessments) performed; some beneficiary assessments performed; stakeholder participation occurred during preparation Tanzania • Implement IPPS • IPPS great success Controls well advanced. Independent • Introduce records management • Sound records management of the Public Service Reform Project, system • Clients’ service charters launched in 15 integrated financial management infor- 73 • Establish clients’ service charters MDAs mation system and MTEF imple- mented and participative public expenditure review performed Zambia • Strengthen MTEF • Developed road map to introduce MTEF in • Expenditure accountability systems • Implement IPPS 2003 (partially achieved) weak but nascent • Establish policy monitoring • Integrated financial management • Administrative controls weak but system information system likely to be introduced under development • Ensure civil society consultation and tested in 2004, supported by PSCAP • Civil society engagement bore little during project preparation • Work on IPPS has commenced success • Conduct public information • Policy monitoring system not yet achieved campaign • Civil society consulted but blurred project design • Public information campaign reached only public servants, not the public IPPS Integrated Personnel and Payroll System. Source: Authors. 74 Comparative Experience with Public Service Reform in Ghana, Tanzania, and Zambia national development plan.8 Typically, these approaches contained a macro- economic framework (often unduly optimistic), sector policies, and a public sector investment program. In theory, the size of the annual budget would be governed by the national development plan’s fiscal framework, the com- position of the recurrent budget would be governed by the plan’s sector policies, and the development or capital budget would be a one-year time slice of the public sector investment program. Dual budgets and national development plans turned out to be dynamically expansionary, however, and the very same factors that were behind the expansion and collapse into dysfunction of public services played out with these budgets. To support performance, budgets must fund operating costs both adequately and pre- dictably, and neither criterion was met. In preparation, budgets became overextended and incremental, and in execution, wildly unpredictable, when the appropriated budget was replaced by the cash release system as finance ministries struggled to stay within aggregate spending limits. In the face of dysfunctional operating budgets, projects became donor-financed enclaves with donor-determined priorities. If the current round of public service reform programs is to achieve the objectives of improving the per- formance of public officials, logic requires that a complementary agenda of budget and financial management reform be pursued. One option is to restore the previous status quo and make the national development plan/dual budget work again as it was once intended. Although undoubtedly some improvement in this direction is possible, a return to the past seems unlikely to resolve problems of the present, particu- larly because national planning systems arguably have been the driving force of the overreach of government. Increasingly, the belief is that govern- ments should make annual budgets within an availabilities-determined MTEF, which is neutral between recurrent and capital spending. Elsewhere the book explores in detail the rationale for putting an MTEF at the center of 8. National development plan/dual budget systems have been dynamically expansionary for at least two reasons. First, the macroeconomic frameworks employed by planners typically used some sort of investment-driven growth model. Pressure placed on planners by politicians to maximize targeted growth led to both overoptimistic revenue assumptions and an associated requirement for a large pub- lic sector investment program. Second, when projects were scrutinized for inclusion in the public sector investment program and the annual dual budget, MDAs rou- tinely underestimated future recurrent costs. From a sector ministry’s point of view, the larger the number of projects, the greater the chance of attracting donor aid. In many countries over time, this was exacerbated by the rents staff anticipated from the procurement that would follow from investment spending. Mike Stevens and Stefanie Teggemann 75 the national budget process. The first objective of an MTEF is to set forth the government’s overall fiscal strategy. In addition, to support improvements in budget planning and execution, governments would modernize financial management systems, facilitating the timely preparation of accounts and regular fiscal reporting and ensuring a constant flow of information on activity costs, commitments, payments, and cash balances to managers. Annually MDAs would report publicly on their programs, including mea- sures of output and outcome achievement, as well as accounts and financial statements covering the use of inputs. How have the three countries fared in the pursuit of budget and finan- cial management reform? Of the three, Zambia is under the greatest fiscal stress. It is experiencing declining public revenues as copper production decreases, an extremely high wage bill (close to 8 percent of gross domestic product [GDP]), and a longstanding inability to meet the separation costs of staff identified for retrenchment. As a result, the country has a long history of cash budgets replacing appropriated budgets, playing havoc with budget predictability. The modernization of the financial management system is currently under way, although strengthening the external audit process has moved slowly. Overall, the absence of a medium-term fiscal strategy had made moving forward with its public service reform program difficult for the Zambian government. Ghana has the longest standing MTEF and, on paper, the most sophisti- cated, because it entails the unification of the previously dual budget and the incorporation of performance improvement elements. As with other parts of Ghana’s public sector reform program, however, the main responsi- bility for developing and introducing the MTEF was given to consultants. Consequently ownership and understanding of key policymakers was never more than superficial, despite strenuous efforts at staff training by the MTEF team and much enthusiasm for the new approach on the part of prac- titioners. Furthermore, at the time the mainstreaming of the MTEF was being attempted, ministers were more concerned about winning the next elections, so fiscal discipline was abandoned. Not unexpectedly, the focus of the new government of President Kufuor was more on the economic strains it inherited, than on implementing the budget reforms designed by the pre- vious government. As a result, budgeting in practice has reverted to the dysfunctional incrementalism that characterized earlier budget efforts. An integrated financial management information system has been designed and installed, but it is still being mainstreamed across government, and staff is still being trained. 76 Comparative Experience with Public Service Reform in Ghana, Tanzania, and Zambia Only in Tanzania does one find an MTEF that functions well, helping to shape budgets that support the restructuring goals of the Public Service Reform Project, and improved performance. The MTEF has completed the installation of an integrated financial management information system, which has been successfully operating for several years, the organic finance act has been overhauled,9 and the external audit process strengthened. Turning next to administrative controls, this element is crucial for keep- ing track of public sector employment and remuneration and thus forms the basis for pay reform and right-sizing efforts. Reform comprises two steps: the computerization of payroll and personnel information and integration of the two sets of data. Tanzania installed the largest and most sophisticated integrated person- nel and payroll system in SSA. Since its inauguration in 2000, it has had a great impact on cost savings and was celebrated as one of the most success- ful parts of the Public Service Reform Project.10 To every public servant, it assigned payroll identification numbers, which remain valid for a lifetime and without which no salaries are paid. Linked to the existing platinum accounting system, the integrated personnel and payroll system proved to be extremely effective at eliminating ghost-workers and double-employment and helps prevent fraud (in one month, the system registered and foiled 127 fraud attempts). Zambia is moving down a similar path to Tanzania, albeit not as far along. Currently the establishment is not computerized and not connected to payroll database, but the British Department for International Develop- ment is funding a payroll and establishment activity to improve and com- puterize payroll and personnel and to integrate the two. Zambia has also just gone through a payroll cleaning exercise to remove ghost-workers from the payroll as a first step toward reducing the ratio of wage bill to GDP to below 8 percent. 9. Most SSA countries have left their organic finance acts largely unchanged since independence. South Africa introduced a new Public Finance Management Act in 2000, and Tanzania and Uganda followed suit. Ghana and Malawi have draft laws under consideration. Updating the organic finance law is important for two reasons. First, it provides an opportunity to clarify the roles and responsibilities of key actors in government financial management. Second, the restructuring of government can lead to changes in the financial management responsibilities of MDAs, and key fea- tures may need to be made statutory. 10. The integrated payroll and personnel system was largely funded by the British Department for International Development and the World Bank Mike Stevens and Stefanie Teggemann 77 Ghana’s weak administrative controls are reason for concern. The exist- ing integrated personnel and payroll system covers only half of the public service, thus effecting little control over staffing levels and opening the door to fraud. A payroll and personnel system audit in 2002 uncovered irregular- ities in the way data are entered and delays in updating of records, allowing employees to remain on the payroll for years after officially leaving the pub- lic service. Turning finally to accountability through civil society participation and monitoring, the focus is on efforts in the three countries to use their public service reform programs to strengthen civil society participation and moni- toring, and the chapters in this book explore a rich variety of other participa- tive mechanisms. Civil society organizations including the media as well as ordinary citizens can play a powerful role in monitoring progress in public sector reform, holding their representatives accountable and providing input into the decisionmaking process. Because they are the ultimate beneficiaries of public service reform aimed at better service delivery, their voices need to be heard in the design and assessment of policies. Their involvement can range from information (through public information campaigns) to consulta- tion (beneficiary surveys, reporting mechanisms, and public hearings) and participation (active monitoring and shaping of policies) in increasing order of participation. Naturally civil society organizations that are actively partic- ipating in crafting and assessing reforms would have to be picked carefully to ensure independence from government and other special interests. The most creative of our three countries in keeping civil society informed and actively involved in the public service reform effort is Tanza- nia. Here the CSD has an explicit mandate to disseminate information on reform efforts widely and to facilitate beneficiary input, thus harnessing the role of the electorate in holding their public servants accountable. This effort is especially important because the Public Service Reform Project seeks to promote greater client orientation in service delivery, the success of which also depends on the existence of a well-informed and empowered public. Specifically, the Public Service Reform Project foresees citizen participation through the following channels, which are operational to different degrees: • A comprehensive information, education, and communication pro- gram drawing on print and electronic media, press briefings, and publicity events. The CSD also receives and responds to complaints and requests from the public. • Organizational and efficiency reviews that encourage sector min- istries and departments to foster partnerships with nongovernmental 78 Comparative Experience with Public Service Reform in Ghana, Tanzania, and Zambia organizations, communities, and the private sector to expand supply and improve access to social services. • Beneficiary participation in shaping reform and assessing perfor- mance through annual clients’ charters, service delivery surveys, and other participatory approaches. Some MDAs have launched a clients’ service charter, which defines and publicizes minimum service stan- dards, and CSD is in the process of developing a national charter. MDAs will be held accountable for the implementation of their per- formance improvement plans through beneficiary surveys on the quality of service delivery and self-assessments by senior manage- ment teams. Because the implementation of performance improve- ment plans has only recently begun, however, experience with service delivery surveys and self-assessments is limited, and their results have yet to be disseminated and action taken by citizens. Overall, it is too early to assess the effectiveness of civil society involve- ment in the Tanzanian Public Service Reform Project because the implemen- tation of this component, especially its feedback part, has been lagging behind. The sea change of traditional culture and roles of public servants and beneficiaries required for civil society empowerment to take hold should not be underestimated and will require time before it bears fruit. In Zambia, the government has a strong record of consulting civil society during program preparations. Numerous workshops with many segments of society were held in all provinces in the preparation of the government’s Public Service Reform Program, the PSCAP, the National Capacity Building Program for Good Governance, the preparation of the Decentralization Pol- icy, and other plans. This effort has, however, resulted in massive wish lists and a failure to make tough choices. Less successful has been the public information campaign, which was supposed to keep the public informed on PSCAP program objectives, activi- ties, and expected and actual results through news releases and press con- ferences as well as a program website. Although it has effectively informed civil servants, the public and even donors were barely reached. Civil society further plays virtually no role in monitoring, with the notable exception of civil society representation on the Steering Committee of the Governance Baseline Survey. The objective at the end of the 13-year adaptable program lending effort is for civil society to participate in the setting of service delivery standards and policymaking. To this end, the participatory development subcompo- nent would build capacity of civil society organizations to enhance their par- ticipation in public service reform dialog. The first phase, however, does not Mike Stevens and Stefanie Teggemann 79 go beyond studies and focus group discussions to ascertain citizens’ views on the quality of service delivery and the potential benefits of decentraliza- tion. The recently conceived second phase now envisions citizens’ charters and service delivery surveys, tapping into an increasingly vocal civil society. The overall role of civil society in PSCAP has been limited. Inputs into the project planning process had been received but civil society involvement in project implementation is weak to nonexistent, leaving untapped a valuable resource for monitoring and public pressure on government performance. Ghana widely engaged the public in the early stages but lost momentum after that. During project preparation, a series of workshops was organized to solicit views of and input from various stakeholders, in particular the Office of the President, the Cabinet, central management agencies, MDAs, local government, public managers, public servants, the private sector, the judiciary, Parliament, customers and society, and other donors. As part of project implementation, social impact assessments were carried out to assess the social implications of reforms, especially the downsizing of the public service, and to ensure broad participation and consultation. These social impact assessments as well as representation of trade unions in the Alternative Employment Committee have helped improve the design of the Alternative Employment Program. In addition, the NIRP Secretariat made available to change teams tools for beneficiary assessments that serve as entry points for reform. The Ministry of Agriculture, among others, chose to conduct a beneficiary assessment, and the results from it will feed into the restructuring of the ministry. A public information campaign targeted at internal stakeholders and the general public seeks to win support and edu- cate the public on the intentions of reform. In the future, the plan is to assess progress in service delivery through annual service delivery reports or sur- veys and annual evaluation reports and surveys. TRAINING. Skills have been lost at all levels in public services in SSA, and new skills must be acquired as the tasks and methods of government change. Thus capacity building in the form of training must be a priority in public service reform programs. Needed are both short-term measures to fill immediate capacity gaps that otherwise would hinder the implementa- tion of reform and long-term measures that aim at building sustainable in- country training capacity. One advantage in investing in public service training is that training is generally nonthreatening and welcomed by civil servants as evidence that public service reform is not solely about organi- zational change and staff reductions, but also about investing in human capital (see table 2.6). Table 2.6. Training—Intent and Results, by Country Country Reform intent Results so far Assessment Ghana • Retrain 1,200 civil servants • Trained civil servants in a few MDAs Long-term needs not addressed under PSMRP Tanzania • Establish Public Service College • Tanzania Public Service College Addresses short- and long- • Conduct training established as EA in 2000; offered 70 term capacity needs 80 • Establish distance learning center courses to 2,750 participants by 2002 • Trained senior managers • Distance learning center operational and moving toward financial sustainability Zambia • Conduct training in management • Seven training activities conducted on Long-term needs not and policy policy coordination, some of which addressed under PSCAP • Train legal staff were not relevant Source: Authors. Mike Stevens and Stefanie Teggemann 81 Short-term training interventions are sufficiently addressed in all three countries although the degree of implementation and its relevance varies. For the longer term, building local training capacity is key to ensuring the continual education of public servants at low cost. Most civil service training establishments became seriously run down during the decline of public ser- vices. Budgets were insufficient to modernize equipment, and trainers them- selves were unable to renew their skills. The emphasis was and continues to be on long-term general courses at the expense of shorter, function-specific courses and workshops, as if the purpose of a civil training institute is to equip staff with credentials they were missing or to otherwise compensate for an inadequate preservice education. Lecture style teaching, as opposed to learning through class discussion and case studies, is also overemphasized. Few countries elaborated coherent training policies, and even if they did, training budgets were always the first to be cut in a budget squeeze. Further- more, training requirements were seldom, if ever, tendered out to noncivil service training institutions, such as universities and business schools. Tanzania has made efforts to refurbish its civil service college and has raised it to EA status. The Tanzania Public Service College (TPSC) now has a strategic plan, annual business plan, and mission statement. It is also required to cover its expenses by charging the full cost of courses. Although this direction is a desirable one in which to go, at present MDAs, the main public sector clients for training, still lack a sufficiently strong accountability framework for results and the consequence is insufficient funds devoted to training and thus deficient demand. TPSC compensates for this by fee-earning secretarial and computer courses for school leavers hoping to enter the workforce either in government or the private sector. Although this move augments income, providing this service is hardly cen- tral to TPSC’s mission and suggests that direct investment by the CSD in the development of relevant course material for short courses in areas criti- cal to capacity building in new methods and processes would be worth- while, using the TPSC more deliberately as an instrument for change in the public service, backed up by periodic training needs assessments. A similar situation seems to exist in Zambia, although the TPSC’s equiv- alent, the National Institute of Public Administration lacks both the formal status of an EA and the close supervision by the parent ministry that accom- panies agency status. In Ghana the Ghana Institute of Public Administration has become autonomous, making its way by catering to the training needs of the private sector as well as government, on a fee-charging basis. Although this plan reduces the direct charge on government funds, again it infers that the Ghana Institute of Public Administration is not addressing the specific 82 Comparative Experience with Public Service Reform in Ghana, Tanzania, and Zambia needs of government transformation in as effective a way as it might if cen- tral funds were made available as a contract for specific training offerings— perhaps even in competition with other training institutions. Conclusions Second-generation reform programs in SSA are tackling a wider range of issues than did first-generation reforms, and clearly they are relevant ones. Success is still elusive in many countries, however. Of the three countries examined in this chapter, Ghana has been the most disappointing case and stands in stark contrast to the early days of structural adjustment, when Ghana was seen as a leader in SSA, with strong growth rates as evidence. The PSMRP, however, did not survive the change in government in early 2001. The timing of regime change was unfortunate. It arrived at the end of the analytic stage but before recommendations could be implemented, leav- ing the PSMRP with few concrete results to show. The Kufuor government put in place a new task force to draw up a fresh reform agenda. In Zambia following initial progress in reducing excess staffing, public service employ- ment has risen again and prompted the wage bill to surpass an alarming 8 percent of GDP. Entering its second phase, the more ambitious performance management components of the program have been, to some extent, scaled back in favor of a greater focus on expenditure accountability. Of the three, only Tanzania seems to be fulfilling its original promise. Putting the Pieces Together: Explaining Results Each project, taken as a whole, has a particular story to tell of why it suc- ceeded or failed. What were some of the ingredients that helped a reform effort succeed in one country but fail in another? As our framework depicts, some factors are very intervention-specific and technical (for example, hav- ing an integrated personnel and payroll system as a prerequisite for pay and establishment control). Others concern the political and economic environ- ment (such as the existence of political space for reform and the reform facil- itating effects of sound economic growth). Still others are specific to the way each program is conceived as a whole and implemented. The following pro- vides a discussion of what we believe made a difference in each country program. Conclusions are necessarily tentative because we find ourselves at a midprogram evaluation point and anecdotal because they are based on observations of Bank staff involved in the supervision, observations of country counterparts involved in the implementation of the programs, and the authors’ own judgment of factual progress (see table 2.7). Mike Stevens and Stefanie Teggemann 83 Table 2.7. Reform Conduciveness and Drivers for Results, by Country Criteria Tanzania Ghana Zambia Political and Favorable Highly unfavorable Challenging economic context (regime change) Reform leadership Favorable Unfavorable (fragmented Favorable and structures public service) Reform components Appropriate Too comprehensive Too comprehensive Source: Authors. In Tanzania, as alluded to earlier, the political environment of uncom- petitive pluralism and the ruling party’s strong popular base were largely favorable to reform. One of the key drivers was committed and skillful reform leadership as encapsulated by the level of competency that the CSD had acquired over years of capacity building and technical advice. Among others, the CSD had benefited from the presence of an expatriate chief technical advisor for a few years prior to the reform program. The Tanzania experience shows that temporary external advisors can be highly beneficial if they build capacity and reform commitment in the process. In addition, the CSD is well-placed and has sufficient authority to lead the reform process. Taken together, the favorable political envi- ronment, reform leadership, and structures permitted Tanzania to get a comprehensive reform program successfully off the ground. Implementa- tion has been positive so far although the government may not live up to its medium-term pay reform strategies, a failure that would put at risk the achievements of the SASE scheme in particular and the ability of the government to staff key posts with highly qualified people at large. Cau- tious optimism that the reform program is sufficiently equipped to deal with policy reversals and other obstacles along the way, is called for, however. Zambia’s political environment contains a high level of political stress given the weak power base of the ruling MMD party and the resistance that strong trade unions offer. The program leadership of PSCAP, however, and especially the recasting of its second phase are cause for optimism that progress can be made even if slow or less comprehensive than in the case of Tanzania. First of all, program leadership and implementation arrangements in Zambia are sound. Zambia’s Project Implementation Unit is made up of well-paid and well-qualified long-term local consultants and intermittent 84 Comparative Experience with Public Service Reform in Ghana, Tanzania, and Zambia international consultants. This arrangement obviously raises concerns about a lack of sustainability and integration into existing institutional structures. The latter is counteracted, however, by the fact that most of the local consultants had been civil servants and still enjoy the access to and credibility among their former colleagues. In addition, the unit has strong links to powerful champions in government. Finally, the head of the civil service is also the cabinet secretary, the principal advisor to the president, and the head of the steering committee, so responsibilities are concentrated in one person. Thus some of the usual drawbacks of project implementation units can be offset if the unit is well-managed and its staff well-connected into the establishment and political leadership. Second, the scope and choice of project components matters. As Engberg- Pedersen and Levy (chapter 3 of this book) show in their review of 15 pro- jects in SSA, 8 out of 9 expenditure accountability projects have been rated satisfactory even in countries that have low political commitment, whereas only 2 out of 8 administrative reform projects were satisfactory. Overall, nar- rowly focused reforms tended to be more successful than comprehensive ones. This finding implies that in difficult country circumstances (low com- mitment or, we propose, where political space for reform is limited) nar- rowly focused expenditure accountability components are more likely to succeed. Consequently, PSCAP II with its narrower focus on expenditure accountability should be more successful than the catch-all approach of its first phase. The logic underlying this recasting is that of sequencing and fea- sibility, in that beginning with expenditure reforms, which are politically eas- ier to implement, this also lays the foundations for the more challenging administrative reforms to follow. Eventually, however, administrative reforms will have to be addressed to sustain progress on the public expendi- ture front. In Ghana barely two years into the project, the change in government truncated midway what had started out as mainly analytic work, preparing the ground for implementation to follow. Reform proponents did not live up to the challenge of convincing the new government of the virtue of the existing reform path, thus leading to its demise. This failure harkens back to the underlying problem of having a highly fragmented public service with no unified institutional leadership and as a consequence, a NIRP Secretariat that lacked authority over and integration into the whole of the public ser- vice as well as political links. It follows that had the ties between the pro- gram team and the rest of the public service and political ranks been stronger, the reform program might have survived the change of govern- ment more easily. Mike Stevens and Stefanie Teggemann 85 Lessons Learned Overall, the experience of Ghana, Tanzania, and Zambia seems to suggest that for a reform program to succeed, the most important prerequisite is a favorable political environment that offers incentives for reform (or at least does not thwart it). Furthermore people involved in the reform program must be well-motivated and trained, and the implementation structure must be sufficiently embedded in existing administrative and political structures. Finally, program components must be appropriate, technically sound, and strategically sequenced. This conclusion reads like a wish list somewhat removed from the reality in most developing countries in need of reform. The country experience also suggests, however, that although shortcomings in each one of these three areas potentially threaten the success of a component or project, as a whole these effects can also be counterbalanced. In countries where the political environment is less favorable reform components must be carefully selected and tailored to start with basic and politically less contentious reforms while creating opportunities for more comprehensive reform and generat- ing momentum. Relying more heavily on technocrats and project staff in pushing reform forward might also be necessary. People engaged in change management must have a solid understanding of the political systems and incentives facing the political elite before entering into reform programs. Reform interventions must budget time and resources to build commitment to reform through policy dialog while resisting cookie cutter approaches. They must have the courage to start (and stay) modest. More advanced reforms should be saved for politically stable and committed countries. Similarly, political transitions could be anticipated through increased policy dialog and couched by engaging the right project and program actors. Finally, mechanisms for transparency and democratic accountability must be built into reform processes, to maintain the momentum of reform, scrutinize performance, and win allies inside and outside of government. References Kiragu, Kithinji, and Rwekaza Mukandala. 2004. “Tactics, Sequencing, and Politics of Public Service Pay Policies in Developing and Middle Income Countries. Lessons from Sub-Saharan Africa.” Draft report for the U.K.’s Department for International Development. World Bank, Africa Region, Public Sector Reform and Capacity Building Unit, Washington, D.C. Ndegwa, Stephen N. 2002. “Decentralization in Africa: A Stocktaking Survey.” Africa Region Working Paper Series no. 40. Washington D.C.: World Bank. 86 Comparative Experience with Public Service Reform in Ghana, Tanzania, and Zambia Office of Public Service Reform. 2002. Better Government Service: Executive Agencies in the 21st Century. London: The Stationery Office. Valentine, Theodore R. 1999. “Pay Reform Implementation Study: Operationalising the Medium Term Public Service Pay Policy: Final Report.” Paper presented to the Public Service Reform Programme, United Republic of Tanzania, Dar es Salaam, Tanzania. _____. 2002. “Revisiting and Revising Tanzania’s Medium-Term Pay Reform Strat- egy: Final Report.” Consultant’s report. Department for International Develop- ment East Africa, Dar es Salaam. 3 Building State Capacity in Africa: Learning from Performance and Results Poul Engberg-Pedersen and Brian Levy Even as capacity building over the past decade has enjoyed much promi- nence as the missing ingredient of African development, analysts disagree about what approaches are most promising, what has been achieved, and what is achievable. This chapter aims to help fill this knowledge gap by exploring what can be learned from the performance of a set of 15 World Bank–supported operations that sought to strengthen public sector capacity in Africa in the context of institutional and governance reform. In this chap- ter, we take the perspective of governance reformers, typically in core min- istries, such as ministries of finance, or in planning departments of sector ministries and local authorities, who seek to design and implement reforms with maximum impact on poverty reduction. Taking 1999 as the starting point, the 15 operations comprise projects managed by the Public Sector Reform and Capacity Building Unit in the Africa Region of the World Bank, namely, the projects for which some mea- sures of performance results are available in the form of project reports pre- pared by Bank staff. For nine completed projects, the performance and results are assessed on the basis of ratings in implementation completion reports (ICRs); for six current projects the performance refers to ratings included in project status reports of all relevant ongoing projects in the World Bank’s Africa Region for which sufficient time has elapsed under 87 88 Building State Capacity in Africa: Learning from Performance and Results supervision to make the supervision ratings meaningful.1 Appendix 3A pre- sents the full set of 15 operations; for this chapter, they are referred to by their respective countries. In the group for which ICRs are available, six of the nine were rated sat- isfactory after adjusting for the judgments of the Bank’s Operations Evalua- tion Department.2 In the group of ongoing operations under supervision, four of the six currently are rated satisfactory. Although this aggregate record—10 of 15 satisfactory—may perhaps best be described as mixed, bet- ter than some of the negative rhetoric surrounding operational performance in this area, but hardly cause for complacency, the more interesting exercise than judging whether the glass is half full or half empty is to explore whether systematic characteristics account for differences in performance and hence might guide future efforts. We look only at two sectors of reform under public sector governance (PSG). We include public financial management and accountability (EXPACC) and public administration reform (ADMIN), but we exclude from consideration institutional checks and balances (CHECK).3 We define operational performance as the achievement of objectives and outcomes assessed by ICRs and project status reports, distinguishing between unsatis- factory, satisfactory, and highly satisfactory. Performance is more than the completion of activity and output targets, and we use results to indicate observable improvements in governance outcomes. Note that our sample of PSG operations is small, and the sophistication of performance assessment is low, distinguishing in most cases only between satisfactory and unsatis- factory. Caution is therefore required in the interpretation of findings. 1. In addition to these six projects under supervision, the active public sector gov- ernance portfolio of the World Bank’s Africa Region included (at the time of writing in mid-2003) three operations—in Congo-Brazzaville, Malawi, and Mozambique—that were only presented to the Board or became effective in 2003; six operations—in Burk- ina Faso, Ethiopia, Madagascar, Nigeria, Tanzania, and Uganda—for which project concept documents had been approved, but that had not yet been appraised; plus a number of project initiation exercises (including in Ghana, Lesotho, Sierra Leone, and Cameroon) for which the concept notes had not yet been written. 2. A seventh (the Malawi operation) was rated satisfactory in the ICR, but margin- ally unsatisfactory by the Operations Evaluation Department. For the purposes of this chapter, it is rated as unsatisfactory, with the ADMIN component rated unsatisfactory (ADMIN was rated marginally unsatisfactory in the ICR), and the EXPACC component rated satisfactory (ECPACC was rated satisfactory in the ICR and the platform for an approved follow-up Malawi Financial Management Technical Assistance Project). 3. Some of the operations examined here included legal and judicial reform components, but these elements are not considered in this analysis. Poul Engberg-Pedersen and Brian Levy 89 Some Determinants of Performance This section lays out three hypotheses as a basis for sorting the 15 projects and examining what might be the relationship between project characteris- tics and their rated performance. The first hypothesis highlights the importance of a good fit between country characteristics and operational design. Countries differ from one another, with respect to their political and bureaucratic commitment and capacity, in two major ways: (a) the degree of political consensus and com- mitment to public sector reform and (b) the extent to which they have basic institutional capacity to sustain reform during implementation. In this analysis, we are not concerned with the degree of competitiveness and democracy in the achievement of political commitment and consensus, only with whether it exists and facilitates reform. We focus on the term “commit- ment” as opposed to the more fashionable term “ownership,” because we want to explore the effects of actual commitment by political and bureau- cratic leaders to implement reforms of the public sector. PSG operations differ from one another in the comprehensiveness of their ambition: • Some operations focus on analytic and advisory activities and limit their investments in capacity building to one part of the public sector (for example, the ministry of finance and related institutions, a partic- ular set of line ministries, or a set of public service providers at cen- tral and local levels) • Other operations aim at full reforms of the structure of the public sec- tor, involving multilevel capacity building and related investments. We hypothesize that fit is good and operations are more likely to show satisfactory performance in one of two situations: (a) where country com- mitment is far-reaching, some institutional capacity exists, and operations are comprehensive; or (b) where political and bureaucratic commitment and capacity is limited and the operational scope of projects is narrow. Fit is bad and we hypothesize that operational performance is likely to be unsatisfac- tory when comprehensive operations are attempted in countries that have weak political and bureaucratic commitment and capacity.4 Although our 4. For completeness, we note that projects of limited scope are likely formally to be rated successful in countries with far-reaching commitment, although any bal- anced judgment would note a major missed opportunity. In practice, operational underreaching has rarely been an observed sin of the World Bank or other donor agencies. 90 Building State Capacity in Africa: Learning from Performance and Results approach produces an easily testable hypothesis, we recognize that the real- ity in Africa, and elsewhere in the developing world, is one of a continuum of situations, ranging from high to no political or bureaucratic commitment to reform; from widespread to ad hoc institutional capacity; and from com- prehensive to focused reforms and operations. The second hypothesis suggests systematic differences across types of public interventions in their likely performance, especially in countries that have limited reform consensus, commitment, and capacity. We propose a distinction among: (a) interventions that focus on administrative process reform within a given state structure, (b) interventions that aim to transform state structures, (c) interventions that aim to strengthen expenditure accountability, and (d) interventions that aim directly to improve public ser- vice delivery. Our hypothesis is that process-oriented reforms, which may threaten the vested interests of stakeholders, such as civil servants or school teachers, within existing state structures and which do not offer immediate and observable results for particular beneficiaries or strong reform champions, are likely to face more constraints, particularly during implementation, than reforms aimed at transforming state structures, improving public services, and strengthening accountability for the use of public resources. We offer four possible reasons to support this hypothesis. First, process- oriented administrative reforms tend to produce losers in the short term and only to show promise in the long term. Second, more radical reforms, such as decentralization through devolution, are likely to face both tougher resistance and greater support; hence success depends on the politics of reform, that is, on the strength of reform champions: if at all launched, they stand a good chance of success. Third, reforms for which it is difficult to observe and monitor results are less likely to perform well in countries with limited commitment and capacity, because accountability cannot easily be enforced by reformers, users, and beneficiaries. Finally, reforms related to public expenditure may have greater chances of success, if and when they are supported by the ministry of finance, which in many countries is the cor- nerstone of public sector reform because of its control over the public purse. These supporting reasons for the second hypothesis concern a mixture of technical and political economy dimensions of public sector reform. Although the political economy dimensions by definition are country spe- cific, the issue of the observability of intended reform results could be a technical design issue of relevance to PSG reforms in all countries. Our third hypothesis therefore highlights the importance of direct observ- ability of intended results as a determinant of operational performance. Other Poul Engberg-Pedersen and Brian Levy 91 things being equal, operational activities for which results are more directly observable are hypothesized to have better prospects for success. Direct observability has two distinct dimensions. One dimension of observability is the extent to which results can be objectively specified and monitored, thus permitting better monitoring of performance by a reform’s sponsors and appropriate midcourse adaptation. A second dimension is the extent to which a reform or project has well defined beneficiaries or clients who can make their own judgments about what is being achieved, independent of whether quantitatively measurable indicators of performance are readily available, and potentially can exert pressure for performance. Through both dimen- sions, the greater the direct observability of results, the better is project perfor- mance hypothesized to be. Taken together, the three hypotheses offer a basis for sorting systemati- cally among the projects according to three criteria: • Project or reform comprehensiveness • Political and bureaucratic reform commitment and capacity in-country • Results-orientation in project or reform design and implementation. Table 3.1 summarizes these hypothesized links. Whether they prove to be robust will be evident in the extent to which a systematic relationship exists between the project sorting, on one hand, and performance assess- ments in the form of project ratings, on the other. The next section lays out the empirical patterns. Table 3.1. Hypothesized Effects on Project Performance of Project Comprehensiveness, Country Reform Commitment, and Results-Orientation in Project Design Country commitment to reform of PSG Scope of Orientation PSG reform of reform 1: Low/narrow 2: High/broad Results (Un)Satisfactory Highly satisfactory A: Comprehensive reform Process Unsatisfactory Satisfactory Results Satisfactory Satisfactory (few B: Narrow reform reforms here due to Process Unsatisfactory low likelihood of this sin of omission) Source: Authors. 92 Building State Capacity in Africa: Learning from Performance and Results The Relationships between Project Design and Project Performance Table 3.2 (presenting project details) and table 3.3 (a summary assessment) sort the 15 projects (and countries) according to the characteristics high- lighted above and report on how they and their component parts are rated. Some projects are focused exclusively on expenditure accountability, on administrative reform, or on service delivery (only one) and are identified as such. Other projects contain multiple components across the distinct cate- gories; these projects are identified separately by component. Appendix 3A, which provides a brief description of each project, reveals how other com- ponents (especially legal and judicial reform) have been excluded from the analysis. Two patterns are evident in tables 3.2 and 3.3. The first pattern concerns the relationship between political and bureau- cratic commitment and the prospects for success of comprehensive efforts at Table 3.2. Evidence of the Effects of Reform Comprehensiveness and Reform Commitment on Project Performance in 15 Bank-Supported PSG Operations Country commitment to reform of PSG Scope of Performance PSG reform rating 1: Low/narrow 2: High/broad Satisfactory Kenya1 EXPACC Uganda A: Comprehensive Zambia2 EXPACC Tanzania2 reform Cape Verde Guinea SERDEL Unsatisfactory Ghana1 ADMIN Kenya1 ADMIN Zambia2 ADMIN Burkina Faso EXPACC Ghana2 EXPACC Malawi EXPACC Satisfactory Tanzania1 EXPACC Zambia1 EXPACC B: Narrow reform Kenya2 EXPACC Kenya2 ADMIN Mozambique ADMIN Burkina Faso ADMIN Unsatisfactory Malawi ADMIN Sierra Leone EXPACC Sierra Leone ADMIN SERDEL refers to service delivery. Source: Authors. Poul Engberg-Pedersen and Brian Levy 93 Table 3.3. Summary of Evidence of Effects of Reform Comprehensiveness and Commitment on Performance in 15 Bank-Supported PSG Operations Country commitment to reform of PSG Scope of Performance PSG reform rating 1. Low/narrow 2. High/broad Satisfactory EXPACC: two countries Projects in three A. Comprehensive countries reform Unsatisfactory ADMIN: three countries SERDEL: one country Satisfactory EXPACC: six countries ADMIN: two countries B. Narrow reform Unsatisfactory EXPACC: one country ADMIN: three countries SERDEL refers to service delivery. Source: Authors. public sector reform. Row A in table 3.2 identifies seven operations (Cape Verde, Kenya1, Ghana1, Guinea, Tanzania2, Uganda, and Zambia2), which attempted comprehensive public sector reform. Although some ambiguity will always exist in an assessment of commitment, we believe, on the basis of familiarity with the political economy in the seven countries that the commitment to reform can reasonably be argued to have been high over the lifetime of the projects in three countries (Cape Verde, Tanzania, and Uganda) and to have been significantly lower in the remaining four coun- tries (Ghana, Guinea, Kenya, and Zambia). As predicted, in all three coun- tries for which commitment to reform was high, the operations were rated satisfactory; for the four countries with lower political commitment, the operations were at best satisfactory only with the expenditure accountabil- ity subset of components. It follows that more careful up-front analysis of political and bureaucratic commitment to reform would enable reformers to be more discriminating about whether comprehensive programs of public sector reform are worth attempting in specific country settings.5 5. The obvious question that arises is whether the levels of political and bureaucratic commitment across countries asserted in tables 3.2 and 3.3 are based on ex post assessments or whether they could have been identified ex ante. As of the time of project preparation, distinguishing ex ante whether Ghana or Tanzania had stronger political commitment might have been difficult. For the remaining five countries, ambiguity is far less, even ex ante. 94 Building State Capacity in Africa: Learning from Performance and Results The second pattern concerns which kinds of public sector reform have a prospect of being rated satisfactory in countries where political commit- ment is weaker. Tables 3.2 and 3.3 distinguish among 18 public sector reform components in these low-commitment countries: 9 focused on expenditure accountability, 8 on administrative reform, and 1 on service delivery. Although no conclusion can be drawn about the efficacy of a ser- vice delivery focus (one project was rated unsatisfactory, but this rating is consistent with a design mismatch between reform ambition and political commitment), clear conclusions emerge about the potential efficacy in low- commitment countries of the other two types of reform. Eight of the nine components that focused on expenditure accountability in low-commitment countries were rated satisfactory (and the unsatisfac- tory rating on the ninth could be explained as a consequence of Sierra Leone’s broader implosion, not weaknesses in the expenditure accountabil- ity component per se). By contrast, only two6 of the eight components or projects that focused on administrative reform in low-commitment coun- tries were rated as satisfactory. This result is consistent with the proposition that process-based reforms, where results are not readily evident and where the vested interests around the status quo may be threatened, are especially vulnerable to the syndrome of “the surgery was successful but the patient died,”—that is, the money was spent on technically competent work, but it had little subsequent impact. In countries where political or bureaucratic commitment to reform is rated as low, efforts to foster administrative reform, the quintessential process-based exercise, should thus proceed with caution. A caveat here is that process reforms that are judged as having little impact at one time could subsequently turn out to have been influential in the context of changing country circumstances. A disagreement along these lines about reform relevance accounts for the disconnect noted in footnote 2 between the ICR rating of the Malawi operation and the more negative rat- ing (based on skepticism about the credibility of the political starting point and thus prospects for impact) accorded the project by the Operations Eval- uation Department. Furthermore the Kenya public administrative reform 6. Note that project status reports do not give separate ratings for individual components; thus, formally, the Zambian administrative reform component has no unsatisfactory rating. The basis for the judgment in the text is the decision under- taken at the time of the midterm review to scale back the efforts on administrative reform and to focus attention over the next period on the expenditure accountability aspects of the operation. Also note that the ICR for Burkina Faso gives a mixture of satisfactory and unsatisfactory ratings for the ADMIN components. Poul Engberg-Pedersen and Brian Levy 95 work done in the 1990s (and rated as unsatisfactory) could turn out to have a high payoff in that country’s new regime. The evidence is not sufficiently clear to support the third hypothesis, that a focus on achieving observable results through reforms and projects increases their prospect of good performance. Appendix 3B summarizes the reasons given in the nine ICRs for ratings of satisfactory. Evidently the vari- ous EXPACC and ADMIN components have delivered a mixture of process- oriented capacity-building outputs and accountability-enhancing outcomes. Therefore, the question remains open: What explains the generally better performance of EXPACC components, if it is not primarily a greater focus on observable results? Most likely the required additional explanations will deal with the politi- cal economy of reform. Large groups with vested interests and legitimate claims in the bureaucracy may perceive the EXPACC activities as more tech- nical and less threatening than the more politically sensitive ADMIN reforms, which may have more immediate redistributional consequences. EXPACC activities may have a more powerful starting point because of their link with the ministry of finance, a link that would ensure adequate staff, resources, and attention to its own activities. Because they are more technical, EXPACC activities may be less transparent, which would run counter to the hypothesis on observability of results. All in all, such expla- nations may suggest that institutional reforms (whether administrative or financial) are likely to be opposed and perform poorly, if in the short run they threaten large organizations that have vested interests without responding to articulated demands by organized interests, who can see early benefits from the reforms. In other words the differences may be related to what could be termed “bureaucratic politics of reform.” Process reforms that are hidden in a tech- nical disguise may face less resistance. Furthermore, transparent reforms that show more immediate losers than winners face more resistance. Finally, transparent reforms responding to explicit demands and showing immedi- ate results have a greater chance of overcoming organized groups with interests in the present. Given the limited sample of projects and their mixture of components and activities, we can conclude with some confidence on the two first hypotheses, but we cannot conclude about the effects of observability and a focus on results in design and implementation. Analysis of the latter would require that project design and reporting focus more clearly on the achievement of results that can be observed by users and beneficiaries, for example, in the form of public services or influence on the use of public resources. At present, our performance criterion (“achievement of the intended project outcomes”) 96 Building State Capacity in Africa: Learning from Performance and Results is insufficiently precise to examine the effects of greater focus on results in reforms. The increasing preoccupation of the international development com- munity with outcomes and results should enable such analyses in the future. Focusing on Results—The Use of Governance Indicators Good governance has emerged as a critical factor in poverty reduction as well as a development objective in its own right. It remains, however, an elusive concept, and its significance (the what and especially the how) lacks empirical evidence. Governance indicators, that is, quantitative descriptors of gover- nance characteristics, challenges, and progress, are needed to build the case that governance matters. They are also needed to enhance aid effectiveness by achieving and demonstrating results. Since the late 1990s, the international community has reached near consensus that good governance is both a pre- condition for aid effectiveness and an objective of development cooperation. The three hypotheses discussed in this chapter all call for the use of much improved indicators of challenges and progress in country governance: • Ensuring a good fit between the design of PSG reforms and projects, on one hand, and reform commitment and capacity, on the other, requires indicators that can describe and monitor the existing and emerging political and bureaucratic commitment and capacity in each country. Governance indicators are needed both before design and during implementation. They should be used to adjust the reform and project design, rather than to exclude countries from reforms and projects. • Governance indicators must be able to capture differences between types of PSG reforms, distinguishing as a minimum between public administration reform (ADMIN), public financial management aimed at expenditure accountability (EXPACC), and institutional checks and balances on the executive (CHECK). Indicators of percep- tions of the quality of public administration are not specific enough to assist in the design of PSG reforms. • Seeking observable results in the form of governance improvements that are felt and appreciated by stakeholders and target groups requires indicators of governance progress that can be logically linked to the reforms and projects without seeking strict causality and attri- bution. Such indicators are also useful to the World Bank and other development partners for accountability purposes, because they can clarify what results have actually been achieved through the interven- tions made and the resources provided for governance reform. Poul Engberg-Pedersen and Brian Levy 97 In the rapidly expanding governance indicator industry, the role that indicators can and should play in public sector governance reforms has caused some confusion. We believe distinguishing between four major pur- poses of indicators is necessary (see also table 3.4), because the scope and methodology of the governance indicators differ according to purpose: • Identifying country-specific problems and challenges in public sector gover- nance, to be addressed by the country and its partners. This purpose requires specificity in problem identification. Country-specific sur- veys, diagnostics, and analyses of political economy and the politics of reform are needed. Because of the situation specificity required, few global, comparative governance indicators can achieve this purpose. • Monitoring the results of governance reforms and development cooperation. This purpose requires indicators of relatively short-term (preferably annual) changes in governance that can be logically linked (although not attributed to) specific reforms or operations supported by donors. Perceptions indicators tend to be too broad for this purpose, record- ing mainly medium- to long-term changes in overall governance. • Allocating aid or debt relief on the basis of assessments of countries’ absolute and relative performance in governance, assessing both level and trend for individual countries. This purpose is part of the World Bank’s coun- try policy and institutional assessments, as well as of benchmarks and indicators of public financial management (used in a Heavily Indebted Poor Countries context). These assessments are made by experts on the basis of explicit performance criteria and allow for both absolute assessments (for example, can a country ensure that debt relief is used for poverty reduction?) and relative assessments (allocation of aid between countries on the basis of where the donor can hope to assist in pulling the largest number of people out of poverty). • Raising global awareness and public debate about the importance of different dimensions of governance and the weak performance of individual countries and regions in this respect. This purpose is the primary one of indica- tors of governance perceptions,7 including Transparency Interna- tional, Freedom House, and World Bank Institute. 7. Although surveys, diagnostics, Heavily Indebted Poor Countries reviews, and country policy and institutional assessments also record perceptions (and hence are not objective), a distinction is maintained between such assessments and the perception-based indicators of the fourth type, because the assessments are under- taken by experts with transparent criteria that can be challenged. 98 Building State Capacity in Africa: Learning from Performance and Results Table 3.4. Major Purposes, Producers, and Types of Governance Indicators 4. External 1. Assessment 2. Assessment 3. External assessment of Governance of weaknesses; of progress for assessment of performance indicators identification documentation performance for for public purposes of solutions of results aid allocation awareness Governance Surveys, Governance Country policy Aggregation indicators specific progress and institutional of perceptions types and indicators indicators assessments; products HIPC benchmarks Main Specialized Monitoring and World Bank; World Bank producers units in evaluation units World Bank/ Institute/non- development in development International governmental community community Monetary Fund organizations Production Ad hoc, Performance Annual country Regular process and country- indicators tied policy and aggregation methodology specific and to operations institutional of multiple thematic assessments; sources efforts regular HIPC assessments Source: Authors. Table 3.4 maps out the primary usefulness of four types and products of governance indicators in relation to the four purposes delineated above. It shows the breadth of governance indicator production and the diversity in production processes and methodologies. The degree of gaps in governance indicator quality, coverage, and avail- ability differs between the four types. Mainly because of intensive efforts within the World Bank, the two sets of external performance assessments are the best developed, documented, and updated.8 The country policy and institutional assessments have evolved since 1977, their main purpose being to allocate International Development Asso- ciation funds effectively among client countries. Several social and gover- nance criteria have been introduced in recent years, reducing the dominance 8. See the following World Bank web sites: http://www.worldbank.org/wbi/governance/index.html, http://www1.worldbank.org/publicsector/indicators.htm, http://web.worldbank.org/WBSITE/EXTERNAL/EXTABOUTUS/IDA/0,,content MDK:20052347~menuPK:116699~pagePK:83988~piPK:84004~theSitePK:73154,00.html. Poul Engberg-Pedersen and Brian Levy 99 of economic criteria. Each country policy and institutional assessment is based on 20 equally weighted criteria, organized into four clusters: • Economic management • Structural policies • Policies for social inclusion and equity • Public sector management and institutions, which comprises five indicators: — Property rights and rule-based governance — Quality of budgetary and financial management — Efficiency of revenue mobilization — Quality of public administration — Transparency, accountability, and corruption in the public sector. The governance indicators developed for research and awareness pur- poses by the World Bank Institute address the following six sets of issues: • Voice and accountability • Political stability and absence of violence • Government effectiveness • Regulatory quality • Rule of law • Control of corruption. International development agencies, including the World Bank and the Organisation for Economic Co-operation and Development; bilateral agen- cies such as the U.K.’s Department for International Development; and acad- emic institutions are gradually intensifying work on the first two types of indicators mentioned aimed at governance diagnostics and progress moni- toring. Local and international development practitioners, who are engaged in PSG reforms, need more indicators that can meet a combination of the first and second purposes from table 3.4. The biggest gap in the current array of governance indicators lies in the assessment of progress of governance reforms and in the documentation of results related to PSG operations. The World Bank is currently seeking to develop governance operations progress indicators to close this gap. These indicators aim to strengthen a results focus in governance reforms and operations, enabling operational staff and their clients to identify and prioritize problems and to monitor and document improvements in governance that are logically linked to Bank- supported governance operations. Governance operations progress indica- tors will be intermediate indicators aimed at closing the gap between activity and output monitoring on one hand and outcome and impact assessment on the other. The indicators must be country-specific but allow cross-country 100 Building State Capacity in Africa: Learning from Performance and Results Table 3.5. Governance Operations Progress Indicators Public financial management and Public administration Institutional checks and accountability, EXPACC reform, ADMIN balances, CHECK • Aggregate fiscal • External accountability • Constitutional balance discipline to oversight bodies and of power • Allocative efficiency: partners • Oversight by legisla- strategic prioritization • Internal accountability ture and supreme audit • Operational and within the executive institutions technical efficiency • Human and organiza- • Rule of law; legal and tional capacity judicial reform • Structural coherence • Institutionalized and alignment within stakeholder executive participation Source: Authors. comparisons of governance progress. Governance operations progress indi- cators monitor both the intermediate outcome and the process of governance operations. They cover issues in the three areas of PSG outlined in table 3.5. Indicators on these governance issues represent an attempt to move for- ward on the achievement of results and the monitoring of progress through both process- and outcome-oriented reforms. Overall, the comprehensive coverage of governance operations progress indicators should enable them to provide independent, regular information on governance changes that can be linked to reforms and operations and hence can point to more robust results than the performance assessments in project status reports and ICRs. Conclusions The performance of nine recently completed and six current Bank-supported operations aimed at strengthening public sector capacity in Africa provides important evidence on what works in capacity building through public sec- tor governance operations. This chapter has examined three hypotheses, the two first of which have been partially validated (despite the limitations of the sample): • A good fit is needed between country characteristics (political and bureaucratic commitment and capacity) and operational design. Comprehensive PSG reforms require a significant degree of political consensus, sustained commitment, and bureaucratic capacity. In the absence of consensus, commitment, and capacity, more focused Poul Engberg-Pedersen and Brian Levy 101 reforms are more likely to perform well and yield immediate results. Operations design should fit into a continuum of country characteris- tics, not just a distinction between good and bad. • The differences in performance ratings between types of public inter- ventions may be explained in part by different degrees of results orien- tation in activities and outputs. Administrative reform within an existing state structure tends to consist of process-oriented activities that cannot be linked easily to observable results of relevance to ulti- mate clients and beneficiaries. Reforms aimed at transforming state structures (for example, through decentralization), strengthening pub- lic expenditure accountability, and improving public service delivery tend to be more directly linked with results that can be monitored by reform sponsors and appreciated as change toward good governance. • Accordingly, PSG activities yielding observable results are more likely to perform well. The focus on observable results is required especially in countries where the political consensus and commit- ment to reform is low or unsustainable. Setting targets that can be monitored allows reformers to manage reform processes better. It enables donors to contribute even in difficult country contexts. It entails, however, a risk of generating negative or perverse incentives. In countries where more comprehensive reforms are possible, focus should be increased on the processes of reform. The third proposition remains a hypothesis, because the operations reviewed here do not provide sufficient evidence for its validation. This chapter offers instead a number of political economy explanations for the difference between the relatively unsatisfactory performance of administra- tive process reforms and the relatively satisfactory performance of expendi- ture accountability reforms. Finally, this chapter suggests how more systematic use of governance indicators can enhance the prospects of suc- cessful public sector governance reforms and state capacity building by allowing good fit to country commitment and capacity, by encouraging a results and accountability focus in reforms, and by enabling reform benefi- ciaries to observe reform processes and demand reform outputs. Appendix 3A. Summary of the 15 Operations Covered Completed Projects BURKINA FASO: PUBLIC INSTITUTIONAL DEVELOPMENT PROJECT. Interna- tional Development Association (IDA) $16 million. Implementation com- pletion report (ICR), May 2001. Burkina Faso EXPACC, ADMIN. Objectives: 102 Building State Capacity in Africa: Learning from Performance and Results Support the development of Burkina Faso’s major public institutions for economic and sector management and transform key public institutions, the laws supporting them, and the personnel staffing them by improving (a) information, (b) budget management and programming, (c) public pro- curement, (d) civil service performance, and (e) the judiciary system. Over time, the emphasis was increasingly put on the development of tools for improving economic management, including budget administration and data collection. CAPE VERDE: PUBLIC SECTOR REFORM AND CAPACITY BUILDING CREDIT. IDA $13 million. ICR, January 2001. Cape Verde. Objectives: (a) Support the government’s broader program to transform and modernize key public institutions, the laws supporting them, and the personnel staffing them and (b) support improvements in economic policy analysis and management, and in the legal, judicial, and administrative environment for investment and for business operations through institutional development and capacity building in public administration, both at the central and municipal levels. The midterm review identified a need to redefine the role of the Reform Advisory Implementation Monitoring Board, splitting it into two boards: a political level and a technical level. KENYA: INSTITUTIONAL DEVELOPMENT AND CIVIL SERVICE REFORM PRO- JECT. IDA $163 million, of which $23 million for components considered here. ICR, March 2001. Kenya1 EXPACC, ADMIN. Objectives in 1994: Set in the context of structural adjustment, (a) overall rationalization of staffing and improved establishment control, (b) wide-ranging compensation reforms, (c) better personnel management practices, (d) staff development, and (e) rationalization of five target ministries in service delivery (agricul- ture, health, finance and planning, water development, roads and public works). Revised objectives in 1998: Reshaping the role of government through (a) reduction of staff, (b) reductions in functions, (c) improvements in ser- vice delivery, and (d) improved budgeting and controls through a medium- term expenditure framework. Revised objectives in 1999: Financing the salaries of four of the government’s six-member change team to address inefficiency and corruption throughout the public service. MALAWI: INSTITUTIONAL DEVELOPMENT PROJECT II. IDA $20 million. ICR, March 2002. Malawi EXPACC, ADMIN. Objectives: (a) Address specific sys- temic management issues that constituted bottlenecks to efficiency and effectiveness in the civil service, (b) help the Department of Statutory Poul Engberg-Pedersen and Brian Levy 103 Corporations better manage the interface between the parastatal sector and the government, and (c) strengthen the Malawi Institute of Management. Five major components: (a) Civil service policy and information framework, (b) the Department of Personnel Management and Training, (c) the Ministry of Finance, (d) the Department of Statutory Corporations; and (e) the Malawi Institute of Management. MOZAMBIQUE: CAPACITY BUILDING—PUBLIC SECTOR AND LEGAL INSTITU- TIONS DEVELOPMENT PROJECT. IDA $13 million. ICR, March 2001. Mozam- bique ADMIN. Original objective: Build and maintain capacity in key public institutions and skills areas by expanding the supply of well-trained senior planners, policy analysts, managers, and technicians and by enhancing pay and other incentives and conditions of employment for senior civil servants. Specific objectives: (a) Strengthening legal institutions and professional capa- bilities and (b) developing public administration and management skills and enhancing civil service personnel systems. The project was revised through several amendments, aiming at more modest practical steps to lay the foundation for more far-reaching reform. SIERRA LEONE: PUBLIC SECTOR MANAGEMENT SUPPORT PROJECT. IDA $10 million. ICR, June 2002. Sierra Leone EXPACC, ADMIN. Main objective: Strengthening the capacity of government institutions responsible for fiscal management. Additional objectives: (a) Increasing public service productivity by initiating civil service reforms, (b) improving the efficiency of public enterprises as well as restructuring and reducing the government’s role as owner and operator, and (c) improving aid management. Components: Reforms of (a) budget and expenditure control, (b) Accountant General’s Office, (c) Auditor General’s Department, (d) economic policy research, (e) procurement, (f) aid coordination, (g) Bank of Sierra Leone, (h) civil service, (i) public enterprises, (j) petroleum unit, (k) social and poverty action pro- gram, and (l) institutional support. Other studies would also be needed. TANZANIA: FINANCIAL AND LEGAL MANAGEMENT UPGRADING PROJECT. IDA $19 million. ICR, January 2001. Tanzania1 EXPACC. Objectives: To improve (a) accounting and auditing, (b) the legal and regulatory frame- work, and (c) the administration of justice. These objectives were to be achieved by strengthening four institutions on the financial side (the National Board of Accountants and Auditors, the Institute for Financial Management, the Institute of Accountancy in Arusha, and the Office of Con- troller and Auditor General) and four institutions on the legal side (the 104 Building State Capacity in Africa: Learning from Performance and Results Attorney General’s Office, the judiciary, the Law Reform Commission, and the Registrar of Companies), as well as by undertaking studies on the legal sector, cost recovery, and the Tanzanian Audit Corporation. UGANDA: INSTITUTIONAL CAPACITY BUILDING PROJECT. IDA $39 million. ICR, June 2002. Uganda. Objectives: (a) Continuation of civil service reform, (b) decentralization, and (c) legal and financial accountability framework and institutions. Five major components: (a) Central government capacity building; (b) local government capacity building; (c) legal sector reform; (d) accoun- tancy profession reform; and (e) establishment of training funds, through a cross-cutting, demand-driven instrument for financing training on a cost- recovery basis. Most funds went to local government capacity building. ZAMBIA: FINANCIAL AND LEGAL MANAGEMENT UPGRADING PROJECT. IDA $17 million. ICR, December 2000. Zambia1 EXPACC. Objectives: (a) Improve the flow of government business, (b) improve the implementation of devel- opment programs, and (c) facilitate reform. The objectives were aimed at removing obstacles through capacity building in accounting and auditing, public sector procurement, legal drafting, the administration of justice, and the provision of business information. Three strategic thrusts: (a) Improving key aspects of public–private interface, including regulatory, legal, judicial, and procurement; (b) increasing the transparency of the legal process; and (c) building the accounting profession within Zambia. Projects under Supervision GHANA: PUBLIC SECTOR MANAGEMENT REFORM PROJECT. IDA $14 mil- lion. Board document, April 6, 1999. Ghana1 ADMIN. Objectives: Phase I of an intended three-phase comprehensive (IDA $84 million) adaptable pro- gram loan aimed at (a) redefining the role and functions of the state, (b) designing appropriate institutions and systems to implement this role, and (c) rationalizing the existing structure and systems to meet the new design. Phase I components: (a) Initiate pilot program for reform of subvented agen- cies; (b) realign central management agencies, for example, the Ministry of Finance and the Public Services Commission; (c) strengthen human resource management system; (d) introduce coherent public sector pay pol- icy; and (e) develop public–private partnership program GHANA: PUBLIC FINANCIAL MANAGEMENT REFORM PROGRAM. IDA $21 million. Board document, December 1996. Ghana2 EXPACC. Objective: Pro- mote efficiency, transparency, and accountability in public financial Poul Engberg-Pedersen and Brian Levy 105 management through rationalization and modernization of the budget and public expenditure management system. Components: (a) Improve planning and budgeting of public expenditures through a medium-term expenditure framework system; (b) strengthen accounting and fiscal reporting through, among other things, the introduction of an information technology–based Budget and Public Expenditure Management System; and (c) reform audit and procurement systems. GUINEA: CAPACITY BUILDING FOR SERVICE DELIVERY PROJECT. IDA $19 million. Board document, November 22, 1999. Guinea SERDEL. Objective: Phase I of an intended three-phase comprehensive (IDA $139 million) adaptable program loan aimed at strengthening centralized and decentral- ized administrative systems to provide effective public services to the rural population, in a way that empowers local institutions to take charge of their own needs. Components: (a) Decentralize public expenditure management, (b) decentralize accountability and capacity for service delivery to local agencies and communities, and (c) introduce a performance-based incentive system. KENYA: PUBLIC SECTOR MANAGEMENT TECHNICAL ASSISTANCE PROJECT. IDA $15 million. Board document, July 6, 2001. Kenya2 EXPACC, ADMIN. Objective: Improve governance by putting in place systems and human resources capacity that are the necessary conditions for achieving greater fiduciary responsibility and reducing corruption in the public service. Com- ponents: (a) Public service reform, including ministerial rationalization, per- sonnel establishment control and harmonization of payroll and personnel data system, new pay and benefits policy, and performance management; (b) public financial management, including implementing a medium-term expenditure framework, strengthening government finance and account- ing, strengthening the auditor-general, and enacting procurement reform; and (c) legal sector reform. TANZANIA: PUBLIC SERVICE REFORM PROJECT. IDA $41 million. Board docu- ment, October 30, 1999. Tanzania2. Objective: Phase I of an intended three- phase comprehensive (IDA $94 million) adaptable program loan to support the attainment of a high rate of economic growth and ensure that delivery of quality public services within priority sectors conforms to public expectations for value, satisfaction, and relevance by end 2011. Components: (a) Install a performance improvement management framework in key ministries depart- ments and agencies; (b) facilitate transfer of functions, services, and opera- tions to enhance private sector participation; (c) implement executive 106 Building State Capacity in Africa: Learning from Performance and Results agencies program; (d) strengthen management information systems; and (e) transform public service into meritocracy through leadership and manage- ment skills development, merit-based recruitment and promotion, ethics awareness, and gender mainstreaming. ZAMBIA: PUBLIC SERVICE CAPACITY BUILDING PROJECT. IDA $28 million. Board document, February 22, 2000. Zambia2 ADMIN, EXPACC. Objective: Phase I of an intended three-phase comprehensive (IDA $61 million) adapt- able program loan intended to help make public service delivery processes more effective and efficient to facilitate economic growth and reduce poverty. Components: (a) Restructure the public sector, by means of private sector participation, more effective management of payroll and establish- ment, and pay reform; (b) improve strategic management of policy and public service; and (c) strengthen financial management, accountability, and transparency. At the midterm review, resources were refocused on the expenditure accountability component, which will also be central in phase II of this adaptable program loan, currently under preparation. Appendix 3B. Reasons Given for Satisfactory Performance in Implementation Completion Reports With the understandable exception of the Sierra Leone project, the ICRs for eight projects found reasons to give satisfactory ratings to components of either EXPACC (in seven projects) or ADMIN (in three projects). The justifi- cations are summarized below. Burkina Faso: Public Institutional Development Project EXPACC COMPONENT. Budget management was rated highly satisfac- tory. The project helped to introduce and ensure application of up-to-date budget and treasury management instruments. These instruments enabled expenditure control, computerized links between the Ministry of Finance and all ministries, and an accounting system involving also regional offices of the Ministry of Finance. Cape Verde Public Sector Reform and Capacity Building Credit EXPACC COMPONENT. The strengthening economic management capacity category, which was rated highly satisfactory, included installa- tion of a macroeconomics model, a debt management system, and a man- agement information system for the multiyear public investment program; Poul Engberg-Pedersen and Brian Levy 107 establishment of the National Institute of Statistics; preparation of a legal framework to integrate investment and recurrent budgets at central and local levels; and completion of a legal framework for economic manage- ment. According to the ICR, however, the component did not target the budget formulation and execution mechanisms that are needed to enhance accountability and transparency of the budget. ADMIN COMPONENT. Modernizing the civil service category was rated satisfactory because of extensive technical and computer training program, a first phase of an early retirement program, and a successful pilot program for voluntary departure of civil servants. The decentralization and munici- pal strengthening category was rated satisfactory because of a legal frame- work for fiscal decentralization and accountability and for delegation of revenue generation and financing authority (EXPACC-related), as well as capacity-building in municipalities. Kenya: Institutional Development and Civil Service Reform Project EXPACC COMPONENTS. All EXPACC components were small, but they were rated satisfactory. Preparation and implementation of a medium-term expenditure framework and capacity building in the Office of the Accoun- tant General were assessed to be more successful than other components in the project. Malawi: Institutional Development Project II EXPACC COMPONENT. The strengthening institutional capacity of the Ministry of Finance category, which was rated satisfactory, included the intro- duction of forward budgeting under a medium-term planning framework, the introduction of a new public sector accounting system, and the establish- ment of a tax policy analysis unit. Mozambique: Capacity Building—Public Sector and Legal Institutions Development Project ADMIN COMPONENT. Notable successes included a public sector reform workshop and two anticorruption seminars, but the main reason for the sat- isfactory rating was the extension of the personnel management system to all sectoral ministries and provincial governments. It represents an important first stage of a comprehensive reform of the civil service. Other activities included capacity building in the Ministry of State Administration. Still, the institutional development impact of the project was rated modest. 108 Building State Capacity in Africa: Learning from Performance and Results Tanzania: Financial and Legal Management Upgrading Project EXPACC COMPONENTS. National Board of Accountants and Auditors, which was rated satisfactory, increased pass rates, issued new accounting and audit standards, and reduced dependence on government budgetary support from 52 percent in 1991 to 21 percent in 1999. Institute of Accoun- tancy in Arusha, which was rated unsatisfactory, received fewer students from the public sector, but maintained production of trained accountants and lowered dependence on government funding. The Office of the Con- troller and Auditor General, which was rated satisfactory, managed to process 60 percent of transactions and accounting records by computer. The Institute of Financial Management, which was rated highly satisfactory, almost doubled student intake, increased pass rate, and reduced depen- dence on government funding. Uganda: Institutional Capacity Building Project EXPACC COMPONENTS. All EXPACC components were small, but they were rated satisfactory. They included the establishment of an intergovern- mental fiscal system and development of local professional accounting qualifications and examining capability at the Institute of Certified Public Accountants of Uganda. ADMIN COMPONENTS. Comprising civil service and decentralization programs, the ADMIN components were rated satisfactory because the pro- ject supported timely technocratic, institutional, and capacity-building responses to a fast-evolving, government-driven process of constitutional change and legislative reform. Zambia: Financial and Legal Management Upgrading Project EXPACC COMPONENTS. All components were rated satisfactory. The Zambia Institute of Chartered Accountants produced 700 new technician graduates and facilitated the regulation of the accounting and auditing pro- fessions. The Zambia National Tender Board computerized tender evalua- tion for faster tender processing and greater document security; adopted and published a set of procurement guidelines, improving the basis for inspecting performance of contractors and suppliers; and decentralized the procurement function, achieving fewer procurement delays. 4 Reforming Pay Policy: Techniques, Sequencing, and Politics Kithinji Kiragu, Rwekaza Mukandala, and Denyse Morin The argument that adequate pay is crucial to sustaining the motivation, perfor- mance, and integrity of public servants has been widely accepted and docu- mented.1 Evidence from around the world indicates that when their salaries are low, government workers either cut back their productivity or on their work hours. The decline in production is greater as the compensation diminishes. As an alternative, they will actively seek to switch to jobs in the private sector. In many countries in Africa and elsewhere, the economic crises of the 1970s and 1980s, and the attendant deterioration in service delivery, brought public service employment and pay reform to the fore of the policy agenda. Yet despite the extensive documentation of the seriousness of the problems and issues associated with pay reform and the repeated efforts at reform, poor pay has remained an intractable policy issue in most public services around the world. Public service pay is indeed a complex policy issue. Stakeholders inter- ests are many and diverse, as are the policy objectives that political groups seek through pay decisions, such as the following: • Pay policy can be used to pursue egalitarian and equity goals. • Labor unions and vested groups in the public service often ignore labor market prices. • Raising pay is difficult when budgetary allocations to operational and maintenance costs are low. • Significant improvements in service delivery entail expansion of the workforce. 1. Throughout this chapter, pay is defined as the sum of salary, allowances, in- kind benefits, and retirement benefits. 109 110 Reforming Pay Policy: Techniques, Sequencing, and Politics • Pay operates in conjunction with other organizational and manager- ial systems. In this context, public service pay issues are obviously extremely politi- cal, and several questions about the political aspects of public service pay policies are pertinent, namely: • What is the political rationale for or against public service pay reform? • What are the critical political (including ideological) and institutional constraints regarding pay reform? • Which interests are affected by pay reform, how, and to what degree? • Which individuals, groups, and institutions are critical players in the pay reform process and why? • What is the political rationale for the choices that have been made or for not making those choices? • What factors have constrained implementation of pay reform decisions? This study examines the tactics, techniques, sequencing, and politics of pay policy in eight countries in Africa: Benin, Botswana, Burkina Faso, Ghana, Tanzania, Senegal, Uganda, and Zambia. The objectives of the study are as follows: • Systematically collect and organize data and information on public service pay reform experiences in the selected countries • Explore and describe the context and limits to which models have guided pay reform strategies and policies in the selected countries • Explain the political contexts and rationales underlying pay reform policies and strategies adopted by the selected countries. These issues are addressed in the sections that follow. Pay Trends and Structures The survey data indicate that six out of the eight study countries experi- enced either an overall decline or stagnation in salary levels between early 1990s and the turn of the century.2 This trend is as illustrated in figures 4.1 2. An overall decline in salary levels does not necessarily translate to an overall reduction in real pay because allowances and benefits could be used to compensate. Salary remains the core element of any pay system, however. In fact, the replacement of salary by allowances and benefits, in the majority of cases, is a choice of tactics and techniques to deal with such external factors as taxation (where allowances and benefits permit tax avoidance) or political considerations (as when such allowances and benefits must be paid selectively, inequitably, and nontransparently). Kithinji Kiragu, Rwekaza Mukandala, and Denyse Morin 111 Figure 4.1. Comparative Trends in Median Salary Levels for the Study Countries, 1992–2001 14,000 Amount (US$ per year, parity terms) 12,000 10,000 8,000 6,000 4,000 2,000 0 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 Year Benin Botswana Burkina Faso Ghana Senegal Tanzania Uganda Zambia Source: Government data. and 4.2. Salary data on Benin were not available except for 2000 and 2001, but the interviews with key officials confirmed that the trend was the same as with other francophone countries of West Africa. In Burkina Faso and Senegal, real salaries declined for all groups, but governments increasingly used allowances and in-kind benefits to raise pay. However, these allowances were selectively awarded to teachers, health workers, and mag- istrates. Apparently the trade unions of these groups had pressured the gov- ernment into this action. At the same time, the public service elite in these countries, for example, ministers, members of the national assembly, and secretary generals, also awarded itself more allowances and in-kind bene- fits. Therefore in Burkina Faso and Senegal, only those public servants in politically weak or disadvantaged categories, that is, nonunionized middle to senior cadres and members of the weaker unions, bore the full brunt of the decline in real salaries. In Ghana and Zambia, real salaries generally continued to drop through the 1990s despite recurring, almost annual, nominally large salary increases 112 Reforming Pay Policy: Techniques, Sequencing, and Politics Figure 4.2. Comparative Trends in the Wage Bill to Gross Domestic Product Ratio for the Study Countries (excluding Botswana), 1992–2000 10 9 8 7 6 Ratio (%) 5 4 3 2 1 0 1992 1993 1994 1995 1996 1997 1998 1999 2000 Year Benin Burkina Faso Ghana Senegal Tanzania Uganda Zambia Source: Government data. that the strong trade unions in these countries achieved by persistently pres- suring the governments. • Ghanaian public servants had salary increases in • 1994 (20 percent) • 1996 (38 percent) • 1997 (10–18 percent) • 1999 (20–42 percent) • 2000 (21 percent in March and 20 percent in November). • Zambian public servants had general salary increases in • 1990 (50 percent) • 1991 (50 percent) • 1992 (27 percent) • 1993 (5 percent in April and 50 percent in September) • 1994 (24 percent) Kithinji Kiragu, Rwekaza Mukandala, and Denyse Morin 113 • 1995 (30 percent first and later another 45 percent) • 1996 (105 percent in April and about 12 percent in September) • 1997 (38 percent) • 1999 (40 percent) • 2000 (20 percent in April and 15 percent in October). These frequencies generally reflect the intensity of pressure from trade unions that governments in these two countries feel. Despite the general decline in real salaries, the government of Benin managed to stick with a mix of (a) a salary freeze in promotions effected in 1987, and (b) modest adjustments based on the salary index throughout much of the 1990s. It apparently achieved this stasis by a combination of two political tactics. First, trade unions were co-opted in various ways, including by maintaining continuous dialog with them. Second, unions were indirectly intimidated or undermined by the promotion of an alterna- tive cadre of nonunionized employees whom the government had employed on contract terms. In the teaching profession, for example, as of mid-2002, up to 60 percent of the work was performed by about 8,000 con- tract workers. Therefore the Benin government is not under severe trade union pressure from strike threats as is the case in, for example, Burkina Faso, Ghana, Senegal, and Zambia. The two countries where real salary levels in 2000 were above those pre- vailing in the early 1990s are Tanzania and Uganda. Consequently, even though by comparison these countries had the lowest salary levels in 1992 (see figures 4.1 and 4.2), by the end of the decade they generally had higher salary levels than the other study countries except Botswana. The turn- around in salary trends in both Tanzania and Uganda was launched with measures to consolidate allowances and other pay benefits into salaries. These measures were strongly influenced by structural adjustment program (SAP) conditions targeted at enabling the effective control of the public ser- vice wage bill, which was considered a crucial element in the strategies to reinstate fiscal discipline and stability. Beyond the consolidation of allowances, however, the two governments have, with varying degrees of success, maintained the policy objective of enhancing their employees’ salaries. The momentum for regular raises in public service salaries has largely been sustained by the commitment of influential technocrats, who have continue to present their political leaders with politically acceptable policy and strategy papers on the subject. The support of key development partners to the agenda presented by technocrats, especially the World Bank and the United Kingdom’s Depart- ment for International Development, has also played an important part in 114 Reforming Pay Policy: Techniques, Sequencing, and Politics persuading political leaders in Tanzania and Uganda to sustain salary raises for public servants over the past decade or so. In Tanzania, these donor part- ners have supported the implementation of an accelerated salary enhance- ment scheme that is performance-oriented and selective. With the development partners’ support, this scheme tops up salaries for specifically tasked and well-performing technical, professional, and managerial staff to levels in line with the medium- to long-term target salary levels for these grades of personnel. In Uganda, progress in enhancing the salaries of public servants has been a condition under SAPs and their successors, that is, poverty reduction support credits. Thus the convergence of a number of inter- nal and external factors has been crucial to the apparent successes of Tanzania and Uganda in sustaining public service pay policies over the past decade. Botswana stands out as paying its public servants high salaries (in U.S. dol- lar parity terms) compared with the other study countries. In 2000 the top salary level in Botswana was estimated at about US$40,235 per year. This extraordinary position of Botswana is in line with the country’s macroeco- nomic status, especially its high gross domestic product (GDP) per capita, high rate of economic growth, and budget surpluses for most years. The next closest salary level was in Uganda, which was estimated at about US$12,908 per year. The lowest of the top salary levels was in Ghana, at about US$3,373 per year. The data from the case studies suggest that in the first half of the 1990s the wage bill to GDP ratio was on the decline in all the study countries except Tanzania and Uganda. These two countries have consistently raised their wage bill to GDP ratio, albeit from rather low levels, between 1992 and 2000 (figure 4.3). Furthermore, because these economies were growing in those years while employment numbers were generally declining or under control, the rise in the wage bill to GDP ratio translated into real increases in public service pay levels. Nevertheless, the trends suggest extraordinary control of the wage bill in the context of each country’s macroeconomic and fiscal framework. Zambia has also had a comparatively high wage bill to GDP ratio: nearly 8 percent in 1992 and just above 6 percent in 2000. Considering that both top and bottom salary levels for Zambia (in U.S. dollar parity terms) are signifi- cantly lower than for both Tanzania and Uganda, the comparatively high wage bill to GDP ratio suggests that either the Zambia public service has a comparatively high number of employees or that actual pay levels are much higher than the salary levels indicate. In the latter case, a much higher por- tion of the compensation for public servants in Zambia would be paid in the form of allowances. Either way, introducing pay reform in Zambia would be difficult, whether because of difficulties in introducing transparency in the pay system or because of controlling growth in numbers. Kithinji Kiragu, Rwekaza Mukandala, and Denyse Morin 115 Figure 4.3. Comparative Nominal Wage Bill to GDP Ratio for the Study Countries (excluding Botswana), 1992–2000 10 9 8 7 6 Ratio (%) 5 4 3 2 1 0 1992 1993 1994 1995 1996 1997 1998 1999 2000 Year Benin Burkina Faso Ghana Senegal Tanzania Uganda Zambia Source: Government data. Fluctuations in the wage bill to GDP ratio over the past decade have been much less pronounced for the three francophone study countries: Benin, Burkina Faso, and Senegal. The ratios for these countries were between 6 and 7 percent in 1992 and between 5 and 7 percent in 2001 (figure 4.4). Nevertheless, this reduction in the ratio translated to reduced levels of pay. Botswana again stands (figure 4.3) out, by a considerable margin and for virtually all the years, as the study country that had the highest nominal wage bill ratio,3 amounting to about 9 percent in 2000. This ratio is consis- tent with the comparatively high levels of salary that also prevail. Except where GDP growth is high, a reduction in the wage bill to GDP ratio would necessarily compel a reduction in either average pay levels or 3. The wage bill to GDP ratio is qualified here as nominal, because in countries that have weak budgeting and wage bill accounting, significant elements of the wage bill are hidden under nontransparent expenditure items. 116 Reforming Pay Policy: Techniques, Sequencing, and Politics Figure 4.4. Nominal Salary Decompression Ratio for the Study Countries, 1992–2001 35 30 25 Ratio (%) 20 15 10 5 0 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 Benin Botswana Burkina Faso Ghana Senegal Tanzania Uganda Zambia Source: Government data. employment numbers. As described in the previous section, in the slow- growth countries, pay levels were allowed to drop. One important political leverage for achieving this reduction in the wage bill to GDP ratio in these countries during the 1990s was the binding article of the Economic Commis- sion of West Africa States that commits all member countries to a maximum wage bill to government expenditure ratio of 35 percent. The governments of the region have invoked this article in negotiating with trade unions. The salary decompression ratio is an important attribute of a pay struc- ture, with a low ratio indicating a commitment to an egalitarian pay struc- ture, one that minimizes the pay differential between the higher and lower echelons of the public service. This minimization is normally achieved through a combination of deliberately suppressing the salaries of those at the top levels, usually to below the market value for their skills, and raising the salaries of the unskilled and semiskilled labor force to above the market wage. The study data suggest that among the study countries, Senegal’s ratio has remained the lowest: it fell from 3.8 in 1992 to 3.0 in 2000. Benin Kithinji Kiragu, Rwekaza Mukandala, and Denyse Morin 117 and Burkina Faso have also kept to a low (below 10) salary decompression ratio. Unskilled and semiskilled workers have been too strong politically to allow the decompression needed to match pay levels with skills and respon- sibilities. Not surprisingly, the top salary levels in these countries (which in 2000 were US$9,026, US$5,537, US$3,373, and US$3,180 for Benin, Burkina Faso, Ghana, and Senegal, respectively) are too low to keep professionals in the public service. Ghana closely follows this pattern with a decompression ratio of about 12 to 13 during 1992–2001 (figure 4.4). Zambia started rela- tively high at about 20 in 1992, but after a steep drop in 1993, it has remained at about 15. Even at this reasonably high decompression, however, salary levels at the top are comparatively very low. Uganda also had a low salary decompression ratio (about 5) in the first half of the 1990s. In 1996 and 1997, however, the salaries of top civil servants (permanent secretaries) were dramatically raised. As a result, by 1997 the salary decompression ratio in Uganda was about the same level as that of Botswana (30). Since then, it has gradually declined to about 25, which is still high and is surpassed only by that of Botswana among the study coun- tries (figure 4.4). However, a Cabinet decision in 2000 endorsed a target compression ratio of 20 for Uganda’s public service. Tanzania’s ratio was high, in the 20s, during the early 1990s, but since then has remained in the 15 to 20 range, which may be regarded as moderate. A unique macroeconomic and fiscal situation sets Botswana apart from the other study countries in terms of the levels, structure, and trends of pub- lic service pay. The country has maintained comparatively very high salaries for all levels of its public service throughout the 1990s and to the present. The pay structure decompression ratio, which has remained at above 30.0, is also high even by international comparisons. Furthermore, even though the country’s GDP per capita is much higher than that of any of the other study countries, the wage bill share of GDP also remained quite high throughout the 1990s, at about 10 percent. In sum, the comparative analysis of trends in pay levels, wage bill, and salary structure shows a steady deterioration in public service pay in the three francophone countries (Benin, Burkina Faso, and Senegal) during the 1990s. At the start of the decade, salary levels in these countries were much higher than they were in the other four study countries (except for Botswana), but by 2000 these four were at the bottom of the group. The con- sistently egalitarian character of the pay system in these countries, as indi- cated by a comparatively low salary decompression ratio of less than 10 is also noteworthy. While the need for long-term reform seems clear, given the political forces that have brought about this situation, the reform process will be a difficult one and the political stakes are likely to be high. 118 Reforming Pay Policy: Techniques, Sequencing, and Politics Among the anglophone countries, public services also started the 1990s with fairly high salary levels in Ghana and Zambia, especially when com- pared with Tanzania and Uganda. By 2000, however, the situation had reversed. Salary levels in Tanzania and Uganda had steadily risen (in U.S. dollar parity terms) over the decade but were sliding in Ghana and Zambia. This outcome is consistent with the observed gradual rise in the wage bill share of GDP in Tanzania and Uganda, whereas that share was generally declining in Ghana and Zambia. These trends suggest the possibility of more political tension in public service pay decisions in Ghana and Zambia than in Tanzania and Uganda. Tactics, Techniques, and Models We found a plethora of tactics and techniques in relation to decisionmaking about public service pay across the study countries and linked them to eight models of pay policy decisionmaking (table 4.1).4 An examination of the decisionmaking models reveals three dominant policy orientations: technically rational, politically rational, and politically reactive. Table 4.2 shows that four of the eight pay decision models fall roughly into the technically rational policy orientation: the performance based, the market-benchmarking, the job evaluation and regrading, and the wage bill and employment rationalization models. The emergence of tech- nically rational models of pay reform in the study countries is relatively recent. The introduction of the models can be associated with neoliberal economics on the back of SAPs and new public management. From the per- spective of concepts of modern public sector management that are in vogue, the technically rational orientation is ideal for the following reasons: • Pay levels are consistent with the productivity of employees because the adjustments are related to performance. • Government is able to recruit, retain, and motivate personnel with the skills it needs by matching market wages for these skills and accepting a decompression of the pay structure. • Pay awards are equitable in relation to the marketability of skills and the responsibilities of employees. 4. The configuration of public service pay decision models that is presented here is figurative and not definitive. Therefore the tactics and techniques are not entirely exclusive to each model. To our knowledge, such systematic identification, categorization, and consolidation of the tactics and techniques has not been attempted before in either the academic literature or in practice. Kithinji Kiragu, Rwekaza Mukandala, and Denyse Morin 119 Table 4.1. Models, Tactics, and Techniques Characteristics approaches models Characteristic techniques and tactics Corporatism • Mandate a statutory national incomes and employment body and task it to define policy targets and limits that bind both public and private sectors in pay adjustments. • Control collective bargaining agreements with a government-appointed salary review commission; trade unions affiliated to the ruling party. Salary indexation • Invoke mathematical and statistical basis for early fix- ing of future pay. • Negotiate and agree to future levels many years in advance. • Preempt future negotiated outcomes. • Ensure government and workers representatives nego- tiate indexes by which salaries for various grades will be adjusted over the years. • Assign task of automatic adjustment of salaries to the ministry of finance. Wage bill and • Define wage bill ceiling for fiscal stability. employment • Reduce and control employment to within wage bill rationalization ceiling. • Adjust salaries within the wage bill envelope. • Enhance transparency of pay system. • Attach a monetary value to in-kind benefits. • Consolidate allowances and attach a monetary value to benefits within a basic salary structure. • Eliminate distortions and nontransparent compensation (allowances). • Reduce external pressure to accord priority to fiscal sta- bilization. (This reduction is usually made as part of structural adjustment.) • Eliminate allowances and in-kind benefits not associated with facilitating specific organizational func- tions or operations. • Prepare government to absorb the additional wage bill and improve wage bill control. • Enhance fairness, equity, and efficiency of the salary structure and improve the post-employment compen- sation structure. table continues on following page 120 Reforming Pay Policy: Techniques, Sequencing, and Politics Table 4.1. (continued) Characteristics approaches models Characteristic techniques and tactics Pressure driven • Encourage powerful or influential groups to put patronage pressure on or threaten the government with industrial action or political consequences. • Allow collective bargaining to become out of control. • Allow government to yield to the pressures and threats usually by an award of allowance outside the salary structure. • Use allowances and in-kind benefits frequently. • Introduce new salary scales for select groups. Cost of living cum • Determine the minimum acceptable standard of living egalitarian for the public servant as the basis for establishing a tar- get minimum salary. • Accord priority in salary adjustment to achieving the min- imum living wage (if necessary, compress the structure). Job evaluation and • Perform comparative analysis and regrading of jobs as salary regrading specified. • Pursue both fairness and equity. • Encourage limited participation. • Use expert opinion. Market-benchmarking • Perform comparative salary survey across sectors. • Decompress salary structure in favor of the senior and skilled staff. • Establish a parallel progression salary structure for skilled professionals. Performance-based • Determine productivity measures. adjustments • Benchmark performance. • Use performance contracts to define employment terms. Source: Authors. • Public service employment and pay levels are managed to ensure a level of affordability of the wage bill that is consistent with fiscal dis- cipline and stability. • Pay structures are generally transparent. In the technically rational school, the perspective of public service issues is comparatively narrow. Analysts have a tendency to oversimplify the conse- quences of the decisions made, especially in terms of how they are experienced Kithinji Kiragu, Rwekaza Mukandala, and Denyse Morin 121 Table 4.2. Three Major Categories of Models, Policy Orientations, and Decisionmaking Models Pay decisionmaking Model category Main policy concerns models Technically rational • Productivity of the workers 1. Performance-based • Ability to recruit, retain, and 2. Market-benchmarking motivate skilled workers 3. Job evaluation and • Fairness of the compensation regrading system 4. Wage bill and employ- • Transparency of pay structure ment rationalization • Affordability of the wage bill Politically rational • As above but moderated by 5. Corporation political considerations 6. Salary indexation • In some countries, 7. Cost of living cum egalitarianism egalitarian Politically reactive • None that is conscious or 8. Pressure-driven coherent patronage • In some countries, regime survival Source: Authors. or perceived by various interest groups. An implicit assumption is that bad politics underlie the problems in public service pay. Therefore ana- lysts further assume that by depoliticizing the decisionmaking, more effi- cient and effective solutions to the problems will be realized. Such rationalist policies pushed Ghana into unsustainable public service pay policies in the late 1980s, only to slide back into deep crisis in the 1990s. The second dominant policy orientation among the pay decisionmaking models is the politically rational model. The policy orientation is political and rational because the government is ready to acknowledge the strength of other protagonists in public pay decisionmaking, especially the trade unions and other organized workers’ groups, and systematically factors political considerations into the decisionmaking processes. In other words, the government chooses to adopt less than ideal (technically rational) solu- tions in public service pay issues. The models that fall into this orientation include cost of living cum egalitarian, salary indexation, and corporatism. The third dominant policy orientation is also political, but it is reactive. This orientation corresponds to the pressure-driven model of public service pay orientation. As elaborated earlier, this model is associated with the absence of a coherent policy and strategy. 122 Reforming Pay Policy: Techniques, Sequencing, and Politics Table 4.3 maps out the study countries’ choice of pay models through five crudely defined political-economic epochs, from 1960 to 2000. The table highlights the three main categories described in the previous section, that is, technically rational, politically rational, and politically reactive. The results of the mapping show the following: • In the 1960s and first half of the 1970s, governments in all the eight study countries generally confined themselves to the use of politi- cally rational pay decisionmaking models. • In the following decade (1975–85), pay decisionmaking in Burkina Faso, Benin, Ghana, and Uganda degenerated to the politically reactive mode. • In the second half of the 1980s, while an SAP propelled Ghana to a technically rational framework, albeit temporarily, Zambia moved down to the politically reactive zone. • During the first half of the 1990s, Ghana not only abandoned techni- cal rationality, but also dipped into political reactive models. During this period, when Tanzania was liberalizing the economy but resist- ing SAPs, it joined Ghana and Zambia in the politically reactive zone. Table 4.3. Study Countries’ Choices of Pay Models through Five Political-Economy Epochs, 1960–2000 Epochs Model 1960–75 1975–85 1985–90 1990–95 1995–2000 Technically Ghana Botswana Botswana rational Uganda Tanzania Politically Burkina Faso Botswana Botswana Senegal Uganda rational Benin Senegal Senegal Burkina Faso Senegal Senegal Zambia Tanzania Benin Burkina Faso Zambia Tanzania Botswana Ghana Uganda Tanzania Politically Ghana Zambia Tanzania Benin reactive Burkina Faso Burkina Faso Zambia Ghana Benin Benin Ghana Zambia Uganda Uganda Note: The ordering of countries within each cell is based on the authors’ sense of the predominant pay policy model prevailing in each of the countries during the period indicated. The time periods (epochs) cannot be definitively established, hence the apparent overlap of years. Source: Authors. Kithinji Kiragu, Rwekaza Mukandala, and Denyse Morin 123 Meanwhile Uganda shifted to technical rationality, as evidenced by massive reductions in public service employment from 320,000 to about 150,000 and the consolidation of all allowances into a basic salary structure. During the same period, through a combination of job evaluation and market-benchmarking measures, Botswana con- solidated its perch in the technically rational zone. • In the second half of the 1990s, Benin joined Ghana and Zambia in the politically reactive model zone. Tanzania rapidly ascended into the technically rational zone, where Botswana remains. Uganda, how- ever, began to drift back into the politically rational zone, where Burkina Faso and Senegal remain. The mapping indicates that with the partial exceptions of Botswana, Tanzania, and Uganda, the sequencing of public service pay policy has gen- erally been haphazard rather than strategic. Since the early years of Botswana’s economic boom, its government adopted the policy objective to compete for skilled human resources at both national and international lev- els. Subsequent decisions that underlie the technical rationality of its public service pay decisionmaking model in the 1990s, however, were not, at least not explicitly, a result of a strategic commitment to a series of sequenced decisions. Therefore, the trends in Botswana depict consistency and not a long-term sequencing decision at some point. Some evidence indicates the use of sequencing in Uganda as it transited from politically reactive decisionmaking in the late 1970s and 1980s to a political rational model in the first half of the 1990s (table 4.3). The SAP- driven strategy that planned for halving personnel numbers was a prelude to the introduction of a transparent salary structure and a hefty raise in salaries, which took place between 1994 and 1996. Beyond that, the intent was that the pay structure would support results-oriented management, which implied a technically rational pay system. This strategy was not sus- tained, however, and the sequencing plan is no longer in place. Sequencing considerations have also been evident in the Tanzanian strategy for public service pay policy adopted by the Cabinet in January 1999, which explicitly stipulated the following: • In the short to medium term, market-benchmarking and the imple- mentation of job evaluation and salary regrading would be the main features of decisions on public service pay. • The adoption of a performance-related model would be a long-term goal pending improvements in public sector capacity and performance. • The sustained reduction, control, and rationalization of public service employment were necessary to achieve the long-term policy objectives. 124 Reforming Pay Policy: Techniques, Sequencing, and Politics Even in the Tanzanian case, however, the sequencing plans were never fully explicit and concrete. These three exceptions aside, the oscillation between models over the years (table 4.3) and the attendant instability and inconsistencies in pay lev- els and trends across the study countries (figure 4.1) underscore the dearth of strategic and sequencing perspectives. Governments have faced difficul- ties in consistently applying a stable policy framework to the issues sur- rounding public service pay determination. Policy objectives, as well as approaches, techniques, and tactics, have varied over the years and have often changed without any degree of consistency or strategic orientation. In several of the countries, and on many instances, decisions with far-reaching and enduring implications have been made to meet the political exigencies of the moment. In any case, as elaborated in the next section, sequencing outside a politically strategic framework may not be sustainable. Pay Reform and Politics The previous section outlined the comparative pay trends, techniques, tac- tics, and sequences adopted, as well as their successes and failures. This sec- tion explains them from the vantage point of politics. Simply stated, do politics matter? Have politics and related variables had any impact on pay reform? If so, how so? To understand the politics of pay reform in Africa note that, in essence, politics consists of the constraints under which political actors operate and the strategic maneuvering that occurs within those constraints. In calculat- ing the costs and benefits of available alternatives, political actors take into account, first, historically determined structural imperatives; second, opportunities and benefits; and third, the risks and constraints provided by the existing means of legitimation, administration, and coercion. Structural imperatives include the country’s wealth (or state of the economy) and social distribution; the nature and strength of power centers in the polity, that is, whether sociopolitical power resides in political parties, trade unions, the military, or the public bureaucracy; and the strength of the media. Opportunities and benefits may include election times and situa- tions of political disturbance. Depending on the nature of a particular regime, organized unions may see election times as an important structural opportunity for the government to make salary and wage concessions. Con- versely, political elites may believe that instituting severe anti-inflationary policies is expedient while elections are still a long way off. Risks and con- straints may include the erosion of political support and legitimacy, and Kithinji Kiragu, Rwekaza Mukandala, and Denyse Morin 125 ultimately the loss of political power through, for example, the loss of elec- tions, a military coup d’état, or the rise of popular movements. Politics is a complex game of strategic calculations. As Geddes (1994) points out, politicians often face the so-called politician’s dilemma in decid- ing among a set of policy preferences. According to Geddes (1994, pp. 132–33), all politicians, particularly presidents, share three main goals, namely, that they must survive in office, that they must govern effectively, and that they must build a loyal political organization. To achieve these goals they must sustain what Weber (1978) refers to as the means of administration and coercion (MAC). Mouzelis (1990, p. 51) elaborates these MAC as macro-institutional, organizational ensembles that, on the level of the whole polity, link the rulers and the ruled, for example, types of state administrations or military apparatuses, techniques of taxation, of national accounting, of mass mobilization . . . and forms of political incorporation. Mouzelis’s statement refers to the following: • What the state is—state institutions and organizations, armies, public service, and parastatal organizations • What the state does and its processes—generation and maintenance of legitimacy, administration, adjudication, tax collection, regulation, provision of critical infrastructure and social services, defense of its boundaries and sovereignty • How it does the above—for example, political parties, means of destruction, techniques of accounting, finances, skilled and experi- enced personnel, technologies and techniques, and knowledge. These means of legitimation, administration, and coercion can be divided into two parts in the context of our discussion: those that are critical to the functioning and continuing existence of the state and those that are important, but not critical, such as developmental activities. To borrow from Thompson (1967), the former activities make up the technical core of MAC. While ruling classes can easily give up on the lesser MAC, they will do everything possible to keep the core means of administration and coercion going. To sustain them they need adequate legitimacy and support, fiscal resources, human resources, and technical and organizational capability (Mukandala 1999). The sustenance of critical MAC is dynamically linked to the three regime goals of survival, effective governance, and political loyalty identi- fied by Geddes (1994). These goals conflict with one another, however, the 126 Reforming Pay Policy: Techniques, Sequencing, and Politics major trade-off being between technical considerations for effective perfor- mance in relation to partisan loyalty. In practical terms, this trade-off may mean a choice between a rational pay and employment model based on optimum choices within an overall financial ceiling, on the one hand, and a pay and employment model driven by political considerations of patronage and clientelism on the other. In most cases, “exigencies of electoral needs often overwhelm other commitments” (Geddes 1994, p. 133). Geddes fur- ther concludes that variations among politicians’ preferences depend largely on the characteristics of the party and the political system. Politi- cians in well-institutionalized parties with well-developed electoral machines and networks for delivering benefits are more likely to invest some resources in developing technical expertise to improve the party’s performance in office. Conversely, politicians from new, and weak, parties that have no effective distribution networks and electoral machines are more likely to invest resources in developing a loyal organizational base for political support. Furthermore, regimes with the domestic budgetary capacity to meet their needs are more likely to pursue independent policies and resist donor pressure. A regime’s bargaining position and capacity to resist donor pressure declines as its capacity to pay for its recurrent budget weakens. This decline in capacity may manifest itself in many ways, the most dramatic being the failure to meet the payroll and to pay for essential foreign imports because of a depletion of the country’s foreign exchange reserves. Within the context of the politician’s dilemma and African realities, development partners have been the force pushing for effective and efficient macroeconomic man- agement. Their increased influence translates into the ascendancy of merit and professionalism in settings where patronage and clientism had domi- nated (and may yet dominate again). Political and Related Variables in Pay Reform Pay reform, like many other public service reforms, is intertwined with the politics of resource mobilization and allocation (or reallocation), because it is mainly, although not exclusively, determined in the budgetary process where the question of who gets what, when, and how is settled. Many institutional actors, both strong and weak, participate in or are interested in the process of resource mobilization and allocation, and therefore the pattern of allocation cannot be determined rationally by a group of experts in defiance of the inter- ests of the main actors. The following are the main institutional actors: • Administrative actors, who include the president or prime minister, ministers, public service bureaucrats, and technocrats Kithinji Kiragu, Rwekaza Mukandala, and Denyse Morin 127 • Representative bodies, which include parliaments, regional assem- blies, and local government councils • Market institutions, which include the private sector or the business community • Civil society institutions, which include the political parties, trade unions, the media, and various civic organizations The donor community, which includes donor agencies and external con- sultants or experts. Figure 4.5 depicts the main actors in the politics of resource mobilization and allocation. All these actors may exert an influence on pay reform in a pos- itive or negative direction. For ease of reference, the main characteristics of these actors are summarized in table 4.4 for each of the eight study countries. ADMINISTRATIVE ACTORS. The strength and commitment of the president makes a major difference to the implementation of pay reform, but this ele- ment can be counterbalanced by other factors. Figure 4.5. The Main Actors in the Politics of Resource Mobilization and Allocation The Political System Administrative institutions • Chief executive The donor community • Ministers • Donor agencies • Public sector • Consultants Bureaucracies and technocracies Civil society institutions Politics of Market institutions resource • The ruling political mobilization and party • Private sector allocation • Political parties in The business opposition community (The budgetary • Trade unions process) • Media • Others Representative institutions • Parliament • Local government • Councils and assemblies Source: Authors. Table 4.4. Strength and Orientation of the Main Actors Institutional actors Benin Botswana Burkina Faso Ghana Senegal Tanzania Uganda Zambia Administrative Very effective Very powerful Strong personal Rawlings was Very powerful President is Very strong President actors checks on presidency; rule by the charismatic presidency; highly executive, per- politically • President executive from very powerful, president with and highly unmotivated committed to sonal rule; insecure; • Ministers other cohesive, shaky committed to bureaucracy reform and is bureaucracy bureaucracy • The Bureaucracy branches; well-paid, legitimacy and reform till the politically and ministers is socially bureaucracy is fairly efficient, political base mid-1990s; secure; the are fairly effec- mobilized and politicized and and loyal bureaucracy is bureaucracy is tive and dissatisfied ineffective in bureaucracy managerially fairly effective efficient reform imple- weak because and efficient mentation of loss of man- power and moonlighting; 128 corrupt but non-partisan legacy Representative Strong multi- Parliament is Weak, divided Strong multi- Strong multi- Parliament Parliament Strong multi- institutions party quiescent and among many party party dominated by weak, loyal to party • Parliament parliament, an uncritical, so parties, subor- parliament parliament one party; with movement and parliament; • Local arena of vested are local gov- dinate to presi- and district with control few exceptions, president; local increases own government interests; weak ernment coun- dent; parlia- assemblies swinging from rubber stamps councils active; pay and perks councils and local govern- cils and ment continues since 1993 one party to executive deci- increases own assemblies ment councils; assemblies to raise own pay the other sions; increases pay and perks government and is opposed own pay and lacks majority to 1998 public perks since 1998 servant– dominated reform Market Public servants n.a. Large informal Business com- n.a. n.a. n.a. n.a. institutions dominated sectors provide munity active • The business through an additional and has been community investments, economic invaded by joint ventures, avenue to low-paid pub- and the like senior public lic servants servants; 55% millionaires are civil servants Civil society Strong compet- Very strong rul- Too many Strong parties Strong compet- Strong ruling Movement has Competitive institutions itive parties ing party; weak political par- since 1993 with itive parties in party, weak monopoly of political par- • Political since 1990; opposition ties (over 60); fierce political the late 1990s; opposition political space ties; factional- parties in power strong unions, parties, unions, dominated by competition; fairly strong parties with allied with ized, strong and in critical media the media, president’s strong unions; unions though weak political army; political unions; critical opposition and nongov- party; unions critical media often co-opted; competition; parties operate media • Trade unions ernmental active, but vibrant media weak unions; underground; 129 • Media organizations divided and student fairly informa- fairly critical • Others movement tive and criti- media and cal media civil society organizations The donor Highly influ- Development Very active; Highly cooper- Highly influ- Cooperative in Development Highly community ential, heavily partners play pushed the ative until 1992; ential when early 1990s; partners ini- supportive, • Donor engaged but very sec- unpopular shuffling feet in their con- unhappy tially very sup- but disap- agencies could not get ondary role in 1998 Law on the 1990s, but ditions have 1994–95; since portive of pointed with • Consultants and government to economy and Public Admin- have admitted been met then highly country and implementa- experts move on the pay reform istration country to supportive reform policies, tion of reform 1998 reform because of Reform against Heavily currently grow- programs law strength of the wishes of Indebted Poor ing uneasy economy unions Countries sta- with trend of tus and accom- politics modative n.a. Not applicable. Source: Authors. 130 Reforming Pay Policy: Techniques, Sequencing, and Politics In Benin, for example, the executive faced strong checks from the other branches (especially the Parliament), and the bureaucracy was too politi- cized to be able to implement pay reform in a consistent manner. Thus after a promising start, reform slowed down on the critical issues of performance- based system of recruitment, promotion, and wage indexing. In Burkina Faso, President Blaise Compaore had a freer hand, but his shaky legitimacy and weak political base diminished his influence on the reform. The bureaucracy was both politicized and vulnerable to corruption. It was therefore a weak instrument of reform implementation. In Ghana, Jerry Rawlings was charismatic and highly committed to reform until the mid-1990s, when plural politics eroded the reform disci- pline that he had enforced and maintained for a decade. The Ghanaian bureaucracy was also managerially weak because of corruption and an inability to attract and retain highly qualified staff. In Tanzania, President William Mkapa has been highly committed to reform and is politically secure, but bureaucratic corruption and the lack of a strong push from the other actors (especially unions and the Parliament) have left the government free to award discretionary and selective pay increases. The story is the same for Uganda, where President Yoweri Museveni exercises personal rule and some ministers and a few in the bureaucracy have faced charges of corruption. Pay reform in Zambia under President Frederick Chiluba suffered from a combination of a relatively weak, politically insecure, and allegedly corrupt chief executive and a corrupt bureaucracy. Even though he had a majority in the Parliament, his party was characterized by in-fighting and expulsions. It also faced strong resistance from opposition parties and civil society. Botswana was in a class of its own on the administration variable. It had a fortuitous combination of a powerful presidency and a well-paid, fairly efficient, nonpartisan bureaucracy. Senegal has had powerful presidents who have implemented World Bank and International Monetary Fund–inspired SAP policies while com- bating trade unions and opposition parties. REPRESENTATIVE INSTITUTIONS. In three of the eight countries, representa- tive institutions were relatively active in the reform process as follows: • Benin, where the multiparty Parliament is active, but councils are rel- atively weak • Ghana, which has an active multiparty Parliament and district assemblies • Zambia, which has an active multiparty Parliament, but weak local councils. Kithinji Kiragu, Rwekaza Mukandala, and Denyse Morin 131 Representative institutions in the remaining five countries are relatively weak compared with the executive. In Botswana, the Botswana Democratic Party–dominated Parliament and the local councils are quiescent and uncriti- cal of government policy, actions, or inactions. In Burkina Faso, with more than 60 political parties, the many party interests represented weaken the Parliament. In Tanzania, the ruling party, which has more than 80 percent of the members of parliament, dominates not only the Parliament, but also most of the local government councils. In Uganda, most members of parliament are loyal to the movement and the president, as are the active, movement- based local councils. One party, the Parti Socialiste, for a long time dominated the Senegalese Parliament, which is currently dominated by the ruling coali- tion of the Parti Démocratique Sénégalais. A feature common to almost all the representative institutions is that they succeeded in raising the pay and perquisites of their own members during the past decade—in stark contrast to the government’s failure to effect substantial increases in civil service pay. They also passed a substantial increase in the salaries and allowances of min- isters and other members of the political elite during the decade. MARKET INSTITUTIONS. The study shows that market or private sector institutions also have an influence on pay reform. First, in some cases they poach human resources from the civil service by offering better pay. Second, in most of the study countries, a large informal sector provides an additional economic avenue to junior public servants. Moonlighting activities supplement meager salaries and governments tolerate these activ- ities for that reason. Experience has shown that junior civil servants tend to prefer a low-paying job in the civil service (considered more secure) to a bet- ter paying job in the private sector (considered less secure) as long as they are free to moonlight or work at secondary jobs. Third, in some countries, for example, Benin and Burkina Faso, senior public servants are major actors in the private sector in which they individ- ually or jointly invest incomes earned justly or corruptly. In Burkina Faso, for example, 55 percent of the country’s millionaires are civil servants. CIVIL SOCIETY INSTITUTIONS. Important civil society institutions that are engaged in the politics of resource allocation include political parties, trade unions, and the media. The eight countries fall into two categories on this variable. The first category has a balanced power situation, with a combination of strong opposition parties, trade unions, and media (Benin, Ghana, Senegal, and Zambia). 132 Reforming Pay Policy: Techniques, Sequencing, and Politics The second category includes countries that have strong ruling parties, but relatively weak opposition parties, unions, and media institutions (Botswana, Burkina Faso, Tanzania, and Uganda). In Tanzania and Uganda, media institutions have continued to become more informative and critical since the early 1990s, but media laws still favor state control, and even though civil society organizations have mushroomed, they have not joined hands with the new political parties. In Uganda, political parties were allowed to exist, but not to organize and mobilize support. Old party identi- ties (Ugandan People’s Congress, Democratic Party, and Kabaka Yekka), however, still linger on and have a substantial influence in electoral politics. THE DONOR COMMUNITY. In all eight countries, development partners have been the major force pushing for pay reform and related public service reforms. In particular, development partners have pushed politicians to adhere to proper macroeconomic management practices, including observ- ing proper ceilings for the wage bill, sometimes by threatening to withdraw aid in general or support for pay reform in particular. For example in Burk- ina Faso, the development partners succeeded in pushing the unpopular 1998 Law on Public Administration Reform against the unions’ wishes. As the case studies show, development partners have not succeeded in all cases. Politicians have complied with donor requirements or conditions only when their political survival is not threatened. In Ghana, for example, Rawlings complied closely during 1980–90, when he had complete control over social forces and the backing of the army. He failed to comply with donor requirements during 1990–2000, however, because his survival under political pluralism required the use of patronage on a wider scale. Donor fatigue and disappointment with the slow pace of pay reform implementation is now evident in some of the study countries. Table 4.5 summarizes progress on pay reform during 1990–2002. Political System and Stability Variables The systemic and political stability variables that are of interest to this study include the following: • Competitiveness of the political system • Institutionalization of political organization • Regime legitimacy • Regime ideology • Social basis of the regime. The variables are summarized country by country in tables 4.5 and 4.6. Table 4.5. Pay Reform, Ideology, and Political Organization Variable Benin Botswana Burkina Faso Ghana Senegal Tanzania Uganda Zambia Progress on pay Initial move- Continuous Initial move- Stalling; Government Initial move- Initial move- Stalling; reform 1990–2002 ment but pay increases ment but slow; increases have has been able ment stalling; ment, stalling, sporadic and stalling on crit- civil service been sporadic to contain discretionary characterized chaotic ical issue of pay structure and chaotic wage bill and and selective by selective increases performance highly com- control size of increases increases (merit) based pressed; pay public service; system of and promotion pay structure recruitment still not based highly com- and promotion on perfor- pressed; wage and wage mance and indexing indexing wage indexing Ideology Neoliberalism Neoliberalism Neoliberalism Neoliberalism Neoliberalism; Neoliberalism Neoliberalism Neoliberalism; SAP 1990; pre- free market since introduc- since introduc- previous ideol- but strong since SAP previous 133 vious ideology: ideology since tion of SAP; tion of SAP; ogy: socialism undercurrent 1997; previous ideology: socialism independence; previous ideol- previous ideol- (Negritude) of socialist ideology: humanism no SAP; previ- ogy: radical ogy: African equity persists Common ous ideology: African social- socialism of Man’s Charter Kagisano ism of Sankara Nkrumah Political Discontinuous, Continuous, Discontinuous, Discontinuous, Relatively sta- Continuous, Discontinuous, Relatively con- organization unstable stable; the Bot- unstable unstable ble; change of stable, Chama unstable tinuous but swana Democ- party in power Cha Mapin- unstable ratic Party in in 2000 duzi/Tangan- power since yika African independence National Union in power since independence Source: Authors. Table 4.6. Political System and Stability Variables Variable Benin Botswana Burkina Faso Ghana Senegal Tanzania Uganda Zambia Political System • Uncompetitive pluralism X X X X • Competitive pluralism X X X X Political stability • Unstable • Fairly stable X X X X X • Stable X X X Regime legitimacy • Low X • Medium X X 134 • High X X X X X Regime ideology • Socialist/leftist • Neoliberal X X X X X X X X Institutionalization • Low X • Medium X X X X • High X X X Social basis regime • Factional basis X X X X X X • National basis X X Source: Authors. Kithinji Kiragu, Rwekaza Mukandala, and Denyse Morin 135 COMPETITIVENESS OF THE POLITICAL SYSTEM. The nature of the political system is important in a study of pay reform, or of resource allocation gen- erally, because analysts normally assume or hypothesize that the more authoritarian the regime is, the more insulated it is from people’s demands, and conversely, the more democratic the regime is, the less insulated and therefore more responsive it is to people’s demands. However, the authoritarian-democratic model lacks precision in a transition situation. All eight countries are in a state of transition from authoritarian to democratic forms of governance, and thus each has elements of authoritarianism and democracy. We therefore prefer to look at the degree of competitiveness of the political system, focusing on the degree to which pressure from a coun- try’s configuration of interests constrains the freedom of action of the regime. On this variable the system may exhibit a monopolistic or mono- lithic structure, uncompetitive pluralism, or competitive pluralism. Some multiparty (institutional pluralist) systems in Africa have proved to be uncompetitive in practice. One of the countries, Uganda, has a unique, uncompetitive, pluralist political system. Political competition, for example, elections, takes place within the movement framework rather than between political parties. Three of the countries, Botswana, Burkina Faso, and Tanzania, have con- ventional uncompetitive pluralism, that is, multiparty systems dominated by long-established ruling parties and strong executives. Two of them, Botswana and Tanzania, also have relatively weak civil society organiza- tions, especially trade unions. The remaining four countries, Benin, Ghana, Senegal, and Zambia, have highly competitive pluralist systems that have vibrant trade unions, busi- ness organizations and other civil society organizations, media, and opposi- tion parties. To varying degrees they also have a good balance of power among the three branches of government. For example, cases of disagree- ment between the executive and the legislature have been referred to the judiciary on several occasions in Benin, the most notable being in 1994 fol- lowing a disagreement between the president and the national assembly on a percentage increase in civil service pay. The case went all the way to the Constitutional Court. The 1998 law on pay reform was also referred to the Constitutional Court. Similar tussles among the three branches have occurred in Ghana and Zambia as well. Moreover, all four countries had a change in the political party in power: twice in Benin and once each in Ghana, Senegal, and Zam- bia during the 1990s. Even though the Movement for Multi-Party Democ- racy retained power in Zambia following success in the 2001 elections, it 136 Reforming Pay Policy: Techniques, Sequencing, and Politics was led by a new person, Levy Mwanawasa, after Chiluba’s attempts to amend the constitution so that he could stand for a third term were soundly rejected. INSTITUTIONALIZATION OF THE POLITICAL ORGANIZATION. Political stabil- ity requires institutionalization of the political organization, including the adoption of an acceptable constitution and other political rules, behavior, and practices that help underpin organizational capabilities. Institutional- ization of the political organization ensures political continuity and orderly change. A politically stable country is not subject to traumatic political shocks, violent changes of government, secession, military interventions, popular insurrections, or civil war. The modern form of political organization is the political party around which politics has been organized for more than three centuries. Political parties have important political functions, such as representing, aggregat- ing, and articulating people’s interests; mobilizing support and inducing compliance; and facilitating political socialization and political recruitment. A strong and well-organized party is thus essential for strong and effective political leadership. Other types of organizations may serve as spring- boards for gaining political power, as in the case of social movements in Burkina Faso and Ghana. These other forms, however, tend to be temporary and eventually wither away. The country cases fall into two categories with regard to the institution- alization of the political organization variable (table 4.6). The first group consists of Botswana, Tanzania, and Senegal. These countries have enjoyed political stability since independence as evidenced by three facts. First, the dominance of one party in each country has brought continuity, namely, the Botswana Democratic Party; Chama cha Mapinduzi and its predecessor, the Tanganyika African National Union, in Tanzania; and the Senegalese Pro- gressive Union Party, which was in power until 2000, but is now in opposi- tion. Second, the presidential successions have been orderly. Third, the army has stayed in the barracks, accepting a role subordinate to that of the political class. Consequently, no serious political dissent or turmoil has erupted. In the case of Tanzania, the Zanzibar crisis, which arose out of dis- puted elections in 1995 and 2000, has been resolved amicably through inter- nally generated processes. The other group includes the remaining five countries, Benin, Burkina Faso, Ghana, Uganda, and Zambia. With the exception of Zambia, they have each undergone no fewer than three military coups d’état and have been subjected to military rule for substantial periods. All the countries, including Kithinji Kiragu, Rwekaza Mukandala, and Denyse Morin 137 Zambia, have had their share of political turmoil and assassinations and continue to experience an undercurrent of political stress and turbulence resulting from a variety of sources, including ethnoregional competition. Political organization in these countries has therefore been characterized by discontinuity and turmoil, and institutionalization of the parties and rules of operation has not yet reached maturity. In Uganda, for example, since 1987 political parties have not been allowed to organize and carry on normal political activity or contest each other for political power. The National Resistance Movement is a quasimili- tary organization based on the National Resistance Army. The movement, which is molded along the lines of a Leninist organization, is headed by President Museveni. It has sought to be all-inclusive while preserving the purity of its inner core, which consists of those who went to the bush to fight the guerrilla war alongside the president. REGIME LEGITIMACY. Regime legitimacy is a complex and relative matter, because it ultimately depends on the perceptions and judgments of the peo- ple. It can be assessed or indicated in three ways. First, in a historical sense, it is partly a function of the institutionalization of political organization, a process that puts in place rules of the political game acceptable to the major political actors and the general citizenry. Second, in a political and legal sense, regime legitimacy can be indicated by the public’s perception of whether the present government came to power in a free and fair election. Third, in a utilitarian sense, regime legitimacy can be gauged by how people judge the performance of the current government, especially in meeting their basic needs. Using an index of stability, Botswana, Senegal, and Tanzania theoreti- cally would be more likely to generate legitimate regimes than the other five countries, which are lower on the scale of stability and political institution- alization. Such an index cannot stand alone, however. Botswana’s good eco- nomic position, for example, would also enable the government to score higher marks on performance than the other countries. Table 4.6 assesses regime legitimacy in the eight countries on a low- medium-high ordinal scale based on an evaluation of the freeness and fair- ness of the most recent elections. The losing parties in virtually all the countries have criticized the elections, although not to the same extent. One country, Burkina Faso, is assessed as low on regime (that is, electoral) legiti- macy; three countries, Benin, Uganda, and Zambia, are assessed as medium; and four countries, Botswana, Ghana, Tanzania, Uganda, and Senegal, are assessed as high. Tanzania qualifies for the high score because the elections 138 Reforming Pay Policy: Techniques, Sequencing, and Politics of 2000, compared with those of 1995, were judged by all local monitors and international observers as being largely free and fair, discounting the elec- toral problems in Zanzibar, but it can also be assumed that the mwafaka (agreement) of October 2001 between the Chama Cha Mapinduzi and the Civic United Front had a legitimizing effect on the Amani Karume regime in Zanzibar. REGIME IDEOLOGY. Following independence in the 1960s and 1970s, African countries came up with home-grown ideological packages, such as Ujamaa (villagization) in Tanzania, Kagisano in Botswana, humanism in Zambia, African socialism in Ghana and Senegal (Negritude), the Common Man’s Charter in Uganda, and radical socialism (Sankaraism) in Burkina Faso. As table 4.6 shows, however, all the countries under study have embraced neoliberalism. Ideological substitution started in the late 1980s and early 1990s, when SAPs implementation started. All the countries have tried to open up their economies, privatize loss-making public enterprises, and implement macroeconomic reforms. Although the pay reforms have been conceived within the neoliberal ideology, which favors decompression and performance-based compensa- tion systems, vestiges of previous egalitarian ideologies also exert an influ- ence on pay policies and practices in some countries. A good example is Tanzania, where equity concerns still resonate in the ruling party (Chama Cha Mapinduzi) and the general population. The ruling party in Tanzania still claims to be a socialist party, and the word Ujamaa is still mentioned in the country’s constitution, although without programmatic content. The continued tendency to increase the wages of public servants in the lower grades may be partly attributed to the Ujamaa legacy. No politician is will- ing to be seen abandoning a concern for equity, fairness, and justice for all. This state of affairs may also explain the unwillingness of the Tanzanian authorities to decompress salary scales openly. The general unwillingness in most of the countries under study to put a monetary value on all allowances and benefits of senior civil servants may in part be a result of the fear of revealing existing inequalities, which might cause political problems. Botswana has been an exception in this. Espousing neoliberal economics since independence, the country accepts steep inequalities in wealth, an acceptance that is reflected in the sharp salary decompression ratios. Most other countries have not followed this example. In some cases, politicians pay lip service to previous ideologies for politi- cal reasons. Burkina Faso represents such a case. In the late 1980s, the mass mobilization institutions of the revolution were abandoned following the Kithinji Kiragu, Rwekaza Mukandala, and Denyse Morin 139 decline of socialism internationally. In 1989, the government issued a first draft of its overall policy framework, which was based on a strategy of state capitalism under which the state continued to claim that it was pursuing socialism, when in reality it was moving toward capitalism. The govern- ment adopted a pragmatic attitude toward the middle class, business lead- ers, and the donor community. Compaore had seen the need for economic liberalism, but was not in a hurry to abandon his perceived stance as a revo- lutionary. After all, he had removed Thomas Sankara, accusing him of betraying the revolution by becoming a mere reformist. In practice, how- ever, the country operates a liberal economy, as is the case in all the other study countries. In Benin, President Mathieu Kerekou accepted the demise of socialism in 1990, but he campaigned on the platform of limiting privatization during the 2001 presidential elections, which he won following former Prime Min- ister Nicephore Soglo’s boycott of the second round. Soglo, a former World Bank official, despite sharing a neoliberal ideology, could not sustain that ideology as a politician. He received much praise in the 1990s, when he was prime minister, for managing to pay the country’s 47,000 civil servants regu- larly (even paying some salary arrears) and for convicting some leaders who had been involved in financial scandals. However, when he tried to implement fundamental reforms with austerity measures, he met with stiff opposition from the unions and members of parliament and backed off. Political leaders, especially insecure ones, are wary of implementing unpop- ular reforms. THE SOCIAL BASIS OF THE REGIME. In socioeconomic terms, all the regimes are ruled by what we may call a compradorial middle class of bureaucrats, the intelligentsia, small business owners, and small property owners who are highly dependent on external development partners and investors. The regime is a coalition of divergent ethnic and regional interests; many resources are channeled to patronage activities deemed necessary to keep the coalition intact. All the study countries, with the possible exception of Tanza- nia, have sharp ethnoregional competition. A north, central, and south divide exists in Benin; a north-south divide in Burkina Faso; a north, central, and coastal divide in Ghana; a west, south-central, north, and east divide in Uganda; and a north-central and south divide in Zambia. A rebellion in the Casamance region continues to this day in Senegal. Balancing these forces and tendencies produces tension and stress in the political system, and affects the pattern of resource allocation. Under normal circumstances, main- taining a coalition may be more important than pay reform. 140 Reforming Pay Policy: Techniques, Sequencing, and Politics Trade Unions Implementation by the government of pay reform partly depends on the strength of the trade unions and whether or not they are inclined to support or oppose it. Table 4.7 summarizes the strength of trade unions, the media, and opposition parties in the countries under study. Trade unions are extremely strong in three of the eight countries—Benin, Burkina Faso, and Zambia—where their influence on pay reform has been substantial. They are fairly strong in Ghana and Senegal but relatively weak in Botswana, Tanza- nia, and Uganda, where their influence on pay reform has been minimal. In Benin, the organization of the entire civil service into trade unions works against civil service reforms in general and pay reform in particular. The result is a vicious cycle in which the more senior civil servants believe they are underpaid, but cannot accept the introduction of a performance- based system, which would weed out poorly performing workers. Benin’s public service is seen as unified, because one salary or allowance must be given to all. The limited decompression of salary scales and the automatic increases seem to have created a unity of purpose among civil servants that some senior officials in Benin’s public service ministry believe will under- mine the whole civil service in the long run. In Burkina Faso, trade unions had been weakened by the Sankara regime. Then the Compaore regime sought to neutralize them through a combination of tactics, leading to their polarization, with one group inclined to dialog with the government and the other inclined to engage in revolutionary activities, including strikes. The power and militancy of trade unions in Burkina Faso increased following a series of assassinations of political opponents. The most radical of the trade unions is the General Confederation of Labor, which has been in the forefront in denouncing assassinations, corruption, police brutality, and hegemony of the ruling party. The power and unity of trade unions were demonstrated in the 1999 general strike, which demanded and obtained salary increases. The military took advantage of this situation to demand and obtain salary arrears. In Zambia, organized labor has been the main challenger of the state since colonial times because of the nature of the colonial (mining) economy, which led to the existence of a large working class. Urban dwellers accounted for 42 percent of the total population by 1990. In an attempt to control the activities of the labor movement, the government engineered the formation of a central body in 1965, namely, the Zambia Congress of Trade Unions, which in 1974 elected Frederick Chiluba as its president. Attempts to co-opt or subordinate trade unions to the United National Independence Party were resisted successfully by the Zambia Congress of Trade Unions Table 4.7. Civil Society Variables: Strength of Civil Society Generally, and Trade Unions, Media, and Opposition Parties Specifically Variable Benin Botswana Burkina Faso Ghana Senegal Tanzania Uganda Zambia Civil society generally • Weak X X X • Fairly strong X X X X • Strong X Trade unions • Weak X X X • Fairly strong X X 141 • Strong X X X Media • Controlled/not free • Fairly free X X X X • Free X X X X Political parties in opposition • Weak X X X X • Fairly strong X X X X • Strong Source: Authors. 142 Reforming Pay Policy: Techniques, Sequencing, and Politics leadership, which preferred autonomy. Thus union chairman Chiluba and secretary-general Newsteed Zimba refused repeated offers to join the Cen- tral Committee of the United National Independence Party. This autonomy has enabled the unions in Zambia to get what they want from the govern- ment, as evidenced by the following: • In 1981, the unions forced the government to release union leaders from jail, including Frederick Chiluba and 15 others. • In 1985, the unions forced the government to ignore an agreement with the International Monetary Fund enforcing a 10 percent wage rise ceiling. Instead, unionized workers ended up with an 18 percent wage raise. • In 1986, violent demonstrations against Kenneth Kaunda’s so-called Radical Reform Program, which left 15 people dead, forced the gov- ernment to cancel the program. To crown it all, the political reform that ended the single party system in 1990 was largely championed by the labor movement, which subsequently assisted the Movement for Multi-Party Democracy led by Chiluba to oust Kenneth Kaunda’s United National Independence Party in the 1991 multi- party elections. Conclusion How do the three distinctive policy responses to Gedde’s (1994) politicians dilemma—technically rational, politically rational, politically reactive—cor- relate with the political characteristics detailed in the previous section? Two variables, competitiveness and institutionalization, emerge as key. Figure 4.6 and table 4.8 map the eight countries according to how they fall on a competitiveness-institutionalization spectrum. Competitive pluralist cases—with high political competitiveness, but low institutionalization—exist in Benin, Ghana, and Zambia. These systems are multiparty, have vibrant opposition parties, and have had a change in the party in power. Politics in all four cases is intense; trade unions are either strong or fairly strong; and the media are vibrant, critical, and unafraid to tell the truth. Society is highly politicized, alert to political issues, and partisan. This mix at times gives rise to a situation verging on instability. This case is unlike contexts of strong democracy, where both institutionalization and competition are high. Pay reform in these countries, as shown in table 4.5, either has not taken off and the process is chaotic and slow, or is in reverse as in Ghana. Senegal, although currently a competitive pluralist case, was for a long time an uncompetitive pluralist case with the Kithinji Kiragu, Rwekaza Mukandala, and Denyse Morin 143 Figure 4.6. Institutionalization—Competition, 1995–2000 High Technically rational— Botswana, Tanzania Politically rational— Institutionalization Burkina Faso, Senegal, Uganda Politically reactive— Benin, Ghana, Zambia Low Low High Competitiveness Source: Authors. Table 4.8. Political Institutionalization and Political Competition Competitiveness Low Medium High High Uncompetitive Mature democracy in Institutionalization pluralism in Senegal Botswana and Tanzania Medium Uncompetitive Weak democracy pluralism in competitive pluralism Burkina Faso in Benin, Ghana, and and Uganda Zambia Low Source: Authors. Parti Socialiste in charge. It has thus enjoyed political stability, legitimacy, and high levels of institutionalization. Lately it has experienced intense political competition. Viewed through the lens of its pay policy, Senegal resembles Benin, Ghana, and Zambia, but unlike these countries it enjoys a legacy of stability, 144 Reforming Pay Policy: Techniques, Sequencing, and Politics high legitimacy, and institutionalization. It thus seems well placed to initi- ate and sustain a process of pay reform closer to those of Botswana and Tanzania. The four noncompetitive pluralist cases fall into two subgroups. One subgroup consists of Botswana and Tanzania. These two countries have enjoyed political stability and continuity in the dominance of the ruling party. The change of presidents has been orderly, and the opposition parties, civil society organizations, and the media are all weak. These countries enjoy high institutionalization with medium competition as shown in table 4.8. Reform has been systematic, driven by the executive, and under little pressure from popular forces. Thus our finding of evidence of consistency in the use of techniques and tactics in public service pay decisions in these two countries is not surprising. The remaining two cases, Burkina Faso and Uganda, belong to the second subgroup of uncompetitive pluralism. Even though both countries hold elections, they are dominated by the all-inclu- sive movement in Uganda and the Popular Front and its allies in Burkina Faso. Civil society organizations, trade unions, and the media are given lim- ited space. But, by contrast with Botswana and Tanzania, the level of institu- tionalization is medium to low. Pay reform has been on and off, responding to political pressure from the larger political body rather than from unions and other organizations. As figure 4.6 suggests, which approach to adopt to pay reform—technically rational, politically rational, or politically reactive—depends on where a coun- try falls on the institutionalization-competition spectrum. Technically optimal options depend on having in place the right mix of a low competitiveness and a high institutionalization environment. Similarly, political reactivity is most likely when competitiveness is high, but institutionalization is low. The figure points to the following questions as especially pertinent in designing a strat- egy for pay reform: • Given where a country is currently located on the institutionalization versus competitiveness scale, what policy options are feasible and which of the feasible options is most attractive? • What are the available options that will strengthen institutionaliza- tion without reducing competitiveness, and thereby make available a more attractive set of feasible options? • To what extent can skillful political leadership and coalition building rotate the rays in figure 4.6 in a southeasterly direction, thereby, expanding the space within which technically rational options are fea- sible and reducing the space or area in which countries are trapped with only politically reactive options? Kithinji Kiragu, Rwekaza Mukandala, and Denyse Morin 145 The Way Forward This study has brought to the fore the need to rethink the technocratic orienta- tion to pay reform that, in some of the countries we studied, has turned out to be what Wildavsky (1979, p. 123) once called “deceptive, self-deceptive, and inattentive to the ‘world of action’.” Countries therefore need an approach that self-consciously seeks creatively to understand and appreciate the politi- cal context and political variables that are inextricably critical to the success of pay reform programs. Any country that initiates pay reform needs to understand and appreci- ate the nature of the political system, its social structure, and the nature and strength of the power centers. Countries must assess the political bases of the political regime to determine if it has a well-established political organi- zation or a weak and fractured one. Countries must also determine if the regime is based on the military, trade unions, populist forces, or a combina- tion of any or all of these. Analysis must gauge the levels of political legiti- macy and political stability of the regime. The extent to which regime change has been anticipated, planned, and orderly—or traumatic and even chaotic—also needs to be taken into account. The nature, role, and strength of the various actors also deserve atten- tion. What are the power centers? What is the role of political parties? Are they equally competitive, or is one party a giant and the rest dwarfs, or are political parties outlawed altogether? What are the nature, role, and strength of trade unions? Are they powerful or subordinate and weak? What are the objective and subjective positions of the bureaucracy in the country’s political economy? Is it a power center? Is it formally insulated from politics? Are civil servants expected to be politically nonpartisan? The purpose of finding answers to these and other questions would be to deter- mine objectively (a) the possibilities of pay reform, the nature and types of models, and the strategies and tactics that may work; (b) the length of time for different kinds of support or resistance to emerge; and (c) the possible allies or foes of the policy. Consequently, one might be able to determine where change is possible and where it is not and plan measures to help the country move toward a technically rational model. Our country cases fit into a continuum, with Botswana and Tanzania at one end; Benin, Ghana, and Zambia at the other end; and Burkina Faso, Sene- gal, and Uganda somewhere in between. As summarized in table 4.8, in Botswana and Tanzania institutionalization is high and political competition is moderate; politics has to a large extent been tamed. The executive, anchored by a strong, well-established party organization drives public policy. The other possible power centers, including political parties, trade unions, civil 146 Reforming Pay Policy: Techniques, Sequencing, and Politics society, parliament, and other representative institutions, are weak and subor- dinated. Regime legitimacy, partly rooted in periodic elections, is high, mak- ing it extremely difficult for other forces in society, including the military, which is well incorporated into the regime, to pose any credible threat. Conse- quently, both countries have implemented quite ambitious pay reform poli- cies with little stress. The two regimes have a leeway and freedom of action that their colleagues in the other case countries can only dream about. How- ever, these tamed politics and the resultant absence of political stress in the system may not go on forever. Policymakers must therefore be aware of the needs or imperatives for long-term institutional sustainability even if forces emerge in the political system that may change the status quo. In this context one needs to be mindful of what will be politically acceptable. Benin, Ghana, and Zambia present a counter scenario of higher competi- tiveness but lower institutionalization. These political systems are in perpet- ual latent crisis. Any bold policy initiative is bound to cause stress in the system as forces mobilize around it to support or oppose it. Inertia and grid- lock are more often than not inevitable in such situations. In such cases, pol- icy initiatives must carefully separate the possible from the desirable; widespread consultation will be needed to arrive at an option that combines both and sells it to a critical mass of stakeholders. Planning pay reforms may mean that one needs to spend time on information, education, and communication activities and on sustained dialog about the issues with political leaders and other stakeholders to increase awareness and under- standing of the need for the reforms. Country cases in this category may thereby move closer to the Botswana-Tanzania end of the continuum if a certain level of stability and a consensus start to emerge. Viewed through the lens of its pay policy, Senegal resembles Benin, Ghana, and Zambia, but unlike these countries it enjoys a legacy of stability, high legitimacy, and institutionalization. It thus seems well placed to initiate and sustain a process of pay reform closer to that in Botswana and Tanzania. Burkina Faso and Uganda present interesting cases of how pay reform can proceed in settings of medium to low institutionalization during a period when competitiveness is also low. Both emerged from crisis situa- tions, which in Uganda involved civil war and the forceful take over of gov- ernment and in Burkina Faso involved a coup d’état and political assassination. Burkina Faso also experienced a period of revolutionary mass mobilization that raised levels of political awareness and partisanship among the population. In the early 1990s, Uganda gradually moved toward an all-inclusive model of politics anchored by the movement. The regime enjoyed a political honeymoon because of several factors. First, the population, especially in Kithinji Kiragu, Rwekaza Mukandala, and Denyse Morin 147 the south, the west, and to a lesser extent the east of the country, was relieved because of the relative peace, and even the economic prosperity that followed. Second, several initiatives taken in the 1990s pointed toward a democratic dispensation. The constitutional commission, the constituent assembly, and presidential and parliamentary elections were all seen as ini- tial tentative steps toward democratization. Third and most important, while alternative political forces were immobilized, exiled (into Sudan, for example), and generally took time to regroup and gather the courage to challenge the supreme leader both within the movement and in the larger society, Uganda enjoyed a lull in tensions. Uganda’s bold initiatives regard- ing pay reform were taken during this honeymoon period. The Ugandan honeymoon ended as a result of the realization that the movement, and even President Museveni himself, were not about to allow others the political space to organize politically; the continued, and even increased, wars in the north and east of the country; the adventure into the Democratic Republic of Congo by Ugandan forces; and a certain perceptible tiredness by the population. These developments have clearly stressed the political system. One consequence is a shift to a less technocratic approach to policymaking, though moderated by the regime’s ongoing willingness to stick to structural adjustment-backed policies. The pay policy area reflects this development with incidents of politically driven awards of allowances or special salary adjustments to select groups, such as the judiciary, teach- ers, and health workers. Burkina Faso is also a case that is neither in crisis nor stress free. Although the system is uncompetitive and the executive dominates all facts of life, a strong current of opposition objects to what it perceives as too much impunity, for example, the Too Much Is Too Much movement of 1998–99. The 1998 political assassination of journalist Norbert Zongo can be understood only in this context. Thus even though the president exercises personal rule, beyond certain limits he could face serious problems. These limits are part of the reason why pay reform has stalled or slowed down. In sum, the political reality that prevails in a particular country at a par- ticular time should guide the choice of the strategy and models to imple- ment public service pay reform. In relatively stress-free systems, as in Botswana and Tanzania, considerable scope exists to introduce technically rational models, as long as the top political leadership is comfortable with or convinced of the efficacy of the model. That appears to be the case with, for example, the selective, accelerated, salary enhancement scheme in Tan- zania. But even in these relatively stress-free systems, technocrats seeking to move ahead with technically rational pay reform models must still be aware of what will be politically acceptable and develop their abilities to elaborate 148 Reforming Pay Policy: Techniques, Sequencing, and Politics balanced strategies that are not only technically rational, but also factor in their political environment, and so are more likely to get support from the political leadership. Among countries with lower levels of institutionalization—where we place most of the countries in our sample, indeed, most of the countries across Africa—the circumstances under which technically rational approaches are likely to be feasible are few. Most of these are likely to be time-bound win- dows of opportunity, for example, a reformist political leader coming to power in an environment of medium-low institutionalization and low com- petitiveness. In this latter setting, using technically rational models may be feasible in the short term, but changes brought about in this way run a high risk of reversal. This risk is exemplified by Ghana’s reversal in the 1990s of public service reforms that had been made in the 1980s and early 1990s. If a political leader temporarily has space to be technically rational, the crucial challenge is to sustain this space in the face of inevitably rising politi- cal competition and the risk of progressive de-institutionalization. This need points to the adoption of pay policies that nurture consensus and build institutions and governing coalitions rather than those that risk being polar- izing. Tanzania represents a case in which institutionalization progressively deepened over the course of the 1990s, with a corresponding expansion in the scope for more technically rational reforms. Viewed from the perspec- tive of the design of pay reform, the key has been technocrats’ readiness to proceed in a politically sensitive way, focusing on strategies that strengthen institutions, thereby making feasible a progressive expansion over time of the technical ambition and quality of the pay options. References Geddes, B. 1994. Politician’s Dilemma: Building State Capacity in Latin America. Berke- ley: University of California Press. Mouzelis, N. 1990. Post-Marxist Alternatives: The Construction of Social Orders. Lon- don: Macmillan Press. Mukandala, R. 1999. “From Proud Defiance to Beggary: A Recipient’s Tale.” In Goran Hyden and Rwekaza Mukandala, eds., Agencies in Foreign Aid: Company China, Sweden, and the United States in Tanzania. London: Macmillan Press. Thompson, J. D. 1967. Organizations in Action. New York: McGraw Hill. Weber, M. 1978. Economy and Society, vol. II. In G. Roth and C. Wright, eds., Economy and Society. Berkeley: University of California Press. Wildavsky, A. 1979. Speaking Truth to Power. Boston: Little, Brown. 5 Cabinets, Budgets, and Poverty: Political Commitment to Poverty Reduction Harry Garnett and William Plowden Much effort in recent years has been put into the preparation of poverty reduction strategy papers (PRSPs). As the foundation of poverty reduction strategy credits and the Heavily Indebted Poor Country debt relief pro- grams, much hangs on PRSPs, which generally have been participatory and raise expectations. The extent to which the policies and programs contained in the PRSPs will be implemented, however, depends on how the PRSP process relates to the ongoing policymaking processes of government. Bud- get preparation is a key policymaking process, to which PRSPs must be directly linked if their objectives are to be implemented. The apex of the pol- icymaking process is the cabinet. PRSPs, for the most part, however, rarely dwell on the policymaking processes of government and only a few make reference to the cabinet. This chapter examines the role of cabinets in the budgetary process. It asks whether this role enhances the making of coherent and sustainable policies, including that key set of policies aimed at reducing poverty. The chapter begins by reviewing a normative framework about why cabinets should play a central role in the budgetary process. Using this framework as a lens, it looks in more detail at the experience of budgeting in Africa, including an in-depth examination of the process in four countries selected as case studies. 149 150 Cabinets, Budgets, and Poverty: Political Commitment to Poverty Reduction Cabinets and the Budget Process The Hypothesis This study tests the following hypothesis: If the cabinet is involved at key strategic points from the beginning to the end of the budgetary process, from setting strategic priorities to approving the detailed budget prior to submission to parlia- ment, then the execution of the budget will be more predictable and policy is more likely to be implemented. What do we mean by this? By “cabinet” we could use the definition in Evans and Manning (2003). They argue that policy is more predictable if gov- ernment is collegial. They distinguish between collegial and hierarchical exec- utives. The three key characteristics of collegial executives are the following: • They comprise a compact group of 5 to 40 senior decisionmakers who, usually, have known each other and worked together for some time. • All major policy matters go to this group for final ratification. • The members of this group publicly support all decisions that have been collectively ratified. Under hierarchical executives, the leader may frequently make decisions alone, and senior policymakers are accountable for the performance of their agencies alone. A survey of 182 countries concluded that two thirds of gov- ernments use a somewhat collegial model. Alternatively we could use the simple definition of cabinet: the group of ministers appointed by the president or prime minister who work together to make major decisions and to coordinate policy. Either way, we are refer- ring to the political leadership, although the Evans and Manning (2003) def- inition is more flexible. Whichever is taken as the definition, the point is that the medium-term expenditure framework (MTEF) and the annual budget are major policy statements and that this group, according to the hypothe- sis, should be involved through the budgetary process. By “budgetary process” we mean the activities required to define the priorities of government and express them in terms of capital and recurrent expenditure. Very often these days the budget takes the form of an MTEF, typically a rolling three-year expenditure plan, constrained by the expected availability of tax revenue and other resources (for example, project and program funding from donors). Taking a longer view makes adjusting expenditure to bring about major policy changes easier. For example, right- sizing programs take some years to implement. Typically, a one-year hori- zon allows little room to maneuver; many expenses are fixed over that period (for example, the wage bill and debt). We could add to this definition Harry Garnett and William Plowden 151 a process by which the implementation of the budget is reviewed and include in the hypothesis the need for the cabinet to be active in reviewing budget outturns as a reflection of policy implementation. Knowing that the cabinet, that is, the political leadership, will review budgetary performance might improve the predictability of the budget and policy. The “priorities” referred to in this definition of budgetary process are the government’s interpretations of the will of the people. They might have been expressed as a manifesto taken to the people during an election or as a national plan. These days governments’ priorities are strongly influenced by, and publicly stated in, the PRSP. Governments treat PRSPs seriously, partly because their preparation and implementation are tied to funding from their cooperating partners’ through debt relief and budget support. Our hypothesis is not easy to test. Few positive cases exist—countries that have predictable budgets and successful policy implementation. Even in those cases, attributing success solely or mainly to cabinet involvement would be difficult. The same attribution problem applies to the mass of the cases in which budgetary outcome bears little relation to the allocations approved by parliaments. Carrying out econometric studies to find the root causes of poor policy implementation or budget execution, or at least that part of the root causes related to cabinet involvement, would be very difficult. The Rationale Why do we hypothesize that the participation of cabinets in the budgetary process is desirable and indeed necessary—and why do we hypothesize that such participation increases the predictability of budgets and budget compliance on the part of some of the most important players? What are some of the main ways that cabinets take part in different countries? The authority of a budget, as a framework within which policy decisions are made, and as a constraint on inappropriate behavior, is based in two quite distinct dimensions: politics and rules. Budgeting, which involves dis- tributing scarce resources in the face of intense competition for resources, is a highly political activity. The reasons for budget indiscipline are simply expressed in the familiar notion of the tragedy of the commons—the fact that the benefits of nonconformity to the nonconforming individual (in this context a spending minister) greatly exceed his or her share of the costs. Acceptance of the budget as having some binding force requires political will, but political will alone is not sufficient. For an orderly process, political will must be channeled by institutions—within the executive and between the executive and the legislature. “Rules and routines,” write the authors of a valuable comparative study of budgeting in Australia and Canada, “shape 152 Cabinets, Budgets, and Poverty: Political Commitment to Poverty Reduction the capacity of ministers to translate political will into action and achieve greater fiscal discipline” (Kelly and Wanna 1999, p. 143). The subject of this inquiry, then, is the processes for agreeing on priorities and policies that are most likely to generate coherent and internally consistent priorities, and policies that will be fully implemented. One theme finds wide support in the recent literature about budgeting and the executive. This theme is that a major cause of budget indiscipline and of consequent budget deficits is fragmentation of spending authority among too many individuals and units. The solution proposed is to mini- mize the number of interests allowed to make autonomous spending deci- sions, by centralization. This term, somewhat confusingly, covers two quite different alternative arrangements. In the first arrangement, total spending is controlled by delegating authority to a powerful individual, which in practice means the minister of finance and the ministry. The minister is thought to be more likely to make disinterested spending decisions than his or her colleagues who head spending departments and are concerned for the welfare of special interests. In the second arrangement, spending may be controlled through a contract among the leaders of the government. Partici- pants start the budget process by agreeing on key parameters, typically the spending limits for individual departments. The theory is that the process of negotiation itself makes those involved understand the marginal costs and benefits of each other’s programs and thus the externalities of excessive spending by any of them. The choice of methods should depend upon the type of government. Single-party governments that have strong internal discipline can choose the former; coalitions in which member parties may be reluctant to give any party representative such authority may choose the latter (von Hagen, Hallet, and Strauch 2002). The basic arrangement must be appropriate to the politics of the local sit- uation, but the detailed procedures, too, must be acceptable politically. If members of the cabinet do not, explicitly or implicitly, assent to the processes that determine their own budget shares, they are less likely to accept those shares if they disagree with the outcomes. If they do not agree that ultimate authority for the budget has been delegated to the minister of finance, they are not likely to find his or her dispensations authoritative. As Schick (2001) has pointed out, the central budget office cannot impose or enforce fiscal norms by fiat: “a fiscal rule that truly constrains government spending or the deficit must be accepted by the government and strong enforcement mechanisms must be in place” (p. 15). Once the contract starts to break down, or the authority of the coordi- nating body (generally the minister of finance, but sometimes the head of Harry Garnett and William Plowden 153 government) is challenged, the whole system can unravel irreversibly. For a spending minister, no incentive to ignore budget constraints is greater than the sight of colleagues doing just that. “If ministers do not perceive the rules of the collective decisionmaking process to be credible and enforceable because the head of government grants priority access to favorite or power- ful individual ministers, or if the finance minister routinely makes side deals, with the effect of pre-cooking, end-running or even overturning the council of ministers’ decisions, then other ministers will have only weak incentives to participate meaningfully in, adhere to, and publicly support collective decisions” (Evans and Manning 2003, p. 27). In some countries, the minister of finance has no special status. In others, he or she is second only to the prime minister or president in the cabinet peck- ing order. The theme of this chapter, however, and the hypothesis underlying the study behind it, is that whatever specific tasks are delegated to the minis- ter of finance, his or her general authority must be reinforced by maintaining consensus within the cabinet. The more that cabinet itself has approved the budget, the more effective the budget will be as an instrument of control. In some countries such approval is secured by submitting the final bud- get, in all its extensive detail, to the cabinet shortly before the budget is pre- sented to the legislature, but such approval is wholly formalistic. If this perfunctory step is the cabinet’s only involvement in the budgetary process, it likely has no significance at all, given the virtual impossibility of making any substantial changes in the budget at such a late stage. The activity is little more than a token gesture toward the principle of the cabi- net’s collective responsibility. What we propose here is that ongoing involvement and consensus-building among cabinet members—both much earlier in budget formulation and at several stages of the ensuing process—is key to both fiscal sustainability and sustainable pursuit of a coherent, prioritized set of policies. Appendix 5A gives some examples for developed countries of how the process works. The remainder of this chap- ter examines the budgeting process in African countries. Budgetmaking in Africa—An Overview Taken together, the approach laid out in the first section suggests that, if it is to wholly achieve its goals, a PRSP process needs to meet three conditions: • The PRSP must be costed and linked to a country’s national budget. • The annual budget must be anchored in MTEFs. • The budgetmaking process needs some consensual cabinet-style decisionmaking. 154 Cabinets, Budgets, and Poverty: Political Commitment to Poverty Reduction Four Country Cases The four country cases in our study span the spectrum of budgetmaking performance. We begin with South Africa, which is not included in table 5.1 but which is often used recently as a model in this arena among developing countries. The second case, Zambia, scores reasonably well in table 5.1, but looking beyond the scores, it turns out to have serious shortcomings in its process. Case number three, Uganda, emerges among the top performers in table 5.1, and Mozambique, the fourth case, is a poor performer. Each coun- try case provides some historical background and recent performance mea- sures, describes the budgetmaking process as it currently operates, and highlights some continuing challenges. South Africa South Africa has one of the model MTEF processes, with heavy involvement by the Cabinet and by other key stakeholders throughout the process. One result is that the budget, and therefore policy, is highly predictable. BACKGROUND AND PERFORMANCE. The post-apartheid regime in South Africa consistently has maintained strong budget discipline and predictabil- ity. Within the context of budgetary discipline, budget prioritization has progressively become more a inclusive process that continues to evolve. The government that came to power in 1994 inherited a highly central- ized system, including a Westminster approach to budgeting, that is, a process whereby the budget is shaped by the executive and is rarely modi- fied by the legislature. The continued dominance of the National Treasury, and specifically of officials there, was consistent with the determination of the new regime to demonstrate its fiscal responsibility. The budget was, for the most part, put together by the bureaucracy. The minister of finance pre- sented his or her budget speech for the information of his or her Cabinet col- leagues with little previous input from them. Still struggling to master their new portfolios, they were disinclined to challenge this short notice. Even the president saw the budget speech only a few days before it was delivered. One signal of budget discipline has been that annual expenditure out- turns have been within little more than a single percentage point of the origi- nal estimates: 1.8 percent in 2001–02, 1.4 percent in 2002–03. (An additional reason for this achievement may have been the Public Finance Compliance Act, which provides for prison sentences for ministers or their permanent secretaries who overspend.) At the sectoral level, too, the budget outturn is close to the budget allocation, which is also close to the MTEF for each sector. Table 5.1. PRSPs, Medium-Term Budgeting, and Cabinet Participation 1. 2. 3. 4. 5. 6. 7. 8. PRSP PRSP Recurrent Capital Cabinet involved Cabinet involved PRSP linked to linked to Have expenditure expenditure throughout only at end Country costed budget MTEF MTEF in MTEF in MTEF budget process budget process Benin Y Y Y Y Y Y Y — Burkina Faso Y Y Y Y Y Y N Y Cameroon N Y N N Y — N — Central African Republic N N N N N — N — Chad N Y Y Y Y Y N Y Congo, Democratic Republic of N N N N — — N — Ethiopia Y Y Y Y Y Y N N Gambia, The Y Y Y Y N Y — — Ghana Y Y Y Y N Y N Y Guinea Y Y Y Y Y Y N — Guinea-Bissau N N N N N Y N N 155 Kenya Y Y Y Y N N N Y Lesotho N N N — — — — — Madagascar Y Y N N N Y N — Malawi Y N N Y Y Y N N Mali Y Y Y Y Y Y — — Mauritania Y Y Y Y — — Y — Mozambique N N Y Y N N N — Niger N N Y Y — Y Y — Rwanda N Y N Y — Y N Y São Tomé and Principe N N N N N N N Y Sierra Leone Y Y Y N — — Y — Tanzania Y Y Y Y Y Y Y — Uganda Y Y Y Y Y Y Y — Zambia Y Y Y Y N Y N Y — Not available. Y Yes. N No. Source: Authors. 156 Cabinets, Budgets, and Poverty: Political Commitment to Poverty Reduction Table 5.2. The MTEF in Relation to the Budget and Actual Outcome, South Africa, 1999–2000 (percent) Sector MTEF (1999) Budget Actual outcome Administration 6.1 6.3 6.8 Defense 6.7 6.6 6.4 Economic 5.5 6.6 5.9 Education 28.6 27.5 28.0 Health 14.5 14.8 14.8 Infrastructure 13.8 12.3 12.8 Justice 13.1 13.0 13.5 Welfare 11.7 11.5 11.7 Source: Le Houerou and Taliercio (2002). Initially, then, discipline was at the expense of consensus-building. As table 5.2 reveals, however, an ongoing series of reforms, initiated in 1997 (the first year an MTEF was considered by the government) is progressively making the process more inclusive. THE BUDGETMAKING PROCESS. Two key episodes in the role of the Cabi- net in the annual budgetary process are the lekgotlas held in January and July. All cabinet ministers, deputy ministers, directors-general, and some presidential staff—a total of 80 or 90 people—attend these two- or three-day retreats. The start of the annual process is a request, early in the calendar year, from the minister of finance to his or her departmental colleagues ask- ing for their contributions to the minister’s budgetary priorities paper for the July lekgotla. This paper goes first to the Ministerial Committee on the Budget (MCB), chaired by the minister of finance. The other seven senior cabinet members who make up this group, although the heads of some of the big-spending departments (for example, education), are selected not to represent and fight for their departments, but rather to represent the Cabi- net, and make decisions in the national rather than in any sectional interest. The president, who has his or her own economic adviser as well as a Policy Coordination and Advisory Services Unit, has input at all stages, especially during the drafting of the budget priorities paper. The budget priorities paper is, in practice, an expression of priorities for the following MTEF round. It is finalized at a cabinet meeting in June and dis- cussed at the July lekgotla. Following these two steps, departments submit their detailed bids for the upcoming budget, plus their spending plans for the next Harry Garnett and William Plowden 157 two years. These plans are discussed first in a committee of officials, the Medium-Term Expenditure Committee (MTEC). The MTEC’s recommenda- tions are reviewed by MCB over the next few months. MCB calls in individual ministers to discuss specific issues arising from their bids. MCB also reviews the macroeconomic framework, the fiscal situation and fiscal policy, and the division of revenue between the three spheres of government—national, provincial, and local. (Recommendations on this last issue are made by the independent Fiscal and Financial Commission.) In late September the MCB’s proposals are considered by an “extended” cabinet meeting (the normal cabi- net plus the nine provincial premiers) and, in October, by the cabinet proper. When the minister of finance submits the detailed budget to the Cabinet in November, he does so with the MCB’s backing. Spending ministers who are dissatisfied with their allocations may raise objections at this late stage, but the objections rarely lead to significant changes. The budget is approved at the January lekgotla. A few last-minute changes can still be made before the budget is presented to Parliament in February. The minister of finance briefs the president two days before the budget is presented, and before the president’s own speech is finalized. The Cabinet receives the printed budget review a week before it is presented. The January lekgotla also begins the process of planning priorities for the next MTEF round and the following year’s budget. Clusters comprising ministers from several departments aim to provide the first statements of medium-term priorities for the group of departments concerned. From the discussion at the lekgotla the president’s staff writes up a governmentwide statement of priorities. After reference back to the clusters for approval, this statement informs the president’s State of the Nation speech, given in Feb- ruary. Normally one adjustment budget is produced each year, seven or eight months into the fiscal year. This budget allows for supplementary appropriations, although only for unforeseen or unavoidable contingencies, and also for the allocation of windfalls. Most new bids are turned down. The adjustment budget is drawn up by the Treasury Committee of Minis- ters, chaired not by a treasury minister but by the deputy president. The net increase to the original budget has normally been around 1–2 percent, although in 2002–03 unexpectedly high inflation drove it up to just over 3 percent. Accounting officers forwarding their departments’ budget submis- sions to the National Treasury are required to summarize “the extent of donor funding received and how these funds are utilized in support of key strategic objectives” (Republic of South Africa 2003, p 18). Throughout the year the National Treasury publishes, on its website, quarterly outturn reports both for departments and for the nine provinces. The cabinet 158 Cabinets, Budgets, and Poverty: Political Commitment to Poverty Reduction receives an unaudited report on the year about six weeks after the end of the year and normally discusses the end-year outturn. In principle and in intention these processes allow for a much greater collective political input, and for a more strategic approach, to budgeting than is found in many countries, developed and developing. The two annual lekgotlas set the tone. The cabinet in various forms (regular, extended, and lekgotla) has at least eight meetings a year on budgetary mat- ters. MCB, which meets four or five times a year, is explicitly intended to act as a substitute for the cabinet, acting in collegial mode. Collegiality, or at least collectivity, is further encouraged by frequent meetings of ministers for other purposes: six cabinet subcommittees meet every two weeks, and the cabinet itself meets in the intervening weeks. The president or his or her staff, or both, have the chance to be involved at all stages of the process. Working relationships between the president and the minister of finance are said to be close. The MTEF, which has reduced the significance of the annual budgeting process by putting it into a three-year context, has encouraged both longer term thinking and planning, on the one hand, and some propensity to prioritize, on the other. We were told the following story about how seriously the budget and the MTEF are taken and how the MTEF has forced changes in behavior. Long accustomed to being a favored department in apartheid days, the police (now sponsored by a structural adjustment program) adapted slowly to changed times. In one of the MTEF years, it requested many vehicles, arguing that they were needed to combat crime. Within its budget ceiling, the capital spending was approved and the structural adjustment program bought its vehicles. The police had made this request, however, at the expense of other needs. After the first quarter they came to the National Treasury and demanded a supplementary budget on the grounds that they no longer had the money to operate the vehicles. In the old days, which were characterized by frequent supplementaries, this request would have been immediately granted, so the police again asked National Treasury for money, arguing that this circum- stance was a budget crisis, and if they did not get a supplementary, crime would explode. No, replied National Treasury, this issue is not a budget one, but rather a management one. You are responsible for achieving your depart- mental targets within the MTEF envelope, the National Treasury asserted, and if you have spent your budget on vehicles you cannot now run and main- tain, that is your management failing. No more funds. CONTINUING CHALLENGES. For all of the strengths of South Africa’s bud- getmaking process, the perception remains among some senior South Harry Garnett and William Plowden 159 African policymakers that the budget still, broadly, drives policy rather than the other way around; the priorities expressed in the budget are those of the National Treasury rather than of the cabinet. Some observers feel that MCB adds little to the proposals put to it by the minister of finance. One commented that he doubted whether members of MCB really acted in as disinterested, collegial a manner as others have claimed. The budget priorities paper that emerges from the January lekgotla, said another, gave little firm guidance on priorities. Similar to other stages of the budgetary process, it tended to produce an undifferenti- ated, lengthy wish list—“not so much a shopping list as a shopping trol- ley,” commented the second observer. This result is perhaps not surprising given the number of people who take part. Treasury guidance to depart- ments for submissions for the 2004 budget points out that departmental requests for the 2003–04 to 2005–06 MTEF baseline were well over 100 per- cent greater than the actual allocations. “National departments are there- fore requested to submit funding requests that are more closely aligned with the trend in baseline increases” (Republic of South Africa 2003, p. 13). The quality of three-to-five year planning varies greatly between depart- ments. Analysts, both in the treasury and in the presidency, clearly believe that the system needs to evolve further if these weaknesses are to be elimi- nated or at least diminished. The year 2003 thus witnessed a further evolution in South Africa’s bud- get system. Described in a local financial weekly under the headline “Mbeki Tightens Grip on Budget,” the change has consisted of involving the presi- dent’s office and other central departments (Public Service and Administra- tion, Local and Provincial Government), along with the treasury, in preparing the budget priorities paper following the January lekgotla. Now known as the medium-term strategic framework, the paper not only sets out the medium-term economic and social framework, but also the govern- ment’s strategic approach to key challenges, the issues and priorities identi- fied by the cluster groups, and the budget choices for the following year’s medium-term strategic framework. The aim is to persuade ministers to dis- cuss and agree on priorities before any specific figures for departmental allocations are brought forward. After discussion in the MCB, and then the cabinet, the final version of the medium-term strategic framework is consid- ered by the July lekgotla. It provides the basis for the medium-term budget policy statement presented in the parliament by the minister of finance in October. It should also inform the bilateral MTEC hearings, in the autumn, at which National Treasury officials scrutinize departments’ medium-term proposals. 160 Cabinets, Budgets, and Poverty: Political Commitment to Poverty Reduction Whether these changes will achieve their aim of what one treasury official was cited as calling “a politically driven process” rather than a “budget . . . determined by bureaucrats” remains to be seen. A virtuous circle can be envisaged in which a stronger and more assertive presidential staff ensures that the preparation for the January lekgotla secures serious ministerial atten- tion, that the subsequent medium-term strategic framework has greater authority than did the budget priorities paper, that it is prominent at the July lekgotla, and that it consequently impinges significantly on the following MTEC hearings and on departments’ internal planning and forecasting processes. Much will depend on developments in the next couple of years and, in particular, on whether the president’s staff can effectively grasp their new, extended role. “If we can’t become credible in that time,” said a mem- ber of the president’s staff, “we might as well back off and leave things to the treasury.” Zambia Zambia introduced its first MTEF process, with strong Cabinet and parlia- mentary involvement, in 2004; however prior to 2004 there had been no Cabinet involvement until the very end of the process and little participa- tion by any other stakeholders. Zambia had one of the most unpredictable budgets in Africa. BACKGROUND AND PERFORMANCE. Among the four cases, Zambia’s bud- getary practice and performance represents the opposite extreme from South Africa. Achieving aggregate discipline has been difficult and, for long periods, dependent on the blunt instrument of a cash budget (that is, autho- rization of spending only on the basis of cash receipts in hand). The Cabinet has not been involved in the early stages of the budget process, and as our hypothesis predicts, budgets and policy have been quite unreliable.1 The draft Public Expenditure Management and Financial Accountability Review for Zambia (World Bank 2003b) states that The lack of effective and credible budget preparation is one of the most serious weaknesses in Zambian public expenditure management. . . . 1. This case refers to the practice observed for the fiscal year 2003 budget cycle, a practice common to many countries. Zambia is now introducing an MTEF process characterized by substantial participation by the Cabinet, Parliament, and civil society. Harry Garnett and William Plowden 161 For years now, the Zambian government has consistently approved budgets that are structurally over-committed and that lack the resources to meet the on-going obligations of the government. The most obvious result has been that actual expenditures at the end of the year bear little relationship with the original budget estimates (p. x). The proximate drivers of Zambia’s endemically weak budgetary system are the rules governing supplementary appropriations. Zambia’s constitu- tion is very lax on supplementary appropriations. Article 103(4)(b) allows “the Minister responsible for finance shall introduce in the national assem- bly not later than fifteen months after the end of that financial year . . . a . . . Supplementary Appropriation bill, confirming the approval of Parliament of such expenditure. . . .” Article 103(5) of the Constitution provides: Where in any financial year, expenditure has been incurred without the authorization of Parliament, the Minister responsible for finance shall, on approval of such expenditure by the appropriate committee of the National Assembly, introduce in the National Assembly, not later than thirty months after the end of that financial year . . . a bill to be known as the Excess Expenditure Appropriation bill for the approval by Parliament of such expenditure. The Parliament, of course, has the right to refuse passage of the bill, but the minister will already have authorized spending. The Constitution also allows for spending when “the expenditure is authorized by a warrant under the hand of the President.” Effectively, this statement negates parlia- mentary control of spending, when set alongside ex post facto supplemen- tary submission. In practice, a spending minister may, at any point during the financial year, approach the minister of finance or the president to press for a supplementary appropriation to his or her original budget, which may already have been fully spent. Given these rules and processes, the fact that supplementary appropria- tions during the year amount to considerably more than the initial program budget hardly comes as a surprise. In recent years supplements have ranged from just over 100 percent of the original budget to nearly 1,400 percent. THE BUDGETMAKING PROCESS. What tends to happen is that the Ministry of Finance and National Planning agrees to relatively high levels of expendi- ture for the sectors that attract the attention of the donors, in particular the PRSP sectors, and holds down allocations for the less popular sectors, such as defense and foreign affairs. The budget thus often ignores data explaining 162 Cabinets, Budgets, and Poverty: Political Commitment to Poverty Reduction where the expenditure has actually been incurred in the past. The action needed to bring the less popular (with donors) line items down is simply not taken: no embassies are closed; no brigades are closed down; and prisoners are actually fed, even though the provision in the approved budget is unreal- istically low. So in the end, historical patterns are followed, and retargeting does not take place. Supplementary appropriations are not even submitted to the Cabinet for prior approval; they are simply reported to parliament for retrospective approval, after the end of the budgetary year to which they relate. Allegedly the level of supplementary appropriations has limits but, as the figures quoted suggest, they do not seem to be very effective in practice. A further factor weakening budgetary control is that ministers do not need to seek external approval for virement between programs in their own portfolio. Another internal factor may be an absence of agreement, between a minister and a permanent secretary, on the ministry’s policy. External influences also work to inflate expenditure over budget. One of these is ratio budgeting, the insistence by the international institutions that only a certain percentage of the total budget should be allocated to, for exam- ple, wages. The activities of donors represent another influence; grant agree- ments with donors are not subject to parliamentary scrutiny, although loans are, and the funding provided under them is often tied to specific projects. In general the budget is what a well-placed observer termed “academic.” The cabinet has played only a small role in budgetmaking in Zambia. Ministers are excluded from the budget both collectively and individually. Collectively the cabinet has normally had no input into the strategic think- ing from which the budget itself should be derived. Typically, this strategy has been worked out between the International Monetary Fund and the Ministry of Finance. Individually, ministers too often do not have close working relationships with their permanent secretaries; the latter are appointed by the Ministry of Finance as accounting officers and work more closely with the Ministry of Finance than they do with their own ministers. The coordination between the president and the cabinet is also inadequate. The cabinet collectively is usually informed about the budget only the day before the minister of finance presents it to parliament. Efforts made in 2002 to improve the process had little impact. The inten- tion had been to submit a budget framework paper to the Cabinet, from which indicative budgetary ceilings could be derived. In fact, so many spe- cific issues came up for discussion that no time was left for consideration of the big picture. Meetings to discuss budget strategy had to be postponed because of the absence of the president, as chairman of cabinet. Furthermore Harry Garnett and William Plowden 163 although four meetings were called at which the Cabinet was intended to consider strategic issues, the meetings took place very late in the season—in November. The budget was due to be presented to Parliament in January. By then only marginal changes could be made; it was too late to discuss or question the strategy. When the framework paper was submitted to Policy Analysis and Coordination division in the Cabinet Office, the division had only half a day to analyze it and prepare briefing on it. This briefing was available only to the president and the cabinet secretary. The final draft of the budget for the current year was presented to the Cabinet three hours before the meeting at which it was to be discussed. In those three hours min- isters were required to read the paper and to prepare their own comments on it, with no opportunity to refer back to their ministries or even to their permanent secretaries. The budget speech was drafted by the Ministry of Finance, read the night before the budget was due to be presented, edited in the early hours of the morning, sent to the government printers, and read to Parliament the same afternoon. One result of this haste was that the budget as actually presented was full of errors, omissions, and inconsistencies and contained no explanation of the underlying objectives or policies. CONTINUING CHALLENGES. Although many ministers express dissatisfac- tion with most aspects of the budgetary process, their dissatisfaction does not inhibit them from exploiting it to their own advantage. Some observers suggest that ministers in fact prefer that both overall and departmental bud- get ceilings should be declared by the Ministry of Finance, working with the International Monetary Fund, because this spares ministers the need to make painful choices about priorities. In practice ministers pay little heed to the budget; they do not respect it either as a guide to action or as a limit on what they may do. No incentives encourage ministers, or permanent secre- taries, to stay within budget, and no penalties punish those who do not. A permanent secretary bold enough to challenge a minister who wanted to exceed his budget appropriation would be likely to lose his job. Some years ago a committee of permanent secretaries recommended that the Ministry of Finance should conduct midterm reviews of expenditure, so that diver- gences from budget could be noted and corrected, but no action was taken. Uganda Uganda has one of the model MTEF processes, with strong Cabinet and par- liamentary participation throughout. Both the president and the minister of finance, development, and economic planning play strong roles. The budget 164 Cabinets, Budgets, and Poverty: Political Commitment to Poverty Reduction is extremely predictable, although instances have occurred when interven- tion from the top changed priorities. BACKGROUND AND PERFORMANCE. Uganda’s budgeting process has been transformed over the past 10 years, to a condition that not only assures bud- get predictability, but also that has been much admired by donors and oth- ers for the rigor of its priority setting. As detailed in this section, the basis for budget predictability seems to be primarily a close and effective work- ing relationship between the president and the minister of finance. It remains uncertain whether this relationship, in an executive which, in the terminology used by Evans and Manning (2003), seems to be more hierar- chical than collegial, will continue to evolve toward greater collegiality. Before the mid-1990s, the Ugandan budget system was anything but a model. In the 1980s the budget was planned and managed by the Ministry of Finance with little input from spending departments or from the cabinet collectively. The Ministry of Finance was often in conflict with the separate Ministry of Planning. The frequent supplementary appropriations could be validated retrospectively by parliamentary approval within four months of the end of the fiscal year. Individual departments, such as Defense and the State House, had huge overruns on spending. The consequences of these overruns, combined with efforts to hold down total expenditure, were unprogrammed cuts in the budget allocations of other less privileged departments. Inflation rose to over 100 percent in 1991. The Ministries of Finance and of Planning were merged in 1992 and, despite a short renewed separation in 1996–98, have been combined ever since as the Ministry of Finance, Planning, and Economic Development (MFPED). Budget discipline was achieved. Table 5.3 shows that, although the budget deviation index (comparing actual expenditure with the alloca- tions approve by parliament) has fluctuated over time and has been quite high recently, the overall trend is downward and, more to the point, over- runs have been small in recent years. In fiscal 2001–02 revenue fell substan- tially below forecasts and, partly as a result, the deficit was historically very high at 11.3 percent of gross domestic product. For only two budget items, however, were overruns significant: public administration at 12.4 percent and security at 4 percent. Supplementary appropriations reached the level of 3 percent, which is the maximum now permitted under the 2001 Budget Act. Previously the only legal requirement was that these should be submitted to parliament for approval within four months of the end of the fiscal year. Now the Public Finance and Accountability Act requires prior parliamentary approval of any supplementaries. The published report on the first quarter of Table 5.3. Budget Deviation Index, Uganda, Fiscal 1987–99 (percent) Budget variations 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 Shortfalls 28.2 9.9 21.9 23.3 11.8 10.3 17.1 20.5 22.0 16.4 8.1 28.9 7.3 165 Overruns 19.8 25.9 16.0 8.4 10.2 5.5 9.6 11.5 4.4 10.9 4.8 0.7 0.5 Budgeted deviation index 48.0 35.8 37.9 31.7 22.0 15.8 26.7 32.0 26.4 27.4 12.9 29.5 7.8 Source: Le Houerou and Taliercio (2002). 166 Cabinets, Budgets, and Poverty: Political Commitment to Poverty Reduction the following fiscal year showed these two items once again heading toward modest overspending, at 1.7 percent and 1 percent, respectively. Beginning in 1997, the budget also has increasingly been aligned with the country’s poverty reduction goals. In 1997 following a 1995 workshop attended by the president. the first poverty eradication action plan was pre- pared, as the overarching document defining the government’s strategy for reducing poverty. It was updated in 2000 and has since been closely linked to the annual budget process. A report by the Overseas Development Insti- tute (ODI) (Foster and Mijumbi 2002) expresses this view: “We know of no other country that has achieved such a dramatic pro-poor change in spend- ing patterns in so short a period” (p. ix). THE BUDGETMAKING PROCESS. How does the priority setting and budget- ing process operate in Uganda? The focus here is on the process within the executive; since 2001 the parliament has also been brought into the process, but the impact of this potentially inclusionary reform cannot be evaluated yet. In outline, the main activities in the executive budgetary process are as follows. The fiscal year runs from 1 July to 30 June. MFPED reviews and updates the MTEF in September and October, identifying the key issues and providing a preliminary macroeconomic framework and projection of aid flows. In October a consultative budget workshop is attended by the presi- dent, members of the cabinet, members of parliament, district chairs, senior officials of national and local government, nongovernmental organization representatives, and development partners. Following this workshop the first budget call circular is issued to four sector working groups. These groups include not only sector officials, but also donors and other stake- holders. They review ongoing programs and produce sector strategy state- ments covering objectives, activities, and outcome indicators, which set three-year resource ceilings for each major program. They also identify any priorities that remain unfunded. Each sector then produces a sectoral bud- get framework paper. These papers indicate how each minister would use his or her budget allocations. After bilateral discussions between MFPED and the ministers to resolve any outstanding issues, the sectoral papers are combined into a comprehensive budget framework paper. This combined paper also takes into account Budget Committee of Parliament comments, which are in turn informed by discussions between individual ministers and the relevant parliamentary sectoral committees. At this stage the minister of finance and his or her team have a one-day meeting with the president; this meeting produces the final input to the bud- get framework paper that is discussed in the cabinet in March. The budget framework paper provides the basis for the second budget call circular, which Harry Garnett and William Plowden 167 sets out the guidelines for ministries’ detailed budget estimates. These are reviewed by MFPED and combined into a final draft budget. The cabinet for- mally approves this very shortly before its presentation to Parliament in June. The ODI’s summary of the Ugandan arrangements, though acknowl- edging that the policies to which they contribute “are still recent and still contested (Foster and others 2002, p. ix) was generally laudatory. It goes on to say that in Uganda, both the capacity [to question budget bids] and the author- ity [to propose modifications] have been given to MFPED, and there is an annual process of reviewing budget proposals that is remarkably open, involving donors as well as civil society. The Cabinet approves budget ceilings based on MFPED advice, with MFPED submitting information on unfunded bids for the Cabinet’s decision on priorities. . . . [D]iscussions in the Cabinet on expenditure proposals are taken as part of a disciplined budget process, respecting the need to keep within an overall resource envelope, and to make choices (p. 18). CONTINUING CHALLENGES. Notwithstanding this praise, two interre- lated questions may be asked about current arrangements in Uganda. First, how well entrenched are they and how effective do they promise to be in ensuring budget predictability? Second, and closely linked to the first ques- tion, what is the relative influence in these arrangements of the president, the cabinet, and the MFPED? A summary view is that Uganda fits the model described by Evans and Manning (2003): “Several Commonwealth govern- ments, particularly in Africa, are . . . more hierarchical than collegial although they formally retain cabinets” (p. 21). In the ODI’s view the key to Uganda’s success in restructuring the budget in favor of the poor has been strong emphasis on poverty by the leadership, that is, the president, and an effective MFPED. The latter has made it possible to bring “finance and planning together within a single institution to facili- tate a process whereby plans prepared to implement the strategy reflect a realistic view of budget constraints, while budgets can be scrutinized for their consistency with plan priorities” (Foster and Mijumbi 2002, p 36). In other words, the key working relationship in this context is that between a strong president and MFPED. One positive consequence of this relationship, in the view of one permanent secretary, is that it is difficult and unusual for spending ministers to try to evade budgetary constraints by seeking to bypass MFPED and making personal approaches to the president or vice president. Even if a ministerial colleague came to the president claiming an urgent need for funds, the secretary thought, the president would be likely to refer him or her to MFPED. 168 Cabinets, Budgets, and Poverty: Political Commitment to Poverty Reduction The downside, in the view of some observers, is that the president, not the cabinet, “owns” the annual budget. The cabinet in Uganda does not seem to be a very effective body either for making decisions or for formulat- ing strategy. It has 21 full members and 67 alternate members in the form of ministers of state. It lacks an effective analytical capability, which would enable it to scrutinize proposals of MFPED and spending ministries alike. (Its secretariat comprises three support staff.). The president and MFPED may well agree beforehand on the essentials of the MTEF to such an extent to which the cabinet, even if its members disagree, can influence decisions. One outside observer felt that, on the one hand, the president normally pre- ferred to secure the cabinet’s agreement and could do so without difficulty, but on the other hand, he or she could, if need be, proceed without cabinet support. The same is said to be true of decisions about specific budget cuts. A donor felt that although from the beginning the poverty eradication action plan and other government policies had been clearly consistent, this uniformity was, for the most part, because of the lead taken by the presi- dent. At the same time, the president’s influence is largely personal, not institutional. An example that illustrates the continuing fragility of the system (one mentioned by almost everyone interviewed for this study) was the presi- dent’s personal decision, early in fiscal 2002–03, to increase the budget allo- cation for the security operation in the north. This decision appears to have been taken without involving MFPED, whose subordination to the presi- dent was thus highlighted. The consequence of making an unbudgeted increase in the budget allocation for defense spending, combined with the constitutional requirement that total supplementary allocations should not be more than 3 percent of budget, was across-the-board cuts of up to 23 per- cent in a wide range of programs. Mozambique Mozambique has a fairly predictable budget, but little participation by the Council of Ministers as a whole in its preparation. The predictability owes much to the budgetary discipline imposed by a strong Ministry of Planning and Finance. BUDGET PERFORMANCE. For much of the past decade, Mozambique’s Ministry of Planning and Finance has been rigorous in maintaining aggre- gate budget discipline to the point that it refuses to borrow, even on the short term, to handle fluctuations in revenue flows. For reasons explored in Harry Garnett and William Plowden 169 this section, however, putting in place any systematic process of prioritiza- tion (let alone an inclusive process) has proven to be quite difficult. Current expenses as a whole tend to be around 10 percent less than approved by the National Assembly. This discrepancy is largely attributable to a shortfall in revenue of about that amount. In 2002 current expenses in only two sectors exceeded the allocation: public administration (104 percent) and public works (205 percent). Investment expenditure is well below the allocation, at 57 percent in 2002. In this case variation among sectors was con- siderable: public administration at 232 percent was the only sector to exceed its allocation; labor and water were both below 10 percent. As explored fur- ther in the following paragraphs, these variations in the investment budget are mainly a result of changes in the expected contributions of donors. THE BUDGETMAKING PROCESS. The budgetary process is typically as fol- lows. Since 1997–98 an MTEF process has been in place, with a gradual increase in the number of sectors covered. In late May the MTEF is pre- sented by the Ministry of Planning and Finance (MPF) and discussed at the Economic and Social Commission, a kind of ministerial subcommittee of the Council of Ministers. Shortly afterward, the MPF informs ministries of their budgetary ceilings. These ceilings are discussed, bilaterally, over the next few months; ministries’ final proposals for expenditure within agreed ceil- ings are submitted to the MPF between July and September. A week or two later the MPF presents the budget as a whole to the Economic and Social Commission and, after perhaps an additional week, to the Council of Minis- ters, where the discussion does not usually last longer than an hour or so. A few weeks later the budget is introduced in the legislature. As participants observe, these processes do not allow for any significant discussion, at any interministerial level, of the context, the strategy, the pri- orities, or even the details of the budget. The Council of Ministers does not discuss the MTEF at all. It discusses the budget at such a late stage that sig- nificant changes are effectively impracticable. Even the Economic and Social Commission’s initial discussion of the MTEF has little impact on the subse- quent distribution of resources between programs, because by the time it takes place, ministries’ ceilings have already been decided by the MPF. The situation was summarized in the World Bank (2001) Public Expenditure Review: Elaboration of the final budget . . . is the sole responsibility of the MPF, on the basis of the proposals received from line Ministries. . . . Although negotiations may in some cases take place, much of the 170 Cabinets, Budgets, and Poverty: Political Commitment to Poverty Reduction discretionary power for resource allocation rests with the MPF which submits the budget for approval to the [Council of Ministers]. Subsequently, overall levels of spending by individual ministries bear little consistent relationship to their budgetary allocations, nor do the indi- vidual projects funded relate clearly to previously determined governmen- tal priorities. Both total ministry spending levels and the selection of individual projects depend partly, first, on the interests of donors and the support which they provide to ministries and, second, on levels of other off- budget receipts, for example from user fees, which in 1998 were equivalent to 37 percent of budget allocation for health to 107 percent for public works. These revenues are not under MPF control and are often not even fully reported to MPF. Even within ministries, the functional classification of expenditure and revenues identifies only broad sectors, such as education. The result is that, as the Foster and Mijumbi (2002) study points out, analysts can determine how much has been allocated to and spent on education in general, but they cannot to do the same for primary education. This deficiency is reinforced by the large degree of independence and authority allowed, within min- istries, to national directors, who are responsible to the minister who appointed them rather than to their nominal superior in the civil service hierarchy, the permanent secretary. CONTINUING CHALLENGES. One word that could be used to describe the Mozambican budget process is “disjointed.” The connections which in a well-functioning system would exist, both between different elements and between different participants in the process, are weak or nonexistent. Thus no well-developed relationships between overall strategy and trends in sec- toral spending exist because, as the 2003 report of the Public Expenditure Review (World Bank 2003a) noted, “The lack of a medium-term perspective results in purely incremental budgeting . . . with no real link with activities or with outputs or outcomes.” We found that in Mozambique the Council of Ministers has no collective ownership of the budget, for two reasons. First is the general inability of the Council of Ministers to play an effective role in thinking and planning strategically for the government, reinforced by the lack of a strong collective or corporate spirit within the government. One result of this deficiency is a fragmentation of the policymaking and policy implementation process; some of those involved comment that departments see themselves engaged in an essentially competitive power struggle in which their own objectives Harry Garnett and William Plowden 171 are more important than the public welfare. Second (and both the cause and effect of this attitude) is ministries’ partial independence of the MPF; this independence is made possible by the significance of donor funding in rela- tion to government spending and, at least until this writing, the consequent strength of bilateral relationships between donors and individual min- istries. In some ministries (for example, Health, Roads, and Water) donors contribute 70 percent or more of the total. Ministries’ propensity to plow their own furrows is reinforced by their lack of confidence in the MPF either as guardian of current budgetary policy or as guarantor of budgetary allocations. Even the MTEF is described by ODI ( Fozzard 2002, p. 34) as “a fragile instrument”; the underlying projec- tions of growth and revenue performance are “extremely optimistic and tend to be set as targets rather than what can reasonably be achieved.” The medium-term financial framework, the formulation of which takes up a great deal of official time, has little influence because it is not generally seen as credible. Ministries too often do not receive, or do not receive in a timely way, the allocations on which their planning should be and sometimes is based. (In early August 2003, nearly two-thirds of the way through the financial year, ministries had received only 30–40 percent of their budget allocations for the year.) In the opinion of one donor, the MPF does not see itself as in any way accountable to ministries nor as responsible for ensuring that promised funding is made available. Ministry officials say that they sometimes face difficulties in planning new initiatives because MPF will not confirm whether the necessary resources would be available. A minister observed that the budget was treated as though it were the private property of the minister of planning and finance, rather than an expression of the government’s collective policy. In sum, Mozambique faces major challenges if it is to strengthen its bud- getmaking process in a way that can facilitate credible prioritization and consensus-building. One challenge is to strengthen the coherence, consis- tency, and legitimacy of the domestic prioritization process. The MTEF has been a valuable start; recognition of the importance of other aspects of bud- get management has also increased, with resources and technical assistance going into these areas as well. A second challenge is to bring more coher- ence and flexibility to the provision of donor support. Moves toward a more collective approach by donors are starting to produce results in the shape of sector programs and, more recently, of budget support. We hope that, in the not too distant future, Mozambique’s processes of budget prioritization will enjoy the same levels of success that have been achieved in maintaining aggregate budgetary discipline. 172 Cabinets, Budgets, and Poverty: Political Commitment to Poverty Reduction Conclusions Any effective government strategy, whether its objectives be poverty allevia- tion or anything else, must both reflect and drive the government’s current priorities. Effective strategy also requires consistency in patterns of resource allocation, horizontally—between activities and programs—and vertically— over time. The budget is the government’s main instrument, first, for ensur- ing that the way in which resources are allocated reflects current priorities and, second, for achieving the necessary consistency. The facts that the department responsible for the budget is the Ministry of Finance and that the budget is the key tool that the Ministry of Finance uses to achieve its specific objectives relating to economic and financial management should thus not obscure the importance of the budget to the government as a whole. All members of the government—that is, in cabinet systems all members of the cabinet—therefore have a legitimate interest in the content and objectives of the budget. These elements should not be left to be decided by the Ministry of Finance on its own. Moreover, the value of the budget as an instrument of policy and as an expression of governmental strategy, increases with its predictability. If out- turns, sectoral and total, have a purely random relationship to budget allo- cations and forecasts, budgeting will have been a waste of time. A well-staffed and well-organized Ministry of Finance is a necessary condi- tion for budget predictability, but it is not a sufficient condition. Budgets will be the more predictable the more they embody priorities and con- straints acknowledged and accepted by all principal stakeholders within the executive (and also, if possible, outside). In some hierarchical systems the only stakeholder of any importance may be the president. In some colle- gial systems the relevant group may be the whole cabinet. In any system in which the cabinet performs the function of co-opting powerful interests whose support is necessary for the government’s survival, however, mem- bers of the cabinet must be involved to some extent. In some systems cabinet members agree, implicitly or explicitly, to leave much discretion to the minister of finance in putting the governmental bud- get together and sign off on the final product in a manner more formal than real. In others they insist on being directly involved at all stages. In many, the precise point on the spectrum between these two approaches may vary with politics, personalities, and circumstances. The survey in appendix 5A to this chapter of budgeting practices in a range of Organisation for Eco- nomic Co-operation and Development (OECD) and other countries shows that the trend is, broadly, toward increasing cabinet involvement. Harry Garnett and William Plowden 173 Our review of practices in African countries, and our four case studies, show a range of variations in approach and in budgetary outcomes. • South Africa shows a trend from firm Ministry of Finance control toward thorough and increasing cabinet involvement, combined with high predictability at all stages. • Zambia shows little cabinet involvement and low predictability, but now with attempts to increase both. • Uganda shows normally high predictability with firm Ministry of Finance control. The basis for this predictability has been active presi- dential involvement, however. One exceptional episode in which pres- idential involvement resulted in damagingly unpredicted outturns provides some negative evidence consistent with the hypothesis. • Mozambique, which has minimal cabinet involvement in the budget process (and even problems in measuring outturns) but which also has substantial cabinet involvement in PRSP, is investing heavily in improving budget management systems. We have not been able to assemble sufficient good quality objective data for us to be able, confidently and scientifically, to affirm the hypothesis with which we began this study. Quite apart from the need to have more cases studies, in particular ones from francophone Africa, we are faced with the small numbers problem. Africa has only two known success stories, South Africa and Uganda, which fit the hypothesis and these countries have other factors on their side, such as strong and mature budget institutions, in the case of South Africa, and a strong president, in the case of Uganda. Even in Uganda, however, the strong leader has found it necessary to develop a process characterized by strong cabinet involvement and participation with those outside the executive branch of government. Indeed, both of these two model countries appear to be thinking in terms of even more cabinet involvement in the budget and MTEF process; in fact, many we spoke to were quite critical of the level of cabinet involvement, even though it is far higher than an other countries. Regardless of the specific modalities of budgeting at the country level, however, our case studies confirm in our minds the crucial importance of linking PRSPs to this budgetary process. An increasingly large proportion of donor support to the poorer developing countries, and thus their escape from poverty, hinges on well prepared and, more important than that, effec- tively implemented PRSPs. We believe (sharing the view of others quoted in this report), setting up parallel systems to prepare, implement, and monitor 174 Cabinets, Budgets, and Poverty: Political Commitment to Poverty Reduction PRSPs is an extremely bad idea. We concur with the view that PRSPs will not be implemented as planned unless the poverty reduction priorities are determined through an MTEF process that includes substantial involve- ment of the political leadership from the cabinet—from setting initial bud- getary ceilings related to those priorities, constrained by realistic estimates of the resources available, to monitoring the implementation of the budget. Future reviews of PRSPs should ask, as in this study, how a country’s policy and budgetmaking processes work in practice. The weaknesses evident from such analysis would give guidance about the institutional changes required to establish a necessary set of conditions to implement poverty reduction programs. Sufficient conditions are of course related to the leader- ship’s political will, skill, and—simply—power. Appendix 5A. Budgetmaking in Countries of the Organisation for Economic Co-operation and Development As the experience in countries of the Organisation for Economic Co-opera- tion and Development (OECD) highlights, no single institutional arrange- ment unambiguously is the best practice for budgetmaking. Relationships within Cabinets In some OECD countries, the minister of finance has no special status; spending may be controlled through a contract among the leaders of the government. This practice has, broadly, been used in Belgium, Greece, Ire- land, Italy, Luxemburg, and Portugal (Alesina and Perotti 1999). In others, the minister of finance is second only to the prime minister or president in the cabinet pecking order. In the United Kingdom in the Blair administra- tion from 1997, the chancellor of the exchequer has been at least as powerful a figure as the prime minister, Tony Blair, and, by 2002–03, arguably more powerful. The basic arrangement must be appropriate to the politics of the local situation; however, the detailed procedures must also be acceptable politically. If members of the cabinet do not, explicitly or implicitly, assent to the processes that determine their own budget shares, they are less likely to accept those shares if they disagree with the outcomes. Whatever the precise balance between the finance minister and the head of government, the relationship between them must be close and mutually supportive. The finance minister will always be outnumbered by the spenders. They will always be restless. The minister should not be regularly out-gunned, as he or she will be if trying to fight a lone battle without the Harry Garnett and William Plowden 175 support of the head of government. No less a figure than J. M. Keynes made the point three-quarters of a century ago: “[N]o chancellor of the exchequer can in the long run enforce a policy which is opposite to the prevailing cur- rent in the Cabinet or contrary to the temperament of an autocratic Prime Minister” (quoted in Monck 1997, p. 279). In systems with strong cabinets and strong prime ministers, the latter’s support for the minister of finance is essential and usually forthcoming. Australia, Canada, France, and the United Kingdom are some examples. One might add the contrary proposi- tion: that no prime minister can in the long run enforce a policy that is oppo- site to the prevailing current in the cabinet or contrary to the temperament of a powerful minister of finance. As a former British Chancellor has said: “A Chancellor who is doing his job properly has few friends. That is one rea- son why it is of vital importance to the successful conduct of government that there is an extremely close and special relationship between the Chan- cellor . . . and the Prime Minister. They do not have to be cronies or soul- mates but they do have to be on the same wave-length” (Lawson 1993, p. 273). In the United Kingdom, their offices are next door to each other. Timing and Extent of Cabinet Involvement in Budgetary Processes Many OECD countries where the level of influence of the minister of finance is high nonetheless involve the cabinets early on in the budgetary process and at several subsequent stages. In some, the cabinet has a general discus- sion, at the very start of the process, of the economic assumptions, broad parameters, budget strategy (often multiyear), often before any figures have been put on the table by anybody. This procedure is followed in Australia, Austria, Canada, Denmark, Finland, Ireland, the Netherlands, New Zealand, Norway, and Spain. In the United Kingdom this task is virtually the only one the cabinet as a whole has, apart from giving final formal approval. Sometimes this general discussion may include a range of stakeholders wider than the cabinet itself. Some of the countries in the foregoing list nor- mally have coalition governments (as has the Netherlands ever since the end of the World War II). In these cases the strategic discussion is an essen- tial part of the contractual approach, in which the several partners in gov- ernment agree, up front, on their budgetary targets and other strategies for the ensuing period. In the Netherlands the agreement applies to the whole four years of the administration. In Finland, another country in which coali- tions are the norm, it is a constitutional requirement that a new administra- tion agree and publish its government program, which sets out the main 176 Cabinets, Budgets, and Poverty: Political Commitment to Poverty Reduction policies agreed between the coalition partners, including the main bud- getary provisions. Multiyear budgeting in Ireland effectively began with the coalition agreement of December 1994 that explicitly set out medium-term fiscal targets (von Hagen, Hallet, and Strauch 2002). In New Zealand the Cabinet is initially involved during the strategic phase at the outset of the annual budget cycle, in determining budget strat- egy objectives for the following three years, in discussing policy priorities, and in agreeing to key themes for the coming year’s budget. In Canada the Cabinet has two two-day planning sessions a year, in spring and autumn. These sessions enable the Cabinet both to plan for the future and to assess progress on key objectives. In Sweden, where the budget is presented to Parliament in September, the Cabinet has a two-day meeting at the end of March to approve the total level of expenditure for the coming year as well as the indicative funding levels for the 27 expenditure areas for which min- isters are responsible. The broadly comparable Cabinet meeting in Finland lasts for a few hours. In effect, in the Scandinavian countries, the cabinet decides the broad budget totals for the ensuing period and indicates the sec- toral subtotals. Cabinets also take on a miscellany of what might be called “sub-strategic” tasks, all of which help to shape the final budget at a stage when decisive choices can still be made. In the Netherlands (whose budgetary process is, in the OECD’s view, as good as the best in OECD), the Cabinet is more tightly constrained by the initial coalition agreement; at the Cabinet’s major spring meeting to agree the annual budget framework, the issues discussed are how to apply underspending in any category to other activi- ties, how to distribute any budgetary windfalls, and whether any changes are needed in the coalition agreement. In New Zealand the Cabinet’s agree- ment is needed both for new policy proposals and for proposals for changes in budget baselines. Negotiating and Agreeing on Budget Allocations Later in the process, the numbers become more specific and disagreements about allocations more focused and, often, more heated. At this stage the question arises of whether and how far the cabinet should be involved in detailed horse-trading between individual spenders and the ministry of finance. This stage is, for many cabinet members, the most political one of all: these maneuvers are in effect zero-sum games, the outcomes of which can have profound effects on the status of winners and losers alike. In most cases the cabinet as a whole stays well out of these detailed and sometimes Harry Garnett and William Plowden 177 acrimonious discussions. Discussions can take place bilaterally between min- isters of finance and spending ministers. This approach involves the former in a long series of confrontations, some of which he or she will probably lose. Cabinet Budgetary Subcommittees In some countries the task is remitted to a powerful cabinet subcommittee of more-or-less disinterested senior ministers. Such a committee has existed in Canada since 1994, in the United Kingdom since 1992, and in Australia since the 1970s. Ideally, such a committee will be seen as in effect representative of the cabinet and as exercising authority on the cabinet’s behalf. As such the cabi- net will usually accept its decisions and appeals against them will be inef- fective. The Australian Expenditure Review Committee, active in various forms for 25 years, “has a broad mandate to review both new and on-going government expenditure, is relatively small and workable, and is uncon- cerned with the questions of hierarchy in cabinet [or with] appearing as an inner cabinet” (Kelly and Wanna 1999, p. 133). A similar committee in the United Kingdom makes allocations to spending departments, adjudicates on disputes, and reports back to the full cabinet. In Finland a subgroup of senior ministers, chaired by the minister of finance, reviews all issues of dis- agreement between the minister of finance and spending ministers. Its deci- sions are final; only if it cannot agree is an issue taken to the full cabinet. A deadlock is extremely rare. Conversely Canada’s equivalent Priorities and Planning Committee was accused of acting as an “inner cabinet” and its authority was thus undermined. (Kelly and Wanna 1999). In general the use of committees will, of course, come more easily in those systems that habit- ually use committees to process business on its way to or from the Cabinet. They are less likely to be acceptable elsewhere. A committee of this kind is less likely to be needed when both constitu- tional and political factors enable the head of government to use his or her personal authority to support the minister of finance against the spenders. Thus in France it is the prime minister who, with the minister of finance, decides budgetary policy for the coming year and informs spending minis- ters of the guidelines setting out the general policy framework and who subsequently informs ministers of the ceilings for their departments. This example illustrates the more general point that in some contexts the head of government can substitute his or her own authority for the collective con- sensus of the Cabinet, in insisting that colleagues respect budget constraints and imposing sanctions on those who do not. 178 Cabinets, Budgets, and Poverty: Political Commitment to Poverty Reduction References Alesina, A., and R. Perotti. 1999. “Budget Deficits and Budget Institutions.” In J. M. Poterba and J. von Hagen, eds., Fiscal Institutions and Fiscal Performance. Chicago: University of Chicago Press. Evans, Gord, and Nick Manning 2003. Helping Governments Keep Their Promises. South Asia Region Internal Discussion Paper no. 187. Washington, D.C.: World Bank. Foster, M., and P. Mijumbi. 2002. How, When, and Why Does Poverty Get Budget Prior- ity? Poverty Reduction Strategy and Public Expenditure in Uganda. London: Over- seas Development Institute. Foster, M., A. Fozzard, F. Naschold, and T. Conway. 2002. How, When and Why Does Poverty Get Budget Priority? Poverty Reduction Strategy and Public Expenditure Reform in Five African Countries. London: Overseas Development Institute. Fozzard, Adrian. 2002. How, When, and Why Does Poverty Get Budget Priority? Poverty Reduction Strategy and Public Expenditure in Mozambique. London: Overseas Development Institute. Kelly, J., and J. Wanna. 1999. “Once More into Surplus: Reforming Expenditure Man- agement in Australia and Canada.” International Public Management Journal 2(1): 127–46. Lawson, N. 1993. The View from No 11: Memoirs of a Tory Radical. London: Corgi. Le Houerou, Philippe, and Robert Taliercio. 2002. Medium-Term Expenditure Frame- works: From Concept to Practice. Preliminary Lessons from Africa. Africa Region Working Paper no. 28. Washington, DC: World Bank. Monck, Nick. 1997. “The Need for a Strong Treasury and How to Make It Work.” In D. Corry, ed., Public Expenditure: Effective Management and Control. London: Dryden. Republic of South Africa. 2003. Medium-Term Expenditure Framework—Treasury Guide- lines: Preparing MTEC Submissions. Pretoria: National Treasury. Schick, A. 2001. “The Changing Role of the Central Budget Office.” OECD Journal of Budgeting 1(1): 9–26. von Hagen, Hughes, A. Hallett, and R. Strauch. 2002. “Budgetary Institutions for Sustainable Public Finances.” In M. Buti and J. von Hagen, eds., The Behavior of Fiscal Authorities: Stabilization, Growth and Institutions. New York: Palgrave. World Bank. 2001. Mozambique: Public Expenditure Management Review. Report no. 22985. Washington, D.C. _____. 2003a. Mozambique: Public Expenditure Review, 2003. Report no. 25969. Wash- ington, D.C. _____. 2003b. “Public Expenditure Management and Financial Accountability Review for Zambia.” Report no. 261621-ZA. Washington, D.C. Draft. 6 Public Expenditure Accountability in Africa: Progress, Lessons, and Challenges Bill Dorotinsky and Rob Floyd Since the mid-1990s the World Bank has provided operational support and policy advice and analysis to Africa on public expenditure accountability, an area that has emerged as a development priority for Africa.1 This chapter reviews the progress of that work. It looks, in particular, at whether expen- diture accountability interventions over the past five to seven years have led to improved outcomes, at whether lessons on good practice are being applied consistently to current interventions and reforms, and at the critical preconditions for and sequencing of reforms to ensure effective and sustain- able expenditure accountability in Africa. The “Expenditure Accountability in Africa and the HIPC Expenditure Tracking Process” section describes the broader context, and outlines the expenditure tracking exercises of the Heavily Indebted Poor Countries (HIPC) Initiative, which provide the empirical backdrop for the discussion. The next three sections focus on the HIPC expenditure tracking data and lessons learned with particular regard to budget formulation, budget execution, and budget reporting. The “Learning Lessons: Budget Execution” section focuses on the overall lessons and outstanding areas for research such as decentralization, pro- curement reform, and participation in the budget process. 1. The International Development Association was slow in building the knowl- edge required to strengthen the control environment at the country level (World Bank 2001c). 179 180 Public Expenditure Accountability in Africa: Progress, Lessons, and Challenges Expenditure Accountability in Africa and the HIPC Expenditure Tracking Process Improving the allocation and execution of pro-poor expenditures in a trans- parent environment is of utmost importance to enable African countries to break the cycle of dysfunctional development. In July 2001, African heads of state adopted the New Program for African Development, in which they publicly committed themselves to pro-growth and pro-poor policies (World Bank n.d.). That same year, the United Nations adopted the poverty reduc- tion focused Millennium Development Goals as development targets. At the Monterrey Conference on Financing Development (Monterrey, Mexico, 2002), donors noted their intent to award larger shares of assistance to coun- tries working effectively in pursuit of the Millennium Development Goals. Furthermore—and important for this chapter—they noted that a country’s own budget should be the framework for development assistance.2 The focus by donors on country budgets as providing the framework for development assistance reflects an evolution in the modalities for resource transfer from donor to developing countries. Before the 1980s, the focus of much donor support was on capital investment, typically through self- standing projects. Although some projects were effective, in many recipient countries the overall quality of public spending had declined, undercutting aggregate development impact. By the mid-1980s the budget and financial management systems in many African countries were characterized by nar- row, politically driven budget formulation and execution, a lack of internal controls, and a degradation of fiscal procedures. Initially, analytical work to diagnose the problem—public investment reviews and later public expen- diture reviews—focused on sectoral allocations, particularly in health, edu- cation, agriculture, and the public service. In time, though, reflecting broader changes in development thinking, the focus shifted increasingly to institutions, political economy, and the underlying expenditure system and controls. By the late 1990s an increasingly holistic approach to public expenditure accountability was being promoted by the World Bank and other donors, with greater links between analytical instruments and operational approaches. Emphasis increased on client-owned and country-specific capacity-building, to put the country’s capacity to manage its own public finances front and center. The interest in public expenditure accountability led to numerous instruments for analyzing different aspects of public 2. For more information, see http://www.un.org/esa/ffd/. Bill Dorotinsky and Rob Floyd 181 expenditure management, including the public expenditure review country financial accountability assessment, country procurement assessment review, and the International Monetary Fund’s report on the observance of standards and codes. Yet despite all this activity, until 2002 few multicountry benchmarks and little consistent data over time existed to assess expenditure accountability progress. The International Development Association (World Bank 2001c) noted that there was no consensus on what actions were required at the country level for public financial accountability. What data did exist painted a worrisome picture of African public expenditure management. A 1999 World Bank, Africa Region, survey noted that only 5 percent of countries rated their public expenditure management as good, 21 percent as moder- ate, 47 percent as fair to poor, and 26 percent as poor. From 1995 to 1999 public expenditure management improved in 44 percent of countries, remained unchanged in 31 percent, and worsened in 25 percent (Kostopou- los 1999, p. i). By 2000 the boards of the World Bank and the International Monetary Fund had become increasingly concerned about the ability of country expenditure management systems adequately to control and track resources flowing through the HIPC debt relief program. The result was the introduction in 2000 of the HIPC expenditure tracking assessment initiative, which built on already extensive work programs on financial management. To initiate the HIPC tracking exercise, staffs of the World Bank and the International Monetary Fund jointly developed a 15-question survey. The process for conducting the assessment was as important as the assessment instrument: the process was designed as a collaborative effort between the World Bank, the International Monetary Fund, and the country, with increas- ing participation over time from other donors as well. The objective was an agreed-upon assessment and action plan and regular monitoring of progress in implementing the action plan, supplemented by reassessment against the benchmarks. In 2001, field assessments were carried out and action plans agreed-upon for the 28 countries that qualified for HIPC debt relief. The assessments were not designed for cross-country comparisons and represent a minimum set of measures deemed indicative of the overall quality of a country’s public expenditure management system. They include both perfor- mance and institutional indicators. The 15 criteria are listed in figure 6.1. The 2001–02 field assessments found that all African HIPC countries needed some or substantial upgrading to track poverty reducing expendi- tures adequately. Figure 6.2 lists the countries, including (in parentheses) the number of benchmarks met in each country. In 63 percent of countries the situation was deemed to require substantial upgrading of public financial 182 Public Expenditure Accountability in Africa: Progress, Lessons, and Challenges Figure 6.1. The 15 HIPC Tracking Criteria Expenditure accountability Benchmark description Comprehensiveness 1. Composition of the budget Meets Government Finance Statistics entity definition of general government 2. Limitations to the use of off- Extra- (or off-) budget expenditure is budget transactions not substantial 3. Reliability of budget as guide Level and composition of outturn is Formulation to outturn “quite close” to budget 4. Data on donor financing Both capital and current donor-funded expenditures included Classification 5. Classification of budget Functional and program information transactions provided 6. Identification of poverty Identified through use of classification reducing expenditure system (for example, a virtual poverty fund) Projection 7. Quality of multiyear expendi- Projections are integrated into budget ture projections formulation Internal control 8. Level of payment arrears Low level of arrears accumulated 9. Quality of internal audit Internal audit function used (whether Execution effective or not) 10. Use of tracking surveys Tracking used on regular basis Reconciliation 11. Quality of fiscal and banking Reconciliation of fiscal and monetary data reconciliation data carried out on routine basis Reporting 12. Timeliness of internal budget Monthly expenditure reports provided reports within four weeks of end of month 13. Classification used for budget Timely functional reporting derived Reporting tracking from classification system Final audited accounts 14. Timeliness of accounts closure Accounts closed within two months of year end 15. Timeliness of final audited Audited accounts presented to legisla- accounts ture within one year Source: World Bank (2002a). Bill Dorotinsky and Rob Floyd 183 management systems to enable reliable tracking of poverty reducing spend- ing. Figure 6.3 summarizes the data by performance across each of the 15 benchmarks. Some performance indicators revealed areas in particular need of support. For example, 92 percent of countries did not provide budget data consistent with the standard Government Financial Statistics (GFS) defini- tions, 83 percent did not have medium-term perspectives integrated into their budget, 88 percent had ineffective internal audit (IA), 88 percent did not close accounts within two months of the end of the fiscal year, and 83 percent did not have final audited accounts presented within 12 months of the end of the year (World Bank 2002a, p. 13). As part of the assessments, each country developed an action plan to remedy identified weaknesses, building upon reforms already under way. The action plans include short- term measures such as broadening coverage of government expenditures, upgrading classification systems, and piloting financial management infor- mation systems (FMISs). Medium-term actions include improving legal and regulatory frameworks, strengthening budget formulation, and improving internal controls. Figure 6.2. Expenditure Tracking Performance in 24 HIPC Countries Source: World Bank (2002a). 184 Public Expenditure Accountability in Africa: Progress, Lessons, and Challenges Figure 6.3. Substantial Upgrading Needed to Meet Benchmarks (percentage of countries not meeting benchmark) 100 75 Percent 50 25 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 Formulation Execution Reporting Indicator number (see figure 6.1) Source: World Bank (2002a). Implementation updates carried out in 2003 found that more than three-fourths of measures in the action plans have either been fully imple- mented or are in the process of being implemented. Progress is similar across budget formulation, execution, and reporting. Countries have been especially active in identifying and tagging poverty reduction spending. An increasing number of countries are now able to report on poverty reduction public spending, and such spending is rising in relation to gross domestic product and total expenditures. These increases must be accom- panied by increased efficiency and better targeting to improve social out- comes. Measures from the action plans are also being incorporated into poverty reduction strategy papers as well as into International Monetary Fund–supported programs and Bank adjustment operations. Eleven of 19 countries included in the update had initiated or implemented more than 80 percent of planned measures. About 20 percent of planned measures had been fully implemented and about 25 percent had not yet started (World Bank 2003b, p. 5). Bill Dorotinsky and Rob Floyd 185 Even with the recent gains, the HIPC Assessment and Action Plan (AAP) process confirms a situation well known to development practitioners: expenditure accountability systems in Africa are not functioning well, and reform measures over the past decade have not resulted in stellar improve- ments. Although in some cases, improvements have been identified, in many more cases, reforms have been under way for some time with little or no positive impact. Learning Lessons: Budget Formulation Budget formulation practices rated better on the HIPC benchmarks than either budget execution or reporting, a finding that correlates with greater rel- ative emphasis being placed on improved budget formulation in Africa over the past decade. It also reflects the fact that making improvements in budget formulation is relatively easier than execution or reporting because budget formulation generally requires fewer or more readily available skills and fewer technical systems. This section (a) describes some key characteristics of an effective budget formulation process and how they have been built into the HIPC tracking indicators, (b) reports on the HIPC tracking results for bud- get formulation among the participating African countries, and (c) highlights some emerging lessons about how to improve budget formulation processes. Some Characteristics of an Effective Budget Formulation Process Operational experience and analytical work on numerous countries point to the following prerequisites for effective budget formulation. • Macroeconomic stability • Good macrofiscal and public revenue projections • A comprehensive budget, that is a single budget that contains no off- budget items and has donor resources integrated • Meaningful hard budget constraints, including adequate internal controls and political will • Capacity and willingness to prioritize, reallocate, and make the diffi- cult trade-offs • Sound accounting and budget data, allowing budgets to be grounded in the prior year ’s actual spending and providing accurate informa- tion for costing existing or new policies and programs. This prerequi- site includes a robust classification system for recording expenditures, analyzing trends and identifying savings options, and controlling spending at an appropriate level of detail. 186 Public Expenditure Accountability in Africa: Progress, Lessons, and Challenges As noted in figure 6.1, the HIPC AAP process evaluated seven aspects of budget formulation: composition of budget entity, limitations to the use of off-budget transactions, reliability of budget as guide to outturn, data on donor financing, classification of budget transactions, identification of poverty reducing expenditure, and quality of multiyear expenditure projec- tions. Although all of these aspects are important, the rationale for two in particular—the composition of budget entity using GFS standards and the quality of multiyear expenditure projections—warrant further explanation. GFS-like standards are key, because effective budget planning requires a consistent accounts classification structure for recording and reporting financial transactions. The classification structure should be determined after identifying the information requirements for expenditure control, cost- ing programs and policies, assessing public finance trends, and carrying out economic and sector analysis. A robust classification system is important for reporting and planning expenditures, allowing different standardized pre- sentations, and analyzing data to meet the needs of policy officials. If the classification is too broad, it does not allow for the identification and moni- toring of preferred policy (for example, pro-poor spending). As for multiyear planning, a well-implemented medium-term expendi- ture framework (MTEF) enables the budget to be a management and policy tool of government because a better developed budget facilitates better bud- get execution. That is, if the budget is realistic and comprehensive, then the variables effecting budget deviation will be (a) unanticipated variations in revenues (donor and domestic revenues by large changes in economic activ- ity) and (b) weak execution systems. Ideally, one would want data on multi- year macroeconomic and public revenue projections for several budget cycles before and after introduction of an MTEF (as well as any concurrent independent economic forecast) to assess whether the MTEF had improved formulation of the fiscal envelope (that is, public revenue and expenditure). Absent such data, analysts generally assess actual variation between approved and actual budgets.3 At least as important as the quality of multiyear projections is the quality of participation in budget formulation. Line agencies should be an integral part of the process and budgeting, and good ownership of budgets requires 3. Although data are difficult to obtain, they are sometimes available in public expenditure reviews. Unfortunately, public expenditure reviews do not always report such data, and even when they are reported, techniques for reporting or mea- suring budget deviation are not standardized. Any given public expenditure review may report data by vote, administrative unit, functional classification, or another category. Bill Dorotinsky and Rob Floyd 187 more than a technician’s involvement. Active involvement of policy officials is crucial. The aim is to use budget formulation as a process for reaching political policy consensus at senior levels and for allocating resources to meet objectives. Good-practice MTEFs thus emphasize the starting point for the annual budget: a fiscal policy paper4 prepared early in the budget process, that embodies the multiyear macrofiscal framework and sets the overall macroeconomic and fiscal policy parameters for the budget. Educat- ing senior policy officials on economic and fiscal policy, and the impact of policy decisions, is thus one of the ministry of finance’s key roles. As chap- ter 5 of this book explores in detail, by reaching a consensus in the cabinet, all cabinet members better understand the constraints facing government and are more inclined to adhere to the policies. The consensus on fiscal pol- icy makes the task of budget preparation for individual ministries easier. The Practice of Budget Formulation in African Countries As figure 6.4 summarizes, the HIPC AAPs showed that 7 of 20 countries (35 percent) achieved four or more of the seven benchmarks and 13 (65 percent) met three or fewer. Furthermore, 92 percent of countries do not provide budget data consistent with the GFS definition of general government; 40 percent have substantial off-budget transactions; less than 30 percent have budget outturn quite close to the budget; more than 50 percent have both capital and current donor-funded expenditures included in the budget; and approximately 35 percent provide functional or program classifications (World Bank 2002a, p. 13). Thirty-five percent of African countries had administrative, economic, and functional classifications, and an additional 15 percent had program classification as well. In total, 50 percent met or exceeded the benchmark for broad tracking of pro-poor spending. Perfor- mance on budget formulation is thus mixed. We highlight first some bright spots, which signal what can be achieved, then turn to some of the more challenging results. One important gain was in the success of some countries in closing the budget deviation gap between budgets as formulated and as actually exe- cuted. Tanzania increased the ratio of executed to approved development budget from 14 to 57 percent from fiscal 1996 to fiscal 2001 and reduced the average deviation from approved budget by sector from 75 to 40 percent 4. Reflecting the fact that these processes have been treated as technical exer- cises, the fiscal policy paper is sometimes termed the “budget call circular,” a term that masks its function and import. 188 Public Expenditure Accountability in Africa: Progress, Lessons, and Challenges Figure 6.4. Budget Formulation: HIPC AAP Benchmarks Source: Authors. over the same period (World Bank 2001b). Similarly, Rwanda increased the executed to approved budget ratio for recurrent expenditures from 92 to 95 percent from fiscal 1998 to 2001, and increased the execution ratio for capital spending from 46 to 88 percent over the same period. In terms of sectoral allocations, the average deviation of actual to approved budget declined from an average of 11 percent in fiscal 1998 to 4 percent in 2001. Moreover, the improved execution to budget for education and social services was achieved in one year; between 1998 and 1999, the deviation was reduced from 20 to 3 percent for education, and from 16 to 6 percent for social ser- vices (World Bank 2002a). Two countries that emerged as overall solid performers on budget for- mulation are Tanzania and Uganda. The reasons for their strong perfor- mance can be found, in part, in broader domestic political processes and programs of public sector reform, which facilitated readiness for more robust budget processes. Additionally, changes in donor assistance in these countries likely also played a role in greater progress being made. Rather Bill Dorotinsky and Rob Floyd 189 than narrow, disparate technical assistance with midlevel finance officials, policy official buy-in and involvement may have added impetus. In Uganda, the president has been directly involved in supporting the approach and the policies incorporated into the budget. Policy officials sup- port the process, use the information to make decisions, and use the process to enforce choices, so budgeting is more than a technical accounting exer- cise. Furthermore, programmatic lending and analytical support linked directly to the government budget cycle have helped support rather than undermine the core government process. Turning to weaknesses in performance, a country may have scored poorly on budget formulation benchmarks for many reasons, but a lack of executive discipline is often the critical element. If budget execution is not aligned with the approved budget, the budget process cannot be taken seriously. (For example, the Malawi health sector was budgeted to receive 20.7 percent of development budget, but it received only 3.6 percent) (Le Houerou and Tal- iercio 2002, p. 26). Moreover, large deviations undermine the budget process, and adversely affect service delivery. More fundamentally, looking at the rel- atively weak areas in formulation—coverage and comprehensiveness, classi- fication of spending, and identification of priorities (pro-poor spending)—the budget formulation process itself is not institutionalized in many countries: the budget covers only part of government activity; budget information is not presented to facilitate policy and program choices or evaluation of impact; and key government policies (pro-poor) are not identified in the bud- get, reducing transparency and accountability. Hence, the budget process is not seen as the primary vehicle for making policy choices under resource constraints and for matching priorities and resources. The countervailing forces that continue to undermine central budget processes, or provide competing means of financing government programs, comprise another set of reasons for weaknesses in budget formulation and include donor funds, earmarked revenues, and off-budget funds (especially roads). Whereas MTEF’s have been established as separate, stand-alone processes, loosely connected to the main budget process, these forces pose a threat to the core process. Similarly, poverty reduction strategy papers may also weaken the budget process as well by providing a completely separate decisionmaking process. Finally, such activities as the Chad petroleum rev- enue oversight work have potential to either reinforce or undermine the government budget process, depending on how its structure and function evolve. These forces are all areas for further monitoring and evaluation. A final, more ambiguous, finding on trends in budget formulation con- cerns the growing use of MTEFs. At least 19 countries have MTEFs in some 190 Public Expenditure Accountability in Africa: Progress, Lessons, and Challenges stage of implementation: Benin, Burkina Faso, Chad, Ethiopia, Gabon, The Gambia, Ghana, Guinea, Kenya, Lesotho, Malawi, Mali, Mozambique, Namibia, Rwanda, South Africa, Tanzania, Uganda, and Zambia. All of these countries, except Namibia, receive some World Bank support on MTEF, and the reform was often proposed in public expenditure reviews or country assistance strategies (Gabon, Ghana, Kenya, Malawi, Rwanda, and Tanzania) (Le Houerou and Taliercio 2002, p. 8). Despite all this effort, a 1999 survey noted that only 21 percent of countries surveyed had an MTEF that projected aggregate expenditures over a three- to five-year horizon in a way that was consistent with macroeconomic targets. In the 2002 HIPC AAP exercise, only 17 percent had a medium-term perspective well integrated into their budget formulation process, although about half did provide some kind of medium-term forecast (World Bank 2002a, p. 13). An important part of the problem is that these MTEFs are almost never participatory. Only Uganda and Tanzania have significant line agency and civil society representation in the MTEF process. Generally involvement of both policy officials and line management in budget formulation is low. Because MTEFs are often not part of the budget, they are not always approved through the cabinet or parliament, and in some countries, for example, Mozambique, it remains only a technical document. Generally, budget formulation—and the MTEF in particular—continues to be treated as a technical exercise. Absent substantive, planned involvement of cabinet and policy officials throughout the process, budgets lack ownership from the senior levels. Budget formulation is treated as an accounting exercise rather than a process for reaching political consensus on policy and the allo- cation of resources to meet objectives. Some Emerging Lessons on Budget Formulation The HIPC action plans have included a variety of measures to achieve gains in the short-term in the quality of budget formulation. Among these mea- sures are broadening the coverage of government expenditures to include, in particular, donor-financed expenditures (Burkina Faso, The Gambia, Ghana, Guinea, Malawi, Mali, Mauritania, Tanzania, and Uganda) and upgrading classification systems (Cameroon, Guinea, Guyana, Niger, Mau- ritania, and Zambia). Other countries (Chad, Mauritania, and Mozambique) are focusing on incorporation of off-budget transactions into the budget. For still other countries, improving budget formulation is a medium-term objective, such as the introduction and formalization of MTEFs. As of this writing, Chad, Ethiopia, Mali, and Zambia planned to develop MTEFs in Bill Dorotinsky and Rob Floyd 191 2003, The Gambia aims for 2005, and Ghana is reactivating its MTEF. A number of countries plan to implement improved program, functional, and economic classifications across their budgets over the medium-term (for example, Burkina Faso, Cameroon, Mauritania, and Tanzania). The following lessons from ongoing operational experience and analyti- cal work can improve the likelihood of success of the efforts to improve budget formulation. • The process of ensuring that the budget entity meets GFS is more complex and time-consuming than expected. Although seen as pri- marily a technical issue, political economy incentives frequently keep certain categories of spending outside formal reporting chan- nels. The increasing role of subnational governments generally and their role in budget execution specifically make developing compre- hensive financial reporting systems more imperative. Furthermore because budget coverage is largely viewed as a technical issue, it is generally not addressed explicitly in lending operations, even dur- ing the development of an integrated financial management infor- mation system. • Closing the gap between the enacted budget and actual spending and establishing the reliability of budget as guide to outturn are possible in a relatively short time, as in Tanzania and Rwanda. • Economic and public finance forecasting competence and predictabil- ity of resources is of particular importance. Ensuring flow of funds predictability is about reducing the gap between forecast and actual revenues. Good forecasting also requires good costing, and the vari- ance between the quality and quantity of budget costings is generally high. Le Houerou and Taliercio (2002, p. 12) found no countries had uniform quality across all sectoral costings, and few countries pro- vided costings to the level of programs and activities—meaning that in reality, the MTEFs are usually only partially completed or used and the likelihood of predictability is hence reduced. Forecasting for bud- get formulation, whether for a traditional budget or a budget linked to MTEF, must be standardized, particularly regarding sectoral inputs. • Budget reforms, particularly the introduction of an MTEF often do not take into account the initial country conditions regarding budget management, in particular budget comprehensiveness, execution, and auditing. The failure to assess budget management systems and processes adequately leads to inadequate design and sequencing (Le Houerou and Taliercio 2002, p. 16). 192 Public Expenditure Accountability in Africa: Progress, Lessons, and Challenges • Many countries are piloting an MTEF that is only a partial MTEF. Le Houerou and Taliercio’s (2002) research assessed the design features of existing MTEFs in Africa and found that of nine MTEFs, four included only priority sectors and five included all sectors.5 Only those MTEFs covering the whole of government provide the neces- sary elements useful in controlling aggregate spending and fiscal deficit. The other finding was that countries diverge on the issue of whether capital and recurrent expenditures are included. Four of the cases included both, but the others either included only recurrent expenditures or separate capital budget, limiting budget coverage. • Senior political officials support for reforms and involvement in the budget process are important factors contributing to successful, sus- tainable reforms. • When MTEFs are being implemented, they need to be integrated into the budget. As of 2002, only five MTEFs in Africa were integrated in a meaningful way into the budget process. In most cases they operate parallel to the general budget process. Only four countries submitted MTEFs to both cabinet and parliament, and in some cases the MTEF remains only a ministry of finance technical document (Le Houerou and Taliercio 2002, p. 13). Where MTEFs are separate from the budget process, they are far less effective. A proper MTEF is the budget process. • Decentralization is posing new challenges to improved formulation. Some progress is being made in improving budget formulation at the national level, including building capacity. The trend toward decen- tralization of resources and responsibility for key sectors is sweeping the continent even before national gains can be consolidated. Capac- ity at the local level for financial management is even weaker than at national level, and few sustained efforts have been made at local capacity-building. Key challenges include how to retain improve- ments at the center while decentralization is under way, how to build new capacity at local levels, and how to align local budget with national budget formulation. Learning Lessons: Budget Execution Budget execution is the link between approved budget policies and actual resource flows. Although one cannot effectively implement a badly formu- lated budget, a well-formulated budget can be undermined through poor 5. Another discussion can follow from this difference: are sectoral MTEFs really MTEFs? Bill Dorotinsky and Rob Floyd 193 budget execution. Whether the budget is well or poorly formulated, improved budget execution can assure more regular and predictable resource flows to line agencies, enabling better service delivery. Operational experience and analytical work suggest the following pre- requisites for effective budget execution: • Well-formulated budgets based on good forecasts • Regular management oversight of budget execution and the execut- ing institutions • Functioning internal controls to avoid arrears, prevent unauthorized virements, and generally reduce the risk of waste, fraud, or abuse of public resources. • Timely and accurate data, including good records management and reconciliation of banking and fiscal data to ensure that accounts are accurate. The HIPC AAP exercise assessed four areas of budget execution: pay- ment arrears, the quality of the IA, the use of tracking surveys, and the qual- ity of fiscal and banking data reconciliation. As figure 6.5 summarizes, the 2002 HIPC AAP exercise noted that 10 countries met two or more of the benchmarks and 10 countries met fewer than two of the benchmarks. Figure 6.5. Budget Execution: HIPC Benchmarks Source: Authors. 194 Public Expenditure Accountability in Africa: Progress, Lessons, and Challenges Eighty-eight percent of countries have inactive or ineffective IA practices (World Bank 2002a, p. 13). For Africa, less than half (45 percent) met the benchmark for a small stock of arrears, compared to half for Latin America. Only 40 percent of African countries regularly reconciled banking and fiscal data; no Latin American countries undertake regular reconciliation. As short-term measures to redress budget execution weaknesses, some countries (Benin, Chad, and Tanzania) are introducing functionally based in-year budget reports. Eight countries (Chad, Ghana, Guinea, Mali, Mauri- tania, Mozambique, Tanzania, and Zambia) are also establishing bridging mechanisms to facilitate tracking of poverty reducing spending. A com- monly used mechanism has been the tagging of certain budget lines that meet or approximate the definition of poverty reducing spending in policy reduction strategy papers. These short-term measures, however, qualify more as temporary band-aids than as sustainable responses to the chal- lenges of improving budget execution. Sustainable gains depend on improvements in both financial management information systems and internal management control systems. Financial Management Information Systems Effective budget execution requires at a minimum: timely information sys- tems to monitor and manage accounts receivable and payable, a basic cash management system, and data on actual agency expenditures and cash bal- ances in government accounts, drawn from the general ledger (Hashim and Allan 1998, p. 13). At independence, these systems were all manual. By the mid-1990s, however, many African countries had computerized numerous aspects of public financial management. These systems were often stand- alone, home-grown ones that did not take full advantage of networking capabilities, improved technical design, or an integrated approach to expen- diture management. Currently, at least 15 countries have programs under way to computerize and systematize their financial management information further (Benin, Burkina, Cameroon, Chad, The Gambia, Ghana, Guinea, Madagascar, Mau- ritania, Malawi, Mali, Niger, Tanzania, Uganda and Zambia); the World Bank is supporting FMIS improvement in one half of African HIPC countries (World Bank 2002a). These FMIS upgrading programs generally include core modules for general ledger. In addition, an FMIS may include budget and warrant control, and together they constitute the government’s core account- ing system. Some programs are very broad and complex, including not only the accounting system, cash management system, commitment control, aid Bill Dorotinsky and Rob Floyd 195 and debt management, and payroll, but also procurement and inventory control and reporting functions (Heidenhof and others 2002). Heidenhof and others (2002) reviewed the FMIS experience in five African countries. All five FMISs were donor initiated, based on prior ana- lytical studies. Four of the five focused on budget preparation, and all focused on accounting, cash management, and commitment control. Three of the five provide an interface linking payroll control and expenditure con- trol systems, and three of the five had a complementary procurement inter- face. The systems range from the complex (Ghana) to the moderately complex (Burkina Faso and Malawi). As Heidenhof and others (2002) pointed out, FMIS complexity poses particular challenges in Africa. As early as 1998 Hashim and Allan (1998) noted that “it is only too common, particu- larly where substantial reforms to the underlying management process are required, for systems to be implemented piecemeal without significant reform and little attention paid to critical flows of information between sys- tem components. The end result is often a set of partial information systems with overlapping and/or conflicting functionality . . . ” (p. 4). Attention to the soft systems that surround and use the FMIS is essential for these investments to have any impact. Technical improvements empower government to manage its own affairs, but the process is built on the assumption that governments want to manage their own affairs well and that policy officials will reinforce the reforms by (a) demanding timely, accurate reports for decisionmaking, (b) monitor spending trends to approved budgets and policies, (c) take corrective action when spending is out of line with policies, (d) actively support internal auditors when con- flicts arise over jurisdiction, and (e) take corrective action when internal or management control weaknesses are identified. Reviews of experience con- clude, however, that all too often FMIS upgrading efforts have focused nar- rowly on technical improvements, neglecting these softer, but nonetheless crucial, considerations. Internal Management Control Internal management control systems comprise a series of procedures and arrangements designed to minimize the risk of fraud, waste, and abuse, as well as provide reasonable assurance that management’s objectives are being achieved and policy direction followed. Robust systems of internal control are vital for ensuring that resources are directed to their intended destination, rules and procedures are respected, records and reports are accurate, and a sound basis for decisionmaking and accountability is 196 Public Expenditure Accountability in Africa: Progress, Lessons, and Challenges assured. Countries with different government heritage organize internal controls differently. Anglophone-tradition systems tend to rely on internal controls imbedded in management practices, with a robust IA office (either central or within each major ministry) to check on the functioning of inter- nal controls ex post. Francophone-tradition countries tend to use central ex ante reviews of documents—an ex ante visa system—as the primary inter- nal control mechanism. In some francophone systems, the internal control may also go beyond checking documents for accuracy to monitor overall system performance, looking behind the documents to ensure the underly- ing contracting and financial systems are operating properly. IAs are a key part of the control system because an effective IA provides a vital check on the robustness of the components of internal control. Although few African countries have effective IA offices, many countries are supplementing ongoing IA work and strengthening IA departments. For example, in Benin the ex ante French internal control system is being rein- forced with supplemental audits by the inspector general of finance, as well as with the accelerated introduction of a computer-based expenditure man- agement system to automate some controls. Where IA is not yet fully established, action plans to complete IA intro- duction are generally in place. In the case of Burkina Faso, IA functions are carried out by the state inspection and treasury inspection offices only upon request; in Ethiopia and Ghana IA is not well developed; in Guinea and in São Tomé and Principe capacity is inadequate; and in Mauritania and Zam- bia IA does not work well because of a lack of resources. In yet a third set of countries such as in Cameroon, The Gambia, Mali, and Uganda, IA is inop- erative (World Bank 2002a). The World Bank is directly supporting IA devel- opment in Ethiopia, Ghana, Kenya, Malawi, Mozambique, and Tanzania. Both Kenya and Malawi reforms involve significant restructuring of IA departments, including reviewing program operations and developing new strategies and schemes of service for internal auditors. Capacity building is also being supported. In Malawi nearly 200 auditors are being trained. In Tanzania financial system automation and internal auditor training are being supported. Some Emerging Lessons on Budget Execution Lessons from operational experience for improving budget execution include the following: • In many countries, the role of IA and the strength of internal and man- agement controls have been neglected and nearly completely degraded. Bill Dorotinsky and Rob Floyd 197 Neither countries nor donors have paid sufficient attention to these components of expenditure management. Additional work is needed to provide tools and gather capacity-building experience. • If an FMIS is part of budget execution reform, it requires substantial reform of existing institutional arrangements and different processes and information flows among different elements of the system. Ghana and Uganda tried to do a comprehensive budget reform, which complicated FMIS implementation. Work under way in reviewing Bank-wide FMIS experience suggests that FMIS imbed- ded in too broad a reform program and stand-alone FMIS projects have higher failure rates (Dorotinsky 2003). An appropriate balance, such as FMIS investment as part of treasury strengthening, may be the best approach. Moreover, realistic expectations of what FMIS can and cannot do and adequate attention to the soft systems surround- ing the hardware (for example, records management, data entry, and comprehensiveness of tracking) may be needed before introduction of an FMIS. • When instituting systems to improve budget execution, size matters. Larger and more complex projects are more likely to fail than smaller, focused projects. Complex projects are difficult to manage, expensive, and not likely to produce quick results, and they have higher risk of failure. The focus cannot be too narrow either: building more than one of the institutions of financial accountability simultaneously is important, so these institutions can support and reinforce one another (for example, ministry of finance, budget office, internal auditors, control general, and the external audit office). In Chad, a more active budget office and Court of Accounts seem to be mutually reinforcing one another, and this synergy may yield a more sustain- able increase in system performance. • Design and implementation of budget execution systems are difficult. In nearly all countries, the timelines for implementation of FMIS have been extended (in Ghana from five to seven years, in Burkina Faso from four to seven years, and in Malawi from three to five years). The Bank-wide average was seven years; the Bank’s Africa Region had the highest regional average at nine years (Dorotinsky 2003). Cost esti- mates for FMIS vary greatly—from US$5 million to US$15 million for introduction costs. The British Department for International Develop- ment (2001, p. 60) estimates the cost of a core budgeting and account- ing system to be—US$10 million to US$20 million. Hashim and Allan (1998, p. 4) estimated the costs of an integrated financial management 198 Public Expenditure Accountability in Africa: Progress, Lessons, and Challenges information system at—US$10 million to US$50 million over a five- year period of implementation. An internal World Bank document indicates that the Bank-wide average was US$12 million. • Lessons of experience are not well translated across the region or across donors. Africa is experiencing a persistent emergence of cash budgeting (cash rationing) systems. The World Bank and other donor efforts to build capacity and strengthen treasury operations to sup- port poverty reduction as well as macrofiscal balance are urgently needed to avoid further missteps in cash rationing. Other persistent trends of concern include creating off-budget funds (for example, roads), pushing autonomous agency concepts in low accountability environments, and advancing more cutting-edge reforms where the basic institutions are weak. Learning Lessons: Budget Reporting Budget reporting, usually treated as part of budget execution, warrants spe- cial treatment as a core element of transparency, empowerment, and accountability. Without timely, accurate in-year reporting, public spending cannot be managed. Without timely, accurate ex post reporting, accountabil- ity for results is compromised. Operational experience and analytical work suggest the following pre- requisites for effective budget reporting: • Accurate public accounts are essential. The HIPC AAP process deter- mined that 71 percent of countries do not close accounts within two months of the end of the fiscal year. Likewise the HIPC AAP process and country financial accountability assessments note the absence of credible improvements in accountability. • Comprehensive accounts are equally important. All African countries maintain off-budget items, including extrabudgetary funds, military expenditures, or resources provided directly to the head of state or government. In addition, many African countries maintain secrecy on intelligence spending and security-related infrastructure, foreign affairs, and research expenditures. In many of these cases, such accounts are never audited and have no parliamentary oversight. • A clear legal framework for supreme audit institutions (SAIs) is important. An SAI must have independence from the executive and operate in a well-defined and transparent environment. This issue is more pressing in Westminster systems, where auditors general may report directly to the minister of finance. Bill Dorotinsky and Rob Floyd 199 • Better understanding of accountability arrangements are critical for well-designed reforms. A better understanding of the institutional and governance arrangements in the country, and appropriate roles and responsibilities in the system, are essential for designing and implementing public expenditure management reforms. Assigning roles to institutions that generally are not designed to carry the responsibility may not be a productive exercise in social engineering. The HIPC expenditure tracking assessment focused on four aspects of budget reporting: timeliness of internal reports, classification in reports, timeliness of accounts closure, and timeliness of final audited accounts. As figure 6.6 summarizes, the 2002 HIPC AAP exercise found that only 8 of 20 African countries met at least two of four benchmarks. Twelve countries met only one benchmark, and three countries (not shown on figure 6.3) did not meet any benchmarks for budget reporting. Of the African countries surveyed, 86 percent did not close their accounts within two months of the end of the fiscal year, and 90 percent did not have final audited accounts presented within 12 months of the end of the fiscal year. That is, in 90 per- cent of the cases, parliament will not have a government financial statement within an adequate time frame. In Westminster public expenditure systems, reporting is the responsi- bility of the accountant general, who maintains government financial data. Figure 6.6. Budget Reporting: HIPC Benchmarks Source: Author. 200 Public Expenditure Accountability in Africa: Progress, Lessons, and Challenges The HIPC exercise identified significant data problems: only 25 percent of countries provide monthly expenditure reports within four weeks of the end of the month; and 88 percent cannot produce timely, functionally based expenditure reports from core accounting data, although about one- third do provide some form of functionally based in-year reporting (World Bank 2002a, p. 13). This lack of internal reporting on budget execution lim- its the government’s ability to manage finances and contributes to the inability to finalize and close accounts. Likewise, this lack of internal reporting directly affects the role of SAIs, which audit both government revenue and expenditures. Supreme Audit Institutions SAIs have two basic models in Africa: Westminster and francophone. In the Westminster system the SAI reports directly to parliament. Audits in Westminster-type systems focus on the financial statements and operations of government. In the francophone tradition the SAI is a cour des comptes, having judicial and administrative authority and independence from both the legislative and executive branches of government. Judgments often focus on government compliance with laws and regulations (World Bank 2001a). A 1998 assessment of African SAIs found that of 25 surveyed, 11 had no qualified professionals, and for 9 determining whether staff were qualified was impossible. Furthermore 24 of the 25 SAIs did not have satisfactory independence from the executive and only one (South Africa) had adequate external quality control. One-fifth submitted reports on a timely basis, but only one country made such reports public. Finally, no country surveyed exhibited satisfactory SAI follow-up (Mbanefo 1998, p. 9). At that time nearly all SAIs depended on ministries of finance for operating budgets, and SAI staff were generally civil servants. In most francophone countries the supreme audit institution reported directly to the prime minister or president, not parliament. A 1999 World Bank, Africa Region, report noted that “disparities between budget allocation and spent amounts are frequent in client countries. Auditing procedures have been instituted, but timing delays render them useless as a disciplining tool” (Kostopoulos 1999, p. 21). Line agency financial accounts were prepared more than one year after the fiscal year ended in nearly half of the cases. Kostopoulos (1999, p. 24) also noted that there were “almost always no sanctions against delays or dis- crepancies. Accounts are usually not made public,” and audits are under- taken 1–3 years after the end of the fiscal year. Financial accounts of line agencies were prepared within six months from the end of the fiscal year in Bill Dorotinsky and Rob Floyd 201 only 9 percent of countries; within one year or more in 45 percent. Only 6 percent of countries levied sanctions for delay, and only 22 percent of coun- tries made accounts public. Likewise line agency accounts were audited only 35 percent of the time; audits took place 6–12 months after end of the fiscal year in 14 percent and more than one year in 86 percent. Audits were made public in only 8 percent of cases (Kostopoulos 1999, p. 45). The 2002 HIPC expenditure tracking assessments found that 71 percent of African countries do not close accounts within two months of the end of the fiscal year. Although countries may have some legitimate reason to hold accounts open after the end of the fiscal year to record expenditures that occurred during the fiscal year, this grace period is frequently subject to abuse. Current year expenses can be inappropriately booked to prior years, raising the prior year ’s apparent deficits and artificially lowering the cur- rent deficits. The 2002 HIPC exercise also found that 83 percent of African countries do not have final audited accounts presented within 12 months of the end of the year. Without credible data, SAIs cannot conduct audits and accountability mechanisms cannot work. Burkina Faso has struggled for some time with an abnormal and unexplained lack of balances for the nomi- nal accounts of government; in Ethiopia and The Gambia the accounting data have been declared unsatisfactory for processing by the auditors gen- eral; in 1994 only 13 of 33 financial statements to Sierra Leone’s Auditor General were in usable form; and the quality of data has been rated low in many countries (for example, Kenya, Malawi, and Zambia). Efforts to improve the performance of SAIs have focused on upgrading their regulatory frameworks and their internal capacities. The regulatory frameworks focus primarily on the independence of the SAI; oversight institutions for the SAI (boards, commissions, and committees); conditions of service for SAI staff; scope of audits (generally expanding to include per- formance or value for money); and the ability to assess audit fees. Ghana passed an Audit Service Law in 1999, which enhanced SAI independence. Malawi has drafted a new Audit Act that will provide for auditor general independence and establish an oversight body, but does not adequately address conditions of service. Kenya is streamlining a National Audit Office with an independent budget process and the power to impose some fees. Rwanda established the Auditor General’s Office in 1998 and has carried out fiscal audits. Senegal created a cour des comptes in 1999 which is indepen- dent and active, and Cameroon is beginning the process to establish a cour des comptes. Nigeria’s Constitution of 1999 required the auditor general to audit accounts and report to the national assembly, but did little to address issue of independence and conditions of service. 202 Public Expenditure Accountability in Africa: Progress, Lessons, and Challenges Although clear framework laws can be a valuable starting point, their impact depends on SAIs having the internal capacity to make use of them. The capacity constraints facing SAIs are particularly daunting—in nearly all countries the capacities are too weak to fulfill the legal mandate. For exam- ple, in Ghana of 1,450 staff, only one has formal qualifications; in Sierra Leone the Office of Auditor General has only five staff; Mozambique has filled only 101 of 335 posts in the Administrative Tribunal, and only eight staff members have formal qualifications; Uganda’s Office of the Auditor General has only eight qualified professionals from a cadre of 296 staff; and Malawi’s Office of the Auditor General has only six qualified accountants of a staff totaling 170 people. Many donors, including the World Bank, are providing support to help build the capacity of SAIs, for example, Ghana receives assistance from the European Union and the United Kingdom’s Department for International Development, Guinea from France, Malawi from the Swedish International Development Authority, Mali from the European Union, Uganda from the United Kingdom’s Department for International Development, and Zambia from Norway. The World Bank also supports SAIs, directly in Ghana, Kenya, Malawi, and Zambia and indirectly in many more countries through operations that support data provision and reporting in ministries, depart- ments, and agencies and in the accountant general offices and through sup- port for integrated financial management systems that enhance the local accounting professions and strengthen regulatory practices. In building SAI capacity, a continuing controversy is whether to encour- age African institutions to use the latest technology and to adopt the audit approaches of developed countries. Even if technology is added, skilled staff are still required to enter the data and maintain the records, and trained auditors are needed to audit accounts and detect errors or fraud. As to the approach to auditing, in some cases, donors have promoted value for money auditing, which represents the leading edge practice in developed countries. For example, the Mozambique Administrative Tribunal has few trained professionals, lacks resources, and depends on the Ministry of Plan- ning for its budget. At the same time, donor interventions emphasize value for money, performance, and forensic auditing. Perhaps even more funda- mental than skills, however, is the prevailing institutional culture. In many African countries, informal practices, rather than compliance with formal rules, often are the norm for public expenditure management. In such set- tings, traditional compliance auditing could yield greater benefits for the short to medium term, with value for money auditing sequenced for a later stage. Bill Dorotinsky and Rob Floyd 203 Parliamentary Oversight Parliament is a key formal institutional player in the expenditure account- ability cycle.6 Chapter 7 focuses in depth on the role of parliament in four countries. Here we provide an overview of broader, Africa-wide patterns, with a particular focus on the role of parliaments in relation to expenditure accountability. All Sub-Saharan African countries have national legislatures or parlia- ments, though with varying degrees of power and voice. Since the early 1980s more than 40 African countries have held national elections, and many have vibrant parliaments. Although the oversight role of Africa’s par- liaments in the expenditure accountability cycle potentially is a large one, in practice it remains limited. Parliamentary budget and finance committees are not active in budget preparation and are often not provided with finan- cial resources nor basic budget information. Public accounts committees (a) are generally charged with considering the financial statements of all central government institutions and executive institutions, (b) are expected to address both the financial statements and the auditor general’s reports, and (c) are then expected to report their findings to the full parliament. Chapter 7 explores in depth the reasons for the uneven performance of parliaments; four reasons of special relevance to their expenditure oversight role are noted here. First, in almost no African country do parliaments have the capacity or resources to carry out these activities. For example, the Par- liamentary Accounts Committee in Uganda, a standing committee (and highly regarded as among the strongest parliamentary committees on the continent), has only a secretary and an intern as staff. In some years the Malawi Parliamentary Accounts Committee holds no meetings, because no resources are allocated for this purpose. Second, a preoccupation with secrecy undercuts the potential for parliaments to play an effective expendi- ture accountability role. By contrast to the practice in developed democra- cies, most public accounts committees in Africa hold meetings in closed chambers, reducing the impact of their discussion and hindering media involvement. Third, perhaps in part because of the circumscribed media role, the presumed executive and judicial response to parliamentary find- ings often is not forthcoming. In Burkina Faso the Finance Commission reviewed the cour des comptes report, and the commission’s report was pre- sented to the National Assembly in the form of a bill, but it was never 6. This chapter does not address the role of the media or civil society beyond parliament. 204 Public Expenditure Accountability in Africa: Progress, Lessons, and Challenges passed because of insufficient interest, capacity, and resources. In Kenya the Public Accounts Committee makes recommendations for disciplinary actions, but these recommendations are not implemented by the executive; in Uganda, the Public Accounts Committee can make recommendations for censure, but these recommendations are not binding. Fourth, the effective- ness of parliament as an oversight institution depends in part on the inter- play in parliament of political parties. A key accountability feature of some parliamentary public accounts committees is that a member of the opposi- tion party chairs the public accounts committee, assuring rival party access to public finance information. This mechanism can work only where the opposition party not only is electorally represented, but is sufficiently orga- nized to disseminate the information among its ranks, interpret the data, and form a position. Given the central role of parliaments in expenditure accountability and in the democratic process more broadly—and the increasingly evident mag- nitude of the challenges to be met if they are to play this role effectively— support is growing from Africa’s development partners to help build the capacity of the continent’s parliaments. During the 1990s donors provided more than US$200 million in support to legislative assistance projects (World Bank 2002b). The World Bank has been somewhat cautious about fully engaging parliamentary bodies because they are, by nature, political. Only the parliaments in Ghana, Malawi, Mozambique, and Zambia are recipients of direct project support. The World Bank Institute has a major program to help build parliamentary capacity; the focus of this program has shifted away from training individual parliamentarians toward strengthen- ing parliament as a key institution of governance and accountability. The World Bank Institute provides training to public accounts committees, strengthening parliaments’ capacity to participate in the budget cycle. Some Emerging Lessons on Budget Reporting Operational experience points to the following lessons. • Capacity constraints are often most acute in budget reporting. Even when the roles and responsibilities are clearly defined and appropri- ate, organizational capacity (human, managerial, and technological) is a major constraint to effective oversight. Shifting some attention from diagnosis or drafting laws to capacity development may yield more sustainable results. • A comprehensive, long-term approach to capacity building is needed for SAIs. Capacity constraints are not new, but neither are they going Bill Dorotinsky and Rob Floyd 205 away. Staff retention is low, creating the need for continual training; expenditure accountability issues are now given higher profile, thus making the role of the SAI higher profile; numerous SAIs have been strengthened with improved regulatory frameworks; and new and more complex auditing systems are being adopted. Nearly all HIPC AAPs and country financial accountability assessments note the need for significantly more SAI capacity including staff, equipment, soft- ware, and motor vehicles. As noted in the last section, SAIs are receiv- ing donor support, but possibly not in a manner and form that best meets the needs of SAIs. Donors are not generally supporting contin- uous local training of audit staff; operating expenses related to capac- ity building; or equipment, software, and motor vehicles in a long-term and comprehensive manner. • Conditions of service must be improved for SAIs. In many cases, capacity building activities have been under way for many years, and yet have little impact because of staff turnover. Although wage policy issues are of interest across government, two factors make them par- ticularly important for SAIs. First, SAIs are generally small organiza- tions in comparison to government and are generally detached from government. Upward mobility is more limited. Second, SAIs can, because of their autonomy, institute their own conditions of service and hire at market rates without undue impact on other sectors of the civil service. Standardized civil service pay rate systems may break- down for highly skilled specialists fields, such as auditors and infor- mation technology specialists. • Improved interaction is needed between SAIs and other institutions, particularly spending units, the ministry of finance, and parliament. A chronic problem facing African governments is the lack of follow- up to audit reports and inquiries. In many cases, audit inquiries are forever logged as “under correspondence” and never resolved. The issues raised in audit reports are not adequately addressed by parlia- ment or rectified by the government. Political realities, capacity for follow-up, elapsed time, and obstruction are partially to blame, but so is the weak relationship between SAIs and these institutions. • Committee departments in parliament can support well-functioning committees. Design issues are informed by two projects that currently support public accounts committees, the Ghana project and the Malawi Financial Management, Transparency, and Accountability Project. Both projects focus heavily on establishing a committee department through which technical, advisory, and research support 206 Public Expenditure Accountability in Africa: Progress, Lessons, and Challenges is provided to all committees in a coordinated and comprehensive fashion. Without a committee department (or similar organizational structure) support to committees is limited and subject to reallocation. Conclusions Well before the introduction of the HIPC AAP exercise, expenditure accountability had been the subject of a great deal of analytical and opera- tional work, and much remains to be done beyond what is being addressed by the exercise. The HIPC AAP excludes from consideration a variety of key areas related to expenditure accountability, including the following: • Fiscal decentralization • Procurement reform in the context of expenditure management • Regulatory frameworks to support effective financial management reform • Specific institutions’ roles, particularly ministries of finance in rela- tion to executive institutions (such as the office of the president, the office of the prime minister, or the cabinet) • Performance management and performance audit • Participation in the budget process in Africa • Political economy in terms of expenditure accountability reforms. The challenges posed by decentralization are of particular importance, because many African countries are rapidly implementing decentralization policies or have recently adopted decentralization plans that have poten- tially significant impacts on fiscal transfer systems, community-level spend- ing, revenue generation, and pro-poverty allocations. Chapters 9 and 10 of this volume explore the challenges of decentralization in more depth. Yet for all of its limitations in providing a limited and consistent set of key indicators, common across countries, the HIPC AAP exercise is a valu- able addition. It represents a key element of a performance framework, against which country progress can be measured as well as the effect of reforms or donor support. The initial results are sobering. The state of pub- lic expenditure management systems in many countries emerged in the AAP as poor and as having weaknesses in some very basic elements of these systems. Moreover, the reform of expenditure accountability systems is turning out to be more difficult than expected. A review of project docu- mentation from the early and mid 1990s revealed an expectation of quick wins and rapid development for fiscal governance, particularly in countries that at one time had robust treasury and accounting systems. Project results Bill Dorotinsky and Rob Floyd 207 and the HIPC AAP exercise point to longer time lines and slower implemen- tation. Apart from tepid political commitment in some countries, the com- plexity of numerous initiatives (for example, an integrated financial management information system, an MTEF, activity-based budgeting, and performance management) quickly drains available capacity and slows all reforms. Although the challenge of strengthening expenditure accountability sys- tems is more daunting than it initially seemed, the intensified focus on the area is beginning to yield some potentially important gains. Three in partic- ular are noteworthy. First, some early signs indicate that the HIPC AAP exercise is generating a response along the lines predicted by the managerial dictum that organi- zations give attention to what is measured. The HIPC assessment defines a core set of indicators and provides a basis for monitoring performance over time. The focus is on the basics of expenditure management, and in compar- ing lending documents and analytical work conducted before and after the HIPC AAP exercise was introduced, a trend back toward basic reforms is evident. Furthermore, despite widespread resistance to producing figures (of the kind reproduced in this chapter) that display the number of HIPC expenditure tracking benchmarks met by each country, such figures can be an effective tool for transparency and motivation. Indeed, the 2003 AAP update found that those countries scoring lowest in terms of number of benchmarks met in the 2001 AAP were among those making the most progress in implementing measures to strengthen their systems. Second, the focus on expenditure accountability is beginning to help address the ubiquitous problem of reform overload. In many countries, given the breadth of dysfunction, and the interdependence of management systems, reformers are tempted to try comprehensive public sector reforms. A recent World Bank strategy document concluded, though, that “experi- ence has shown that a broad frontal approach, i.e. broad and comprehensive program of public sector reform simultaneously along many fronts—may not be the most effective strategy in some setting if it meets too much politi- cal resistance or encounters too many institutional, technical, or capacity related barriers” (World Bank 2000, p. 11). The HIPC AAPs have reinforced this conclusion. They often pointed to overly ambitious agendas within too narrow a time frame. How to narrow the reform load in expenditure accountability is best discussed in the country context, within the frame- work of ongoing reforms, with available financing in mind, and within a reasonable time frame. What is becoming clear is that public expenditure accountability is emerging as a key entry point to reforms in many African 208 Public Expenditure Accountability in Africa: Progress, Lessons, and Challenges countries and may serve as a prerequisite for broader public sector reforms such as decentralization or civil service reform (World Bank 2000). Three major public sector reform projects in Africa supported by the World Bank—in Ghana, Malawi, and Zambia—have been designed or retooled to focus largely or solely on expenditure accountability as a stepping stone or backdrop for deeper reforms. Third, as noted in the earlier discussion of budget formulation, part of the reason for reform overload was the multiplicity of distinct, often discon- nected, initiatives on the part of donors. The HIPC AAP appears to be help- ing to make inroads in addressing this problem. Most immediately, the HIPC AAP process has enhanced collaboration between World Bank and International Monetary Fund teams, as well as governments and other donors, through the need to come to an agreement on the assessment of country performance in managing public expenditures, an action plan, and surveys of donor support (and gaps) in support of the action plans. By com- ing to agreed-upon assessments and plans, the process heightened the con- sistency of follow-up and support. More broadly, some of the countries scoring best in the HIPC assessment were those where the World Bank and other development partners have adopted more innovative approaches to their engagement. One example is the coordinated approach in Chad to strengthen the expenditure control systems surrounding oil revenues. Uganda and Tanzania offer more far-reaching examples: analytical work by the World Bank (mainly public expenditure reviews) is annual and directly in support of the country’s own budget development process; many donors provide aid in the form of budget supplementation. This assistance is pre- cisely the type of ongoing, supportive role that can help strengthen national management processes. The increased international attention to expenditure accountability res- onates powerfully with the growing domestic expectations within African countries for the transparency and accountability that are accompanying that continent’s democratization. Taken together, the combination of democratization and international attention offers hope that—if the current focus can be sustained—over the course of a decade, many countries could achieve (incrementally, step by step) substantial gains in the quality of their expenditure accountability systems. References Department for International Development. 2001. Understanding and Reforming Pub- lic Expenditure Management, Guidelines for the Department for International Develop- ment. London. Bill Dorotinsky and Rob Floyd 209 Dorotinsky, Bill. 2003. “Implementing Financial Management Information System Projects: The World Bank Experience—Preliminary Results.” PowerPoint pre- sentation, World Bank Budget Execution Course, April 8–9, Washington, D.C. Hashim, Ali, and Bill Allan. 1998. Information Systems for Government Fiscal Management. Washington, D.C.: World Bank. http://www1.worldbank.org/publicsector/pe/ JanSeminar/Course%20Readings/04.Treasury%20and%20FMIS/gfmap35.rtf. Heidenhof, Guenter, Helene Grandvoinnet, Daryoush Kianpour, and Bobak Rezian. 2002. “Design and Implementation of Financial Management Systems: An African Perspective.” Africa Region Working Paper Series no. 25. World Bank, Washington, D.C. Kostopoulos, Christos. 1999. “Progress in Public Expenditure Management in Africa: Evidence from World Bank Surveys.” Africa Region Working Paper Series no. 1. World Bank, Washington, D.C. Le Houerou, Philippe, and Robert Taliercio. 2002. Medium-Term Expenditure Frame- works: From Concept to Practice. Preliminary Lessons from Africa. Africa Region Working Paper no. 28. Washington, D.C.: World Bank. Mbanefo, Uche. 1998. Strengthening African Supreme Audit Institutions—A Strategy Paper and Action Plan. Washington, D.C.: World Bank. World Bank. n.d. Africa: A Regional Perspective. World Bank Brief. Africa Region. Washington, D.C. _____. 2000. Reforming Public Institutions and Strengthening Governance. A World Bank Strategy. Washington, D.C. _____. 2001a. Features and Functions of Supreme Audit Institutions. PREM Notes no. 59. Washington, D.C. _____. 2001b. Tanzania Public Expenditure Review. Report no. 2381-TA. Washington, D.C. _____. 2001c. Public Expenditure Management and Accountability: Evolution and Current Status of World Bank Work. Washington, D.C.: Poverty Reduction and Economic Management Network and Operation Policy and Country Services Network. http://siteresources.worldbank.org/IDA/Resources/Seminar%20PDFs/ evolution.pdf. _____. 2002a. Actions to Strengthen the Tracking of Poverty-Reducing Public Spending in Heavily Indebted Poor Countries (HIPC). Washington, D.C. _____. 2002b. Strengthening Oversight by Legislatures. PREM Notes no. 74. Washing- ton, D.C.: World Bank. _____. 2003a. “Rwanda Public Expenditure Reform for Sustaining Poverty Reduc- tion: A Public Expenditure Management Review.” Africa Region, Poverty Reduction and Economic Management, Washington, D.C. Draft. _____. 2003b. Update on Implementation of Action Plans to Strengthen Capacity of HIPCs to Track Poverty-Reducing Public Spending. Washington, D.C. 7 Emerging Legislatures: Institutions of Horizontal Accountability Joel D. Barkan, Ladipo Ademolekun, and Yongmei Zhou with Mouftaou Laleye and Njuguna Ng’ethe A decade after the resumption of multiparty politics across Africa, the legis- lature is emerging as an institution to be reckoned with in some countries. Once a rubber stamp of the executive, or nonexistent during periods of mili- tary rule or civil war, these bodies have begun to assert their independence as players in the policymaking process, as watchdogs of the executive, and as organizations that respond to demands by civil society. Put differently, they are becoming institutions of horizontal accountability—institutions of the state or of quasi state institutions that both check each other and are mutually supportive of each other to make government work. Although most African legislatures remain weak in relation to the execu- tive, they are arguably more powerful today than at any time since indepen- dence. That power is likely to increase as some legislatures across the continent strive to enhance their authority and independence from the exec- utive branch. The extent of legislative power, however, varies from country to country. In some countries, the legislature remains very weak despite the return to multiparty politics and the holding of regular legislative elections. In others, the legislature has asserted itself forcefully, both as a check on the executive branch and as a player in the making of public policy and policy implementation. In still others, the reality lies somewhere in between—the legislature remains weak, but it has begun to aspire to a more significant role. Where the legislature has begun to emerge as an independent branch of government, its impact on the nature of governance could be profound. 211 212 Emerging Legislatures: Institutions of Horizontal Accountability However, our knowledge of how and why legislatures do so is limited. Lit- tle is known about the legislative process in Africa. Indeed the study reported in this chapter is the first comparative analysis of the legislative process on the continent. Approach and Method The paucity of literature on legislatures in developing countries, and espe- cially in Africa, defined our study as a benchmark effort.1 Only after gaining an initial understanding of the legislative process can we hypothesize with confidence about why African legislatures operate in the manner that they do and how they are likely to develop in the future. Three broad areas of inquiry guided this study. • In what ways have legislatures evolved across Africa since the return to multiparty politics in the early 1990s? Given the succession of multi- party elections that occurred in most countries, how were the composi- tion and internal organization of legislatures and their relationships with the executive branch and with civil society affected? • What is the role and significance of African legislatures in terms of five basic functions associated with these bodies, namely: policymak- ing, oversight of the executive, representation, constituency service, and political recruitment? • What incentives, both within the legislature and outside it, structure the behavior of individual legislators and the evolution of the bodies of which they are members? What is the significance of these incentives in terms of the legislature’s impact on the policymaking process and its oversight of the executive branch? Can these incentives be reconfig- ured to enhance legislative performance? If so, how and by whom? 1. The academic and comparative literature on legislatures around the world, and especially on legislatures in developing countries and emerging democracies, is surprisingly thin. In their review of the state of legislative studies for the state of the discipline volume, Gamm and Huber (2004) reported that 94 of the 110 articles on legislatures that appeared in the American Political Science Review between 1993 and 2001 and in the American Journal of Political Science and Journal of Politics between 1996 and 2001 were limited to the American experience. Although most articles in these journals are written by and for American political scientists, the paucity of lit- erature on non-American legislatures is replicated in more specialized journals. Between 1995 and 2002, only 22 of the 87 articles published by Legislative Studies Quarterly dealt with the legislative experience outside the United States; none dealt with legislatures in Africa. Joel D. Barkan, Ladipo Ademolekun, and Yongmei Zhou 213 We sought the answers to these questions through a series of four field studies of the legislature experience in Benin, Ghana, Kenya, and Senegal.2 The studies were conducted between April and July 2002 and consisted of interviews with 50 to 75 individuals in each country. We carried out inter- views with five groups to obtain multiple perspectives of the legislative process: (a) a representative (though not a random) sample of approxi- mately 25 members of the national legislature;3 (b) senior professional staff (that is, clerks and deputy clerks) and presiding officers (that is, the speaker or president) of the national legislature; (c) government ministers or senior civil servants, or both, in the executive branch; (d) leaders of civil society organizations; and (e) a small sample of local elites (for example, leading businesspersons, church leaders, and heads of secondary schools) in each of three to six parliamentary constituencies. The interviews consisted of a series of open-ended questions that were grouped into a list of topics appro- priate for each sample. We chose the four countries that served as sites for this study on the basis of several assumptions that suggested that the legislatures in these countries had expanded their authority more than legislatures elsewhere. Where multiparty elections had been held on a regular basis, and where an alternation of power or an intense contestation for power had occurred, we assumed that the development of the legislature was most extensive and therefore a good candidate for inclusion. Because of the different constitu- tional traditions that structure the role of the legislature in anglophone and francophone Africa, the sample included two legislatures in each context. These assumptions were only partially validated by our research. We found a greater variation in both the capacity and authority across the four legislatures studied than anticipated. This variation yielded several expla- nations of legislative performance that would not have been apparent had all four legislatures been equally powerful as initially assumed. Put differ- ently, our faulty assumptions resulted in a greater variation of the depen- dent variable (that is, the emerging authority of the legislature) than 2. The field study in Benin was conducted by Ladipo Adamolekun and Mouf- taou Laleye. The field study in Ghana was conducted by Yongmei Zhou assisted by Sully Gariba. The field study in Kenya was conducted by Joel Barkan and Njuguna Ng’ethe. The field study in Senegal was conducted by Melissa Thomas of the Uni- versity of Maryland and Oumar Sissikhou of the University of Dakar. Joel Barkan served as the overall coordinator of the study, which was funded by the World Bank. 3. The samples were drawn on a quota basis to reflect the composition of the legislature in respect to party, regional distribution of members, and gender. In addi- tion, we sought interviews from the chairs of the finance committees, and from the committees on agriculture, education, and health. 214 Emerging Legislatures: Institutions of Horizontal Accountability anticipated and thus provided an opportunity for identifying the key inde- pendent variables that explain this variation. African Legislatures in the 1990s The evolution of African legislatures since the early 1990s is largely the result of the broader process of political liberalization and democratization that swept the African continent. Without a significant measure of liberal- ization and the return of multiparty politics, the legislature would still be nonexistent or a rubber stamp of the executive in most countries. That said, not all countries that experienced a significant measure of liberalization and democratization have seen their legislatures develop into institutions that play a greater role in the making of public policy or oversight of its imple- mentation.4 Although 43 of 47 African countries have held multiparty elec- tions and liberalized their politics, only 15 to 20 are regarded as “aspiring” or “consolidated” democracies (Barkan 2002; Freedom House 2002; van de Walle 2002). An equal number fall into the gray area of semidemocratic or semiauthoritarian regimes—polities that share elements of both democratic and authoritarian regimes and that are likely to remain so (Carothers 2002; Ottaway 2003). The four countries included in this study are examples of this variation. Over the last decade, and especially since 1998, the Kenya National Assem- bly has established itself as an independent branch of government and taken several bold steps to enhance its power in relation to the executive branch. By contrast, the National Assembly in Senegal remains weak, notwithstand- ing the election of a new president and new majority party in the legislature. In Benin and Ghana, where an alternation of government has also occurred, the extent to which the national assemblies play an independent and signifi- cant role in the governmental process, falls somewhere in between. Table 7.1 lists the four legislatures in order of increasing authority. We array these legislatures on an ordinal basis with respect to three measures of legislative authority: (a) our summary subjective assessment of the 4. We must distinguish between political liberalization and democratization at the outset of this study because they are not the same nor considered as such in the literature. By liberalization, we refer to the absence of human rights abuses, the emergence of civil society and an independent press, and in some instances, the holding of multiparty elections. By democratization, we broadly mean that the gov- ernment of a country is accountable to the governed and that the latter have the means and power to replace the government of the day periodically to ensure accountability. See also Dahl (1989, p. 221). Joel D. Barkan, Ladipo Ademolekun, and Yongmei Zhou 215 Table 7.1. The Emerging Power of the Legislature Measure of legislative authority Senegal Benin Ghana Kenya Independence and strength of the Moderate- legislature Weak Emerging Emerging significant Constitutional role Nominally Semi- Semi- Fully independent independent independent independent Ability to amend or modify budgets Rarely Occasionally Occasionally Occasionally Source: Authors. independence and strength of the legislature in relation to the executive branch, (b) the formal powers accorded to the legislature by the country’s constitution, and (c) our findings on the extent to which the legislature as a corporate body affects the policymaking process, including the budgetary process. These three measures of legislative authority are at best ordinal scales for what we will treat as our dependent variable. The processes of political liberalization and democratization in the four countries not only share certain features, but also differences that partially explain why the legislature has developed more in some countries than in others. The decade of the 1990s was a decade of increasing political liberaliza- tion across Africa in states not torn by civil war. In Benin, Ghana, and Kenya, the process of liberalization began in 1990–91 when the then authoritarian leaders of the three countries—Mathieu Kerekou in Benin, Jerry Rawlings in Ghana, and Daniel arap Moi in Kenya—reluctantly consented to holding multiparty elections after a hiatus of many years.5 They did so in response to a combination of domestic political forces and the changing international cli- mate following the end of the Cold War, especially the former.6 In Senegal, the process was more protracted having begun much earlier in 1974. The process unfolded somewhat differently in each country. In Benin, President Kerekou had presided over an Afro-Marxist regime that led to 5. Prior to the 1990s, multiparty elections were last held in Benin in 1969, in Ghana in 1982, and in Kenya in 1969. 6. The best summary discussion of the democratic opening in Africa is Democ- ratic Experiments in Africa by Bratton and van de Walle (1997). Using cross-sectional data from more than 40 countries, they demonstrate that the process was largely dri- ven by domestic rather than international forces. 216 Emerging Legislatures: Institutions of Horizontal Accountability economic decline and a curtailment of civil liberties. He was ultimately forced by civil society to hold a sovereign national conference in February 1990 at which the modalities for a new political framework were negotiated. The conference was followed by the promulgation of a new constitution in December 1990 and by presidential and parliamentary elections held in 1991. These elections resulted in Kerekou’s defeat and his replacement by Nicéphore Soglo. The opposition also gained control of the legislature. Mathieu Kerekou regained the presidency in 1996, however, and won again in 2001.7 Benin thus passed what Huntington (1991) has identified as the lit- mus test for a successful transition to democracy—a double alternation of power in which the incumbent government is defeated in a free and fair election and the party that defeated the incumbent government is itself sub- sequently defeated—in this case, by the former authoritarian ruler, Mathieu Kerekou. Presidential elections are held every five years in Benin. Parliamentary elections are held every four. Since the return of multiparty politics, three presidential elections and four elections for the National Assembly have been held. These elections have been marked by a splintering of the vote among the candidates from more than a half dozen weak political parties. The result is that no single party has ever controlled a majority of seats in the National Assembly, and the coalition of parties that support or oppose the incumbent president constitutes the majority at any one time has shifted. This fluid situation has retarded the development of the National Assembly as an independent corporate body and political force. At the time of the field work for our study in April and May 2002, a coalition of opposi- tion parties controlled a modest majority of seats in the National Assem- bly—45 to 38. By the end of March 2003, however, a coalition that supported President Kerekou was in the majority by roughly the same amount. This coalition then expanded its majority to 54 seats in parliamentary elections held for the fourth parliament on March 30. Following the elections, one of the parties in the opposition announced that it would join the president’s coalition raising its total to 65. Notwithstanding its double alternation, the opposition within parliament is now the weakest since 1990 when the for- mer single-party system came under attack. In Ghana, Flight Lieutenant Jerry Rawlings led a military government from 1981 for nearly 11 years after overthrowing an elected civilian 7. Benin follows the French practice of a two-round runoff system to ensure that the ultimate winner obtains a majority of the vote. Kerekou won 45 percent in the ini- tial contest. He then won the second round with an outright majority of 57 percent. Joel D. Barkan, Ladipo Ademolekun, and Yongmei Zhou 217 government by military coup. During this period Ghana became an African success story by reversing a long period of economic decline after embrac- ing policies of structural adjustment. His government also bowed to grow- ing domestic pressure for a return to multiparty politics by promulgating a new constitution and overseeing the return to civilian rule. In the process Rawlings transformed the Provisional National Development Council into the National Democratic Congress, donned civilian clothes, and ran for the presidency. He was elected, in December 1991, under questionable proce- dures. When the opposition boycotted the parliamentary elections that fol- lowed the presidential poll, the National Democratic Congress then won 193 of the 200 seats in the National Assembly. The net effect was that the military retained power for another five years. Rawlings won reelection in 1996 in a poll that was considered substantially freer and fairer than that of 1991. The National Democratic Congress won 133 seats, a clear majority, but much less than when the opposition boycotted the elections. Following the elections, the National Assembly became a more credible forum for the deliberation of public policy. Ghana’s third round of multiparty elections were held in 2001, but Rawlings was barred from seeking a third term. The presidential election was narrowly won by John Kufuor, the leader of the opposition National People’s Party. The National People’s Party also won a close victory in the parliamentary elections by gaining 100 seats to 92 for the National Democratic Congress. The remaining parties have since sided with the National People’s Party giving it a narrow control of the National Assembly. The pattern in Kenya closely resembles that of Ghana although the country has never experienced military rule. As in Ghana, the incumbent authoritarian regime was challenged by domestic pressures for democratic reform and sought to orchestrate the process. In November 1991, the sus- pension of more than US$250 million in budgetary support by Kenya’s international donors forced then President Daniel arap Moi to amend the Constitution to permit the return of multiparty politics. With elections not scheduled for another year, Moi encouraged the opposition to splinter through a variety of carrots and sticks. Moi and the ruling party, the Kenya African National Union (KANU), were returned to office but with less than 40 percent of the vote. The elections were also marked by periodic out- breaks of state-sponsored violence and harassment of the opposition. The opposition’s defeat, however, was mainly the result of its failure to unite behind a single party or a single slate of candidates. KANU won 118 seats of 210 seats in the National Assembly to 92 seats for the combined opposi- tion. The largest of the opposition parties won only 31 seats. For the next 218 Emerging Legislatures: Institutions of Horizontal Accountability five years, Moi and KANU governed as if nothing had changed from the single-party era. Parliament was thus systematically ignored. Kenya’s second round of multiparty elections occurred in 1997. The opposition again splintered, and Moi and KANU again retained power with only two-fifths of the vote. As in Ghana, however, the allocation of seats between government and opposition was much closer than after the first multiparty elections. KANU won only 107 seats while the combined opposi- tion won 103. This fact, combined with a substantial turnover of membership in the legislature, including the turnover of half of the members of the ruling party, gave rise to a very different parliament than after the first elections. The Eighth Parliament, which sat from 1997 through 2002, began to assert its independence from the president. By the end of its term in December 2002, it had become one of the most independent legislatures on the continent. That independence may have been institutionalized for a variety of reasons dis- cussed later in this chapter. Although the Ninth Parliament is only 10 months into its term at this writing, it appears to be asserting itself in much the same way as its predecessor. Yet with a majority of 131 seats in the National Assembly, the new ruling party, the National Rainbow Coalition, has a larger majority than KANU had in the Eighth Parliament. This large majority may not auger well for the legislature’s continued independence.8 The return of multiparty politics in Senegal was also orchestrated from above, but began much earlier in 1974. The Constitution was modified at that time to provide for three parties—one socialist, one liberal-democratic, and one Marxist-Leninist although such distinctions had little meaning in the countryside where patron–client politics mediated by powerful Muslim religious leaders was the norm and remains so today. The first multiparty elections for president were not held until 1978 when the incumbent presi- dent, Leopold Senghor of the Parti Socialiste won 83 percent of the vote. His principal opponent, Abdoulaye Wade, who had created the Parti Démocra- tique Sénégalais (PDS) in response to the new political opening, won 17 per- cent. Over the next 23 years, Wade contested each of five subsequent presidential elections, finally winning in 2001 when, at the age of 75, he cap- tured 50 percent of the vote. His coalition of parties, which was known as the Sopi (“change” in the Wolof language) Coalition and included the PDS, also won control of the National Assembly. As a result of Senegal’s hybrid electoral system, however, the coalition won 74 percent of the seats giving it 8. The National Rainbow Coalition is itself a coalition of two broad alliances, the National Alliance of Kenya, which is a grouping of 14 parties that controls 77 seats in the legislature, and the Liberal Democratic Party, which controls 54 seats. Joel D. Barkan, Ladipo Ademolekun, and Yongmei Zhou 219 a large majority in the legislature. From 1978 through 2001, the vote of the ruling Parti Socialiste declined steadily from over 80 percent to under 20 percent of the electorate. The PDS was created by President Wade and remains his personal vehi- cle. The PDS is Wade, and Wade is the PDS. The result is that although Sene- gal has experienced an alternation of government, the political system resembles the one-party and one-party-dominant states that flourished across Africa before 1990. The ruling party remains the vehicle of the presi- dent, and power is concentrated in the executive branch. Notwithstanding its robust civil society, Senegal is a good example of a country that has expe- rienced considerable political liberalization, but limited democratization, if the litmus test for democracy is accountable and transparent government and the alternation of power. This qualified democratization has led some observers, such as Ottaway (2003) to place Senegal in the category of “semi- authoritarian” regimes—states that regularly hold multiparty elections and that have experienced liberalization, but that are not yet true democracies and unlikely to become such in the near term. The Determinants of Legislative Behavior Whether the legislature emerges as a robust institution and political force appears to be a function of two sets of variables: a set of structural variables that exist both inside and outside the legislature that shape the develop- ment of the body and a set of individual-level variables that affect the extent to which members of the legislature are more or less inclined to seek an expansion of their roles and the authority of the body to which they belong. Included in the first set are (a) the structure of the society and political system within which the legislature is embedded; (b) the formal rules that define the nature, authority, and operations of the legislature; and (c) the institutional resources available to the members of the legislature. Our find- ings suggest that these variables create a tripartite configuration of incen- tives (and disincentives) that explain why some African legislators allocate their time to some aspects of the legislative process more than others. This configuration of incentives varies from country to country and is only partly susceptible to change by the legislators themselves. Included in the second set of variables are the characteristics of individ- ual legislators that explain their variations in behavior within a common configuration of incentives. These include the political generation to which the members of the legislature belong, their level of education, their occupa- tional background and personal experience, and their value orientation 220 Emerging Legislatures: Institutions of Horizontal Accountability toward politics, that is, why they entered public life, why they sought elec- tion to the legislature, their career aspirations, and other motivators. Both sets of variables shape the behaviors of individual legislators and thus the operations of the institution. The configuration of incentives in a given country is a structure faced by nearly all members of the legislature, whereas their individual characteristics vary from one member to the next. The configuration of incentives varies across countries, but is usually a con- stant within each. The characteristics of individual legislators, however, varies both within and across country settings. Structural Variables This subsection considers in turn the impact of each of the three structural variables identified above: the structure of society and the political system, the formal rules, and the institutional resources of parliaments. It concludes with a discussion of the limitations of structural variables. STRUCTURE OF SOCIETY AND THE POLITICAL SYSTEM. All but a handful of African countries are agrarian societies. Most of their populations— between 60 and 90 percent—live in the rural areas and are peasant farmers. Most are poor: between one-third and one-half live below the World Bank standard of extreme poverty. Rural populations exhibit little occupational differentiation though the importance of off-farm employment is rising. In this context, most people identify strongly with their place of residence and with the members of their local communities. Most people consequently define their relationships with the state (that is, their political interests) in terms of their local community. They view the state not in terms of its nationwide programs of domestic or foreign policy but rather as a potential provider of basic services at the local level. Consid- erations of ideology are likewise of little importance. Put simply, one’s place of residence defines one’s political interests. This perspective is heightened by the fact that all but two African countries, Botswana and Somalia, are plural societies. The inhabitants of one region or locality thus distinguish themselves from the inhabitants of others not only in terms of their respec- tive geographic identities, but also in terms of their different cultural and linguistic traditions. This context shapes African politics in at least four ways. First, political competition and conflict are organized mainly around considerations of place, ethnicity, and language rather than on the basis of occupation, socioeconomic class, and economic interests as in advanced industrial Joel D. Barkan, Ladipo Ademolekun, and Yongmei Zhou 221 democracies.9 Second, political conflict does not usually occur over ideo- logical or policy differences, but rather over the geographical distribution of state resources to groups that reside in different areas. The familiar “left–right” and “liberal–conservative” distinctions do not apply. Third, political mobilization normally occurs on a community-by-community and region-by-region basis through a series of spatially adjacent clientelist net- works that link hierarchies of leaders and followers and, thus, the state to society. These networks are held together by relationships of personal loy- alty between superordinates and subordinates at each level of these hierar- chies and by flows of patronage that cement these bonds. When the flows of patronage dry up, these relationships are tested and often break. Fourth, political parties are usually very weak, because the political coalitions that emerge under different party labels are fluid alliances of clientelist hierar- chies that are continuously forming and reforming as the situation demands. Neither party alignments nor the meaning of partisanship remains stable over several elections. The combination of all politics being local and clientelist politics gives rise to the following incentives for the members of legislative bodies. First, and most important in the minds of the public, legislators are elected to ser- vice the collective and individual needs of the communities they represent. This purpose is particularly true where parliamentarians are elected from single-member districts, the procedure in all anglophone countries except Namibia and South Africa. It is also true in francophone Africa where legis- lators are elected from either single-member or small multimember dis- tricts. In this context, elections are essentially referendums on the records of incumbent legislators at delivering the goods to their constituencies or pub- lic estimations of the ability of aspiring legislators to do so. Even in coun- tries that elect their legislatures on the basis of proportional representation (for example, Namibia and South Africa) or countries that have mixed- member proportional systems (for example, Lesotho) or where proportional representation is used in combination with other methods (for example, Benin and Senegal), public expectations of legislator performance are usu- ally defined in terms of addressing local needs. The concept of delivering the goods takes two forms: personal favors by legislators for individual constituents (for example, payment of school fees or a small hospital bill, help in gaining admission to a secondary school, or 9. This structuring is why civil society, apart from the church, tends to be weak across rural Africa although producers of cash crops are becoming better organized to protect their interests. 222 Emerging Legislatures: Institutions of Horizontal Accountability assistance in obtaining public sector employment) and the provision of pub- lic and semipublic goods that are enjoyed collectively by the local commu- nity as a whole (for example, the construction of a local feeder road; the provision of certified teachers or textbooks to the local secondary school; drugs for the local health clinic; or a well, borehole, or cattle dip for the resi- dents of a rural hamlet). The ability of African legislators to meet these expectations varies over time and by country. Public expectations for legislator performance, how- ever, have been largely unchanged since the 1970s, because the majority of the population continues to reside in the rural areas where basic social wel- fare services are in short supply. A 1974 study of the Kenyan National Assembly found that constituency service was uppermost on the minds of the public regardless of where they lived or what their ethnic background was. Kenyans voted their member of parliament (MP) in or out of office depending on their evaluations of their performance (Barkan 1976 and 1979).10 A study of the Zambian legislature from 1964 to 1996 reported a similar finding (Alderfer 1997). These expectations were confirmed yet again in all four countries included in this study during the interviews con- ducted with local elites and with the legislators themselves. Public expecta- tions for legislators are a constant within and across political systems. These expectations have had a great impact on how African legislators allocate their time. Because the public focuses on the performance of indi- vidual legislators in relation to their communities, the collective functions of the legislature—for example, the deliberation, amendment, and passage of legislation and legislative oversight of the executive—have, until recently, been largely ignored. That African legislatures remained weak throughout the one-party era and that their effectiveness varies greatly since the return of multiparty rule come as no surprise. During the one-party era, MPs had no incentive to strengthen these bodies; indeed, just the opposite. Their incentive was to remain members of organizations that purposely avoided the collective functions of the legislature. With the return of multiparty poli- tics, this incentive structure began to change, albeit unevenly from one country and legislature to the next. This seeming oxymoron of what it means to be a legislator, that is, to be a member of a corporate body established for the purpose of crafting legislation 10. From the mid-1960s through the 1980s, several countries, including Kenya and Tanzania, regularly held competitive elections for members of the national legis- lature even though they were one-party states. These elections took the form of party primary elections in the United States and were invariably referendums on incumbent performance. Joel D. Barkan, Ladipo Ademolekun, and Yongmei Zhou 223 and oversight of the executive, but not engaged in these activities, is explained by two aspects of African politics. First, as discussed above, the basic structure of African political systems in the context of agrarian and mul- tiethnic societies, shapes public expectations of the legislature and, more important, of legislators. Second, the pervasiveness of clientelist politics gave rise to neopatrimonial rule. In country after country power was centralized in the hands of a single senior patron who brought together previously compet- ing clientelist hierarchies and maintained their loyalty through liberal distrib- utions of jobs, cash, and other handouts. Patronage was dispensed to MPs in a manner that ensured their depen- dency on the executive. Three common practices illustrate this entanglement. • Official pay and fringe benefits provided to MPs were purposely kept low making it very difficult for most to meet their basic expenses in the capital city let alone travel back to their constituencies. Yet with- out direct contact with one’s constituents and engagement in con- stituency service, their prospects for reelection declined. • Appointment to a ministerial or assistant ministerial positions gave legislators added income. Such appointments overcame the problem of low pay, because appointees received significantly higher salaries and enjoyed a broad range of perks such as offices, housing, and automobiles. Ministers also had access to the budgets of their depart- ments, some of which they could funnel back home. They could also reciprocate (that is, “log roll”) with other ministers by trading ser- vices for each other’s constituencies, for example, the minister of works would ensure that a road would be maintained or built in the constituency of the minister of education who would in turn provide support for a secondary school in the constituency of the minister of works. Yet once an MP became a member of “The Government” his or her ability to act as an independent check against the executive evaporated. MPs were also co-opted through appointments to statu- tory boards, state-owned corporations, and other entities. • For MPs who were not appointed to the government or other posi- tions, the president and senior political leaders simply handed out cash as the need arose—for example, on Friday afternoons so that MPs could travel back to their constituencies over the weekend dur- ing an election campaign. Backbenchers were particularly susceptible to handouts, because they could not visit their constituencies without them. Cash handouts were also effective for buying off members of the opposition—to secure their quiescence or desertion from their parties to join the ruling party. In countries where the ideological and 224 Emerging Legislatures: Institutions of Horizontal Accountability policy distinctions between parties are limited or nil, few MPs worry about defecting. Indeed, the failure to defect risked one’s access to resources for the people one was elected to represent. Weak parties go hand in hand with weak legislatures. This combination of incentives and disincentives has retarded the devel- opment of African legislatures. It has persisted since the return of multi- party politics and the so-called “Third Wave” of democratization swept the continent a decade ago because the underlying structure of most African societies and political systems has not changed. Clientelist politics remains alive and well in Africa, because the structure of African society that gives rise to this type of politics has not changed. The puzzle thus arises: if the structure of African society has not changed radically, why have some African legislators sought to redefine their roles to resemble those in estab- lished democracies whereas others have not? And why have some African legislatures started to take on the collective functions of a modern democra- tic legislature for the first time? These questions provide an important back- drop for considering the impact on legislative behavior of the variables considered in the rest of this section. FORMAL RULES. A second set of structural variables that determine the scope of legislative authority and the opportunities for expanding it are the formal rules under which the legislature operates. These rules take two forms—the constitutional framework for the political system as a whole and the internal rules of procedure of the legislature itself. Constitutional rules are presented in table 7.2; rules of procedure are presented in table 7.3. Sum- maries of these rules are presented in each table from left to right beginning with the country with the weakest legislature considered in this study, Sene- gal, followed by Benin, Ghana, and Kenya in order of our assessment of their increasing authority overall. The constitution defines the role of the legislature within the political system. It specifies the nature and scope of legislative authority with respect to making law and amending the constitution, and it defines the relation- ship of the legislature to the executive branch. The constitution also speci- fies the method by which the legislature is elected. It may also specify the relationship between the legislature and the judiciary, particularly if legisla- tion is subject to judicial review (as in Benin and South Africa). If the politi- cal system is a federal system, the constitution will also specify those areas of public policy for which the central government (and hence the national legislature) has primary jurisdiction and which are devolved to subnational authorities. Joel D. Barkan, Ladipo Ademolekun, and Yongmei Zhou 225 Table 7.2. The Constitutional Authority of the Legislature Factor Senegala Beninb Ghanac Kenyad Scope of legislative Restricted Broad Slightly Slightly authority restricted restricted Rules that entrench presidential dominance President can determine agenda of the legislature? Yes Sometimes No No President can force legislature to vote on a bill? Yes No No No President can rule by decree? Qualified No No No President can amend laws unilaterally? Yes No No No President must assent to all laws? Yes Yes Yes Yes Legislature can override — Yes by Yes by Yes by 65% presidential veto? majority of 2/3 vote vote of all all members members President appoints most ministers from within the legislature? No No Yes Yes A ministry for legislative relations exists? Yes No Yes No President can adjourn and dissolve legislature? Yes No No Yes Rules that enable the legislature to limit executive authority Legislature can initiate Yes but Yes but legislation? restricted Yes restricted Yes Legislature can amend legislation? Qualified Yes Yes Yes President and ministers must answer questions and provide documents to legislature? — Yes Yes Yes table continues on following page 226 Emerging Legislatures: Institutions of Horizontal Accountability Table 7.2. continued Factor Senegala Beninb Ghanac Kenyad Legislative involvement in the budgetary process Legislature can introduce money bills including budget? No No No Yes Legislature can amend (that is, reallocate) money bills including budget? — — Yes Yes The legislature’s power to control own affairs Who sets salaries of MPs and deputies? Executive Executive Executive Legislature Who sets parliamentary budget? Executive Executive Executive Legislature Other rules that shape legislative autonomy Type of electoral Hybrid Proportional Single- Single- system (block and representa- member member proportional tion (but 24 districts districts representa- multimember tion) districts) President and legislature are subject to judicial No Yes No No but review? proposed Parliament has power to ratify all government borrowing No No No Proposed — Not available. a Data are for the sixth legislature elected since the resumption of multiparty politics in 1978 and whose term is from 2001 to 2006. b Data are for the Third Parliament, which sat from 1999 to 2003. c Data are for the Third Parliament whose term is from 2001 to 2006. d Data are for the Eighth Parliament, which sat from 1997 through 2002. Source: Authors. Although the extent of legislative authority is not solely a function of the formal rules, a rough relationship between the content of the rules and the level of authority is apparent.11 The legislature is obviously weakest where the configuration of constitutional and internal rules limit its authority 11. With only four cases we cannot, of course, determine whether the relation- ship is statistically significant. Joel D. Barkan, Ladipo Ademolekun, and Yongmei Zhou 227 Table 7.3. Internal Structures And Rules Of Procedure Factor Senegala Beninb Ghanac Kenyad Number of housekeeping committees — — 9 6 Number of ministerial committees 11 5 17 8 Number of public accounts committees and other oversight committees None Qualified 1 2 Estimated effectiveness of Moderate- ministerial committees Very weak Weak Emerging strong Estimated effectiveness of committees concerned Weak- with budget Very Emerging emerging Moderate Estimated effectiveness of Weak- oversight committees — — Weak emerging Power to hold hearings — Yes Yes Yes Parliamentary budget office None None Proposed Proposed Legislative opportunities for deliberating and amending budget — Extensive Some Extensive — Not available. a Data are for the sixth legislature elected since the resumption of multiparty politics in 1978 and whose term is from 2001 to 2006. b Data are for the Third Parliament which sat from 1999 to 2003. c Data are for the Third Parliament whose term is from 2001 to 2006. d Data are for the Eighth Parliament which sat from 1997 through 2002. Source: Authors. explicitly and strongest where the rules do not. A review of the two tables indicates that the structure of formal rules is highly complex, consist- ing of nearly 30 elements. The configuration in each country reflects the particular historical circumstances that brought them into force, and mak- ing generalizations about the impact of each item across the four countries is difficult. We have therefore clustered the various constitutional provi- sions presented in table 7.2 into five categories: rules that entrench presi- dential dominance, rules that enable the legislature to limit executive authority, rules that specify the degree of legislative involvement in the budgetary process, rules that specify the legislature’s power to control its own affairs, and other, that is, miscellaneous rules that determine the extent of legislative autonomy. The relationship between the content of the rules and the extent of leg- islative authority is not pronounced, perhaps because all four countries 228 Emerging Legislatures: Institutions of Horizontal Accountability considered in this study have presidential systems in which the president and executive branch have long dominated the legislature. Although some of these systems may be “more presidential” than others, the constitutional framework is more or less constant across the four cases and variation is limited. This pattern of presidential dominance is a function of the perva- siveness of clientelist politics and the neopatrimonial practice of incumbent presidents discussed in earlier in this chapter. This informal structure of governance has resulted in the enactment of constitutional provisions that entrench presidential hegemony. Put differently, the agrarian nature of African society and the informal structure of politics to which it gives rise determine the formal rules. Notwithstanding this tendency, three categories of constitutional provi- sions appear to be linked to the extent of legislative authority: • The powers of the legislature to shape the national budget • The powers of the legislature to determine its own budget including the salaries of its members • The form of the electoral system for members. The greater the role of the legislature over budgetary matters, both national and internal, the greater the overall power of the legislature. The Kenya National Assembly—the most powerful of the four legislatures—is particularly strong with respect to these provisions. The Kenya National Assembly and its Ghanaian counterpart are also more independent from the executive than are the legislatures of Senegal and Benin because of the electoral rules. In the former, members of the legislature are elected from single-member districts. In the latter members are elected or partially elected by party list proportional representation. Because the party leader- ship determines the rank order of names for list proportional representa- tion, party discipline among legislators of the same party is usually greater in legislatures that elect their members via some version of proportional representation than in legislatures that do not, so the president has an addi- tional power over members from the ruling party that he or she does not possess in district based electoral systems. This dominance results in a weaker legislature. The relationship between the extent of legislative authority and the internal rules of the legislature, which are presented in table 7.3, is arguably stronger than the relationship between legislative authority and the consti- tutional rules. As suggested by the entries in the table 7.3, the ability of the legislature to contain and oversee the operations of the executive branch is a function of the quality and capacity of the committee system rather than the Joel D. Barkan, Ladipo Ademolekun, and Yongmei Zhou 229 number of committees. Here again, however, the rules reflect the informal relationships between individual members and the president. When mem- bers are more tightly bound as clients to the president, the internal rules permit less legislative oversight of the executive than where these bonds are weak. The rules are thus both independent and dependent variables. Although they certainly determine the scope of legislative authority, they are themselves the products of other forces. The variations in these rules—both constitutional and internal—and the efforts to change them can be explained largely in terms of the presence or absence in the legislature of a coalition of members who desire change. Moreover, the rules and structures of these institutions are changing in some countries, but not in others. They do not, by themselves, explain varia- tions in either the activities of individual legislators or the collective behav- ior of the bodies to which they belong. We are therefore reluctant to embrace a so-called “new institutionalist” perspective to explain either the weakness of African legislatures in the past or their emerging authority in some African countries. Our search for the determinants of changes in legislative behavior must thus proceed to the remaining sets of variables highlighted in our analytical framework. THE INSTITUTIONAL RESOURCES AVAILABLE TO PARLIAMENTS. The level of resources available to its members makes up the third set of determinants of legislative behavior. Such resources are of two types. The first, and no doubt most important, is the level of compensation paid to members. The second is the level of institutional support members receive to support their work. Two related issues are (a) whether the legislature determines its own budget or whether it is entirely dependent on the executive for its budget allocation and (b) whether it has the power to appoint its own staff and set their terms of service or is simply another department of the national public service. When the legislature has the power to set its own budget, the chances are good that both the institutions and its members will be better endowed with respect to salaries, professional staff support, and physical infrastructure. The power to set its own budget is also an indicator of the independence of the legislature as a distinct branch of government and an indirect measure of legislative authority. The single most important resource is the provision of appropriate salaries for members and staff. As discussed early in this chapter, MPs who are poor are extremely vulnerable to blandishments by the president or other members of the executive who can provide MPs with the resources they need—to make ends meet in the capital city and especially with the 230 Emerging Legislatures: Institutions of Horizontal Accountability resources they need to maintain good relations with their constituents. Most MPs, with the notable exception of the Kenyans, reported that their salaries were insufficient to cover their basic costs in the capital while attending leg- islative sessions. Even more problematic was the cost of returning to one’s constituency even though the failure to do so on a regular basis is a recipe for defeat. Nearly all MPs in all four countries therefore strive to visit their constituencies at least twice each month. Some do so weekly, whereas a few limit visits to once a month. Such visits are very expensive relative to the compensation and allowances MPs receive for being members of the legislature. Including transportation costs and lodging, MPs in Ghana reported that each week- end visit home cost approximately US$250; MPs in Kenya reported spend- ing between US$300 and US$400 and sometimes double that. In the case of Ghana, a monthly expenditure of US$500 on constituency service means that the typical MP spends his entire monthly salary on constituency ser- vice. Until Kenyan legislators increased their salaries to US$4,800 per month in 2001, their situation was similar. Indeed, some reported that the expecta- tions of their constituents was so great, that they now visit their districts less frequently than in the past. The pressure is particularly great in Kenya, where the Harambee self-help development movement has institutional- ized efforts at community development to the point that MPs who do not participate in Harambee face near certain defeat. Similar pressures, how- ever, are found in each of the other countries included in this study, particu- larly Ghana where MPs are also elected from single-member districts and are thus visible targets of opportunity in the eyes of the public. Given these pressures, a common complaint voiced by legislators in all four countries was that they were poorer now than before their elections to the National Assembly! Some stated that they were in debt or simply subsi- dized the cost of being a legislator from their personal savings or the income of their spouse or both. One legislator complained that his wife was grow- ing weary of the financial pressures of the job. Members who are profession- als stated flatly that their salaries in the private sector were several times that of what they received as legislators and that without this income they could not serve. Notwithstanding these complaints, none stated that they would not seek reelection. Indeed, their most common response was “as long as the people want me to serve, I will do so.” The prospect of indebted MPs is troubling. The most obvious concerns are who the MPs are indebted to and what favors they provide to their bene- factors. Such entanglements raise the specter of corruption—that MPs will engage in corrupt acts out of a desperate need for cash to finance their polit- ical obligations. Joel D. Barkan, Ladipo Ademolekun, and Yongmei Zhou 231 Table 7.4. Resources Available For MPs and Deputies Resource Senegal Benin Ghana Kenya Monthly salary $810 $1,081 $500 $4,800a $6,063b Constituency fund? No No Yes Yes Daily record of debates? No No Yes Yes Office space for — — Only for Yes members? leaders Sufficient meeting space for committees? — — No Yes More space being constructed? — — Yes No Parliamentary staff Professional staff 300 167 — 300 Support staff 300? — — — Committee staff — — Some Some Parliamentary library and resource center? — — Small Medium Internet and World Wide Web access None None Some Some — Not available a As of the end of the Eighth Parliament in 2002. b As of March, 2003 following the election of the Ninth Parliament. Source: Authors. As table 7.4 details, we found large differences across the four countries in the formal provision of public resources to parliamentarians. Kenyan par- liamentarians receive overwhelmingly better provisions than their counter- parts in the remaining three countries. Among the three countries, Ghanaian parliamentarians are somewhat better off than their Beninese and Senegalese counterparts. Kenya’s largesse can be traced to the passage of a constitutional amend- ment in 1999 that formally de-linked Parliament from the executive; a key impetus for the amendment was the desire by MPs to set their own salaries at a level sufficient to finance their activities. The constitutional amend- ment was followed by subsequent legislation that permitted a fivefold raise in salaries in 2001 and another raise following the election of the Ninth Parliament at the end of 2002. Kenyan MPs now earn US$6,063 a month, the highest in Africa, and high by international standards. By comparison, legislators in Benin, Ghana, and Senegal are poorly paid. 232 Emerging Legislatures: Institutions of Horizontal Accountability Ghanaian MPs are most likely to emulate their counterparts in Kenya because they are well aware of the compensation there and in Uganda where salaries have also been raised (to roughly US$2,500 a month). Kenyan and Ghanaian MPs are also seeking to emulate their colleagues in Uganda, where the National Assembly has established a Parliamentary Budget Office to control both the financial affairs of the legislature and play a greater role in setting the national budget. In contrast, Senegalese MPs have made little progress toward asserting the financial independence of the legislature there. MPs in Ghana and Kenya are also more successful than their counter- parts in establishing two funds that provide direct financing to their con- stituencies. In Ghana, Parliament mandated that 5 percent of the District Assemblies Common Fund, the principal source of funding for government at the district level, be set side for projects at the constituency level and over which MPs have considerable control. In Kenya, Parliament created the Constituency Road Fund, which is controlled by a committee of local gov- ernment officials chaired by an MP. MPs in both countries reported that they valued these funds and wanted to see the amount of money allocated to them increased. Those in favor of expanding the constituency funds argued that doing so would both ease the informal pressures by constituents to pro- vide handouts when back in their constituencies and lower their vulnerabil- ity to incurring debts and favors from benefactors. If coupled with a ban or regulation on MPs contributing to community development projects such as Harambee, the expansion of constituency funds could greatly reduce cor- ruption. Over half, although not all, of the MPs questioned on the subject agreed that such regulations are desirable. Others wanted to be free to sup- port their constituents in any way they liked.12 Resources are also required at the institutional level if African legisla- tures are to become significant in the making of public policy. Here again, the Kenyan National Assembly has been the most successful in providing for its needs. Kenyan MPs have provided themselves with substantial infra- structure and are in the process of upgrading parliamentary staff to support their fledgling system of departmental committees. The Parliamentary Service Commission established in 2000 has become the main vehicle for providing such resources. The commission has drawn 12. Since the election of the Ninth Parliament in Kenya in December 2002, the demand for banning MP participation in Harambee development projects has steadily risen with the result that several temporary bans have been ordered. Joel D. Barkan, Ladipo Ademolekun, and Yongmei Zhou 233 up a 10-year plan for strengthening the legislature and has received consid- erable donor support or pledges of support (for example, from the U.S. Agency for International Development,13 European Parliamentarians for Africa, the European Union, and the World Bank). The result is an ambitious program to provide a trained senior clerk and secretarial support to each departmental committee and transform the existing library into a resource center with Internet access. Since 2002, each Kenyan MP has had his or her own office, and the number of committee rooms have been increased. Increased funding for the departmental committees, including support for travel to outlying areas to facilitate committee investigations and the hold- ing of hearings, is also on the agenda. The Parliamentary Service Commis- sion has also hired its own legal draftsmen to facilitate the preparation of bills and free itself from depending on the Attorney General’s Office for such services. Funding for these and other improvements is now possible, because the National Assembly is no longer dependent on the Ministry of Finance and the executive to set its budget, nor dependent on the Public Ser- vice Commission to set the terms of service for its staff.14 Although increased funding will not guarantee a more technically competent and powerful legislature, the development of the legislature is unlikely without such support. The legislatures in Benin, Ghana, and Senegal lack the resources of the Kenyan National Assembly. Their members are increasingly aware of what members of other African legislatures have and are doing to provide for their needs. In Ghana MPs have established a bipartisan Special Budget Committee in 2001 to address the issue of financial dependence on the exec- utive and better plan for their needs. However, the National Assembly can- not as yet vote its own budget. Under pressure from MPs, the Ghanaian 13. The U.S. Agency for International Development has developed the most extensive and formalized program of support for African legislatures. Such pro- grams consist of multiyear efforts to train legislative staff and build institutional capacity. Although some support is provided for information technology, almost none is provided for physical infrastructure. By contrast, the World Bank, under its Public Sector Capacity Building Programs, has provided funds that can be used for constructing or refurbishing office space for members of the legislature and legisla- tive staff. 14. African legislatures face a perennial problem of staff retention, because their salary structures are not competitive with the private sector. The power of the legis- lature to set the terms of service for its staff may be as important as the power to determine its own budget. 234 Emerging Legislatures: Institutions of Horizontal Accountability government secured a US$25 million loan from the World Bank in 2000 to remodel a high-rise building next to the National Assembly into an office block that would provide office space to individual members, committees, and staff. Unfortunately, the minister of finance diverted the money to other priorities, but following complaints by members, disbursed half of the loan in 2001. The other half, however, has yet to be disbursed, and the physical infrastructure remains poor. The lesson of these examples is clear. When MPs come together to form a viable coalition for change and demand resources for themselves and their institution, such resources can be pried out of the executive branch. How and when Ghanaian MPs will obtain the power and resources of their Kenyan colleagues is, of course, unknown but the direction of future change is clear. A major factor in this regard is awareness. Parliamentary leaders and senior staff across anglophone Africa appear to be more aware of developments in peer legislatures than parliamentarians and staff in fran- cophone Africa. Whether this awareness is a function of donor efforts such as those by the World Bank Institute periodically to bring parliamentarians together for regional training sessions or a result of efforts by the Common- wealth Parliamentary Association or the new Association of African Parlia- ments is unclear. What is clear is that MPs are fast taking note of increased resource flows and statutory changes at peer institutions. THE LIMITATIONS OF STRUCTURAL EXPLANATIONS. Although our explo- ration of structural variables has helped identify some reasons for cross- country variations in the influence of legislatures, it has not been especially useful in identifying the drivers of change over time within countries. To be sure, we have identified one critical determinant of influence that has seen substantial change within countries where legislatures have become rela- tively more powerful, namely the extent to which legislatures can deter- mine their own budget, including the salaries paid to members. The puzzle remains, however: increased resource flows to legislatures are not the cause of increasing legislative authority, but rather a reflection of efforts made by members to expand the role of the institution. The amount of resources available to legislators and their institution is unlikely to rise unless force- fully demanded by the legislators themselves. Their availability is thus a function of the presence or absence of a coalition for change that in turn reinforces the desire by reformers to develop their institution into a player in the political process. The analysis of changes in the characteristics of leg- islators in the next section helps explain why, in some countries, the influ- ence of reformist legislators might be growing. Joel D. Barkan, Ladipo Ademolekun, and Yongmei Zhou 235 The Characteristics of Legislators (Individual-Level Variables) During the single-party era, the overwhelming proportion of MPs accepted authoritarian rule so long as they continued in office and the system gave them the resources to service their constituencies. Some members of this group also believed in the superiority of authoritarian rule and are best described as “incumbent authoritarians.” Others simply accepted authori- tarian practice because they were patronage seekers—for themselves and for their communities. Along with incumbent authoritarians and patronage seekers, we define a third group of legislators, namely reformers committed to the process of democratization, including the creation of a legislature that plays a greater role in the policymaking process. Almost all African countries typically had a small group of dissenters that argued for a more open and democratic polity, including the regular holding of competitive, multiparty elections. Until the early 1990s, members of this group of aspiring reformers were invariably harassed and sometimes jailed. When reformers held no public office, but were academics, religious leaders, or leaders of civil society, they were harassed but tended to be ideal- ists who persisted in advocating democracy as a long-term project. When reformers were members of the legislature or the government, they were harassed and usually run out of office. Still others had been members of the regime, but became advocates of reform after being dropped by the regime and vowing to get even. Whether the members of this last group were reformers or patronage seekers or both is unclear. Determining with precision the proportion of members in the four legisla- tures that are incumbent authoritarians, patronage seekers, and reformers was difficult. What we present are impressions, first of some drivers of the chang- ing balance among members and second of some key demographic character- istics of parliamentarians in the four countries as of 2002. We note that reformers are more likely to be younger, more educated, and serving in their first legislative term than patronage seekers. Reformers are also more likely to have a strong, that is, successful private sector background than patronage seekers.15 Not all reformers, however, are members of the youngest cohort. 15. Not surprisingly, members of the legislature who have successfully run their own businesses are more likely to be reformers than those who have not. They have struggled with the challenges of satisfying clients, organizing staff and production, and other demands of business operation, and in the process they developed some minimal level of managerial skills without which their businesses would have failed. These individuals expect that the government should be able to be run with at least the same level of efficiency. 236 Emerging Legislatures: Institutions of Horizontal Accountability DRIVERS OF CHANGE. Changes in the characteristics of legislators are reflective of changes in society at large. African societies have remained agrarian societies, but the urban enclaves in several countries have seen sig- nificant changes, especially the reemergence and growth of civil society from the late 1980s on. Suppressed for two decades, the emergence of civil society was the product of a new and younger cohort of political activists that demanded an end to authoritarian rule. This cohort was composed of activists in their late twenties to early forties. It emerged across the conti- nent, although its members were most numerous in countries that have the largest urban enclaves, especially those that have historically supported a continuous flow of educated elites to and from the West, for example, in Côte d’Ivoire, Ghana, Kenya, Nigeria, Senegal, and South Africa. Donor support for civil society organizations that promote democratization and good governance has also been a factor in the rise of civil society. The introduction of multiparty political systems led to a shift in the mix of parliamentarians. Incumbent authoritarians were confronted by ever larger numbers of patronage seekers and reformers. These new players did not benefit from the clientelist system of the 1980s and had nothing to lose by challenging the system. They were joined by a few senior members of the old system who had soured on its ways. Either way, the defenders of authoritarian rule lost ground. By the late 1990s the pace of change in the composition of some legisla- tures quickened as the regular holding of elections resulted a high turnover of membership. Between 40 and 60 percent of incumbents typically lost their seats. In the process, politics became a three-cornered struggle in which patronage seekers and a small number of reformers first displaced incumbent authoritarians and then competed against each other for control of the new order. The question now is, what is the balance between patron- age seekers and reformers? Where the latter are prominent, even if not dom- inant, the prospects for political reform including the strengthening of the legislature are the greatest. The new cohort of political activists is better educated than its predeces- sors.16 It comprises a distinct political generation that emerged at the end of 16. This generalization has some exceptions or variations. For example, in Sene- gal many of the PDS MPs elected for the first time in 2001 were younger than the Parti Socialiste members they replaced. In marked contrast to the younger genera- tion of MPs in Ghana and Kenya, however, these MPs were less rather than better educated than their predecessors and represented lumpen elements of Senegalese society. They were nominated by the PDS precisely because of this characteristic on the grounds that they would appeal to the electorate. Joel D. Barkan, Ladipo Ademolekun, and Yongmei Zhou 237 the Cold War in response to the failure of authoritarian rule.17 Another dis- tinguishing feature of this group is that its members are plugged into the global community and supportive of its social, political, and macroeco- nomic policy norms to a much greater extent than their predecessors. They are also computer literate and more likely to be regular users of the Internet than their elders. They communicate frequently with colleagues abroad. They follow world news and know what is going on outside their country. They are familiar with how the world runs in the twenty-first century. They are painfully aware that Africa has fallen further behind the developed world and more likely to support reforms, both economic and political, to close the gap. They are more proactive in their orientation, but also highly pragmatic. A decade later, the members of this age cohort, and civil society, are beginning to shape the evolution of African legislatures, though the sequencing of what is occurring is not fully understood. As civil society has become more sophisticated and turned its attention from opening up politi- cal space to shaping public policy, some civil society organizations now tar- get the legislature. Some groups have done so by offering their skills to inform and train parliamentarians. Others have lobbied the legislature or mobilized citizens to do same. Some civil society activists have been elected to the legislature although this trend was not apparent until after the second and third round of multiparty elections. The magnitude of their presence, both outside and inside the legislature also varies greatly from country to country. Of the four countries considered in this study, their presence is felt most in Kenya, but limited in the other three. COMPOSITION. Consistent with African culture and society, African legis- latures are overwhelmingly male organizations (see table 7.5). Ninety per- cent or more of the membership in three of the four legislatures is male, and the figure is over 80 percent for the fourth. The Senegalese Parliament has the highest percentage of women, because of its hybrid electoral system. Where the members of the legislature are elected from single-member dis- tricts, or where members are elected from small multimember districts, few women are elected. By contrast where a party-list system of proportional 17. Although most members of this political generation are younger than the average legislator in terms of their chronological age, the new generation of legisla- tors often includes several older members—those who share the characteristics of younger members and who have at times been labeled “young turks” because they were atypical of their own age cohort. 238 Emerging Legislatures: Institutions of Horizontal Accountability Table 7.5 The Composition of the Legislature (percent) Measure Senegala Beninb Ghanac Kenyad Male 82.5 92.8 91.0 96.8 Average age — 53 49 53.8 39 and under — 2.4 7.3 1.8 40–49 — 26.5 51.3 35.1 50–59 — 47.0 27.7 38.8 60–69 — 20.5 13.0 18.9 70+ — 1.2 0.5 5.4 Presence of new age cohort Few Some Some Many Education None (0) — 4.0 0.2 0.0 Primary (1–8) — 16.0 — 1.8 Secondary (9–12) — 20.0 31.5 49.1 University (13–16) — 48.0 40.8 36.0 Advanced (16+) — 12.0 26.1 13.1 Not fluent in English/French 12.0 20.0 <5.0 3.6 Occupation Peasants and traders 14.5 7.7 4.7 0.9 Intermediate civil servant 8.0 0.0 2.6 8.1 Senior civil servant 6.0 26.9 7.9 39.6 Business or managerial 7.0 15.4 11.0 12.6 Teachers 25.0 10.3 21.6 5.9 Professional 28.0 39.7 44.7 32.0 Other 11.5 0.0 7.4 0.9 Term in office First 86.1 65.1 42.0 58.6 Second — 30.1 — 30.6 Third — 4.8 — 10.8 Minister or deputy minister 0.0 14.5 21.5 37.8 Number 120 83 200 222 — Not available. a Data are for the sixth legislature elected since the resumption of multiparty politics in 1978 and whose term is from 2001 to 2006. b Data are for the Third Parliament, which sat from 1999 to 2003. c Data are for the Third Parliament whose term is from 2001 to 2006. d Data are for the Eighth Parliament, which sat from 1997 through 2002. Source: Authors. representation exists or where a hybrid system is used, such as in Senegal, where 55 deputies are elected on the basis of proportional representation from a national list and 65 are elected from 30 multimember districts on a block basis, the percentage of women legislators is significantly higher. Fifteen of Joel D. Barkan, Ladipo Ademolekun, and Yongmei Zhou 239 Senegal’s 21 women legislators were elected via proportional representation. Only six were elected from electoral districts. This pattern appears to be uni- versal and not confined to Africa. Because the proportion of MPs who are women is very low, however, variation in the gender composition of the four legislatures does not affect their operation. Our presentation of the age profiles of the four legislatures is incomplete, but suggestive. First, being a legislator is neither a young person’s nor an old person’s game. The proportion of MPs who are under 40 or over 70 years old, is quite small. Roughly three-quarters are between 40 and 59 years of age whereas up to a fifth are in their 60s. The average age of all MPs in Benin, Ghana, and Kenya is between 49 and 54. The typical legislator is thus substantially older than the electorate. Although we are unable to pre- sent an age profile for Senegal, we expect that it more closely resembles the age profile for Ghana than the profiles for Benin and Kenya. The PDS nomi- nated a large number of young candidates before the 2001 elections, and its large majority in the Senegalese National Assembly resulted in a younger legislature. The PDS, however, did not select these candidates from pool of highly educated prospects. Because the presence of the younger generation of politicians is usually marked by the combination of relative youth, educa- tion, and sophistication, we present a separate estimate of the number of MPs who are members of this cohort in table 7.5. The notion that MPs, whether young or old, are members of a so-called “establishment” is supported by the data on education, linguistic ability, and occupation. The educational profile of MPs in Benin, Ghana, and Kenya speaks for itself. Over 60 percent are university graduates or holders of advanced degrees in Benin and Ghana, while nearly half hold such degrees in Kenya. Most of the remaining MPs are secondary school graduates or graduates of technical schools of various types. Almost none have only pri- mary educations or no education in Ghana or Kenya, but a fifth fall into this category in Benin and—we suspect—Senegal. The membership of the two anglophone legislatures are thus far less likely to have an uneducated ele- ment than their francophone counterparts though it is doubtful this distinc- tion can be explained in terms of the colonial experience of the four countries. Indeed, the legislature in Senegal was to some extent “dumbed down,” so to speak, as a result of the PDS victory in 2001 and the defeat of the Party Socialiste. The educational background of members in the previ- ous legislature, in which the Parti Socialiste was in the majority, was sub- stantially higher. The contrast of educational profiles between Senegal and Benin, on the one hand, and Ghana and Kenya, on the other, is further illustrated by the estimates of the percentage of MPs in each legislature that are not fluent in 240 Emerging Legislatures: Institutions of Horizontal Accountability the official language of the country.18 All four parliaments have members who are not fluent in the official language and who are therefore unable to participate fully in their proceedings. They are, in essence, second-class members. Their presence limits the capacity of the legislature to perform its policymaking and oversight functions. Their lack of fluency in the official language, however, does not limit their ability to engage in constituency service.19 The estimated proportion of members who are not fluent in the official language is very small in Ghana and Kenya, but is a significant minority in Senegal and Benin where it is cited by assembly staff as a serious impedi- ment to the development of the legislature. The presence of a poorly edu- cated and inarticulate element is a significant obstacle in Benin where a fifth of the legislature is not educated beyond the primary level and not fluent in the official language. Limited education and the lack of critical language skills are also likely to inhibit the development of a strong committee sys- tem, because a significant number of members are incapable of reading gov- ernment reports, questioning officials at committee hearings, or amending proposed legislation. The occupational backgrounds of MPs reflect their educational back- grounds. What is striking is the large percentage of MPs—between 28 and 45 percent—who are professionals, for example, attorneys, engineers, doc- tors, university professors, and others with specialized university training or advanced degrees. Approximately half of all professionals are attorneys. The percentage of MPs who are professionals appears to reflect the varia- tion in educational attainment across the four countries. The percentage of MPs who are professionals is highest in Benin and Ghana. Beyond this fac- tor, no consistent occupational profile emerges for all four legislatures. A quarter of the Benin legislature and two-fifths of the Kenyan legislature had been senior civil servants, but the proportion was much smaller in Senegal and Ghana. A quarter of Senegalese legislators and a fifth of the Ghanaians had been teachers, but the proportion was much smaller in Benin and Kenya. Nearly a sixth of Senegalese members were peasants or traders, but few legislators had similar occupations in the other three 18. Although Swahili is classified as an official language in Kenya and its use permitted in the National Assembly, it is rarely used other than to utter a colloquial expression, make a joke, or speak informally. The record of debates, the Hansard, is published only in English. 19. When MPs return to their constituencies they are more likely, indeed often must, communicate in the principal indigenous language of the people they represent. Joel D. Barkan, Ladipo Ademolekun, and Yongmei Zhou 241 countries. How these aggregate profiles affect the performance of the four legislatures is unclear, but we would expect that the presence of a large number of professionals is a necessary, although insufficient, condition for strengthening these bodies. Two additional characteristics of individual members are relevant for understanding why some African legislatures are becoming significant players in the policymaking process whereas others are not. The first is the level of turnover among members from one election to the next. The second is the extent to which the ruling party draws from members of the legisla- ture to appoint ministers and deputy ministers. When turnover is very high, especially of those pressing for reform, the prospects are less that the legisla- ture will strengthen itself by establishing a committee system that shadows government ministries or that it enacts other reforms. The evolution of the so-called “professional” legislature in the United States, both at the federal and state levels, was directly linked to a reduction in the turnover of the membership from one election to the next (Squire 1992). Turnover in the U.S. House of Representatives is now less than 20 percent at each election, including retirements, whereas the average turnover in American state leg- islatures is 23 percent (National Council of State Legislatures 1996). Turnover in the four legislatures considered in this study follows the expected pattern: it is lowest in legislatures that have made the greatest progress toward becoming independent and professional bodies that spend substantial time crafting legislation and exercising oversight of the execu- tive branch. Conversely, turnover is highest where progress has been lim- ited or nonexistent. Turnover in the Senegalese legislature is thus very high; 86.1 percent of the members are serving in their first term, that is, they were elected in the most recent election in 2001. Benin is next with 65.1 percent serving their first term, and Ghana with 42.0 percent. Although turnover for Kenya’s Eighth Parliament (1997–2002) was also high at 58.6 percent, the figure dropped to 43.8 percent in elections held in December 2002 when the number of incumbents who won reelection rose substantially. Most likely, the percentage of incumbents who were reelected rose not because they were good legislators in the conventional sense, but rather because they were good at constituency service. The important point is that turnover declined at the same time MPs began to make serious efforts to assert the authority of the institution to which they belonged. In the near term, low turnover is thus an impetus to the professionalization of the legis- lature rather than the professionalization of the legislature being the cause of low turnover. This relationship, however, may be reversed over time. The more the legislature begins to matter in terms of the making of public policy 242 Emerging Legislatures: Institutions of Horizontal Accountability in the eyes of the public, especially civil society, the more likely interest groups seeking to influence the legislature will support the reelection of incumbents they believe are sympathetic to their demands. Similarly, the more the legislature matters, the more incumbent members will draw atten- tion to their records as lawmakers in addition to their records of constituency service—as is true in established democracies that have district-based sys- tems of representation. In Africa, however, constituency service still comes first. We conclude this discussion of the composition of the legislature by con- sidering the proportion of MPs appointed to be ministers or deputy minis- ters. As suggested above, these appointments have long been used by African presidents to co-opt members of the legislature. Presidents such as Kenya’s Daniel arap Moi were particularly skillful—often brazen—at increasing the number of ministerial appointments to achieve such ends. One would thus expect that the greater the percentage of MPs who received such appointments, the less the independent authority of the legislature. Surprisingly, our findings show just the opposite relationship or, more likely, that no relationship exists. As indicated at the bottom of table 7.5, the lowest percentage of MPs who are members of the executive is in Senegal and Benin where the consti- tution requires that all government ministers are appointed from outside the legislative branch. This requirement has been the historical practice across francophone Africa though some countries have modified the provi- sion in recent years. The proportions are much higher in Ghana and Kenya. In the first, the legislature is beginning to make serious efforts to enhance its authority and may be more powerful than the legislature in Benin. The authority of the legislature is highest in Kenya where the National Assem- bly, between 1998 and 2002, transformed itself into one of the most powerful legislatures on the continent notwithstanding the fact that more than a third of its members held appointments in the executive branch. These findings suggest that some configuration of variables other than patronage drives or retards the process of legislative reform. Although min- isterial appointments may ensure the loyalty and hence the quiescence of those who receive them, other factors override or work independently of this form of patronage to determine the extent of legislative authority. Possi- bly this form of patronage is less effective than in the past. In the context of the one-party era, the prospect of becoming a minister, contrasted with low pay for backbenchers, kept most members quiescent in relation to the exec- utive. Those who received a ministerial appointment were loyal to the exec- utive. Those who did not contented themselves with being quiet followers Joel D. Barkan, Ladipo Ademolekun, and Yongmei Zhou 243 of the president and good providers of constituency service. Such behavior would ensure an MP’s renomination and thus his or her reelection at which point he or she might then be promoted to a ministerial position. In multiparty era, however, the game has changed. Some MPs who are members of the ruling party are now willing to press for reforms because they have yet to benefit from the flow of patronage and now have the option of defecting to the opposition. This turn of events is exactly what happened in Kenya where a coalition of ruling party backbenchers joined members of the opposition to create a viable coalition for change. This coalition then passed legislation, including a constitutional amendment, that formally de- linked the National Assembly from the executive. The coalition also raised the salaries of MPs. Both reforms have greatly enhanced the independence and authority of the assembly. Legislatures in Action This section explores from two distinct angles how legislatures operate in the four countries. First, it examines how they allocate effort across five generic tasks or functions that are common to legislators around the world. Second, it examines in some detail their role in relation to the budget process. Five Competing Roles and Functions African legislators are faced with at least five generic tasks that are common to legislators around the world: representation, constituency service, policy- making, oversight of the executive, and political recruitment. The first and second of these tasks are closely linked and depend in part on the form of electoral system used to elect the legislature. When members are elected from single-member districts, whether by a plurality or majority of the vote, the formal role of the legislator is to represent the inhabitants of his or her district. Whether the member actually strives to represent all inhabitants or only those elements that supported his or her election varies, of course, with the individual member and the characteristics of his or her district such as the social characteristics of its population or the presence of powerful interests. In proportional representation systems, MPs are, in the- ory, elected to represent the voters that supported their political party including the policy and ideological positions of the party. As discussed at the beginning of this chapter, however, the agrarian nature of most African countries, when combined with proportional representation in the context 244 Emerging Legislatures: Institutions of Horizontal Accountability of small multimember districts as in Benin or bloc representation in the con- text of small multimember districts as in Senegal, creates an incentive struc- ture that still encourages MPs to define their constituency in terms of the inhabitants of a specific geographic area. This incentive in turn means that constituency service is their primary task for the simple reason that their political careers depend on it. Representation of interests that are geographically defined and con- stituency service tend to crowd out representation of economic and other interests that may be important for the national economy and society, but which are not located or are only partially located within one’s district. Put differently, constituency service often crowds out the time African legisla- tors spend on policymaking, because policymaking deals with issues that are not salient among the constituents they represent. Devoting a substantial amount of time to policymaking is motivated by one of two considerations: either the issue in question affects one’s local constituency directly or one is concerned about the issue for its own sake. Some legislators undoubtedly have a greater ability or interest than others in “connecting the dots,” so to speak, between policy considerations at the national level and their indirect impact on one’s district. For example, in June 2002, the Kenyan National Assembly was the scene of an intense and well-attended debate over whether the government should honor its pledge to raise teachers’ salaries. The Kenya National Union of Teachers threatened a nationwide strike unless their pay was raised. Opposition MPs questioned the minister of education about the issue, because a teacher’s strike would have shut down all of Kenya’s public schools. The issue was also discussed at length within the education committee of the National Assembly and later became an issue in the general elections. Legislators who are predisposed toward playing a greater role in the policymaking process are more likely to participate in parliamentary debates. They are also more likely to participate in the work of the depart- mental committees that shadow government ministries. Indeed, our field studies confirmed that attendance and participation in parliamentary debates and committee meetings was greatest among a core group of legis- lators who want most to increase the authority of the legislature. Participa- tion in the work of parliamentary committees was likewise highest on the part of a small number of MPs who were personally committed to the task. These members tend to be better educated and younger than the member- ship as a whole though this generalization is not always valid. They are also more likely to be professionals by occupation. The result is that the higher the proportion of MPs who have these characteristics, the more Joel D. Barkan, Ladipo Ademolekun, and Yongmei Zhou 245 likely the legislature will develop a strong committee system and build its capacity for scrutinizing the policy implications of legislation proposed by the government. The same is true with respect to the likelihood that mem- bers will seek to engage in the budgetary process and call for the establish- ment of a parliamentary budget office as a core group of reformers have done in Ghana and Kenya. Several other factors also affect the propensity of MPs to spend more time on policymaking, particularly in the work of parliamentary commit- tees. MPs in both Ghana and Kenya stated that interest in national policy increased as the balance in numbers between the ruling party and the oppo- sition in the house narrowed. Once the opposition could block or come close to blocking government legislation, especially when opposition MPs were joined by backbenchers of the ruling party, debates took on added impor- tance and the government was more sensitive to opposition views. MPs in both countries also noted that because departmental committees meet in closed chambers, the committee system has been particularly effective in stimulating dialog and bargaining across party lines. During the Seventh Parliament in Kenya from 1992 through 1997, government and opposition MPs rarely spoke to one another. Indeed, MPs of the ruling party were under orders not to fraternize even in the parliamentary lounge with mem- bers of the opposition. During the Eighth Parliament from 1998 through 2002, however, such inhibitions faded following the establishment of the departmental committee system in 2002. The same was true in Ghana fol- lowing the 1996 parliamentary elections. Once the opposition became credi- ble in terms of numbers, dialog across party lines increased, especially within the departmental committees. Members also recognized that because parties rarely distinguished themselves from each other on the basis of ide- ology or policy, working with each other was relatively easy for members of different parties once a conducive venue for doing so was established. The result has been an increasing commitment to making the departmental com- mittee system work in both legislatures. MPs’ interest in engaging actively in committee work is also closely tied to the amount of resources available to both individual legislators and to the national assembly as a corporate body. Such resources are critical if legisla- tors are to allocate more time to policymaking and oversight. As noted in the previous section, the provision of more resources is no guarantee of a more competent legislature, but it is hard to imagine the development of an effective committee system without it. The propensity of members to devote time to oversight of the executive branch is more complicated. Oversight occurs mainly in two contexts—in 246 Emerging Legislatures: Institutions of Horizontal Accountability the departmental committees or technical commissions, which now exist in nearly all African legislatures, and in special parliamentary audit commit- tees, which are found only in anglophone Africa (that is, public accounts committees, and in a few countries, public investments committees). Parlia- mentary involvement in the audit process in francophone Africa, including Benin and Senegal, is more limited and indirect. In Benin, the National Assembly works with the Office of Accounts of the Supreme Court, whereas in Senegal the legislature’s role is limited to passing the Settlement Law, a report on the execution of the previous year’s budget, before voting a new budget. The Settlement Law, however, is prepared by the Ministry of Finance, and transmitted to the Accounts Office (Cour des Comptes), an independent body similar to the Auditor General’s Office, which reviews the execution and drafts a report. Both the law and the report are then sub- mitted to the National Assembly for a vote, but the last Settlement Law was passed in 1996. Indeed, the legislature elected in 2001 has never passed a Settlement Law, and most of its members do not know what it is (Thomas and Sissokho forthcoming, p. 20). Members are more or less likely to engage in oversight by departmental committees for the same reasons that they are likely to participate in these committees for the purpose of writing legisla- tion. Whether members are active members of the audit committees, how- ever, is also a function of the timeliness of the audits, and how the executive has responded to past audits. Where the audit process is current because the auditor general or the Accounts Office is able to complete audits within a few months after the end of the fiscal year, the motivation of legislators to serve on the audit commit- tees is much higher than when the auditor general is three to five years behind schedule in conducting the audits. The motivation is likewise higher where the executive (that is, the attorney general or the director of public prosecutions in anglophone countries) has demonstrated a willingness to prosecute those identified by the auditor general as responsible for irregu- larities in the expenditure of public funds. Unfortunately, the record is poor with respect to both considerations. MPs frequently refer to the audit com- mittees as “the postmortem committees,” because the committee considers reports by the auditor general that are several years out of date. The result is that little in the way of effective remedial action can be recommended by the committee. The same is true when the government drags its feet on prose- cuting those identified by the auditor general as likely wrong doers. Although audit committees can in principle be effective mechanisms for legislative oversight, they are unlikely to become such so long as these prob- lems persist. Joel D. Barkan, Ladipo Ademolekun, and Yongmei Zhou 247 Turning to the fifth function of political recruitment, clearly, substantial changes have occurred as a result of the political opening of the past decade. Although multiparty elections have been held in Senegal since 1974, and although Kenya held competitive elections for the National Assembly within a single-party format every five years before the return to multiparty politics in 1992, the elections of the second half of the 1990s resulted in a higher turnover in the four legislatures as shown in table 7.5. More than 60 percent of incumbents have been defeated in some of these elections, and few holdovers from the single-party era are still in office. Whether these high rates of turnover will be a persistent feature of multi- party politics or whether the rate will fall as time passes is hard to predict. A new generation of political leaders has been brought into power, even though the mean age of the legislature remains in the low 50s, and will rise if the rate of turnover falls. If this age breakdown is in fact the emerging pattern of parliamentary elections in Africa, then two contradictory out- comes are likely: first, the high turnover of membership of the legislature will, in the short run, accelerate the recruitment of some combination of reformers and patronage seekers to displace what remains of the political elite of the single-party era. Indeed, this process of displacement has proba- bly already occurred in three of the four countries considered in this study—all but Benin where Mathieu Kerekou is once again in power and has apparently gained the support of most members of the National Assembly. The recruitment of a new generation of politicians will increase demands for greater formal powers for the legislature even if the propor- tion of members who patronage seekers remains significant. Second, the high turnover of membership, which is rarely less than 40 percent, will slow the process of professionalization and the concomitant strengthening of the legislature. The process of institution building and capacity building, particularly the strengthening of the system of departmental committees requires a more stable membership of both the house as a whole and its constituent parts. Research on the development of American state legisla- tures (Squire 1992; Squire and Hamm forthcoming) confirms that the pro- fessionalization of these bodies did not occur until the level of turnover dropped and their memberships became more stable. Conversely, the impact of term limits on state legislatures appears to have undermined capacity if for no other reason than the level of institutional memory mem- bers is reduced. The push for strengthening the committee system and transforming the legislature into a player who matters is both accelerated and undercut by high turnover rates. Optimally, the high turnover experi- enced in recent elections would be followed by falling rates. Turnover rates 248 Emerging Legislatures: Institutions of Horizontal Accountability are likely to remain high, however, so long as elections are mainly referen- dums on incumbents’ records at constituency service. The experience of the legislature with respect to these functions varies in a manner consistent with our general thesis that legislative authority is weakest in Senegal and strongest in Kenya with Benin and Ghana falling in between, because the authority of the legislature as a corporate body increases only when the functions of policymaking and oversight of the executive take on greater importance relative to constituency service and when the nature of representation changes. The National Assembly of Sene- gal is weak, because the combination of the large PDS majority and the new constitution have reduced the legislature to a minor player. Members of the legislature thus devote substantial time to visiting their constituencies and relatively little to policymaking through participation in the parliamentary commissions (that is, committees). Participation in the budgetary process, as discussed in the next section, is nil. Participation in oversight of the exec- utive branch is also insignificant. Members also have little incentive to change how they allocate their time. The use of the proportional representa- tion list system of elections acts as a disincentive to challenge the party lead- ership, especially for members of the ruling party. Last, but not least, the resources available to legislators in terms of salary and institutional support are limited. The situation in Benin and Ghana is somewhat better. Although con- stituency service is equally important for these parliamentarians, the oppor- tunities for legislators to participate in policymaking and oversight are greater than in Senegal. The personal inclination of MPs to devote time to these functions also appears to be greater, especially in Ghana where the ruling and opposition parties are in near parity in the National Assembly. MPs in Benin and Ghana have greater constitutional authority than their colleagues in Senegal. The ruling party is not the personal vehicle of a presi- dent seeking to monopolize power. MPs are much better educated. Further- more although resources are limited, they receive greater support, both personally and institutionally, than in Senegal. For all these reasons, the members of these legislatures play an arguably greater role in the policy- making process, that is, in deliberating and amending legislation than do their counterparts in Senegal. Their role with respect to oversight of the executive, however, is limited, particularly in Ghana. The allocation of effort to policymaking and oversight is greatest among members of the Kenya National Assembly. Although constituency service probably remains the single most important function to which MPs give their time, their interest in the defining functions of the legislature is Joel D. Barkan, Ladipo Ademolekun, and Yongmei Zhou 249 increasingly important. Kenyan MPs are highly educated, younger, and more tightly connected to civil society, both urban and rural, than their counterparts in the three other countries. As such the proportion that are reformers as compared to patronage seekers is also greater. The balance of numbers between government and opposition was also relatively narrow from 1998 through 2002 when a core group of members embraced an agenda of building legislative capacity and asserting the independence of the National Assembly. Kenyan MPs have gone the furthest in strengthen- ing the committee system to help them deliberate and amend bills. Partici- pation in the audit committees is also higher than in Ghana, although less than participation in the departmental committees. Kenyan MPs have also gone the furthest in providing resources for themselves and their institu- tion. Indeed, this reason is the main one given by MPs for devoting more time to the core functions of the legislature itself. The Budgetary Process From the standpoint of the legislature, the budgetary process plays itself out in four stages: the preparation of the budget, deliberation and amendment, implementation, and audit. On average, MPs put virtually no effort into budget preparation, some into the deliberation and amendment and the audit, and most into implementation. The reason is a combination of the for- mal rules and the incentive structures they face. The preparation of the budget is an exercise done solely by the executive branch led by the ministry of finance. The constitutions of all four countries largely preclude the legislature from initiating or substantially amending the annual budget, a constraint reinforced by executive procedures. As such, these countries maintain a tradition inherited respectively at indepen- dence from Britain and France. In marked contrast to the United States, the legislature does not and is not expected to play a significant role as an initia- tor, co-initiator, or protagonist in the budgetary process. Rather, the consti- tutional charge to the legislature is to pass the budget, as proposed by the executive, into law. Moreover, in Benin, Ghana, and Kenya, the legislature is precluded to varying degrees from passing any legislation that incurs expenditures on the part of the government that does not provide sufficient revenues to cover its cost. In Benin and Senegal the president is empowered by the constitution to pass the budget by the decree if the legislature fails to do so within a specified period (60 days in the case of Senegal) or attempts to make major changes in what the government has proposed. During the past five years, the president of Benin has exercised this power on three 250 Emerging Legislatures: Institutions of Horizontal Accountability occasions, a move that has created considerable friction between the execu- tive and the legislative branches. The legislature is also at a distinct disadvantage with respect to its delib- eration and amendment of the budget. A major constraint is time. In all four countries, the budget is formally presented to the legislature by the minister of finance, and then the legislature is given a relatively short period of no more than two months deliberate and act on what the executive has pro- posed. Following its transmission to the legislature, the budget is referred to the departmental committees or technical commissions for deliberation and possible amendment after which it is reported back to the legislature as a whole, like ordinary legislation. The exercise is largely a formality in most cases and few if any amendments are made. In addition to having limited time, members are largely ignorant of the budgetary process with respect to how the budget is formulated within the executive branch and thus igno- rant on how to make effective input into the process. This shortcoming is especially true in Senegal and again reflects both the composition of the National Assembly and executive dominance overall. The limitations of time was a particular complaint of some Ghanaian MPs, including members of the finance committee, who noted that although the executive usually completed its preparation of the budget and transmitted it in written form to the National Assembly by the end of December, the minister of finance did not formally present the budget until March, thus keeping MPs in the dark. Not surprisingly, they want this procedure changed. Notwithstanding these limitations, MPs in Ghana and Kenya are increasingly interested in making their input felt on budgetary issues. For the last four years, Kenyan MPs including members of the finance commit- tee, have attended workshops organized by three think tank organizations to explain the budget. These workshops are highly appreciated by the MPs who attend and are becoming an institutionalized feature of parliamentary life. Those who have participated in these exercises are now proposing that the recently established Parliamentary Serviced Commission establish a parliamentary budget office to provide an ongoing liaison with the Ministry of Finance. The proposed budget office, which is modeled roughly on the Congressional Budget Office in the United States, would employ its own analysts to monitor the Ministry of Finance and explain features of the bud- get to members. Not surprisingly, this proposal has been resisted by the executive, first by former president Moi and now by the minister of finance for the new government. They view demands to establish a Parliamentary Budget Office as infringing on affairs that have historically been the prerog- ative of the executive. The same is true in Ghana where members have also Joel D. Barkan, Ladipo Ademolekun, and Yongmei Zhou 251 called for the establishment of a parliamentary budget office, but where the immediate concern is not participation in determining the national budget but rather the assertion of parliamentary control over its own finances as has occurred in Kenya. In Ghana, Parliament has made no changes in recent years to the budget allocations presented by the Ministry of Finance. In 2000, however, MPs substantially amended a proposal by the minister of education to set aside a 2.5 percent of an increment in the value added tax to establish a special general education trust fund. The purpose of the amend- ments was to guarantee that the new revenues would be spent on education and that the fund would be operated autonomously from the ministry. Although this action was not strictly speaking, a change of the budget, it was a milestone in that the National Assembly specified how revenues were to be spent. The National Assembly in Benin has periodically amended budgets pre- sented by the executive branch, so several recent budgets have been enacted by President Kerekou by decree. Resorting to this practice, though permissible under the Constitution, has created considerable friction between the executive and legislative branches and represents something of a throwback to the single-party era when the legislature was a rubber stamp of the executive. In the previous section, we discussed the involvement of MPs in the audits of government expenditures and need not repeat that discussion here. Suffice to say, members are unlikely to devote substantial amounts of time to serving on audit committees until the accounts to be audited are rel- atively up to date (that is, that they are no more than six months to a year in arrears) and evidence is clear that those suspected of wrong doing will be thoroughly investigated and, if warranted, prosecuted by the state. Failure to meet these conditions will, for all practical purposes, nullify the powers that legislatures have for maintaining fiduciary accountability on the part of the government. MPs devote considerable time and effort to stage three in the budget process: implementation. Involvement at the implementation stage, how- ever, is not through collective action with other members, but in the form of individual lobbying of ministers and assistant ministers by MPs to ensure that the programs the ministers control are implemented in the constituen- cies of the MPs. This goal is particularly significant with respect to the con- struction of physical infrastructure such as roads, but also applies to the full range of government services, including water, education, and health. Back- benchers of the ruling party seek to convince ministers that their reelec- tions—both the backbenchers’ and the ministers’—depend on the provision 252 Emerging Legislatures: Institutions of Horizontal Accountability of such services, because the future of the government majority is at stake. MPs of the ruling party may thus have more clout than those of the opposi- tion. One National People’s Party member in Ghana, however, noted that now that his party is in power, his constituents will not be as forgiving if he fails to deliver the goods as they were when his party was the opposition. With all members making similar arguments, however, the result is a con- stant scramble by backbenchers for resources for their districts. As previ- ously noted, ministers and assistant ministers also lobby their counterparts in other ministries and are sometimes in a position to trade favors, whereas backbenchers are not. Conclusion: How Rapidly Can Legislatures Evolve? The wave of political liberalization and democratization that swept across Africa during the 1990s gave rise to an expansion of legislative authority in some countries, but not all. Indeed, the extent of legislative authority varies greatly across the continent. With respect to the four legislatures considered for this study, we found that the authority of the legislature ranged from being very weak in Senegal to moderately strong and expanding in Kenya with Benin and Ghana falling somewhere in between. In other words, the pilot study captured a greater variance in legislative authority than antici- pated at the outset of our inquiry. The question then is, what determines the nature and power of African legislatures within the context of liberalization? Our field studies suggest that the extent of a legislature’s authority is largely a function of the evolving incentive structure confronting individual members of the legislature. We now hypothesize that this incentive struc- ture is the changing of three sets of variables: • The structure of African society (agrarian and plural, that is, multi- ethnic), which gives rise to clientelist politics and weak political par- ties with little or no programmatic or ideological orientation • The formal rules of the political system, that is, the constitutional pro- visions regarding the role of the legislature, the method by which its members are elected • The internal structure of the legislature and the resources available to its members, especially their level of compensation. Of the three sets of incentives, only the second and third are susceptible to change in the near term. The third set of incentives is especially interesting, because its con- stituent variables exhibit the greatest variance and because they are most Joel D. Barkan, Ladipo Ademolekun, and Yongmei Zhou 253 likely to be altered by the members of the legislature. The most significant (and controversial) of these is the level of compensation of MPs. Where MPs are able to raise their pay significantly, the structure of incentives changes so that individual members can now devote more time to legislating and exec- utive oversight. Where MPs also provide greater resources in support of their collective duties, they are more likely to spend more time on these duties. This time commitment in turn begins to transform the legislature into one that matters within the political system and that draws greater attention from both the executive, on the one hand, and civil society, on the other. Neither the rules constraining the legislature nor the resources available to its members and its internal structure are likely to improve until a “criti- cal mass,” so to speak, of reformers and others are elected to these bodies and then come together in a viable coalition for change. Put differently, just as democratization is a process driven mainly by forces within the political system, so too is the strengthening of the legislature mainly a function of actors within these bodies.20 A key research question, therefore, is why in some countries some members are more assertive and collectively seek to overcome the structural constraints that limit legislative authority. We have sought to provide some tentative answers to this question by suggesting that legislatures are most likely to provide the resources required to both change the incentive structures for their members and build capacity when they have a critical mass of members who are reformers rather than incum- bent authoritarians or patronage seekers. As noted at the outset, liberalization or democratization is a necessary, but insufficient, condition for the emergence of reformers within the legisla- ture. At best, it constitutes an improvement of the enabling environment. Civil society and the donor community can also encourage and provide use- ful technical support to those members of the legislature who are commit- ted to enhancing its performance, but the onus is on the MPs themselves to change the institution to which they belong. If they do not embrace reform it will not occur. The modernization of the African legislature, indeed of any national legislature cannot be orchestrated from outside. The experiences of the four legislatures considered in this study support this conclusion. Where, as a result of local conditions, a group of relatively 20. Thus the impact of donor support for strengthening the legislative process, by itself, is unlikely to have much impact without the active cooperation, indeed invitation, on the part of key leaders within the legislature, for example, the presi- dent or speaker, as well as a core group of MPs. 254 Emerging Legislatures: Institutions of Horizontal Accountability young and highly educated members are presented with an opening to enhance both their own situation and legislature, they will do so. Where, by contrast, such opportunities do not exist or can be exercised only at high personal cost or where the number of legislators seeking to strengthen their organization are few, the status quo of the legislature will continue indefi- nitely and the body will remain a weak player. The contrast between Kenya and Senegal illustrates this generalization. In Kenya the combination of a situation of near parity between government and opposition in the National Assembly and a relatively large group of young reform-oriented MPs led to demands for strengthening the legislature which in turn gave rise to efforts to change the rules, raise salaries, build capacity, and undertake other reforms. By contrast, in Senegal, where the ruling party has an overwhelm- ing majority and where few members seek reform, the legislature remains weak—and promises to remain so for the foreseeable future. In Benin and Ghana, the interest expressed by some members to enhance the legislature is genuine, but it does not have the intensity or backing found in Kenya. Whether the number of reform-minded MPs, and hence the demand for change, will rise in all four countries, and especially in those where the leg- islature remains weak, is hard to predict. We have already suggested the conditions that give rise to this type of politician, but these conditions are not easily brought forth simply because they provide a supportive environ- ment. We are, therefore, confronted by a dilemma that may be beyond solu- tion. Our field studies have specified the configuration of incentives that have led to the strengthening of a number of legislatures in the African con- text and the type of legislators that have created these incentives where they did not previously exist. We have also suggested the structural context within which such reformist legislators are most likely to emerge and where they are not. Our findings may mean that the emergence of a sufficiently large group of reformers is itself a function of conditions beyond manipula- tion, although we are reluctant to reach such a deterministic conclusion. References The word “processed” describes informally reproduced works that may not be commonly available in libraries. Alderfer, Philip W. 1997. Institutional Development in a New Democracy: The Zambian National Assembly, 1964–1996. Ph.D. dissertation. East Lansing: Michigan State University. Barkan, Joel D. 1976. “Further Reassessment of the ‘Conventional Wisdom,’ Political Knowledge and Voting Behavior in Rural Kenya.” American Political Science Review 70(2): 452–55. Joel D. Barkan, Ladipo Ademolekun, and Yongmei Zhou 255 _____. 1979. “Bringing Home the Pork: Legislative Behavior, Rural Development, and Political Change in East Africa.” In Lloyd Musolf and Joel Smith, eds., Legis- latures and Development. Durham, NC: Duke University Press. _____. 2002. “The Many Faces of Africa: Democracy Across a Varied Continent.” Harvard International Review 24(2): 72–6. Bratton, Michael, and Nicolas van de Walle. 1997. Democratic Experiments in Africa. Cambridge, U.K.: Cambridge University Press. Carothers, Thomas. 2002. “The End of the Transition Paradigm.” Journal of Democ- racy 13(1): 5–21. Dahl, Robert A. 1989. Democracy and Its Critics. New Haven, CT: Yale University Press. Freedom House. 2002. Freedom in the World. Washington, D.C.: Freedom House. Gamm, Gerald, and John Huber. 2004. “Legislatures as Political Institutions: Beyond the Contemporary Congress.” In Ira Katznelson and Helen Milner, eds., The State of the Discipline. Washington, D.C.: American Political Science Association. Huntington, Samuel P. 1991. The Third Wave: Democratization in the Late Twentieth Century. Norman, OK: University of Oklahoma Press. National Council of State Legislatures. 1996. Total Legislative Turnover 1994–96. http://wwws.ncsl.org/programs/legman/elect/tottrn.htm. Ottaway, Marina. 2003. Democracy Challenged: The Rise of Semi-Authoritarianism. Washington, D.C.: Carnegie Endowment for International Peace. Squire, Peverill. 1992. “The Theory of Legislative Institutionalization and the Cali- fornia Assembly.” Journal of Politics 54(4): 1026–54. Squire, Peverill, and Keith Hamm. Forthcoming. 101 Chambers: Congress, State Legis- latures, and the Future of Legislative Studies. Columbus: Ohio State University Press. Thomas, M. A., and Oumar Sissokho. Forthcoming. “Liaison Legislature: The Role of the Legislature in Senegal.” Journal of Modern African Studies. van de Walle, Nicolas. 2002. “Elections Without Democracy: Africa’s Range of Regimes.” Journal of Democracy 13(2): 66–79 8 Process Interventions Versus Structural Reforms: Institutionalizing Anticorruption Reforms in Africa Sahr Kpundeh Over the course of the 1990s, Africa’s development discourse has been pro- foundly reshaped from one that focused narrowly on the technical dimen- sions of economic policy, to one that has brought to center stage notions of good governance: democracy, participation, transparency, and accountabil- ity. This shift can be traced in large part to the wave of democratization that swept the continent in the early 1990s. An important moment in the process came, however, in a speech in October 1996, when World Bank President James Wolfensohn targeted the “cancer of corruption” as a central obstacle to development effectiveness. At least since that time, virtually every African country and every development agency have pledged their support for the fight against corruption. At this writing, six years later, much of the early anticorruption rhetoric seems increasingly hollow. As chapter 1 summarized, data show no system- atic evidence of a decline in corruption between 1996 and 2002. Instead, some suggest that the single most evident consequence of the attention to corruption has been a resurgence of cynicism among Africa’s citizens—a result of the day-by-day media drumbeat of evidence of nefarious activity with few signs that things are getting better. The questions arise: Was this emphasis on fighting corruption misplaced? Or, even more troubling, was the dire character of the disease correctly diagnosed, only subsequently to discover that the disease is, in fact, incurable? 257 258 Process Interventions Versus Structural Reforms The central arguments of this chapter are that shining the spotlight on corruption was not a mistake and that the situation is not hopeless. Rather, the error was in presuming that the cure lay in narrowly targeted anticor- ruption, data-collection, and consciousness-raising campaigns—process interventions that exhorted citizens and politicians to transform their behavior. The crucial lesson over the past six years has been that the roots of corruption lie in dysfunctional state institutions. Anticorruption cam- paigns are not a substitute for the difficult tasks of public sector reform and capacity building, which are the subject of this volume. Even so, as the next section develops schematically, an emphasis on fighting corruption can play a valuable role as part of a virtuous spiral of mutually reinforcing process and structural interventions. In practice though, as “The Impact of Dedicated Anticorruption Institutions” details, little has been done across the African continent to encourage links between process and institutional interventions. The failure to recognize the benefits of a two-pronged reform approach has impeded the ability to build a legitimate, credible, and institutionalized effort to address corruption and ultimately to sustain change. Process Interventions and Structural Change—A Framework Process interventions or awareness-raising campaigns, which inform and educate the citizenry about the impact of corruption, are often the first steps in changing a culture of corruption and weak service delivery. As the exam- ples in table 8.1 highlight, they can be grouped into three categories: national efforts, local or citizen efforts, and populist initiatives. Over the past decade, each of these has seen extensive use across Africa. Table 8.1. Types of Process Interventions Intervention Examples National efforts Corruption inquiries, awareness raising, training within state and public institutions Local or citizen efforts Community oversight, media, coalitions with stakeholders Populist initiatives Military coups d’etat, moralization campaigns, civil servant purges, public humiliations and executions, quasi-official tribunals, property seizure, heavy fines, and imprisonment Source: Adapted from Kpundeh (2004). Sahr Kpundeh 259 National Efforts to Fight Corruption National efforts have generally aimed to change public officials’ attitudes through education, investigation, and prevention. The most familiar exam- ples are the anticorruption agencies that today dot the African landscape. Examples include the Commission on Human Rights and Administrative Justice and the Serious Frauds Office in Ghana, the Kenya Anti-Corruption Commission, the Anti-Corruption Bureau in Malawi, the Prevention of Cor- ruption Bureau in Tanzania, the Anti-Corruption Commission in Sierra Leone, and Inspector General of Government in Uganda (to mention just a few). This chapter will examine these agencies, and the reasons for their mixed performance, in more detail. Local or Citizen Efforts to Fight Corruption Local or citizen efforts aim to pressure public officials to use public resources for civic, not private, purposes; they are an important part of the mix of activi- ties needed to fight corruption effectively. Establishing civil society organiza- tions is a homegrown strategy, which demonstrates ownership and improves chances for sustainability. A fervent civil society, which is essential for the kind of structured political competition that can remedy problems inherent in systems that lack civil liberties, political freedoms, and incentives, reinforces the political will needed for reform (Johnston 1998; Kpundeh 1998). Although civil society in many African countries traditionally has been weak and divided, and its contributions to improving governance spo- radic, in the early 1990s the pro-democracy movements in Africa drew large numbers of people into political debate and organization. Civil soci- ety organizations proliferated over the next decade. These included dedi- cated corruption-fighting organizations. The influence of Transparency International—local chapters of which have been established in several countries including Ghana, Kenya, Uganda, Zambia, and Zimbabwe—has been especially noteworthy. Some organizations have also adopted the role of civic monitor, conducting locally designed service delivery surveys to determine the efficiency of government services. These surveys have been useful in Ghana, Kenya, Tanzania, Uganda, and Zimbabwe to send strong messages to the government and have resulted in specific strategies to improve service delivery. Despite this proliferation of organizations, and despite a formal commit- ment to openness, many African governments do not encourage genuine political competition and citizen participation. Institutions such the Com- mission on Human Rights and Administrative Justice in Ghana, as the 260 Process Interventions Versus Structural Reforms Ombudsman in Malawi, and the Public Complaints Commission in Nigeria that investigate citizen complaints or protect people against arbitrary state harassment have been in constant conflict with government officials. Although many are constitutionally established, most governments view them as adversaries and attempt to stifle their progress by intimidating their officials and controlling their budgets. Furthermore the inevitable budget cuts have led to greater financial dependence on bilateral and multilateral donors. Populist Initiatives to Fight Corruption Populist initiatives such as civil servant purges, arrests, detention, and pub- lic humiliation of senior government officials often have accompanied mili- tary coups d’état. In 1979, corrupt officials were executed in Ghana during the Jerry Rawlings administration and in Liberia under Samuel Doe in 1980. Following the 1992 coup in Sierra Leone, the military regime confiscated corrupt public officials’ property. Military governments characteristically institute commissions of inquiry to investigate deposed government offi- cials for embezzling and mismanaging government funds. In Sierra Leone, the Valentine Strasser government immediately established three commis- sions of inquiry charged with the investigation of former senior government officials in the Siaka Stevens and Joseph Momoh regimes—all public offi- cials, members of the board and employees of parastatals, the armed forces, and the police force (Kpundeh 1994). In the Sierra Leone case, and in others such as The Gambia, Ghana, Liberia, and Nigeria, quick-fix measures, anticorruption purges, and moral- ization campaigns have not been successful. These campaigns fail to focus on institutionalizing structural reforms or establishing durable and lasting legitimate political processes. Each coup weakens an already fragile civil society, which unfortunately is one of the military’s prime targets. Olowu (1992) argues that “the existence of political will to genuinely tackle corrup- tion must not be assumed on the basis of policy pronouncements by politi- cal leaders. Those most vocal often turn out to be the ones who aid and abet corruption most” (p. 13). Structure–Process Interactions Analysts increasingly recognize that the disabling weakness of an exclu- sive reliance on process interventions to combat corruption is the presup- position that exhortation and moral suasion will be sufficient to transform Sahr Kpundeh 261 the behavior of both public officials and citizens. As table 8.2 summarizes, however, the effectiveness of efforts to combat corruption depends at least as much on the quality of the core public institutions and processes, which have been the focus of the other chapters in this volume. Table 8.2 distinguishes between the three central elements of the fight against corruption: prevention, detection, and enforcement. As the first row of table 8.2 highlights, core public institutions play a central role in all three. Institutionally structured check and balance arrangements, supported by Table 8.2. Arrangements for Combating Corruption Entity Prevention Detection Enforcement Core public Checks and balances— Police Justice system institutions separation of legislature, Parliamentary administration executive and judiciary oversight (horizontal accountability) Community Budget management oversight (treasury, procurement, Ombudsman audit) Auditor general Declaration of public officials’ Free media assets Political competition and accountability Freedom of information; Meritocratic civil service/ adequate pay Decentralization (vertical accountability) Dedicated Ombudsman Anticorruption Anticorruption anti- Anticorruption agencies out- agencies— agencies—seize corruption reach programs and semi- conduct property and programs nars for business searches, incarcerate leaders—help public and examine bank individuals nonstate actors identify accounts, strategies for reducing cor- audit private ruption; fund studies of assets, moni- corruption; establish com- tor lifestyles munity relations highlight- ing awareness, prevention and sanctions Source: Author. 262 Process Interventions Versus Structural Reforms transparency, serve to prevent corruption. Governance structures, which oversee how public resources are used, combined with policing and prose- cutorial services, which investigate criminal activity, detect corruption. Anticorruption enforcement is the responsibility of the justice system. The second row of table 8.2 summarizes how dedicated anticorruption agencies engage each of these efforts. As table 8.2 implies—and as is explored further in the next section—if the core public institutions are in place, these dedicated agencies can add value on the margin, buttressing prevention by intensifying public awareness and strengthening detection by adding to the supply of specialized anticorruption investigatory capabil- ities. The more difficult question posed by table 8.2 is, What is the value of these dedicated agencies if the core public institutions are underdeveloped? One possible answer could be that these agencies add no value because they lack any impartial and credible means of enforcement, and indeed they might even have a negative impact because they detract from the need for systemic reforms. Alternatively, anticorruption agencies might add value as a stopgap, an interim measure enabling a newly empowered political lead- ership that genuinely is committed to stamp out corruption to demonstrate some results in the short-term, even as it embarks on the long-term process of institutional reform. A third answer derives from considering the interac- tion between process and structural interventions from a more dynamic perspective. Viewed in this way, dedicated anticorruption programs emerge as an intermediate hybrid—more than process, but less than institutional- ized structures, in a sequence of change in which process interventions are linked with structural change in a virtuous cycle—a means to educate and move changes forward. Increased participation generates pressure for struc- tural interventions that improve public performance, and the resultant structural interventions sustain change by encouraging more participation, education, and hence another round of calls for more refined structural interventions (figure 8.1). Put differently, political leaders committed to building state capacity over the long-term, might nonetheless choose to embark on complementary process reforms. The regime leader may be committed to anticorruption, but may lack the leverage and power to implement change because of clien- telism, patronage politics, and the active opposition of political elites who want and need to maintain the status quo. For example, Kenya’s recently elected president, Mwai Kibaki, must calculate the delicate balance of power between satisfying the patronage demands of his diverse coalition and reducing the flow of favoritism, one of the major sources of corruption. Sahr Kpundeh 263 Figure 8.1. Virtuous Spiral Versus Burn-Out Populism Structural institutions Goal of process intervention is Structures participatory empower structural participation reforms Process interventions Source: Author. The previous government’s reliance on patronage to distribute the spoils resulted in an elite political culture and set of norms that may prove extremely difficult for President Kibaki to dismantle. President Olusegun Obasanjo’s government in Nigeria and the government of Ghana both face similar challenges. In such settings, process interventions can be useful to push and shake the dynamics within the public sector by raising civic awareness about the damaging effects of corruption in society, as well as highlighting the bene- fits of good governance and improved service delivery, and consequently empowerment of the citizenry and local government. Such interventions can be foundations or entry points for structural reforms, because they engender participatory governance that raises awareness about misgover- nance in societies and hence helps accelerate momentum for change. As this cycle of change proceeds, it can target broader structural complexities, including the internal organizations of political systems, relationships among core state institutions, the interaction between state and nonstate actors, deregulation, and civil service reform. The Kenya example in box 8.1 illustrates the consequences of a cumulative change process that is both complex and contradictory, but ultimately leads to some improvements in transparency and accountability in governance. 264 Process Interventions Versus Structural Reforms Box 8.1. Kenya: Linking Process and Structural Intervention for Cumulative Change Challenges to President Daniel arap Moi’s rule emerged in the late 1980s, spear- headed by church activists and joined by disaffected politicians and activists who began to agitate for change to a multiparty system. The donors, who were acutely aware of domestic calls to suspend further aid to Kenya, were watching events closely. In 1991, donors coordinated their actions to cut off nonhumani- tarian balance of payments support pending economic and political reforms. To restore critical aid flows, Moi capitulated and allowed reforms that paved the way for the multiparty general elections of 1992 and the formation of opposi- tion parties. Following commonwealth parliamentary rules, the opposition members of parliament found themselves chairing key parliamentary watch- dog committees such as the Parliamentary Accounts Committee. These com- mittees reviewed the controller and auditor general reports, which were constantly given prominence in the media. For example, parliamentary over- sight bodies along with opposition parliamentarians, civil society, and non- governmental organizations debated the reports while at the same time calling for suspension of further aid to Kenya. This turmoil forced the government to form a tribunal to investigate the Goldenberg’s scandal, a costly financial scan- dal that had rocked Kenya’s economy since 1992. (The scandal, in which the country lost up to US$600 million through questionable exports of gold and diamonds between 1990 and 1993 even though the country does not actually produce either commodity, is Kenya’s biggest corruption scandal.) In 1994, the International Monetary Fund suspended its lending to Kenya forcing the gov- ernment to commence legal proceedings against Goldenberg’s owners. The Goldenberg investigation was reopened by the Kibaki administration and is still ongoing. During the early 1990s, the official response to the problem of corruption in Kenya came in the form of administrative and disciplinary measures under the guidance of the attorney general and through the judiciary. Because of the nature of patronage politics and political dominance of President Moi and his close allies, however, corruption was not only endemic, but also served as the glue to hold the system together, allowing subordinates and close allies to embezzle public resources in return for loyalty and support. Consequently, after the return of multiparty politics the world witnessed what is referred to as the “precursor” or “mobilization” phase, when several prominent nonparty organizations and activists protested against the authoritarian nature of the state. The second phase (1997–2000) in this cumulative change process was the brief phase when the so-called “dream team” was put in place in response to external pressure from donors and internal pressure from a small group of par- liamentarians and from civil society organizations such as the Kenya chapter of Transparency International. The dream team was appointed in 1999 to implement a good governance program and lay the foundation for reforms in the public service. Their services were terminated in 2001. Another important Sahr Kpundeh 265 development during this period occurred when on December 22, 2000, the Kenya Anti-Corruption Authority (KACA) was declared unconstitutional by a constitutional court. The court ruled that the establishment of the KACA vio- lated Kenya’s Constitution, which states that only the attorney general and the police could prosecute criminal offences. The KACA situation illustrates that the Prevention of Corruption Act in Kenya, upon which such initiatives as KACA depend, does not give the resultant anticorruption bodies sufficient powers to enforce its provisions. The Parliament during this period was almost evenly balanced: Moi’s ruling party, the Kenya African National Union, had 136 members including those nominated by Moi, whereas the six opposition parties combined had a total of 86 members during the Seventh Parliament. Together with the donors, the Parliament put strong pressure on the Moi gov- ernment to address problems of corruption and improve accountability and transparency within his administration. Some of the donors withheld dis- bursements and scaled back programs until the government took concrete actions. The Kenya chapter of the African Parliamentary Network Against Corruption, the Kenya chapter of Transparency International, and other civil society bodies in Kenya also were visible in demanding accountability and transparency from the Moi administration. The third phase (2002 election phase) brought in a new government, the National Rainbow Coalition that has pledged to make the fight against corrup- tion a priority crusade. The African Parliamentary Network Against Corruption still continues to play a role in the fight against corruption. Twelve African Par- liamentary Network Against Corruption members got reelected to the house and eight of them are in cabinet—a reason for the strength of the anticorruption crusade in the government. Additionally, the former head of the Kenya chapter of Transparency International is the current permanent secretary of ethics, reporting to the president. In collaboration with the newly created Ministry of Justice and Constitutional Affairs, they are working to mainstream the anticor- ruption fight in the country. This phase has also seen several changes in lan- guage of legislation and the institutional framework to address corruption: the passage of the Economic Crimes Law and the Ethics and Code of Conduct Law, which mandates declaration of assets; the establishment of a new Ministry of Justice and Constitutional Affairs to coordinate the anticorruption efforts; and the reconstitution of the Kenya Anti-Corruption Commission with prosecutor- ial powers. Thus, the Kenya case highlights both the complexity of addressing corruption as well as the contradictions of linking the process and structural interventions in a cumulative change process beginning in the 1990s. The Impact of Dedicated Anticorruption Institutions Several African countries have developed anticorruption agencies as the cen- tral platform for controlling administrative corruption. Although a few have produced significant and sustained reductions in corruption, most have 266 Process Interventions Versus Structural Reforms mixed records of success—some have demonstrated limited progress but many are simply a façade, a way to appease donors who threaten to reduce or eliminate aid to countries that fail to address unbridled, growing corrup- tion. Because many African nations have adopted the specific elements that have proven effective in Australia, Hong Kong (China), and Singapore, one must ask why these anticorruption reform efforts in Africa have been unable to replicate the success these measures have enjoyed elsewhere. This section considers this issue through the lens of the framework devel- oped in the last section. It explores, first, the extent to which the modest record of Africa’s anticorruption programs can be linked to weaknesses in complementary core state institutions and, second, the extent to which these programs have played a positive catalytic role in a dynamic sequence. Before turning to these issues, however, I will set the stage by looking more closely at the key characteristics of some successful agencies outside of Africa. Characteristics of Well-Functioning Anticorruption Agencies Successful anticorruption agencies share five characteristics: independence, budget predictability, a workable core, a free media, and accountability. This section discusses these characteristics using illustrations from agencies out- side of Africa as benchmarks for assessing their African counterparts. First, independence is fundamental for an effective anticorruption agency. Genuine anticorruption measures cannot initiate or cancel investi- gations; impose, withhold, or modify punishments; or fail to take action. They stand free of the influence of powerful individuals or factions. Addi- tionally, these agencies investigate corruption anywhere it is suspected and follow the trail of evidence wherever it leads. Second, good anticorruption agencies are well-financed operations and have budget predictability. For example, the budget for Hong Kong’s Inde- pendent Commission Against Corruption (ICAC) was US$94 million for 2002 and US$90 million for 2001 (ICAC 2002). The New South Wales, Aus- tralia, ICAC operates with an annual budget of about US$8.7 million (John- ston 1999). In Singapore, the Corrupt Practices Investigation Bureau’s (CPIB’s) budget is not separately published, but it was reported to be US$4.33 million in 1986 (Quah 1999, p.10). Third, key complementary institutions—for example, justice systems, courts, police, and prisons—work relatively well. Additionally, adequate laws and procedures have been established, and basic features of the rule of law and other capable institutions, such as supreme audit institutions and other oversight bodies, are in place. In Hong Kong (China), for example, the Sahr Kpundeh 267 authorities recognized the value of a legal framework within which the ICAC should operate and revised existing legislations to ensure the agency was set up with a broad mandate to investigate allegations of corruption as well as take effective preventive measures. Similarly, the CPIB in Singapore and the New South Wales ICAC also have the backing (after repeated amendments) of strong enabling legislation. Fourth, the media plays a critical role in raising citizen awareness of anti- corruption efforts and practices fair and objective reporting. Hong Kong’s (China) ICAC Community Relations Department, for example, is noted for its unique outreach program. It influences television programming (the ICAC is written into dramatic series and also presents its own public service messages); develops educational programs; and sponsors sporting, cultural, and entertainment events that are aimed at youth and that emphasize anti- corruption themes. Community relations’ activities also include a special hotline and ever-evolving effort to find special opportunities to build public support for reform efforts. Such an elaborate outreach program can be effec- tive only if the media is free from brutal suppression and is allowed to con- duct investigations and publish stories without fear of reprisals. When the media is free and objective they can function as an instrument to help raise, and even help address, issues about the ills in society. To expose corruption, economic mismanagement, and abuse of political power, the freedom of expression is fundamental and can help in the search for internal solutions. Fifth, effective anticorruption agencies are themselves accountable for performance and have checks and balances built into that accountability to ensure that independence is not compromised—the so-called process of guarding the guardians. A powerful independent agency is also capable of committing abuses. Anticorruption entities must carefully maintain accountability and due process for the public and private sectors and, most important, for themselves. The Hong Kong (China) ICAC, for example, has oversight from three advisory committees, the Corruption Prevention Advi- sory Committee, the Citizens Advisory Committee on Community Rela- tions, and the Operations Review Committee. Composed of specially appointed members drawn from the larger community, these committees meet regularly to review the ICAC’s activities. The New South Wales ICAC is held accountable to the citizens through the multiparty Parliamentary Joint Committee, which monitors its activities; through the Operations Review Committee, which includes members of the public and various gov- ernment agencies; through its obligation to report to the public annually as well as on its major investigations; and through the State Supreme Court, which can review commission findings on certain grounds. 268 Process Interventions Versus Structural Reforms The Role of Anticorruption Agencies in Africa Table 8.3 uses the five characteristics highlighted above to assess the diffi- culties and challenges of anticorruption agencies in seven African countries in terms of the various roles they can play in improving governance and addressing corruption. The first case, from Botswana, illustrates how the Directorate for Corruption and Economic Crimes (DCEC) works to comple- ment the state institutions primarily because the core institutions are fairly well developed, and a political dynamic within the country promotes democratic values and processes. The second set of country cases, from Sierra Leone and Malawi, will illustrate how anticorruption agencies are used merely as phony populism—a way to appease donors while the agency’s operations are doomed from its establishment because of political interference and legal constraints. The third case studies, from Uganda, Tan- zania, and Zambia, will reveal how anticorruption agencies can operate as a useful short-term substitute to address corruption when state institutions are weak. In such situations, these agencies are used by the political leader- ship to demonstrate results in the short term by using their power to lever- age change by building momentum for structural reforms. A fourth case on Kenya, illustrated in box 8.1, describes the cumulative effect of anticorrup- tion efforts that have gone through several phases and spanned a number of years, highlighting the links between process and structural interventions. An Anticorruption Agency as a Complement to State Institutions The DCEC in Botswana distinguishes itself among African anticorruption agencies primarily because it has not only replicated the three-pronged Table 8.3. Some Features of Anticorruption Agencies in Selected African Countries Budget Workable Free Country Independence predictability core media Accountability Botswana — + + + + Kenya — — xx xx — Malawi — — — — — Sierra Leone xx — xx xx — Tanzania — — — — — Uganda + — — — — Zambia — — — xx — + Strong. — Fair. xx Weak. Source: Anticorruption agencies; various reports; media ratings from Freedom House (2002, 2003). Sahr Kpundeh 269 approach of investigation, prevention, and education from the success stories mentioned earlier, but it also has produced some successful outcomes in its role of complementing the state institutions established to improve gover- nance and deepen accountability and transparency in the country. Several reasons could be advanced for such successes. First, DCEC’s statute specifies that it is an independent agency that provides community outreach pro- grams to public and private sectors on the costs of corruption. The DCEC was established by a series of public information campaigns aimed at civil society groups and grassroots organizations, resulting in high public awareness about its role and responsibilities. Second, the complementary core institu- tions work relatively well—a consequence of its long tradition of rebuilding its democratic state structures to deepen accountability, as well as to establish a check and balance schematic within their governance structures. The DCEC has been relatively successful in dealing with cases of corruption, most of which have been brought before the courts and resulted in prosecutions and sentences. Table 8.4 gives a breakdown of some of the output measures show- ing the effectiveness of the DCEC in terms of cases investigated, prosecuted as well as the number of convictions (DCEC 1999, 2000, 2002). Table 8.4. Measures of Anticorruption Agencies, Selected African Countries and Years Number Percentage Percentage of reports of reports of cases Percentage Percentage Country investigated concluded prosecuted of convictions of acquittals Botswana 1,388 49.4 12.5 5.8 0.1 (1997–2002) Kenya 345 12.2 24.3 1.7 3.5 (2002–2003) Malawi 2,697 50.6 2.2 0.3 0.3 (1997–June 2003) Sierra Leone 383 31.6 3.7 0.5 0.3 (September 2000– September 2003) Tanzania 2,086 100.0 11.7 1.4 3.1 (1997–2003) Uganda 2,420 45.6 1.1 0.2 0.2 (January– December 2002) Zambia 1,896 — 14.8 6.5 — (1997–2002) — Not available. Source: Selected reports. 270 Process Interventions Versus Structural Reforms The political and incentive structure in Botswana also contributes to the important role the agency plays in complementing the core institutions. For example, the media reports objectively, is free from political repression, and plays a visible role in raising awareness about the ills of corruption. The DCEC receives a lot of its complaints and leads for investigation from the media, other government departments, Parliament, and its own in-house intelligence department (Olowu 1999). Table 8.4 shows that the directorate has done relatively well in concluding and prosecuting cases reported to it. In addition, the DCEC’s public educational program is one of the most important achievements of the organization. It has taken its crusade against corruption to schools, universities, ministries, churches, and other organiza- tions through activities such as talk shows, television and radio phone-in shows, posters, and newspaper advertisements. In terms of incentives, the civil servants in Botswana, including officials of DCEC are relatively well paid in comparison to private sector employees and are reputed to be gener- ally efficient and incorruptible. An additional reason for the DCEC’s operational success in comple- menting anticorruption efforts of state institutions derives from its pre- dictable operating budget from the government. The total budgetary provision for fiscal 2000–01 was approximately US$2.4 million, as compared to US$2.2 million, the previous year. Some observers have generally cred- ited the DCEC with playing a significant role in increasing accountability in the civil service. An Anticorruption Agency as Phony Populism The examples from Malawi and Sierra Leone highlight how an anticorrup- tion agency can be used as a political tool to appease donors, while not mak- ing much of a contribution to address the pervasive corruption problem. In Sierra Leone, for example, although the Anti-Corruption Act of 2000, which established the Anti-Corruption Commission (ACC), grants it operational independence by stating that it cannot be subject to the control of any per- son or authority, in reality, the provision requiring the ACC to submit inves- tigative reports to the attorney general for prosecutions has been alleged to be a source of tension between the ACC and the judiciary. The ACC’s annual report of 2002 acknowledged the rather “lukewarm attitude of the judiciary, which after two years has yet to take a single matter of adjudication to a final conclusion” (ACC 2002, p. 7). Some analysts argue that the supremacy of the rule of law in Sierra Leone has seen progressive erosion through polit- ical interference, leading to a growth of corruption and general mistrust of the system of justice in the country. Sahr Kpundeh 271 An additional constraint for the ACC is the efficacy of the act itself, which has been found to be defective in certain areas after two years of operations. For example, the attorney general insists that any statements taken from witnesses in support of the prosecution should be made under oath, so they are valid for preparing an indictment. Because the majority of witness statements in cases had not been taken under oath, however, inves- tigators had to revisit those witnesses to obtain further statements or to take an oath swearing to the veracity of their previous statements. This effort has caused considerable delay because 119 statements had to be retaken, 46 of which were from individuals living outside of the capital city (ACC 2002). The general weakness of the core state institutions has rendered them insignificant in the fight against corruption. Such weakness is evident in the lack of transparent procurement, the scarcity of adequate audit systems, the weakness of judiciary and enforcement agencies, and the general disregard for accountability mechanisms. Examples are numerous of police, custom officers, senior politicians, judges, bailiffs of the high court, and other offi- cials who have been accused of corruption and are either waiting hearings in court or are out on bail pending appeals of their cases. Parliament, in its capacity as an oversight body, has been ineffective or uninterested in taking steps either to discuss the ACC’s annual report or to demand accountability for the persistent allegations of corruption against top government officials. The media in Sierra Leone is relatively weak and have not played much of a role in helping to combat corruption through investigative stories. An environment for a freer flow of information to the public must be created. The current law discourages releasing information to the public, hence the reluctance of both public officials and private sector operatives to answer reporters’ inquiries and questions, leaving the door open for speculations and sensational reporting. The ACC’s Prevention Department, however, continues to use the broadcast media to deliver messages on the fight against corruption through documentaries, theater plays, discussions aired on national television, and radio jingles with anticorruption messages. In terms of budget predictability, the ACC is a relatively well-financed opera- tion, with a combination of funds from both the government of Sierra Leone and the Department for International Development in the United Kingdom. Their 2002 annual report shows their income of approximately US$6.5 mil- lion for 2002, a decrease from the 2001 budget, which was approximately US$8.5 million (ACC 2002). The Anti-Corruption Bureau (ACB) in Malawi shares similar features with its Sierra Leone counterpart. The Corrupt Practices Act no. 18 of 1995 established the ACB as an independent and autonomous government body comprising a director, a deputy director, and any other officers of the ACB 272 Process Interventions Versus Structural Reforms who may be appointed for the effective performance of the functions of the act. The functions of the ACB as defined in section 10(1) of the Corrupt Prac- tices Act, 1995 are to “(a) take necessary measures for the prevention of cor- ruption in public and private bodies, including in particular, measures for examining the practices and procedures of public and private bodies to facilitate the discovery of corrupt practices and secure the revision of meth- ods of work or procedures and (b) advise public and private bodies on ways and means of preventing corrupt practices and on changes in methods of work and procedures of such bodies that are compatible with the effective performance of their duties.” The formation of the ACB attracted donor interest, and as a result the ACB received financial support from numerous donors such as the British, Danish, and Japanese governments. Before the ACB came into being, it was allocated MK 7 million by Parliament. In fiscal 1997–98 the Bureau was allo- cated MK 10.7956 million by Parliament. The ACB has done relatively well in raising awareness, focusing mainly on the process interventions, but because the core institutions are generally weak in the country, the resulting effect was burn-out populism. The Malawi government used the ACB to appease the donors, and its first direc- tor was removed from office allegedly because some of the investigations carried out by the ACB under his watch were beginning to reach top-level government authorities. Considering that the director ’s position is subject to the direction and control of the minister of justice on all matters of policy, he was not insulated from political interference or influence in the perfor- mance of his duties. Cabinet has yet to take action on the draft Corrupt Prac- tices Bill and the Declaration of Assets Bill after the Malawi Law Commission and the cabinet committee on legal affairs submitted them with recommendations. Analysts hope that the draft Corrupt Practices Bill, when passed into law will help the ACB effectively and efficiently investi- gate cases of corruption in the country. One recommendation in the draft is to remove the powers of the director of public prosecutions to give consent to the ACB on all corruption cases because officials at the ACB feel frus- trated that its mandate has been undermined by the director of public pros- ecutions, who is accused of withholding consent for unsatisfactory reasons. Similarly, the draft Declaration of Assets Bill, which proposes among other things, the impeachment and possible removal from office of the head of state, members of parliament, and heads of government departments if they do not declare their assets after having been reminded twice, also faces delays (The Nation-Blantyre 2003). If one contrasts the Malawi and Sierra Leone cases with the Botswana case, one can argue strongly that the lack of supportive institutional state Sahr Kpundeh 273 structures, such as the judiciary, parliaments, and other democratic account- ability mechanisms, in Malawi and Sierra Leone, coupled with the general lack of effective leadership and commitment to leverage change in the anti- corruption fight, resulted in the anticorruption agencies being used as pop- ulist measures to appease donors. In reality their operational constraints seem to indicate that they are meant to be token efforts or populist moves by politicians to give the impression they are taking steps to improve gover- nance and address corruption. Consequently, one cannot be surprised by the low percentage of prosecutions and convictions in both countries (table 8.4). The next set of cases will illustrate how such agencies can be used as a short-term substitute to effect change when the core institutions are weak. The Anticorruption Agency as a Useful Short-term Substitute The anticorruption campaign in Uganda began in the mid-1980s and was strengthened by provisions included in the 1995 Constitution. Laws have been enacted to curb corruption including the Penal Code Act of Uganda, the Prevention of Corruption Act no. 8 of 1970, and Leadership Code of Conduct. The Inspectorate of Government in Uganda is the constitutionally mandated body to investigate and prevent cases of corruption. Its indepen- dence is provided for under article 227 of the Constitution which states that “the Inspectorate of Government shall be independent in the performance of its functions and shall not be subject to the direction or control of any per- son or authority and shall only be responsible to Parliament.” Article 231 of the Uganda Constitution also puts in place accountability mechanisms to ensure the Inspectorate of Government is accountable to Parliament, requir- ing it to report once every six months. The report should contain informa- tion on the performance of its functions and any other information Parliament may require. The inspectorate can also boast of budget pre- dictability because Parliament appropriates to it an independent budget that the inspectorate controls. In addition to the Inspectorate of Government, other institutions play a role in the fight against corruption, including the Directorate of Ethics and Integrity, which was created in 1998 with a mandate to coordinate the govern- ment’s efforts to fight corruption, and an auditor general who has the author- ity to examine all government accounts and ensure that revenues are spent wisely. Government departments and agencies are required to submit their accounts to Parliament for scrutiny and publication. The Directorate of Public Prosecution handles fraud and corruption-related cases and has preventive, investigative, and prosecutorial functions. The Public Accounts Committee reviews the auditor general’s reports and inquires into improprieties. 274 Process Interventions Versus Structural Reforms The Uganda case demonstrates the need to have an institutional frame- work in place that complements state operations in the fight against corrup- tion. Despite the plethora of institutions, however, success has been limited because the core institutions are relatively weak. Consequently, the anticor- ruption interventions led by the Inspectorate of Government are used as stop- gap measures to raise awareness about the ills of corruption. Much still needs to be done in the areas of detection and enforcement, which are also responsi- bilities of the Inspectorate of Government and other core institutions. For example, perenial problems include the procurement systems; weak financial management; cases of diversion of budgetary funds; poor audit coverage; overregulation; corrupt police and judiciary, including court clerks; and a general perception that senior-level people are above the law, especially when wrongdoing occurs in relation to privatization. In addition, some evidence points to selective impunity ranging from cases involving the former vice- president and cabinet ministers who were censured by Parliament for wrong- doing, but later were given positions in the movement or were retained as ministers. Parliament made no attempt to recover allegedly stolen funds. The coordination between the media, the Inspectorate of Government and Parliament has proven decisive in making some inroads against corrup- tion. With the help of various training workshops followed by ongoing capacity building initiatives, these three institutions have become signifi- cant checks on corruption even at the highest levels of government. The Inspectorate of Government which now reports to Parliament, has won a central role in awareness raising and conducts preliminary investigations into allegations of official misconduct. The press, which is fairly free, and the Parliament have proven themselves to be capable of systematically pre- senting strong cases against alleged corrupt officials, thereby raising the stakes on grand corruption as exemplified by the sanctioning of some mem- bers of Parliament for unexplainable enrichment. Although the picture painted here of the anticorruption efforts in Uganda seems bleak, some accomplishments have been made within the context of a weak institutional capacity. Within the governments’s Strategy and Plan of Action to fight corruption, several positive steps have been taken to strengthen the core institutions, for example: • The legal framework and the capacity of law enforcement for better investigation, prosecution, and judgment, particularly of corruption cases, has been strengthened with the enactment and application of the Leadership Code 2000. • The Inspectorate of Government Bill 2000 has been enacted and applied. Sahr Kpundeh 275 • The revised Local Government Act has been enacted and applied. • The Auditor General’s Act has been revised and strengthened. • Case management by the director of public prosecutions has been computerized. • A new set of Public Finance Procurement Regulations has been approved and is in use. • A reformed Central Tender Board has commenced operation. • Procurement Units have been established in all procuring organs of government. • A pay reform proposal is being considered by cabinet. The most salient characteristic of Uganda’s anticorruption efforts is that Uganda has executed its anticorruption program while implementing a broader economic and public sector governance reform endeavor. There- fore, since the beginning, its anticorruption program has been indirectly supported by a relatively large framework of policy reforms that have yet to bear fruit. In other words, the government has done well in raising aware- ness about the ills of corruption, as well as in pushing for transparency as part of a broader environment of processes and programs for improving public sector reforms. However, as table 8.4 illustrates, the numbers in terms of prosecutions and convictions do not look very good. Tanzania’s Prevention of Corruption Bureau (PCB) was established in 1991 and strengthened in 1995. Its predecessor, the 1975 Anti-Corruption Squad, came into being following the Prevention of Corruption Act of 1971. The PCB is in charge of prevention, education, and investigation; is under the president’s office; and reports directly to the president. By law, the bureau is free to investigate all forms of corruption, whether grand or petty, involving either senior or junior officials. With the director of public prose- cution’s approval, it may also prosecute cases. Like the Directorate of Ethics and Integrity in Uganda, which coordi- nates the anticorruption efforts, the PCB is supported in the Office of the President by the Good Governance Coordinating Unit, a body charged with ensuring coordinated government action on the anticorruption strategy and with providing policy guidance to PCB. In terms of accountability, an inter- nal and external system both assess the work of the PCB. Internally the PCB is assessed by an evaluation committee in the Office of the President and the auditor general. Externally, the auditor general’s annual report is subject to public evaluation. Despite its efforts to show operational independence, the PCB operates under the Office of the President and is not insulated from directives from it. For example, the PCB director is appointed by, and is responsible to, the 276 Process Interventions Versus Structural Reforms president on all executive matters of the bureau and reports to the chief sec- retary in the state house on matters of administration and finance. Having no fixed term, the director serves at the will of the president. The president alone has the power to remove or sanction the director. A Control and Eval- uations Committee, chaired by the chief secretary oversees the performance of the bureau, but its findings are not published. In addition, neither Parliament nor citizens’ committees have formal oversight authority for the PCB nor does the law provide any formal guar- antees to safeguard its independence. Some observers have argued that the PCB has not gained public support and credibility to act independently because of apathy, fear, and lack of awareness of its mission. More impor- tant, the PCB’s work is not supported by other government agencies, and its links to agencies such as the judiciary are weak. The low number of cases prosecuted and convicted, although higher than in Uganda, nevertheless highlights the PCB’s failure to address corruption. The Anti-Corruption Commission in Zambia shares similar features with the Uganda and Tanzania cases. The 1996 Anti-Corruption Commis- sion Act establishes the board of commissioners appointed by the president to whom the commission reports. Unlike the Uganda and Tanzania cases, where the leadership allegedly supported the anticorruption efforts, the commission under the administration of former President Frederick Chiluba was considered to be a tool of the president and had regular inter- ference from politicians and relegated its focus only on petty corruption. During this period, the level of investigations and prosecutions was low. The commission’s resources were insufficient for anticorruption efforts, and it had a serious staff shortage. A weak institutional framework further ham- pers the commissions efforts, especially because the commission must get the consent of the director of public prosecution before prosecuting a case, and the director can refuse consent without citing reasons. The relatively high ratio of prosecutions and convictions (compared with Tanzania and Uganda) as illustrated in table 8.4 shows that the commission was prosecut- ing large numbers of people for corruption, but under former President Chiluba, the commission was criticized for only prosecuting the “small fry” while the big fish went untouched. Some observers have argued that the commission was ill-equipped to address high-level corruption for a variety of reasons. For example, the commission’s budget has suffered from a gen- eral lack of predictability over the years. From 1998 to 2000, the commis- sion’s budget was reduced from approximately US$3.5 million to US$2.9 million. The executive always drastically reduces its proposed budget, and even the little money allocated in the budget is not released in full. For Sahr Kpundeh 277 example, the commission’s estimated budget for 1999 was K 6.3 billion, but only K 1,759,479,000 was approved. Recurrent departmental charges were even reduced from K 792,871,000 in 1998 to K 567,571,00 in 1999. Only K 344,069,042 was actually released by the ministry of finance (Transparency International 2003). With the change of government in Zambia, the budget of the commission has increased tremendously, from US$5.2 million in 2001 to approximately US$18 million in 2002. This picture of the Zambia Anti-Corruption Commission clearly makes the case for classifying it in the category of “phony populism,” where it belonged under former President Chiluba. With the election of President Levy Mwanawasa, however, the commission has received a much higher profile, because the president has used his leadership to leverage change in the fight against corruption. Soon after his election in January 2002, President Mwanawasa bluntly stated a policy of zero tolerance for corruption. He dis- missed or demoted a number of senior government officials associated with corruption under the previous and his own government, including his for- mer vice president, the attorney general, the minister of information, the for- eign minister, and the permanent secretary of the Ministry of Finance. The budget of the Anti-Corruption Commission was increased significantly and its facilities were upgraded. Complementary institutions such as the director of public prosecutions and the judiciary also responded to these zero toler- ance signals by moving aggressively on corruption cases, putting close allies of former President Chiluba under investigation. In July 2002, President Mwanawasa set up an anticorruption task force with a mandate to recover resources allegedly stolen by the previous admin- istration and to investigate and prosecute former officials, including former President Chiluba. That same month, the president addressed a special and unprecedented session of the National Assembly to deliver a stinging indict- ment against his predecessor ’s administration, in which he tabled evidence of grand public sector corruption. His decision to present the case to the National Assembly was significant, giving credence to what had been largely a rubber stamp institution. The following week, the National Assembly voted unanimously to lift the former president’s immunity from prosecu- tion. After the Supreme Court upheld the decision, former President Chiluba was charged on 59 counts. Several of his closest associates have also been charged and some were jailed while standing trial on nonbailable offenses. The president’s campaign against corruption in Zambia is not taking place in a political vacuum. Rather it builds on civil society’s fight against a third term bid. The anti–third term campaign of 2001, in which the churches, nongovernmental organizations, the law association, opposition 278 Process Interventions Versus Structural Reforms members of parliament, former ministers, and many citizens joined to pre- vent President Chiluba’s attempt to stay in office, has remained strong through the so-called “Oasis Forum,” named after the venue of the first meeting of the movement. The Oasis Forum has continued to lobby the cur- rent government on constitutional reform and anticorruption issues, and civil society organizations have now undertaken a number of joint policy initiatives with government, for example: • The law association drafted the Arbitration Act of 2000. • Civil society and government formed an anticorruption association in 2002; the Anti-Corruption Commission is working closely with the National Youth Movement Against Corruption. • In 2002, three media associations significantly influenced content of new media bills prior to adoption by Parliament. The Zambia example also illustrates how an anticorruption effort can be used as a short-term measure to improve governance when the core institu- tions, including the judiciary, are weak. Zambia’s chief justice was forced to resign when it was revealed he had received US$168,000 in bribes from for- mer president Chiluba. The judiciary still has most of the same judges that were appointed by former President Chiluba, but they now seem to be much more vigilant in dispensing justice, which says a good deal about the degree of judicial independence that operated under the Chiluba regime. Thus, the actions taken by the current government have raised the profile of anticorruption efforts. These efforts have culminated in a much more vibrant civil society, in which the media play a significant role to raise awareness and in which the citizenry have hope that the fight against cor- ruption has no sacred cows. The case studies have illustrated three different roles that anticorruption agencies have performed in Africa—the good case of complementing the state institutions from Botswana, the bad cases of being used as phony pop- ulism in Malawi and Sierra Leone, and the in-between case of acting as an effective short-term substitute when state institutions are weak, as we have seen in Tanzania, Uganda, and Zambia. Table 8.4 illustrates some results and output measures of the various agencies discussed in this chapter. The important question then becomes, under what conditions do anticorruption commissions contribute to building momentum against corruption or become phony populist moves to appease donors or domestic critics? One possible answer may be when strong leadership uses its power to leverage change. In the case of Tanzania, Uganda, and Zambia, Presidents Benjamin Mkapa, Yoweri Museveni, and Levy Mwanawasa, respectively, Sahr Kpundeh 279 have succeeded in using their political clout to keep the anticorruption cru- sade front and center. The regular political pronouncements about their com- mitment to the crusade against corruption, coupled with their support for strengthening the capacity of institutional mechanisms, serve to leverage reform efforts toward improved governance. Alternatively, in situations where the anticorruption efforts are used to stir up activism that leads to burn- out, the leadership, as in Malawi and Sierra Leone, lacks the political clout to leverage change primarily because of the nature of the political dynamics as well as the immaturity of the existing democratic accountability mechanisms. In other words, crony politics, nepotism, patronage politics, and the presence of an elite group who may want to maintain the status quo rule the day. In such situations, the political leadership may want to do something about corruption based on pressure from internal constituencies and exter- nal conditionalities, but may lack the political base to make tough, effective decisions. Furthermore, the level of involvement of nonstate actors in play- ing a role in the reform efforts may contribute to sustaining the agenda for reform. Variations in the level of participation of civil society and other stakeholders in reform efforts can also contribute to defining whether the reform efforts build momentum as effective short-term substitutes or are merely phony populist moves. Conclusion The experiences of African anticorruption agencies summarized in this chapter suggest that in a number of these countries, the state lacks either the know-how or the political will to control corruption. The cases examined here reveal that the decisionmaking power lies outside the formal parame- ters of the public sphere. Power is in the hands of select elites. The ineffec- tiveness that characterizes anticorruption institutions are artificial political problems created deliberately as instruments of political management whose intentions are twofold: (a) to ensure structurally ineffective reform agencies and (b) to demonstrate openly to donors that the government has taken action. Corruption reform in Africa must evolve among a broad coali- tion of stakeholders, building and fostering a relationship with civil society and progressive elements of the private sector. Only a broad coalition of this kind will have the power and momentum capable of dislodging powerful elites and institutions with vested interests in the status quo. Coalition building is most effective when government treats civil soci- ety and nonstate actors as participatory allies that have a stake in reforms (Johnston and Kpundeh 2002). What we are witnessing in several African 280 Process Interventions Versus Structural Reforms countries, however, is not coalition building, but separate reform move- ments—government driving its anticorruption and public management reform agenda with little or no real empowerment of nonstate actors. The state–civil society relationship has been adversarial and has driven a wedge between the various stakeholders, resulting in the lack of a cohesive strategy that links ad hoc measures with the structural and institutional changes that improve core government institutions’ capacity (Kpundeh 2004). For example, many governments introduced anticorruption mea- sures with awareness campaigns and an ombudsman’s office. Citizens were supposed to be sensitized to the affects of corrupt politicians, civil servants, and activities and report them to the ombudsman, who investi- gated the offense and recommended prosecution, censure, or some form of punishment. No real change came about, however. This last conclusion emphasizes a glass-half-empty view of the impact of anticorruption activities in Africa over the course of the 1990s. Viewed from this pessimistic perspective, the 1990s began with a wave of popular revul- sion against corruption and misgovernance, culminating in the emergence in many countries of new political leaders who asserted their readiness for a new, clean beginning. By the mid-1990s, many international organiza- tions—spearheaded by the clarion call of the World Bank’s James Wolfen- sohn to fight the “cancer of corruption”—also signaled their readiness to bring the challenge of fostering good governance to the top of the develop- ment agenda. Yet this energy was channeled largely into process interven- tions, high profile, media-friendly initiatives that did little more than dramatize the scale of the problem and facilitate hypocritical exhortations with little effect on business as usual. This pessimistic conclusion is too negative. A glass-half-full perspective highlights the dynamic, cumulative nature of the governance change process. The clamor against corruption was a first necessary step, which transformed the development discourse across the continent: extravagant efforts at self-enrichment could no longer hide behind soothing, empty affir- mations of selfless commitment to national development. Certainly, the ini- tial expectation was naive that merely exposing corruption and campaigning against it, even to the point of committing to zero tolerance, would solve the problem. The emerging lesson, however, that reducing corruption depends on stronger core state institutions is an important one and, as this volume exemplifies, continues to shape the development agenda of the early twenty- first century. The title of figure 8.1 highlights both the risk and the opportunity of bring- ing the fight against corruption to the top of the development discourse. The Sahr Kpundeh 281 risk is that an exclusive reliance on process interventions can lead to burn-out populism and apathy. In the absence of the sustained results that can come only from structural reforms, citizens increasingly will become disenchanted with the emptiness of their leaders’ exhortations and withdraw again from the public arena, leaving the terrain free for a new round of misgovernance. The opportunity is that, if used skillfully by political leaders, a commitment to fight corruption can unleash the civic energy needed for sustained institu- tional change. References ACC (Anti-Corruption Commission). 2002. The Sierra Leone Anti-Corruption Commis- sion Annual Report 2002: Fighting Corruption—A National Concern. Freetown, Sierra Leone: Reffo Printing. DCEC (Directorate for Corruption and Economic Crimes). 1999. Annual Report. Gabarone, Botswana: Government of Botswana. _____. 2000. Annual Report. Gabarone, Botswana: Government of Botswana. _____. 2002. Annual Report. Gabarone, Botswana: Government of Botswana. Freedom House. 2002. Annual Survey of Press Freedom, 2002. http://www.freedom house.org/pfs2002/pfs2002.pdf. _____. 2003. Freedom of the Press 2003: A Global Survey of Media Independence. http://www.freedomhouse.org/pfs2003/pfs2003.pdf. ICAC (Independent Commission Against Corruption). 2002. Annual Report 2002. http://www.icac.org.hk/text/eng/main/index.html. Johnston, Michael. 1998. “Fighting Systemic Corruption: Social Foundations for Institutional Reform.” European Journal of Development Research 10(1): 85–104. _____. 1999. “A Brief History of Anti-Corruption Agencies.” In Andreas Schedler, Larry Diamond, and Marc F. Plattner, eds., The Self-Restraining State. Boulder, CO: Lynne Reinner. Johnston, Michael, and Sahr Kpundeh. 2002. “The Measurement Problem: A Focus on Governance.” Forum on Crime and Society 2(1): 33–44. Kpundeh, Sahr J. 1994. “Limiting Administrative Corruption in Sierra Leone.” Jour- nal of Modern African Studies 32(1): 139–57 _____. 1998. “Political Will in Fighting Corruption.” In Sahr Kpundeh and Irene Hors, eds., Corruption and Integrity Improvement Initiatives in Developing Countries. New York: United Nations Development Programme. _____. 2004. “Corruption and Corruption Control.” In E. Gyimah-Boadi, ed., Democ- ratic Reform in Africa. The Quality of Progress. Boulder, CO: Lynne Reinner. 282 Process Interventions Versus Structural Reforms The Nation-Blantyre. 2003. Corruption, Assets Bills Still in Limbo, September 17. Olowu, Dele. 1992. “Governmental Corruption and Africa’s Democratization Efforts.” Paper presented at the conference on Limiting Administrative Corrup- tion in the Democratizing African States, September 11, U.S. Agency for Interna- tional Development, Washington, D.C. _____. 1999. “Combating Corruption and Economic Crime in Africa: An Evaluation of the Botswana Directorate of Corruption and Economic Crime.” Paper pre- sented at the Ninth International Anti-Corruption Conference, October 10–15, 1999, Durban, South Africa. Quah, J. 1999. “Singapore’s Experience in Curbing Corruption.” In Arnold Heiden- heimer, Michael Johnston, and Victor LeVine, eds., Political Corruption: A Hand- book. New Brunswick, NJ: Transaction Publishers. Transparency International. 2003. National Integrity Systems 2003. Study Report: Zambia 2003. Berlin. 9 The Politics of Decentralization in Africa: A Comparative Analysis Stephen N. Ndegwa and Brian Levy Alongside the familiar economic and democratic–electoral reforms that have occurred in African countries since 1980, and especially after 1990, another significant reform, if less visible and less celebrated, has been the progressive decentralization of the state. Although analysts and practition- ers have long noted the extreme centralization of the state in Africa since independence, the situation today is decidedly different. Whether arising from dramatic political reforms and donor pressure and programs or as part of evolutionary administrative change, recent local governance revival has been one of the most significant facets of state restructuring in Africa since independence. In a recent World Bank study, 13 of 30 countries surveyed showed high or moderate levels of decentralization as measured by a composite index of political, administrative, and fiscal devolution indicators (Ndegwa 2002). Another 13 showed at least some degree of decentralization, and several were in the process of change. Although not all countries have fully revived local governance, the study also indicated no country in Africa today pro- pounds a preference for the centralized state. Indeed, nearly all countries claim in one way or another to be decentralizing power, resources, and accountability to local levels. Yet, as present evidence indicates, the process of reform and results of recent decentralization are widely varied, and the underlying processes are not fully understood. The purpose of this chapter is to contribute to our understanding of the different pathways to decentralization by focusing on the political dynamics 283 284 The Politics of Decentralization in Africa: A Comparative Analysis of the process. To be sure, the technical challenges associated with a move to decentralization are daunting. They include the creation of new political tiers of government, the introduction of intergovernmental fiscal systems, and the realignment of administrative systems. Correspondingly, a rich technical lit- erature explores the options within each of the administrative, fiscal, and political dimensions as well as the strengths and weaknesses of alternative sequencing options among the dimensions. The technical literature, however, tends to neglect the political underpin- nings of decentralization, even though state transformation to shift resources and authority closer to the civic frontline is a profoundly political process. By contrast, this chapter aims to help redress the imbalance between technical and political analysis. The intent is not to imply that tech- nical analysis has no place or that no degrees of freedom are available to policymakers. Rather, the implication is that the room for maneuvering is constrained and that options arise within the context of specific political realities prevailing in individual countries. Only if technical analysis is grounded in these realities can it be useful in identifying and evaluating alternative feasible options. The chapter proceeds as follows. The next section of the chapter lays out a simple framework for a comparative assessment of the politics of decen- tralization within individual countries. The two sections that follow apply the framework to three countries: Malawi, Senegal, and Uganda. The cases were deliberately selected to capture a wide range of reform efforts and out- comes in the region.1 Thus, Senegal exemplifies an enduring democracy that has a long-running effort to decentralize, but an ultimately stalled process. Uganda exemplifies radical and early decentralization following a revolu- tionary takeover and one that is past the stage of installing structural reforms and now contends with consolidation. The political drivers of the very different Senegalese and Ugandan dynamics are explored in the sec- tion on “Commitment and Contradiction: Senegal and Uganda Compared.” This is followed by a discussion of Malawi as an intermediate case. Malawi’s experience with decentralization is not as long-lived as in the other two countries; its current progress appears to be midway between that of Senegal and Uganda; and it evinces an ebb and flow in its decentralization reform, which is often threatened by broader issues of an ongoing political 1. Besides these case studies, the analytical framework and the broader discus- sion is informed by a review of existing case studies covering several other coun- tries, including Mozambique, Namibia, Tanzania, South Africa, and Zimbabwe. Stephen N. Ndegwa and Brian Levy 285 transition. In addition, the Malawian case raises some interesting questions about the interaction between decentralization and grass-roots participation and whether the latter can actively be nurtured in support of decentralized governance. The chapter’s conclusion uses the cases to draw some lessons for a politi- cally sensitive approach to decentralization reform. Assessing the Politics of Decentralization: A Framework Even though our point of departure is the familiar distinction between the so-called “big bang” and incremental reforms, the framework established in this section goes beyond this simple duality. Traditionally, analysts identify big bang reforms as those in which local political authority is created or revived, intergovernmental fiscal governance systems are deployed, and personnel are transferred from the center to localities in a massive program rolled out across all sectors and regions, all within a short period.2 Most recently typified by the decentralizations in Indonesia and Pakistan, this mode of reform tends to be driven by intensely political imperatives, usu- ally linked to a drive to recover or reinvent political legitimacy. Among the most prominent African examples are South Africa and Uganda and, in an ongoing case, Ethiopia. The vast majority of African countries that have explored decentraliza- tion have done so incrementally, and changes have ranged from barely per- ceptible to more dramatic but with sequential reforms over time. The list includes Côte d’Ivoire, Ghana, Malawi, Mozambique, Senegal, Tanzania, and Zimbabwe, and several of them trace their evolutionary devolution (or intentions) to the 1980s especially. However, most subordinated local gov- ernments to the centralization that attended various shades of authoritarian rule. In addition in the more recent phase of decentralization revival, few countries have encountered the significant break with the established model of politics that would allow for a more energized pursuit of devolution given its implications for political control. The narrow preoccupation with big bang versus incremental reform has at least two shortcomings. First, the universe of cases in the latter cate- gory is so broad, and so varied, that it renders the term “incremental 2. Although no firm definition exists, how quickly a reform program moves from the first to the second stage, how broad the spectrum of decision and action is seems to distinguish big bang reform from incremental reform. 286 The Politics of Decentralization in Africa: A Comparative Analysis decentralization” a meaningless catch-all, especially because the incre- mental mode typifies the large majority of decentralizing countries and hence deserves closer attention and differentiation based on the context that reform confronts. Second, although much of the analytical literature and the internal World Bank discussion on the choice between big bang and incremental decentralization focus on the technical strengths and weaknesses of each option, declarations of the pace of reform as causal, desirable, or critical to success may overstate its relevance compared to the state–society structure confronted by a reforming (or nonreforming) regime. Rather, the pace and substance of reform may be linked as depen- dent variables to the previous status quo as defined by the incentives, con- straints, and opportunities confronted by the principal stakeholders in terms of any move toward decentralization. Table 9.1 lays out a two-dimensional framework for assessing the dynamics of decentralization in individual country settings. The first Table 9.1. The Roles and Influence of Stakeholders Political elite Bureaucratic Community Phase stakeholders stakeholders stakeholders Engaging How strong is the To what extent is How strong is decentralization elite political consen- the decentraliza- bottom-up sus in favor of tion discourse pressure for local decentralization? underpinned by empowerment? technical and com- parative analysis? Detailing How engaged is How cooperative is How involved are decentralization the political elite the bureaucracy in civil society organi- in ensuring that developing and zations in defining the details of implementing new their entry points decentralization decentralized and their level of are consistent with systems of involvement in the political intent? governance? contemplated tech- nical details? Sustaining To what extent do elite political and How capable are decentralization bureaucratic stakeholders seek to communities of reassert central control over authority enforcing downward and resources? accountability on local elites? Source: Authors. Stephen N. Ndegwa and Brian Levy 287 dimension distinguishes among three key sets of stakeholders in the decen- tralization process: • Political elites, comprising in particular the political leadership (which may be unified or may be a coalition) which controls the levers of governmental authority. • Bureaucrats within central government, at all levels of authority • Communities, both local elites and grass-roots stakeholders (the term “community” is preferred over “civil society” because it better cap- tures the essentially local dimension of civic activism that is espe- cially relevant for an analysis of decentralization). Although external actors, especially development aid organizations, have significantly shaped decentralization policies in several countries, we do not analyze them separately for two reasons. First, we focus on internal constituencies, which although influenced by external actors, respond prin- cipally and in the long run to internal dynamics. Second, the actions and interests exhibited by internal stakeholders presumably already reflect com- promises with or reactions to the influence exerted by external actors. Arrayed along the other dimension are three distinct (though potentially overlapping)3 phases of decentralization: • Engaging decentralization, the initial phase when fundamental deci- sions about state structure are on the national agenda, specifically to what extent and how to shift resources and responsibilities down- ward to communities (including the establishment of elected local governments) • Detailing decentralization, the phase in which the specific fiscal and administrative mechanisms needed to give life to local empower- ment are clarified and rolled out in an initial phase of implementation • Sustaining decentralization, an ongoing phase of learning-by-doing in which the institutional arrangements of the intergovernmental sys- tem are continually fine tuned. Table 9.1 poses a series of questions about the role and influence of each stakeholder at each phase of the process. As table 9.1 suggests, these roles 3. As one example of overlap, the move to engage decentralization could involve from the outset intensive technical discussions about the details of the pro- posed system. As another example, the process of progressively refining the details of decentralization can continue indefinitely, well into an era when the system has irreversibly been consolidated. 288 The Politics of Decentralization in Africa: A Comparative Analysis vary both across stakeholders and, for a given stakeholder, across different phases of the decentralization process. Furthermore, some questions are italicized on table 9.1, signaling that the roles of different stakeholders may be more or less crucial at different phases of the process. Consider, for example, the italicized role of political elites during the ini- tial phase of engaging decentralization. Arguably, big bang reforms are pos- sible and desirable only when both political consensus for reform and local grass-roots organization are high. When reform is attempted while both the political consensus and the level of grass-roots organization are low, decen- tralization is likely to stall or achieve only moderate and easily reversible gains. Thus in settings where decentralization reforms appear stillborn, where grand-sounding rhetoric and detailed planning never translate into changes on the ground, reformers might have paid excessive attention in the engaging decentralization phase to technical questions of sequencing and capacity and neglected the political dimension. Yet, at the initial stage, the extent to which a broad consensus is reached and champions are empowered both to push the agenda and to survive backlashes determines the extent to which the process goes forward. Now consider the role of the bureaucracy, italicized on table 9.1, during the detailing phase. In some settings, decentralization reforms appear to move forward rapidly, only to disappoint down the road. At first impres- sion the cause of this disappointment might be viewed as technical, and the perception of failure might be interpreted as a consequence of the complex- ity of what is being attempted and the difficulty of aligning the many facets of decentralization. Even at this stage, however, sequencing and pace are not simply technical issues. They are continually conditioned by the endur- ing political consensus and commitment because at each stage new negotia- tions must be undertaken as old institutions and networks are unraveled and as new communities of interest are affected. Central bureaucrats in par- ticular may find themselves in an especially ambiguous position, given that only they have the detailed knowledge of administrative systems needed to make decentralization work. Whether they embrace the reforms or resist them as potentially threatening may thus be key to the technical coherence, and thus effectiveness, of implementation. Finally, consider the role of community stakeholders, highlighted on table 9.1 in terms of the ongoing sustainability of decentralization. When the reform consensus is high but the level of grass-roots organization is low then the path to reform may be a short-lived big bang with a longer period of struggle to make installed components effective, especially if installed components require an active and empowered citizenry at the local level. Stephen N. Ndegwa and Brian Levy 289 When the grass-roots organization is strong, but political consensus for reform is weak or nonexistent, explosive social tensions may arise. Indeed, in such instances, the problem, typically strictly political, is much larger than centralization. Commitment and Contradiction: Senegal and Uganda Compared In both Senegal and Uganda, decentralization has long been central to the development discourse. Uganda’s National Resistance Movement (NRM) pursued popular rule immediately upon acceding to power in 1986 and worked to give concrete substance to this decentralized vision throughout the 1990s. Senegal has been experimenting since the early 1960s with initia- tives to counteract its francophone legacy of strong centralization; high pro- file public sector reforms to deepen decentralization were initiated in 1972, 1984, 1990, and 1996. Yet, notwithstanding Senegal’s long track record, much more progress has been made in actually transforming the state in a decentralized manner in Uganda than in Senegal. Why this is the case will be explored later in the chapter, using the framework delineated earlier. Progress in Decentralization Senegal and Uganda are among the more decentralized countries in Sub- Saharan Africa, ranking second and eighth most decentralized of a group of 30 countries (Ndegwa 2002). This aggregate ranking, however, conceals sharp differences between the countries in decentralization performance. On a 0–4 scale, Uganda is rated at least 3 for the categories of political, administrative, and fiscal decentralization. By contrast, Senegal achieves a score of 3 only for political decentralization and scores only 1.2 (20th in the 30-country sample) for administrative decentralization. POLITICAL DECENTRALIZATION. Both Senegal and Uganda have in place local authorities embedded in a comprehensive, well-defined framework of territorial governance. In both countries, the lowest tier of government is directly elected; turnout is reasonably good, and the elections are perceived to be fair. Under its 1995 Constitution, Uganda established the district as the most significant local government unit and stipulated that it have a directly elected district chairperson and council. At present Uganda has 48 districts, a steep rise from the 39 initially established in the early phase of decentral- ization. In the initial phase of its decentralization in the late 1980s, Uganda 290 The Politics of Decentralization in Africa: A Comparative Analysis had in place a five-tier local governance structure. Each village elected a nine-member resistance council (the first tier), which in turn made up an electoral college that elected a member to the next level resistance council at the parish (the second tier), and it in turn elected a representative to the sub- county, county, and district (the third, fourth, and fifth tiers), and ultimately the National Resistance Council, effectively the parliament.4 Senegal’s 1996 reform put in place an institutional and territorial land- scape comprising 434 local governments: 11 regions, 103 communes (60 ordinary legal communes and 43 district communes), and 320 rural com- munes. The communes are elected bodies and have elected executives whereas the regional authorities are elected, but their executive is appointed by the president. Both the communes and regions have enjoyed legal per- sonality and financial autonomy dating back to 1972. FISCAL DECENTRALIZATION AND FORMAL ASSIGNMENT OF RESPONSIBILITIES. Without responsibilities, and without fiscal resources to follow through on formally assigned responsibilities, the creation of an elected local tier of gov- ernment becomes an empty gesture. Although both Senegal and Uganda have assigned far-reaching responsibilities to local government, only Uganda backs this assignment up with the provision of adequate fiscal resources. Uganda’s 1993 Local Governments Act assigned a wide array of responsi- bilities to district-level local authorities. The competencies devolved to district councils included several duties previously overseen by ministries in charge of trade and industry, agriculture, lands and housing, education and sports, health, information, labor and welfare, internal affairs, women and youth, and tourism, as well as several functions of the Ministry of Local Government. The assignment of a broad range of responsibilities was undergirded by the provision of fiscal resources. As table 9.2 shows, between 1995 and 1998, transfers to local governments increased 148 percent from nearly U Sh 118 billion in 1995–96 to over U Sh 291 billion in 1998–99. The share of Uganda’s consolidated national expenditures for which grants were made to local gov- ernments, already over 11 percent in 1996, rose to nearly 20 percent by 1999.5 4. For the constitutional layout of Uganda’s local government see http://www. government.go.ug/constitution/detail.php?myId=11. With progressive democrati- zation, these councils were renamed to drop the resistance designation. 5. For more details on the progress of Uganda’s decentralization reforms and especially the trend in the devolution of responsibilities and fiscal resources as well as the role of World Bank support, see Girishankar (2002). The analysis reported here draws substantially from this review of World Bank support to Uganda’s decentral- ization effort. Stephen N. Ndegwa and Brian Levy 291 Table 9.2. Transfers to Local Governments, Uganda, 1995–96 to 1998–99 (U Sh billions) Description 1995–96 1996–97 1997–98 1998–99 Unconditional grants 40.575 48.718 51.783 64.388 Conditional grants 17.935 37.200 51.150 83.461 Conditional salaries 59.244 90.597 101.484 143.646 Total 117.755 176.514 204.417 291.495 Unconditional as share of transfers (percent) 34 28 25 22 Total as a percentage of all expenditures 11.1 15.8 16.8 19.9 Source: Republic of Uganda (1998). Note, though, that as the share of total expenditure transferred to local governments increased from 11.1 percent in 1995–96 to 19.9 percent in 1998–99, the degree of expenditure autonomy (the share of transfers that were not earmarked) decreased from 34 to 22 percent. This drop resulted from an increase between 1995 and 2000 in the number of conditional grants to 15. The proliferation of conditional grants in Uganda is illustrative of a strategy on the part of sectoral ministries to use the intergovernmental fiscal system in ways that protect expenditures in priority areas (often at the expense of expenditure autonomy). As Uganda illustrates, intergovernmen- tal fiscal transfers on a large scale do not automatically guarantee expendi- ture autonomy for local government tiers. Although the Ugandan fiscal system is something of a compromise between centralization and devolution—that is, it provides money, but with strings attached—the Senegalese system goes nowhere near even that far. To be sure, in terms of the administrative allocation of responsibilities, Sene- gal’s 1996 decentralization reforms appear initially almost as far reaching as Uganda’s. The 1996 law formally transferred to communes nine areas of responsibilities: planning, land planning, public land administration, urbanization, health, education, environment, youth, and sports and cul- ture. Finance did not, however, follow function. The 1996 law also made provision for the establishment of two funds to provide budget support to enable local governments to take on these expanded responsibilities: the Decentralization Fund to finance current expenditures and the Capital Fund for Local Collectivities to finance local investment. In practice, however, the transfer of fiscal resources has been far below a level commensurate with the expanded scale of responsibilities. Unofficial Senegalese government estimates were that CFAF 80 billion to 292 The Politics of Decentralization in Africa: A Comparative Analysis CFAF 100 billion (about 25 percent of consolidated public expenditures for the year 2000) were required to finance the expanded responsibilities. In practice the total combined amount budgeted in 2000 for the two funds was only CFAF 9 billion (under 3 percent of consolidated expenditures), and commune-specific data suggest that the actual amounts transferred were below the allocated amount, and indeed had declined after 1997. ADMINISTRATIVE DECENTRALIZATION. Administrative reform is needed to actualize the reassignment of responsibilities and associated fiscal decentral- ization. Reform has multiple dimensions, potentially including the following: • Reassigning staff from of the central government to the local govern- ment (or, at a minimum, taking direction from elected local authorities) • Loosening control over the terms and conditions of employment at the local government level, potentially to the point of allowing them to be set by individual local governments themselves • Realigning central ministries to take on strategic, regulatory, standard- setting, and monitoring functions, not the now decentralized function of direct service delivery. Note that deconcentration—the shifting of resources and personnel within central government from the capital city to outlying offices—is not decentralization, but it can facilitate decentralization. Whether it does so depends in large part on whether deconcentrated staff help build the capac- ity of local governments and their bureaucracies, with a view to eventually working themselves out of a job, or whether they compete with local authorities and work to assert central government control at the deconcen- trated level. In the context of this chapter therefore, we view decentraliza- tion as having at its core a devolution of a full complement of power, resources, and responsibilities to local entities—what others have called “democratic decentralization.” As with the fiscal dimension, Uganda’s approach to administrative decentralization has been simultaneously bold and constraining. It has been bold both in its willingness to reassign staff from central to local gov- ernment and in its readiness to realign central ministries to reflect these transformed realities. Thus, the 1993 Local Government Act transferred administrative power from a deconcentrated district administrator to the chair of the downwardly accountable district council government and mandated the transfer of civil servants in the decentralized ministries to the district councils. In part as a consequence of these changes, between 1992 and 1997 total employment in Uganda’s central government fell from 320,000 to 147,000. Stephen N. Ndegwa and Brian Levy 293 Yet alongside these bold, decentralizing actions, the new Ugandan insti- tutional order also built in a variety of mechanisms to ensure that the cen- ter retained the capacity for discretionary intervention at the local level. Under the 1993 legislation, budgetary control of deconcentrated ministries remained in the hands of district executive secretaries, who reported to the center, not to district councils. Moreover, the post of chief central govern- ment representative was created. Appointed by the president and charged with overseeing all central government activities and supervising political mobilization and security matters locally, though formally politically neu- tral, the central government representative position bore a close resem- blance to that of the former district commissioner (Dicklitch 1998, pp. 82, 165–67). The central government also retained the power to suspend or dis- solve district councils and to change the parameters of prerogatives ceded to local authorities by action of the minister in charge without recourse to Parliament. Finally, although district service commissions were created to take responsibility for setting and overseeing the terms and conditions of employment of subnational public employees, norms and membership in the district service commission had to be vetted by the national Public Ser- vice Commission. The district service commissions thus lack key aspects of the authority needed to carry out their role autonomously. If the Ugandan approach to implementing administrative decentraliza- tion combines boldness with residual control, in Senegal, by contrast, no boldness whatsoever is evident. Consistent with francophone practice, vir- tually no independent, or even semi-independent, bureaucracy is recruited by and accountable to local authorities. Rather, the intent of Senegal’s 1996 decentralization law (and the earlier reforms) was to institute a multilevel, deconcentrated structure for central government that represented the line ministries at the central, regional, and district levels and that had multipur- pose technical service centers intended to provide support for local govern- ments at the district level. The deconcentrated offices of central government would act on behalf of elected local governments. This design has not worked as formulated. As a recent study has shown (World Bank 2003), the deconcentrated realm remains small, weak in imple- mentation, and accountable upward to principals at the central government level. Constraints inhibiting effective operation at the deconcentrated level include the following: • Staffing at the deconcentrated level remains insufficient. • Personnel recruitment and management continue to be centralized. • Resources provided are inadequate to operate at the deconcentrated level, and when resources are provided, the disbursement process remains centrally controlled and cumbersome. 294 The Politics of Decentralization in Africa: A Comparative Analysis • Investment decisions continue to be centralized. • Friction is pervasive between the staff of deconcentrated services and locally elected officials. PROGRESS IN DECENTRALIZATION: AN OVERVIEW. In sum, while both Senegal and Uganda have in place the political architecture needed for decentralization, only in Uganda has this design been complemented with fiscal and administrative systems capable of giving life to the formal struc- tures. These variations in performance can be explained by differences in the political and bureaucratic incentives that shaped the ways in which each country engaged and detailed decentralization and in which civil society participated in the process as a whole. Engaging Decentralization—With, and Without, an Embrace The following paragraphs contrast the extent to which political elites in Senegal and Uganda embraced decentralization during the initial, engaging phase identified in the framework laid out in table 9.1 and described in the text. On first reflection, embracing decentralization would seem to be some- thing of an unnatural act for political elites at the center of power: it seem- ingly involves the surrender of power on the part of politicians who are likely to have devoted their lives to its accumulation. Understanding the extent to which political elites embrace an agenda of decentralization thus requires careful attention to the underlying incentives and constraints gov- erning their behavior. These factors were radically different in the Sene- galese and Ugandan settings. UGANDA’S EMBRACE OF A DECENTRALIZATION WINDOW OF OPPORTUNITY. Decentralization in Uganda was intensely political and very much tied to the NRM’s pursuit of legitimacy after capturing power in January 1986. Conditions that Uganda confronted in 1986 provided three factors that were important to the success of the decentralization program. First, the collapse of a brutal central state made possible a critique (and for the incoming regime already a praxis) that made decentralization desirable. Second, the significant collapse of the state provided institutional fluidity that made decentralization less resistible by vested interests. Third, the NRM proved itself to be a regime both strong and astute enough to insist on decentraliza- tion and to author compromises that neutralized friction and thus furthered its purposes. The context in which the NRM came to power and the nature of the state it found explain much of why decentralization was central to its political Stephen N. Ndegwa and Brian Levy 295 agenda. The NRM under President Yoweri Museveni was the revolutionary descendant of a political party defeated in Uganda’s problematic 1980 elec- tions. Following widespread complaints that the election had been rigged, the NRM formed and began a guerilla war that eventually overthrew the second of several short-lived regimes that had intervened in the interim. Once it took power, the NRM pursued its Ten-Point Program of national reconstruction (Museveni 2000). Significantly, the very first goal of the pro- gram was to restore power to the people by first holding elections at the vil- lage level. Indeed, instituting village-level elections and creating so-called resistance councils were tactics that the NRM had adopted in its insurgency as it progressively conquered villages in its march to the center. The introduction of this level of democratic decentralization emanated from a critique of the decline of the democratic state, especially at the local level, since independence. Although the most graphic and memorable decline was under Idi Amin’s military dictatorship (1971–79), the decline of local government (and governance in general) had begun under Uganda’s first president, Milton Obote. A brief review of this history helps explain the verve with which the NRM pursued and Ugandans embraced decentraliza- tion in the late 1980s. After independence and until 1986, centralization pro- ceeded in tandem with the rise of authoritarianism in Uganda and was marked by a series of events that indicated the decline of constitutional rule. The years following independence in 1962 were characterized by a tension between the kingdoms and the central state. In 1966 President Obote sus- pended the independence Constitution that had provided for a quasi-federal arrangement between the central state and the precedent traditional king- doms, the most important of which was Buganda (Tukahebwa 1998). The abrogation of the independence Constitution and the enactment of the 1967 Constitution (essentially by putting Parliament through extreme duress) underscored the great centralization of power under Obote. Under the 1967 Local Administration Act, local councils became largely appointed bodies. According to Hansen and Twaddle (1998), since independence and especially after 1967, local governments “became not only instruments of central and party control but also sources of patronage and personal enrich- ment, trends for the most part reinforced under Amin and Obote II regimes” (p. 162). As Tukahebwa (1998) summarizes: The UPC [Ugandan People’s Congress] government in the 1960s cen- tralized power while the military regime in the 1970s accentuated centralization and militarized local administration up to the lowest level [sic]. Local administrations were thus avenues through which military directives could filter from the top to the lowest levels in the villages (p. 14). 296 The Politics of Decentralization in Africa: A Comparative Analysis In reaction to this lack of local representation and the excessive brutality of the central state under all the successor regimes following Obote’s rule (1962–71), the NRM pursued a preference for local power as a foundation for its own regime’s legitimacy. The institutional fluidity that confronted the NRM once it assumed power was also an important opportunity for it to rewrite the rules of state power. Thus, as Kasfir (1992) notes, when the NRM assumed power it proclaimed the 1967 Constitution—the last workable legal framework—valid, but it circumvented its deeply centralizing notions by “suspending portions that granted executive and legislation powers to the president and parliament” (p. 154.) Instead, the NRM redesigned the legislative organs around the five-tier, village-anchored, electoral college structure described earlier. SENEGAL—ENGAGING DECENTRALIZATION WITHOUT CONSUMMATING. Upon independence, Senegal inherited the radically centralized state that was the legacy of colonialism across Africa, especially francophone Africa. As with most newly independent African countries, the Senegalese rest- lessly experimented with a variety of options for restructuring the state bequeathed by colonialism; engagement with options for decentralization was an important part of that exploration. Yet for all its experimentation, in practice Senegal has been among the most stable countries in independent Africa. Unlike Uganda, where the 1986 NRM regime marked a profound break with the colonial legacy, which had itself decayed terminally over the previous 15 plus years, Senegal is in many ways a model of continuity and gradual change. This stability has yielded some important benefits: a progressively more pluralistic polity and slow, but relatively steady (and certainly better than the African average), eco- nomic gains. The environment has not been conducive to substantial state restructuring, however, not only in terms of decentralization, but also more broadly. According to a 2004 internal World Bank report, after 42 years of independence, “the administration is still excessively centralized,” “its per- formance is low,” and “it costs too much.” Reflection on the disconnect between the imperatives of decentraliza- tion, on the one hand, and Senegal’s institutional legacy, on the other, helps account for this modest progress. In Senegal, as in other postcolonial African states, the character of the state has been shaped by two critical con- ditions that prevailed at the dawn of independence and that gave rise to intense intervention in society and economy, for which centralization of decisionmaking was viewed as necessary. The first of these was an agenda of transforming and developing society, which it embraced as its overriding Stephen N. Ndegwa and Brian Levy 297 project even as it confronted scarce resources. The second was the shallow legitimacy of the state, which arose from the brevity of its historical exis- tence and which created an imperative for the Senegalese state to entrench itself in fairly short order after independence. Not unlike that of its French colonial precursor, the postindependence Senegalese state’s intervention in society was founded on the need to control the interior tightly and in particular arrest the potential centrifugal tendencies that, across the continent, threatened the very existence of the state. As a result, the central state relied heavily on a top-down system that governed the interior rather than allowing autonomous local governance structures, which it saw as potential independent sources and arenas of power likely to chal- lenge the weak center, to exercise control. Projecting state presence, securing its supremacy, and when possible, imposing its prerogative for taxation, became critical goals for the state. In this context, public order, often under- scored by the concurrence and convergence among elites that is possible only within centralized decisionmaking, leads to the dominance of the executive, who then becomes the principal and often the only mover of state action. The state’s quest for dominance over and legitimacy in society can fur- ther be seen in the longstanding relations it has had with the Islamic broth- erhoods, which have been central to politics in Senegal since before independence, in particular in mediating state presence at the local level.6 Even though it is a secular state, Senegal has had to secure support and compliance from the Muslim sufi orders to which an estimated 90 percent of the population belongs and which hold significant influence in individual and community relations with the state. For instance, in electoral politics, edicts from religious leaders historically played an important role in main- taining support for the Parti Socialiste and, in critical elections, the with- holding of such support often turned the fortunes of politicians. At the local level, the sufi brotherhoods are even more significant interlocutors of state power and resources. Religious leaders are often the principal local notables through whom the state must negotiate legitimate entry into communities. That the Christian president Leopold Senghor was able to remove his Mus- lim prime minister in a showdown shortly after independence was largely because Senghor was willing to recognize (and thus empower) the sufi orders as critical anchors of state legitimacy in society. That the current pres- ident made a very public pilgrimage to consult his religious teacher in one of his first public acts as president-elect is telling of the influence of the 6. For a detailed discussion of the role of Islamic society in state elaboration in Senegal, see Villalón (1995). 298 The Politics of Decentralization in Africa: A Comparative Analysis orders in Senegalese politics. Given that their power is most significant at the local level, the sufi are likely to influence decentralization because it will necessarily change the nature of the relationship between the state and reli- gious teachers. More recently, the reduction of the legitimate purview of state action through adjustment has made politicians and bureaucrats particularly sen- sitive to preserving the remaining areas, especially those critical to impera- tives of statehood such as revenue raising, expenditure, and projecting control (that is, exercising power) (see also Economist Intelligence Unit 2001, 2002). The further constraints on central power implied by decentral- ization therefore invite blockage rather than support from state elites. Fur- thermore, resistance to shedding further power or prerogatives to local authorities is likely to have been exacerbated by the transition associated with the 2000 regime change. In sum, attention to the incentives of Senegal’s political elites suggests that what is remarkable is not so much the lack of progress, relative to Uganda, in effecting decentralization, but the fact that, at least by compari- son with other African countries, any progress has been made at all. Detailing Decentralization: Harnessing the Bureaucracy to a New Order Detailing decentralization is defined in this chapter as comprising both the technical work of designing the fiscal and administrative nuts and bolts of the intergovernmental system and the practical work of implementing these designs. The role of the central bureaucracy is key in both technical design and implementation; however, its incentives are mixed, especially when it comes to implementation. Indeed in the phase of detailing decentralization, the crucial difference between Senegal and Uganda is not so much in atten- tion to the technical dimensions of the challenge, with which Senegal has engaged over a period spanning 40 years, as it is in beginning to implement workably a variant of the technical models. This section explores why the process of harnessing the central (and deconcentrated) bureaucracy to the task of workably implementing decentralization was able to proceed fur- ther in Uganda than in Senegal. UGANDA’S BUREAUCRATS AND THEIR POLITICAL PRINCIPALS. The terms of the relationship between Uganda’s political leadership and its central bureaucracy were defined by the accession to power of the NRM in 1986 and the means by which it occurred. That process left no room for doubt that the new political leadership was in the driver’s seat, although the Stephen N. Ndegwa and Brian Levy 299 process through which this assertion of political authority was effected took some years to complete, and ultimately the new leadership itself proved willing to embrace genuine decentralization only up to a point. When it came to power, the NRM initially accommodated many of the remnants of central state authority, especially the district administration, which tended to be the portion of the bureaucracy that survived several regimes because it acted, albeit powerfully, at the margins of the state: heads of government departments and chiefs were ex officio members of resis- tance councils, and councils could be suspended by the central government for certain offences. A district administrator was a presidential appointee who headed the district administration, but former district commissioners were placed as district executive secretaries overseeing deconcentrated gov- ernment ministries. The integration of this vestigial officialdom was largely shaped by the NRM’s inexperience in governing, the fact that it did not have enough peo- ple to ensure total dominance of the system, and an enduring propensity to compromise in the interest of a broad-based government. According to Kas- fir (1992), the “NRM leaders viewed this principle as an NRM reform intended to reverse political disorder by inviting opponents of the govern- ment to share power, instead of following the previous practice of monopo- lizing power in the hands of the victors” (p. 156). Thus, for example, the NRM invited several of its opponents to join the cabinet when it came to power. In these compromises, the regime’s early drive for legitimacy and fundamental concern to retain power is clearest. Set against this initial retention of the deconcentrated structure was the NRM’s commitment to establishing village-level resistance councils to gov- ern at the local level. Although in “no other respect during its first four years did the NRM government achieve as much progress in implementing the political program,” by 1990 the resistance council system seemed to reach its limits as an effective system of decentralized government (Kasfir 1992, p. 162). In the absence of fiscal and administrative levers, local coun- cils were limited, in practice, to the mundane, such as the distribution of commodities in times of shortage (Dicklitch 1998). Inevitably tensions emerged between resistance councils and deconcentrated civil servants, schooled as the latter were in longstanding authoritarian and centralizing practices. Also inevitably given the new realities of power in Uganda, the political leadership moved to restructure the relationship between its newly created five-tier political structure and the deconcentrated authorities. The 1993 Local Governments (resistance councils) Act marked the launch of the detailing phase of decentralization—structuring, legislating, 300 The Politics of Decentralization in Africa: A Comparative Analysis and implementing the specific political, fiscal, and administrative arrange- ments described earlier. It moved the decentralization agenda beyond dis- persing political power, albeit with continued central tutelage, and worked to concretize decentralization through “localized control over policymak- ing, implementation and financial management” (Brett 1995, cited in Dick- litch 1998, p. 69). The focus shifted from preoccupation with creating village-level councils to technically facilitating the districts and providing them with substantive power to make decisions. Yet, for all that, power had shifted decisively to the NRM, and for all the NRM’s commitment to decentralized participation, how far Uganda’s polit- ical leadership was ready to release central bureaucratic control over the decentralized realm had its limits. In part, these limits reflected a residual legacy of the past. For example, although a Commission of Inquiry had been appointed in 1987 to review decentralization and recommend further mea- sures to advance it, not until 1993 were the formal changes adopted. This delay reflected bureaucratic resistance, opposition from politicians espe- cially of the old order, the lack of precedent, and the champions outside the political establishment (Hansen and Twaddle 1998, p. 166). The 1993 legislation, however, also signaled a continued commitment on the part of the new leadership to maintain, through the presumably now loyalist central bureaucracy, discretionary influence at the decentralized level—evidenced by the continuing roles at the district level of the decon- centrated district executive secretary and of the central government repre- sentative, as well as the continuing right of central government to suspend, dissolve, or alter the prerogatives of local authorities. Given these measures, asserting that Uganda’s decentralized realm has decisively freed itself, even within its formally mandated sphere of authority, from the threat of veto from central political and bureaucratic leadership would be inaccurate. Uganda’s state, though certainly reconstituted radically since 1986, remains one in which strong residual controls continue to radiate from the center. SENEGAL’S ELITIST BUREAUCRACY. Bureaucrats everywhere are likely to find fault with reforms that erode their power and privilege. In Uganda, as we have seen, this conservative bureaucratic instinct was trumped by the country’s radical political transformation. Senegal, by contrast, occupies the opposite end of the African spectrum in terms of the stability and stature of the bureaucracy. In such a setting, overcoming natural bureaucratic resis- tance to decentralization calls for a deliberate program to transform the skills and attitudes of bureaucrats so they can find relevance, influence, and prestige in the new decentralized state. With no such program forthcoming, the Senegalese bureaucracy, not surprisingly, presumed that the cumulative Stephen N. Ndegwa and Brian Levy 301 effect of decentralization of authority, resources, and personnel from the center to lower jurisdictions would be a permanent loss of their preroga- tives and—judging by the fiscal shortfalls and administrative bottlenecks described earlier—proceeded accordingly. Bureaucratic bottlenecks to decentralization in Senegal, however, did not arise solely from devious machinations of a disgruntled civil service. Whether genuine or deployed simply as a protective ideology, the elitist attitudes bureaucrats hold toward Senegalese society, especially the rural masses, place severe constraints on their acceptance of decentralization as a good idea. Although not peculiar to the Senegalese case, the sense of superi- ority to society and of a special responsibility to transform society is pro- nounced among civil servants, especially those trained in the National School of Administration and the Judiciary (Ecole Nationale d’Administra- tion et Magistrature), the elite administration and management school.7 In much the same way as the colonial state administrators viewed the interior, bureaucrats trained in this school do not envision the local authorities as venues for self-motivated progress. Full decentralization would of necessity shift not only the arena of decisionmaking and implementation to the local level but would also challenge the development discourse that has long underscored the prerogative of the technically advanced center bureaucrats to develop the so-called “backward interior subjects” of the state.8 Perhaps determined efforts to enhance the skills of bureaucrats so that they could provide the kind of support local governments will need if they must carry out their new mandates under decentralization—combined with attrition at central levels and options for redeployment of staff to deconcen- trated and local government levels—could overcome vested and ideological bureaucratic inertia.9 Significantly, Senegal’s 1996 decentralization plan 7. The classic model which most of francophone Africa has followed is the National Administration School in France. See also, for example, Moudoud (1989) on Tunisia for similar attitudes among elite bureaucrats seeking to transform society through state action, including promoting decentralization through central planning. 8. Although we found no focused writings or discussions about these attitudes, the perception that the training state bureaucrats receive and their career progres- sions create an esprit de corps is broadly accepted (see Villalón 1995, pp. 84–86). 9. So far programs to support decentralization (for example, the U.S. Agency for International Development–sponsored Decentralization and Local Governance Support Program) focus on the demand side by training local leaders to learn more and respond appropriately to the opportunities created by the 1996 decentralization law. Although they recognize that training central bureaucrats to adjust to new roles is equally as important, this aspect has not been pursued (interviews with Abdoulaye Barro, Awa Diouf, and Abderhamane Djirie). 302 The Politics of Decentralization in Africa: A Comparative Analysis does not empower localities to hire their own staffs nor does it provide for the deployment of staff to deconcentrated levels. Instead, the current decen- tralization law creates new layers of relationships and organizations for ser- vice delivery. These new layers at times add to the existing number of institutions. The lack of clarity in some of the provisions has allowed central state functionaries to subvert the spirit of decentralization. In the health sector, for example, local-level committees on health are supposed to be created to coordinate budgetary allocations and service delivery. Because the law does not stipulate who should create these com- mittees, however, they have not been set up in most areas. Therefore deci- sions continue to be made by Ministry of Health functionaries who have no platform obligating them to share information and decisionmaking with local authorities.10 Similarly, although funds allocated to localities are trans- mitted through the bureaucracy (principally through the regional prefects and underprefects), the information is not automatically transmitted to the relevant regional council presidents or commune mayors, who are charged with expending it. Mayors and council presidents are thus obliged to refer to supposedly subordinate bureaucrats for this information although legally they are supposed to be in charge of the funds allocated to each locality.11 Thus in the complex web of authority that arises given new insti- tutions created by the new decentralization law and the parallel layers of administration from before, decentralization is easily derailed by the inertia inherent in such complexity and ambiguity. Grass Roots Participation and the Sustainability of Decentralization: Uganda and Senegal Table 9.1 highlighted two distinct roles for grass-roots participation as a country transits to decentralization. At the initial phase, when the question of whether to restructure state institutions along decentralized lines remains open, a politically activated grass roots can be an important spur to shift resources and authority downward, closer to communities. Furthermore, at the mature phase, when the challenge becomes one of sustaining an inclu- sive system, community activism can help immunize decentralized govern- ments against the risk of local capture, ensuring that the decentralized system remains inclusive. 10. Interview with Awa Diouf. 11. Interview with Abderhamane Djirie. Stephen N. Ndegwa and Brian Levy 303 UGANDA—FROM AN EPHEMERAL TO INSTITUTIONALIZED GRASS ROOTS. When the NRM swept into power in January 1986, local organizations in Uganda were basically timid, powerless, and marginalized, having suffered through systematic subjugation under the first post-independence regime of Obote, virtual elimination through brutal public governance under Idi Amin, and subordination to political parties and proto-parties in the unsta- ble democratic years after the fall of Amin’s regime. Few independent civic organizations, mostly small commercial (for example, coffee growers associ- ations) and professional interest groups and small-scale village or women’s organizations, were active. The vast majority of the rural organizations had been forced to retreat from engagement with the state and were unable to act in support of community life as the state retreated from active gover- nance. As a result when the NRM secured dominance, it confronted a dual reality in state–society relations: a civil society that, because of state brutal- ity, was deeply suspicious of the state, on the one hand, and that, because of state collapse, had managed to survive and retain relevance and abilities to pursue local governance imperatives in the absence of or in spite of the state but not with it, on the other. As such the NRM’s pursuit of local-level participation found ready embrace in this environment, especially as citizens in the rural areas discov- ered its lower tolerance for brutality against civilians. In this context the NRM instituted a countrywide resistance council system that provided for a five-layered representational structure paralleling a similar administrative structure. This structure, which represented not only a fundamental over- haul of the state but also important compromises with existing local institu- tions and patterns of power, established several subnational levels of participation and shared power but also maintained the center’s fundamen- tal hold on politics at all levels, underscoring the guarded devolution underway. From very early on, Uganda’s decentralization was therefore met with a strong local embrace by communities across the country. This acceptance was a result of both the structural change in operating conditions for civic activity at the fall of the repressive regime and the direct action taken by the new NRM regime to activate local civic activity. These actions were integral to the NRM’s pursuit of capturing the state and rebuilding it anew as reflected in its ideological platform, the Ten-Point Program (Museveni 2000), which was focused on community ownership of the revolution through extensive participation in governance. In power the NRM’s effort to decentralize was thus a direct result of this commitment to local partici- pation in governance. In the absence of formal governmental organizations, 304 The Politics of Decentralization in Africa: A Comparative Analysis when it captured power and before elaborating a functional local govern- ment system, the NRM used the resistance councils, organizing villagers and neighborhoods into cells that built upward to the national assembly. In its initial efforts to reconstruct the state, the NRM used these cells to accom- plish basic state functions such as local security, local adjudication, and food distribution in times of drought and crisis response. The response by the previously disenfranchised civilians was an intense embrace. As one commentator stated, “as soon as the RCs (resistance coun- cils) were introduced by legal instruments, the ordinary population began to gain confidence in the locally based institutions and to undertake local initiatives.”12 The activation of civil society was important because it also produced a new political class for local councils that had been schooled in the local accountability imposed by the resistance councils. Although, in general, local district councils were more representative and citizens partici- pated actively in elections, their participation was limited in other facets that required more intense involvement or greater knowledge of adminis- trative details (for example, financial oversight). SENEGALESE SOCIAL CONSERVATISM. As noted earlier, a major buttress of Senegal’s long-standing stability has been the alliance forged between puta- tively modernizing politicians and bureaucrats, on the one hand, and the religious teacher-led sufi orders, to which more than 90 percent of the popu- lation belongs, on the other. Although this alliance has provided an impor- tant degree of civil society check and balance on the exercise of state power, the check and balance turns out not to be of a kind that is supportive of decentralization. Indeed, the Muslim brotherhoods do not seem engaged in the policy discourse on decentralization (Villalón 1995), even though they have traditionally held sway in several local elections and in some areas practically nominate the list of candidates running for local authority assemblies.13 Rather, the orders seek to maintain both autonomy and access to state resources and only rarely challenge state elites. As leading analysts have noted, in return for access to resources, the religious teachers help secure legitimacy for the state, and smooth pathways for bureaucratic prac- tice for the civil service (see especially Villalón 1995). To the extent that this 12. Sabati Makara, “Linking Good Governance, Decentralization Policy, and Civil Society in Uganda.” http://www.afrst.uiuc.edu/Makerere/Vol_2/chapter_ five.html. 13. Some analysts view the relationship between the state and religious teachers as a form of devolution of state power. Although this observation may hold true in holy cities (headquarters of certain orders, for example, Touba), it is not universally true, and the state retains dominance. Stephen N. Ndegwa and Brian Levy 305 relationship works, even suboptimally—and is what the Senegalese have known since before independence—all actors are likely to seek to retain it. At a minimum, any decentralization scheme would have to address how to engage this relationship between state and society at the local level. Other than the religious orders, civil society organizations, although active in other realms (for example, culture), remain particularly weak in their engagement with governance issues, a consequence of both the histori- cal dominance of the state over society and the privilege given to Muslim brotherhoods as mediators of collective interests. Civil society is thus not strong nor developed enough to seek interventions that forcefully articulate the need for decentralized governance. As a result, decentralization in Sene- gal remains very much a central government initiative rather than one that is pushed by civil society. More pointedly, as one respondent suggests, what is weak in Senegal is a sense of active citizenship among the people that prods them to engage the state at all levels not as supplicants, but as empowered citizens.14 Given its legacy of openness and stability, in the 1990s Senegal experienced a different kind of transition to democracy than other African countries where a whole- sale overturn of authoritarian regimes was often sought and therefore pro- vided opportunity to revise notions of citizen engagement. In Senegal, the demands for a greater democratic opening were predominantly stated in the language of alternation in political power, especially the presidency, captured by the popular clarion call for sopi (“change” in the Wolof lan- guage). Although clearly the masses embraced and the politicians encour- aged the understanding that sopi would not only bring change, but also would fundamentally reform the state (by then viewed as steeped in clien- telism, patronage, and corruption), the reality is that sopi did not envision a change of how citizens relate to the state. In its most radical expression, sopi signified changes in the state to be undertaken by the new ruling party, but the most widely understood end was an alternation of parties in power. Thus, the momentum for change—now long dissipated with the achieve- ment of alternation of presidential power—helped little to propel efforts to remake the state through decentralization through a civic awakening in local government. At the Threshold: The Politics of Malawian Decentralization This section analyzes the politics surrounding a third decentralization process, that of Malawi. We add this country for three reasons. First, 14. Interview with Penda Mbow. 306 The Politics of Decentralization in Africa: A Comparative Analysis examining it provides additional means of gauging the analytical utility of the framework presented in “Assessing the Politics of Decentralization: A Framework.” Second, in terms of elite momentum for decentralization, Malawi occupies an intermediate position, between Senegal and Uganda. Including it thus enriches our understanding of the political dynamics sur- rounding decentralization. Third, Malawi has engaged in a far-reaching ini- tiative, the Malawi Social Action Fund (MASAF), complementary to the institutional process of decentralization, to foster bottom-up participation in local public action. Social funds of this kind have been initiated in close to a dozen other African countries, as well as in Latin America and elsewhere. Understanding the opportunities and challenges for decentralization posed by this parallel bottom-up initiative helps enrich the utility of the compara- tive analysis for other countries. Progress So Far As with the previous two cases, a description of changes in Malawi in the direction of decentralization provides the necessary backdrop for the subse- quent analysis. A democratically elected government came to power in 1993, drawing to a close 30 years of one-party rule under President Hastings Banda. The new government sought to revamp the machinery of government and in particu- lar to restart poverty alleviation programs. Although decentralization and local governments were not mentioned explicitly in its Poverty Alleviation Policy Framework (1994), the plan placed a great deal of emphasis on acti- vating area and village development committees to propel development henceforth and also enshrined the principles that eventually propelled the MASAF (Republic of Malawi 1995). In July 1994, the government initiated studies of the District Focus for Rural Development and other decentralization initiatives and ultimately approved a Draft National Decentralization Policy in January 1996. Follow- ing this effort, it established a Cabinet Committee on Decentralization to oversee the policy development and, later on, legislation. In 1998, the formal policy was adopted and a Local Government Act (1998) passed giving effect to the policy (Republic of Malawi 1998a,b, 2000; UNDP/UNCDF 2001). Sub- sequent changes can usefully be grouped into three distinct categories: con- crete changes that have affected the reality “on the ground’” so to speak, steps to strengthen the institutional platform for the progressive implemen- tation of a democratic state, and detailed plans for rolling out fiscal and administrative decentralization. Stephen N. Ndegwa and Brian Levy 307 On the ground, two key actions have been taken, one on the political front and one on the administrative side: • Politically, the territorial structure of a decentralized Malawi was clarified, and elections were held for all the 39 district assemblies, in November 2000; the assemblies are democratically elected and functional. • Administratively, the position of district commissioner, formerly the local representative of the central state overseeing deconcentrated ministries and public administration, has been merged with that of the clerk of council to create a chief executive officer responsible to the assembly and hired by the Local Authorities Service Commission. To strengthen the institutional platform, the following initiatives have been undertaken: • The Local Government Finance Committee, an independent body to oversee intergovernmental fiscal issues and in particular financial probity in assemblies has been established and is functional. The Local Government Finance Committee was fully operational in the last two quarters of fiscal 2001–02 and worked closely with assem- blies on 2002–03 budget submissions. It is expected to serve as a con- duit for all central ministry funds that are allocated to the district assemblies. It is charged with the responsibility of transmitting these funds, setting the distribution formulas, and auditing the expendi- tures of local assemblies. • An accounting and financial management system has been devel- oped, reflecting the priority given to financial probity issues. All dis- trict assemblies had received a computer for financial uses, an integrated financial management information system has been deployed, and training has been effected for a majority of assemblies. Indeed, the interministerial technical committee on decentralization has recommended the expansion of the central level integrated finan- cial management information system to include local assemblies (Republic of Malawi 2002, p. 8). • The Local Authorities Service Commission has been established and has begun recruiting staff. All the district chief executive officers already serve under terms issued by the Local Authorities Service Commission; recruitment for other senior officers (for example, direc- tors of finance and health) was to be complete by July 2002. (It was delayed, however.) The Local Authorities Service Commission most 308 The Politics of Decentralization in Africa: A Comparative Analysis immediate task is to manage the conversion of central government employees at the district level to Local Authorities Service Commis- sion employees responsible to the local district assemblies. • The Malawi Local Government Association is operational, replacing the old Association of Local Government Authorities, which was the association of the former councils. Although only a few years old, the Malawi Local Government Association has found its tracking as a mouthpiece for district assembly issues. It is also a member of the interministerial technical task force on decentralization. Among Malawi Local Government Association’s most recent achievements on behalf of local assemblies is securing an agreement in principle by the government to provide monthly ward allowances for councilors who at present work without pay (The Nation 2002). The Malawians also have made substantial progress in planning how administrative and fiscal decentralization will be implemented: • All 27 district assemblies (excluding the city, town, and municipal assemblies) have written their three-year development plans in which they lay out their operational and development program and indicative budgets.15 • Seven of 10 ministries slated for devolution in the first phase have submitted their devolution plans, which have been reviewed and approved by an interministerial task force on decentralization and the Cabinet (Republic of Malawi 2002). The plans indicate a consider- able spread of activities. For example, the Ministry of Education has devolved management and administration of schools and deploy- ment, though not employment, of teachers and the Ministry of Agriculture has devolved district level planning, budgeting, and inspection and extension services. • The current budget (fiscal 2002–03) submitted to and approved by Parliament includes indicative figures for each decentralizing min- istry on how much of their budgets for operations and for invest- ment is to be spent at the locality. According to revised procedure, these funds will be transferred directly from the Ministry of Finance to the Local Government Finance Committee, which will then dis- burse them to localities. For instance, the Ministry of Education has 15. Although the Local Government Act (1998) recognizes cities, towns, and municipalities as “districts in their own right,” they have generally been excluded in certain things (for example, the District Development Fund and donors look at them separately). They are, however, also in many respects part and parcel of the process. Stephen N. Ndegwa and Brian Levy 309 allocated 40 percent of its budget to devolved activities; Water Development, 58 percent; Natural Resources, 47 percent; and Com- merce and Industry, 15 percent (Republic of Malawi 2002, p. 19). • A number of revenue bases have been transferred to local authorities (for example, business licensing) whereas others (for example, fuel levy) have been ceded to localities through a yet to be determined transfer mechanism after collection at the center. These new revenue sources have had significant effect on the revenues of assemblies; for example, revenues shot up by 15 percent in Dedza and 25 percent in Zomba.16 In addition, the District Development Fund has been expanded from the initial 6 pilot districts to 27, and the Ministry of Finance has allocated 5 percent of the national revenue base to assem- blies for their General Resource Fund. • Several bridging actions and policies are proposed to initiate or main- tain functionality while permanent policies are worked out or while transitions are made from one mode to another or from low capacity toward high capacity. For example, the assemblies are mandated to employ their own staff, but as a transition, the central government will pay the salaries of district-level staff for 12 months as they are converted to employees of the Local Authorities Service Commission. Another example is with regard to fiscal decentralization—always difficult even under the best of circumstances—in which the central government employs a formula adopted under the District Develop- ment Fund, which was inspired by the United Nations Development Programme, while a formal policy of intergovernmental transfers makes its way through approval channels (the technical committee, the cabinet committee, and Parliament).17 The question naturally arises, Will these elaborate plans be imple- mented? Both the framework presented in “Assessing the Politics of Decen- tralization: A Framework” and the contrasting experiences of Senegal and Uganda laid out in “Commitment and Contradiction: Senegal and Uganda Compared” suggest that the answer to this question lies in a careful look at 16. Interviews with financial officers of Dedza and Zomba district assemblies. 17. In late 2002, this policy had already been approved by the interministerial technical committee on decentralization and by the cabinet, and it now awaits leg- islative sanction by Parliament. The development funds are separate from the Gen- eral Resource Fund, which is expected to cover administrative costs. The policy (but not the act) provides for an allocation equal to 5 percent of the national revenue for this transfer. In fiscal 2001, this transfer amounted to MK 250 million. Less than half of it was actually disbursed, however. 310 The Politics of Decentralization in Africa: A Comparative Analysis the extent to which progress thus far is rooted in robust domestic political momentum. We turn now to this issue. Engaging Decentralization: Technocratic Entrepreneurs in a Time of Political Fluidity Decentralization in Malawi has both internal and external origins. The ini- tial impetus for its adoption came clearly from outside Malawi (that is, donors). The government embraced the agenda gradually yet increasingly strongly, however, especially after the transition to a democratically elected regime in 1993. Decentralization dates back to the late 1980s when Malawi adopted its second 10-year national development program, DevPol II (1987–96), which sought to provide for local-controlled development to combat endemic poverty.18 Following a poverty assessment in 1988–89, the United Nations Development Program recommended that decentralization be adopted as a policy framework. Before this time, local governance had been emaciated by the absorption of most of the functions of local authorities by district development committees existing parallel to elected, albeit in a single-party context, councils and accountable only to the center. Not until 1993 did the government adopt a policy of District Focus for Rural Development pro- gram with United Nations Development Program and United Nations Cap- ital Development Fund support. Under this policy, a District Development Fund was established and district level planning and local participation was to be encouraged. The implementation of the District Focus for Rural Devel- opment program was stalled, however, because the broader context of an authoritarian and overly centralized government remained inimical to real, locally driven development. In any case, the program floundered as the 30- year-old authoritarian regime faced massive challenge echoing the conti- nentwide transition to competitive politics. Decentralization took on a new verve with the change of regime in 1993 as the new, democratically elected government sought to revamp the machinery of government and in particular to restart poverty alleviation programs. The new government’s focus on poverty alleviation and its quick 18. DevPol, short for Statement of Development Policies, was the Government of Malawi’s 10-year development plan that appeared twice between independence in 1964 and 1987. Given the lack of presidential interest—President Hastings Banda reportedly showed no interest in any of the three attempts at the DevPol II in con- trast to his very direct influence in DevPol I—the local-level proclivities in DevPol II fizzled out for lack of implementation (Pryor 1990, p. 61). Stephen N. Ndegwa and Brian Levy 311 embrace of decentralization meshed well with the thrust of the nascent United Nations Development Programme, which saw decentralization as the antidote to the anti-poor policies of the former centralized regime whose institutional structures had yet to be dismantled. Under its Fifth Country Program, the United Nations Development Programme in 1992 had pro- posed to launch a District Focus for Rural Development pilot program in six districts to showcase the possibilities of truly decentralized development. With the change in government, this effort was accelerated and three dis- tricts were selected in the southern region (the most populous area), two in the central region, and one in the north.19 These districts were provided with direct development project financing, training to build administrative and policy skills, and targeted technical studies on various aspects neces- sary to further devolution. By 1997, the United Nations program had blos- somed into a US$23 million program covering all (then 26) districts,20 providing demonstration of the likelihood of success for decentralization and providing especially important policy and demonstrative thrust to the evolving decentralization new policy (UNCDF 2002; UNDP 1997). Yet these pilot initiatives, useful as they have been as pilots, will remain parallel until the full set of plans listed earlier for administrative and fiscal reforms are actually implemented. Just as decentralization never arose under the previous regime for very specific reasons, it is also not neutrally pursued in the current context. It is ultimately propelled by events, priori- ties, and individuals within the political realm making very deliberate cal- culations and emboldened by the institutional flux in a post-transition context. The fact that the decentralization program is explicitly tied to the democ- ratic transition in policy documents, legislation, and public pronounce- ments by politicians underscores its deeply political nature in Malawi. Malawi is one of a handful of countries that have consistently made the claim of decentralization being part and parcel of democratization. The heavy tie-in between decentralization and democratization provides a posi- tive force as is evident from its broad-based support. It also presents a num- ber of perils, however, that arise from what we know of democratic transitions in Africa over the last decade, from some of the late indications of trouble in the Malawi transition and, related, from the vestiges of the ancient regime. 19. The pilot districts were Dedza, Mangochi, Mchinji, Nkata Bay, Nsanje, and Thyolo. 20. Districts did not include cities, towns, and municipalities, which were only later recognized as districts in their own right for the purposes of decentralization. 312 The Politics of Decentralization in Africa: A Comparative Analysis One of the vestiges is the personal dominance of the president and of the center in most major policy issues. Thus the decentralization experiment is rolled out with the particular stamp of the presidency and remains centrally directed. For example, the Decentralization Secretariat and the Department of Local Government (once an independent ministry) are structurally situ- ated within the presidency. (For that matter, oversight for MASAF, the flag- ship local development actor, including appointment of its board, is similarly located in the presidency.) The location of the secretariat in the executive is often viewed as a positive indication of the regime’s support for decentralization, as in Uganda. This affirmation cannot be assumed, how- ever, because location in the presidency does not necessarily insulate the program from reversals within the presidency. Indeed, in Malawi the decen- tralization program is imperiled by some of the reversals witnessed in the democratic transition; in particular, where the regime and its supporters have conflated the survival of the democratic experiment with the perpetu- ation of the United Democratic Front government and, previously, of the presidency of Bakili Muluzi beyond the constitutional limit of two terms. Even for the most committed decentralizing countries, the state that chooses to decentralize must confront the patterns established by the status quo or the previous status quo. In the Malawi case, the heritage of a central- ized bureaucracy and a civil society and citizenry conditioned by that her- itage (in spite of the new language of people empowerment) as well as the heritage of a very weak local government structure, which must now be revived and transformed, present significant challenges. Within the political economy of Malawi, decentralization ought to be judged not only based on the installation of institutions for devolving power, and how well insulated from central power they are, but also on the discourse of development and a sustained shift from the heritage of anti-poor policies associated with the previous regime. In our estimation, the discourse has not changed substantially by virtue of decentralization, although the formal decentralization program may pro- vide the seed for its evolution. Among public officials, local politicians, and community members the expectation remains limited to decentralization bringing an appreciable substantive difference to their immediate infrastruc- ture needs rather than as an institutional buffer against bad governance or as a tool for participative development. These enduring attitudes and the lack of genuinely decisive reforms still left many observers, even in early 2003, to doubt the credibility of Malawi’s commitment to decentralization. The decentralization policy framework was slow to emerge in part for three reasons. First, decentralization became intertwined with the logic of Stephen N. Ndegwa and Brian Levy 313 deepening democratization closely associated with the successor regime. This twinning meant that decentralization would be pursued in tandem with, and therefore necessarily subordinated to, the logic of retaining con- trol and power and extending it without undermining the center and espe- cially the ruling party (see, for example, Republic of Malawi 1998a, pp. 2–5). The new regime saw decentralization as critical to its expansion of legiti- macy as a signifier of the end of centralized authoritarianism and also as a pathway to deal with the deep and broad agenda for development, which had been suppressed after three decades of an elitist centralized autocracy. Second, because neither the local assemblies, which were dissolved in 1994, nor civil society organizations, undeveloped because of the three decades of authoritarian rule, were able champions of decentralization, the policy and process remained insulated from institutional pressure from below for faster movement. Third, and related, the rapid pace and evident success of MASAF—the implementation of which bypassed both the central ministries and the defunct local councils—deflected pressure for rapid decentraliza- tion, which might otherwise have resulted from the dire need for immediate poverty alleviation. At the same time, MASAF also helped empower com- munities in their relationships with the central government. Whether the net impact of these two MASAF effects on decentralization is a plus is the subject of the next section. MASAF and Decentralization: Regenerating Participatory Communities? For all of Malawi’s engagement with the details of transforming a central- ized state into a decentralized one, on the ground the most significant and far-reaching change in the discourse and practice of development in Malawi has been the introduction of the MASAF. MASAF has activities in every dis- trict and a development expenditure that dwarfs that of all district assem- blies combined. The success of decentralization is thus bound to be affected by how the fully operational MASAF and the fledgling decentralized struc- tures interact with each other. This section will highlight some of the politi- cal opportunities and challenges for decentralization that emerge from this interaction. As a prelude, though, we must provide some detail on MASAF’s role on the ground. Following goals set out in its postelection Poverty Alleviation Policy Framework, the government of Malawi established MASAF in 1995 as a vehicle to fund community-level projects directly to address poverty across the country. As the primary instrument of reinventing government follow- ing the democratic transition, the Poverty Alleviation Policy Framework 314 The Politics of Decentralization in Africa: A Comparative Analysis (like the decentralization program arising from it later) was integral to the new government’s efforts at securing legitimacy and putting its own dis- tinct stamp on public processes. Under MASAF, development projects were to respond to self-articulated needs by communities rather than to a master plan at the national level (MASAF 1999, 2002). As such, it was a fundamen- tal departure from the previous development planning practice that was centered on national development plans designed by ministries, rearticu- lated in national budgets with no public input, and often imposed on local communities through the supposedly deliberative district development committees. Approved by the World Bank in 1995, with funding to the tune of US$56 million, plus a contribution of US$2.9 million from the government of Malawi and in-kind contributions from communities estimated at US$8.3 million, MASAF commenced operations in 1996 and completed its first phase in 2001. In 1998, a replenishment of US$66 million was effected through MASAF II, which is still operational; another replenishment of US$67 million under MASAF III is in preparation as of this writing. With an operating base of 14.5 percent or US$6,765,480, it has a reasonable overhead and is touted across the country and within the Bank as a very well-run pro- gram making a visible impact (MASAF 2002, pp. 15 and 59). Over the first five years, the project allocated US$46 million to 1,864 com- munity projects and a safety-net employment program tied to public works. By all accounts, MASAF has made an appreciable impact on communities across the country through new or upgraded social infrastructures, such as clinics, schools, and boreholes, and provided incomes to the most vulnera- ble. Indeed MASAF activity is ubiquitous; even a short-term visitor to Malawi is bound to hear of MASAF on radio, see its activities on television, and encounter the standout school blocks or boreholes that dot the country- side. Besides the physical infrastructure, MASAF has also regenerated com- munity institutions for self-help, expanded management and participation capacity in local villages, and opened up significant roles for women, who make up 46 percent of the trained project management committee members (MASAF 2002; see also a summary in the business magazine Malawi First (2002). What is the impact of MASAF on the momentum for decentralization in Malawi? On the plus side MASAF’s performance and its practices provide powerful and possibly enduring patterns that help sustain decentralized governance and development interventions that directly affect the poor. In its operations over the last five years, MASAF has successfully demon- strated the notion that local communities can be the drivers of their own Stephen N. Ndegwa and Brian Levy 315 development and have the capacity and willingness to devise, supervise, and sustain their own development projects in an efficient and transparent manner. As such, the success of MASAF has not only activated community self-help spirit but also it has also promoted new relationships between communities and local governments enshrining an empowered, participa- tive spirit and community ownership of public assets. This ethic of develop- ment backed by the unquestionable success of MASAF has minimized the claims that central government ministries tend to make about the lack of capacity, probity, and sustainability at the local level. Indeed this spirit may partly explain the relative lack of bureaucratic resistance that has character- ized the decentralization program so far. Yet for all of these potential benefits, MASAF is not a natural part of the landscape of sustainable decentralized institutions. MASAF arose within and in response to a vacuum in local government and in local-level, poverty-focused development after the collapse of the centralized authori- tarian regime. Upon its establishment, MASAF worked with the institu- tional structure of the deconcentrated ministries represented by the district executive committee and the district commissioner overseeing the rump district councils. At the same time it was explicitly set up in a way that did not leave it beholden to the deconcentrated structures (or, for that matter, the recently revived district assemblies). Structurally, MASAF remains a central government instrument, operating directly out of the Office of the President and mandated to bypass the dense network of ministerial institu- tions and interests and work directly with communities. The revival of functional local governments through the decentralization program presents a challenge to MASAF’s original reason for being. The conditions that necessitated an intervention such as MASAF, as a temporary injection of funds bypassing a weak central state steeped in authoritarian- ism and center-dominated development, have substantially changed. The reemergence of local governments has eliminated the institutional vacuum that existed at the local level. District assemblies, validated by democratic elections and a new policy framework, now stand poised to reclaim their role as legitimate actors in development administration at the local level. At best, MASAF could help energize local democracy, helping nurture relation- ships of accountability between communities and local authorities and thereby serving as a bulwark against local capture. At worst, MASAF could become an obstacle to the sustainable deepening of democracy, offering the illusion of bottom-up empowerment, even as it helps perpetuate top-down centralized control. Clarifying what role MASAF will play in Malawi’s evolving landscape of state decentralization remains a work in progress. 316 The Politics of Decentralization in Africa: A Comparative Analysis Malawi at the Threshold A simplistic conclusion from the comparison of decentralization in Senegal and Uganda might be that the centralized status quo of African states is suf- ficiently robust that it cannot be broken decisively (and, even then, only incompletely) with only sustained, and not necessarily peaceful, bottom-up pressure. Malawi emerges as a setting capable of rebutting this conclusion and pointing the way toward a more constructively incremental path. Over the past decade, virtually all of the building blocks for a successful transi- tion to decentralization have fallen into place: • The 1993 transition to democracy resulted in the election of a govern- ment with a strong apparent ideological commitment to popular participation. • The process of setting up an elected, political tier of local government has been successfully completed. • Detailed plans for administrative and fiscal decentralization have been drawn up, with the full participation, and seeming cooperation, of a rich array of central government actors. • The ground for bottom-up empowerment has been thoroughly plowed by MASAF. Yet the fact remains that, for all of the thoroughness of the preparatory process, none of these actions had yet shifted authority and resources from the hands of the central apparatus to the decentralized tier of government by the end of 2002. The 2002–05 period will be decisive in signaling whether the political will to cross the threshold of state transformation is indeed pre- sent in Malawi—and thus, potentially, also in many other African countries that are contemplating a proactive, uncoerced state transformation toward decentralization. Conclusion—Navigating the Real World of Decentralization Even as decentralization has moved toward the top of the discourse on state institutional reform in Africa and elsewhere, debate has remained rather remote from the evolving realities on the ground. Analysts have engaged in heated discussion of the merits of a decentralized state relative to a putatively well-functioning centralized alternative (notwithstanding the dismal realities of the past 30 years) and have offered much technical exposition about how a well-functioning decentralized state should function administratively and fis- cally. Advocates of decentralization have put forth little effort, however, to learn from actual country-specific pathways to decentralization. This chapter Stephen N. Ndegwa and Brian Levy 317 Box 9.1. Five Propositions for Nurturing the Process of Decentralization • In most settings most of the time, that is, as in Senegal and perhaps also Malawi, movement toward decentralization will be incremental at best. • Efforts to support movement toward decentralization should focus at least as much on the process of building coalitions of support as on the details of technical design. • The process invariably will be a messy one and will have uneven movement across different dimensions of decentralization and an uneven, stop-and- start rhythm. • Reformers should not presume that decentralization is always the preferred alternative for effecting change in the short and medium terms, but should consider the desirability and feasibility of a broad range of alternative strate- gies for strengthening downward accountability. • Reformers committed to strengthening downward accountability should work to ensure that a vision of a democratically decentralized polity com- prises the long-run backdrop—the guiding North Star—to which more prag- matic and diverse strategies adopted in the short and medium term are intended to converge over the longer term. has sought to help fill that gap. Building on the analyses of Malawi, Senegal, and Uganda, we conclude by highlighting five propositions about how the process can best be nurtured in directions that are both feasible and likely to sustain continuing forward momentum. The first of these propositions addresses the enabling environment for decentralization; the second and third relate to the design of decentralization; and the last two focus on the dynamics of decentralization and empowerment (see box 9.1). The Enabling Environment for Decentralization Taken together, the three country experiences illustrate vividly the vast gap between, on the one hand, rhetorical advocacy for decentralization, which is evident in all three countries, and, on the other, a genuine readiness on the part of central governments to devolve or delegate authority and resources to local governments, which is evident thus far only in Uganda and even there is subject to important constraints. Indeed, the country experiences suggest that a country is likely to make a rapid and sustained transition from a centralized to a decentralized state structure only if all of the follow- ing three enabling political conditions are in place: • Fluidity in the macropolitical discourse is sufficient to enable basic issues of state structure to find their way onto the political agenda. 318 The Politics of Decentralization in Africa: A Comparative Analysis • A powerful political coalition exists and has both the incentive and the authority to push through policies of decentralization. • Stakeholders at the community level are sufficiently engaged to be supportive of and responsive to initiatives by political elites to shift resources and accountability downward. Only in rare circumstances—as in Uganda after the NRM’s and Muse- veni’s accession to power—will these three enabling political conditions be in place. As a first conclusion, it follows that in most settings most of the time, that is, as in Senegal and perhaps also Malawi, movement toward decentralization will be incremental at best. The remaining four conclusions developed in the following paragraphs are intended to guide engagement in those many settings where the way forward on decentralization is likely to be an incremental one. The Design of Decentralization The three country experiences offer important insights into the balance between political and technical influences in shaping the design of decen- tralization. In Uganda, political momentum led with the momentum of change. Technocratic fine tuning caught up. By contrast, in both Malawi and Senegal, reformers have been preoccupied with the technical dimensions of the process, with at best modest attention to the political dynamics, and so they have little to show for the efforts so far by way of changes on the ground. Taken together, these experiences suggest that the drivers of decen- tralization are primarily political and only secondarily, at most, techno- cratic. As a second conclusion, it follows that efforts to support movement toward decentralization should focus at least as much on the process of building coalitions of support as on the details of technical design. The implications for design of this second conclusion go beyond simply timing investment in technical work to ensure that it does not get too far ahead of the change process itself. The country experiences suggest that in any given setting, movement toward decentralization emerges as the result of the give-and-take among political elites; this give-and-take engages diverse interests, diverse visions of where the process might be going, and diverse strategies for taking advantage of windows of opportunity. As a third conclusion, it follows that the process invariably will be a messy one and will have uneven movement across different dimensions of decentral- ization and an uneven, stop-and-start rhythm. This assessment is not intended to be a prescription that anything goes. Rather, the intent is to caution for patience and generosity in judging Stephen N. Ndegwa and Brian Levy 319 whether some specific country’s decentralization process is on track: only after a longer period (typically in excess of a decade) of evolving experimen- tation and practice are the observed configurations of institutional arrange- ments for political, administrative, and fiscal decentralization likely to be entirely consistent with one another and aligned well with theoretical notions of best practice—or, alternatively, be consistently contradictory over and above the contradictions inherent in institutional transitions to be judged clear failures. The Dynamics of Decentralization and Empowerment The first three conclusions highlight the importance of not becoming preoccu- pied with advocacy of some preconceived notion of best-practice decentral- ization. Rather stepping back is key as is focusing more broadly both on the larger developmental purpose and on the time path of change, viewed from the perspectives of both shorter run feasibility and longer run sustainability. Viewed from a broader developmental perspective, decentralization through systematic reforms of intergovernmental political, fiscal, and administrative systems is one way—albeit an especially far-reaching one— of strengthening the downward accountability of a state to its citizens, beyond what is achievable through representative national democracy. Other possibilities for strengthening downward accountability include par- allel social-fund–like mechanisms (as in Malawi’s MASAF), participatory deconcentrated service delivery (of which an especially celebrated example is Uganda’s empowerment of parents committees of schools), and project- level support for local governments (as in Senegal, which has had an effec- tive, and expanding, pilot program to build the capacity of selected urban municipalities). Given their short-term focus, such arrangements must be properly calibrated not to undermine the long-term institutional develop- ment of local governance. Countries embarking on a process of deepening democracy are likely to vary in terms of the relative feasibility and efficacy of these alternatives. As a fourth conclusion, it follows that reformers should not presume that decentralization always is the preferred alternative for effecting change in the short and medium term, but should consider the desirability and feasi- bility of a broad range of alternative strategies for strengthening downward accountability. Yet even as multiple options indeed exist for strengthening downward accountability in the short and medium term, they vary in their institutional sustainability. By virtue of both its institutional coherence and the way in 320 The Politics of Decentralization in Africa: A Comparative Analysis which it locks-in a transformation of how power and resources flow through the polity, decentralization through intergovernmental systems reform has special advantages for sustainability. Finally, it follows that reformers committed to strengthening downward accountability should work to ensure that a vision of a democratically decentralized polity com- prises the long-run backdrop—the guiding North Star—to which more pragmatic and diverse strategies adopted in the short and medium term are intended to converge over the longer term. Put differently, for both tactical and strategic reasons a purist approach to decentralization can often be counterproductive in the shorter run. At the same time heterodox, more short-term initiatives (such as MASAF) need to be designed in ways that keep in mind the longer run, building-in mechanisms that facilitate, rather than hinder, their eventual incorporation into a holistic and coherent inter- governmental political, fiscal, and administrative system. References Dicklitch, Susan. 1998. The Elusive Promise of NGOs in Africa. New York: St. Martins Press. Economist Intelligence Unit. 2001. Country Profile Senegal 2001. London. _____. 2002. Country Report Senegal March 2002. London. Girishankar, Navin. 2002. Implementation Completion Report: Institutional Capacity Building Project (Uganda). Washington, D.C.: World Bank. Hansen, Holger Bernt, and Michael Twaddle, eds. 1998. Developing Uganda. London: James Currey. Kasfir, Nelson. 1992. “Government and Politics.” In Rita M. Byrnes, ed., Uganda. Washington, DC: U.S. Government Printing Office. Malawi First. 2002. “MASAF” 3(1) (Special Supplement): 33–48. MASAF (Malawi Social Action Fund). 1999. Getting Acquainted with MASAF: An Ori- entation Booklet for Stakeholders in the MASAF. Blantyre, Malawi. _____. 2002. MASAF: A Review of Aims and Achievements. Reprinted from Malawi First (2002). Moudoud, Ezzeddine. 1989. Modernization, the State and Regional Disparity in Develop- ing Countries. Boulder, CO: Westview Press. Museveni, Yoweri. 2000. What is Africa’s Problem? Minneapolis: University of Min- nesota Press. The Nation. 2002. “Councilors to Get New Perks,” September 13. http://www.nation audio.com. Ndegwa, Stephen. 2002. “Decentralization in Africa: A Stocktaking Survey.” African Working Paper no. 40. World Bank, Washington, D.C. Stephen N. Ndegwa and Brian Levy 321 Pryor, Frederic. 1990. Malawi and Madagascar: The Political Economy of Poverty, Equity, and Growth. Washington, D.C.: World Bank. Republic of Malawi.1995. Policy Framework for Poverty Alleviation Programme. Lilongwe, Malawi. _____. 1998a. Decentralization Policy. Lilongwe, Malawi. _____. 1998b. Local Government Act, 1998. Lilongwe: Malawi. _____. 2000. Decentralisation Process in Malawi. Lilongwe. Malawi. _____. 2002. Sector Devolution Plans. Lilongwe, Malawi: Decentralisation Secretariat. Republic of Uganda. 1998. Background to Budget, 1998. Kampala, Uganda: Ministry of Finance. Tukahebwa, Geoffrey B. 1998. “Political and Administrative Relations in Decentral- ization.” In Apolo Nsibambi, ed., Decentralization and Civil Society in Uganda: The Quest for Good Governance. Kampala, Uganda: Fountain Publishers. UNCDF (United Nations Capital Development Fund.) 2002. Malawi. Local Gover- nance. http://www.sdnp.org.mw/undp/Undpmlw/Infokit/lg.htm. UNDP (United Nations Development Programme). 1997. http://www.sdnp.org. mw/undp/Undpmlw/programme/CCF/ccf.htm. ———. 2002. Malawi Information Kit. Local Governance and Development Management. http://sdnp.org.mw/undp/Undpmlw/Infokit/lg.htm. UNDP/UNCDF (United Nations Development Programme/United Nations Capi- tal Development Fund). 2001. Program Agreement Document. Decentralized Gover- nance (2002–2006). Lilongwe, Malawi: UNCDF. Villalón, Leonardo. 1995. Islamic Society and State Power in Senegal. New York: Cam- bridge University Press. World Bank. 2003 Senegal: Decentralization and the Challenges of Intergovernmental Rela- tions Reform for Better Service Delivery. Report no. 24384-SEN. Washington, DC: World Bank. 10 Education Decentralization in Africa: A Review of Recent Policy and Practice Alec Ian Gershberg and Donald R. Winkler This chapter surveys the changes occurring in the finance and delivery of basic education in Africa. It focuses, in particular, on the decentralization of basic education functions and responsibilities from central government ministries to subnational governments, to communities, and to the schools themselves. It also focuses on recent decentralization experiences, which appear to be deeper and more significant than those reviewed in a similar survey carried out five years ago (USAID 1997). African education decentralization occurs in the context of severe defi- ciencies in educational access (and quality), but growing financial resources for basic education, strong country commitments to use additional resources toward that end, and rigorous monitoring by donor countries (in the context of the Millennium Development Goals and the Education for All initiatives) to ensure both that funding for basic education increases and that access improves. The additional funding, the new and serious commit- ments, and the continuous international monitoring translate into a demand that countries find more effective and cost-effective means of delivering basic education. Based on international experience, decentralization is viewed by many as offering the promise of a new and more effective mode for organizing the delivery of education—under certain conditions. Hence it 323 324 Education Decentralization in Africa: A Review of Recent Policy and Practice is timely to ask what has been African experience to date with education decentralization. The rationale for education decentralization involves improving effi- ciency, effectiveness, and democracy. Improved equity, too, is a rationale for decentralization, although analysts also often acknowledge that because decentralization makes localities more reliant upon their economic and social endowments, some aspects of equity may suffer in the absence of ade- quate compensatory mechanisms. No silver bullet exists: what is equitable may not be efficient; what is efficient may not be democratic; and what is democratic may not be equitable. In practice, reform strategies must attempt to optimize the sometimes inevitable trade-offs between efficiency, equity, and democracy while seeking to improve on all three. The actual design and implementation of decentralization reforms are inherently polit- ical processes; thus the decisions about making these trade-offs rightly occur in the political arena. Here we discuss the rationale, as opposed to the political reason, for decentralization. Politically, decentralization may be pursued as a strategy to, for example, shift expenditure obligations to lower levels of government or resolve ethnic conflicts by giving different language groups or tribes greater autonomy. This chapter first lays out a framework for assessing education decen- tralization in Africa. Thereafter we use the framework to review African experience to date with the devolution of educational finance and delivery to subnational governments, both regional and local. We examine an espe- cially important experience for education decentralization in Africa—the community school. Finally, we look at the lessons learned from education decentralization in Africa in the context of international experience with education decentralization and ask what these experiences tell us about how best to make the transition from the centralized to the decentralized delivery of education. Education Decentralization: An Analytical Framework This section lays out the framework that will be used to assess education decentralization in Africa. First, a typology of alternative approaches to education decentralization is presented, highlighting some key questions that can help measure the degree to which real decisionmaking power is decentralized in practice across a sample of eight African countries. After that, as a guidepost to assessing good practice across the African continent, some key lessons from international experience, which has been longer and deeper than in most of Africa, are highlighted. Alec Ian Gershberg and Donal R. Winkler 325 A Typology Most decentralization typologies begin with the requisite references to deconcentration, devolution, delegation, and privatization.1 This premise is particularly true when education decentralization is part of a general gov- ernment decentralization—often also part of a public sector reform effort to improve democracy and the legitimacy of the state. Recently attention has focused equally if not more on decentralization to schools and school man- agement committees (commonly called school autonomy and school-based management) as opposed to decentralization to (or of) governments. • Education deconcentration is the transfer of decisionmaking from the central government ministry of education (MOE) either to the regional or local offices of the MOE or to the regional offices of the central government. This transfer typically entails giving those offices increased autonomy both in terms of recruiting, evaluating, and promoting personnel and in terms of allocating and reallocating budgets. It also often includes the decentralization of payroll and other administrative matters for teachers and other school staff. It may include some degree of political decentralization, too. Some- times the election of local or regional political officials is introduced at the same time that decisionmaking is deconcentrated to the MOE’s regional or local offices. In this way, local politicians may gain some influence over local administrative decisions even though they have no direct authority in education. • Education devolution is the transfer of decisionmaking from the central government to popularly elected regional or local governments. Key management decisions, including naming school principals and allo- cating regional or local education budgets reside with the governor and legislature or the mayor and city council. In some cases these decisions may in turn be delegated to schools or school councils. In most cases the revenues of the newly empowered regional or local governments are almost totally derived from central government transfers, thus limiting their fiscal autonomy. Fiscal autonomy and, arguably, fiscal accountability is higher when regional or local gov- ernments must raise a significant share of their own revenues. As 1. Rondinelli (1981); Rondinelli and Nellis (1986); and Rondinelli, McCullough, and Johnson (1989, pp. 57–87) created the original basic vocabulary for describing the various ways in which governments may be expected to pursue administrative decentralization. Winkler (1989) adapted the typology to education. 326 Education Decentralization in Africa: A Review of Recent Policy and Practice with deconcentration, administrative and personnel functions are often transferred. Devolution can be part of political decentralization or a way for central governments to offload service responsibility. • Education delegation is the reversible assignment by the central or regional government MOE, or in rare cases the municipal department of education, to public school principals or (usually elected) school councils, or both. The powers of these school officials vary greatly by country. In some cases they do no more than maintain the physical plant, whereas at the other extreme, school councils may name school principals, help prepare and approve school development plans, and approve school spending plans. • Implicit or de facto delegation to community schools is a special case of education delegation. It sometimes results from the failure of the state to provide educational opportunities in remote areas, so the community takes upon itself the finance and provision of schooling. In other cases, the government actively subsidizes and supports com- munity schools as an especially cost effective means of expanding educational access or as part of a strategy to improve accountability through local involvement. Table 10.1 extends the familiar typology in a way that highlights each of the administrative, fiscal, and political dimensions of education decentral- ization. Decentralization efforts differ not only according to the approaches delineated in table 10.1 but also according to the distribution of real deci- sionmaking power. Table 10.2 provides one classification and listing of the kinds of decisions that may be decentralized (OECD n.d.). Although it is quite difficult to measure the distribution of real decision- making power, international experience reviewed suggests a few key indi- cators that one can use. These indicators can be expressed in terms of questions: • Who determines marginal changes in teacher compensation? • Who makes the decision to recruit or transfer a teacher to a specific school? • Who selects the headmaster? • Does the school community or local government partly finance the school? • Who decides how to allocate the school’s annual budget? The next section uses these questions to explore the pattern of decentral- ization in eight African countries: Burkina Faso, Ethiopia, Mali, Nigeria, Senegal, South Africa, Tanzania, and Uganda. Alec Ian Gershberg and Donal R. Winkler 327 Table 10.1. General and Education Decentralization Matrix Education/general Administrative Fiscal Political Deconcentration Managerial deci- Regional mana- Regional, elected to regional gov- sions and manage- gers are given bodies are created ernment offices rial accountability greater authority to advise regional and regional are moved to to allocate and managers. MOE offices regional offices of reallocate budgets. central govern- ment and MOE. Devolution to Education sector Subnational gov- Elected regional or regional or local managers are ernments are local officials of governments appointed by given the power to general purpose elected officials at allocate education governments are local or regional spending and, in ultimately level. some cases, to accountable both determine spend- to the voters and to ing levels (that is, sources of finance by raising for the delivery of revenues). schooling. Delegation to School principals School principals School councils schools and/or and school coun- and school coun- are elected or school councils cils empowered to cils receive appointed, some- make personnel, government fund- times with power curriculum, and ing and can to name school some spending allocate spending principals. decisions. and raise revenues locally. Implicit delega- School principals Self-financing is School councils tion to commu- and community used with some are often popu- nity schools school councils government subsi- larly elected. make all decisions. dies, especially in remote areas where public schools are not present. Source: Authors. 328 Education Decentralization in Africa: A Review of Recent Policy and Practice Table 10.2. Types of School-Level Decisions That May Be Decentralized Decision area Specific decisions Organization of instruction Select school attended by student. Set instruction time. Choose textbooks. Define curriculum content. Determine teaching methods. Personnel management Hire and fire school director. Recruit and hire teachers. Set or augment teacher pay scale. Assign teaching responsibilities. Determine provision of inservice training. Planning and structures Create or close a school. Select programs offered in a school. Define course content. Set examinations to monitor school performance. Resources Develop school improvement plan. Allocate personnel budget. Allocate nonpersonnel budget. Allocate resources for inservice teacher training. Source: Authors. Guideposts from International Experience By now significant international experience with education decentralization has accumulated. This rich experience provides several lessons learned that help guide the assessment that follows of education decentralization in Africa. Some key lessons include the following. • Efficiency and effectiveness are most likely to improve under decen- tralization when service providers—schools, local governments, or regional governments—are held accountable for results (that is, they suffer the consequences of poor performance or receive rewards for good performance). In the case of decentralization to subnational governments, accountability occurs at the ballot box; in the case of decentralization to schools, principals or teachers may not have their contracts renewed when performance lags far behind expectations (for example, El Salvador; Minas Gerais, Brazil; and New Zealand). Accountability is arguably stronger when the local government or school community provides a share of school financing. Alec Ian Gershberg and Donal R. Winkler 329 • Accountability requires clear delineation of authority and responsi- bility and transparent, understandable information on results (both educational and financial). When responsibilities are shared by more than one level of government or when a school principal has only limited managerial powers, identifying who is responsible for poor performance may be difficult. The clients of education—parents and citizens—need reliable and timely information on their school’s per- formance and how it compares to national standards and comparable schools. School assessments may take the form of annual, censual tests of students’ academic knowledge (for example, Chile and the Netherlands) or periodic, in-depth assessments of school perfor- mance (for example, New Zealand). • Decentralization of real decisionmaking power to schools or school councils is a means of increasing the voice of education’s clients and can significantly increase parental participation in the school; alterna- tively, school councils that are only advisory in nature cannot sustain parental participation. High levels of parental and community partic- ipation are associated with improved school performance (for exam- ple, El Salvador and Nicaragua). In several cases, parents play a role in monitoring teacher absenteeism in rural schools, and sometimes they have the power to authorize payment of teacher salaries or salary supplements. In such cases, teacher absenteeism largely disap- pears as a problem. • Decentralization of education to subnational governments—a policy that is usually part of a larger reorganization of government—does not in and of itself empower parents and improve school perfor- mance (for example, Argentina and Chile). Further decentralization to schools (school councils or school boards) or local communities—a policy that is often initiated within the education sector itself—does empower parents and can improve school performance, especially when changing the organization of education is simultaneously accompanied by attempts to improve teaching and learning (for example, Memphis, Tennessee, and Minas Gerais, Brazil). • For decentralization to schools to be successful, principals must acquire new skills in leadership and management—of finances, of teachers, and of community relations. In developing countries where principals often receive no special management training, acquiring these skills alone is an immense task, which can be facilitated by cre- ating principal networks, by identifying and disseminating examples of successful principals and successful practices, and through devel- opment of formal training programs. 330 Education Decentralization in Africa: A Review of Recent Policy and Practice • Most decentralization includes the transfer of financial resources to subnational governments or schools. The design of transfers has powerful effects on both efficiency and equity. Formula-based capita- tion transfers ensure predictable revenues and can be designed to give schools serving the poor or disadvantaged higher levels of fund- ing per student. When the capitation unit is average daily student attendance, as opposed to the number of registered students, power- ful incentives for schools to attract and retain students come into play (for example, Chile; Memphis, Tennessee; and New Zealand). • Decentralization requires that national and regional ministries of education be restructured to provide the new functions that they should provide to subnational governments and schools: diagnosis of problems and policy formulation, design of policies to ensure equity, student assessment, collection and dissemination of reliable informa- tion, and training and technical assistance to those now responsible for delivering schooling. Failure to restructure and reorient ministries impedes the implementation of decentralization and constrains its results. • The single largest obstacle to education decentralization is often the teachers’ union, which fears a loss of negotiating power and a loss of income if salaries are paid by poorly financed subnational govern- ments. Several countries have designed teacher pay and transfer poli- cies that have won the acquiescence of unions, however (for example, Chile, El Salvador, New Zealand, and Nicaragua). • Teachers are the most important factor in delivering instruction to children. Thus if the teacher management—recruitment, evaluation, transfer, and salary supplements—is not decentralized along with other responsibilities, the potential benefits of decentralization are highly constrained. • The single largest fear expressed by national education ministries is that subnational governments, communities, and schools lack the capacity to manage education. Although numerous actors—for example, principals, school councils, and municipal education secre- taries—require training to acquire new skills and knowledge, evi- dence shows that in practice even poorly educated parents and communities can manage community schools (for example, El Sal- vador and Nicaragua). • Decentralization is a long, evolutionary process. Although legislative and constitutional changes may radically change responsibilities overnight (for example, Argentina, Chile, and Mexico), real changes Alec Ian Gershberg and Donal R. Winkler 331 in governance, accountability, and impact in the classroom take much longer. Experience with education decentralization around the world suggests an idealized model of decentralization consistent with schools effective in enrolling and teaching students. In this idealized model, the national min- istry of education is a proactive agent for change trying to ensure that decentralization is accompanied by an increased focus on teaching and learning; communities are empowered to manage their own schools, irre- spective of which level of government is constitutionally or legally respon- sible for kindergarten through 12th grade education; principals take on significantly enhanced roles in leading and managing schools; parents are annually provided with information on their school’s performance relative to others and provided with a mechanism (for example, school or municipal education council) to strengthen their collective voice; and financing instru- ments are introduced to provide predictable revenues with incentives for enrolling and retaining students. Africa’s Education Decentralization in Practice Just as the approaches to public sector institutional reform vary from coun- try to country across the continent, so too do African countries exhibit a wide diversity of approaches to decentralization of education. This section examines three different types of approaches: decentralization of control over education to regional or provincial authorities, decentralization to local governments, and decentralization to community schools.2 The country examples to be reviewed are listed in table 10.3, grouped according to the three distinct approaches. The grouping is somewhat loose and is based on 2. Deconcentration is not considered a separate category, even though most countries in Africa have attempted to shift responsibilities from MOE offices in the capital city to MOE offices at the regional or district level. Ghana, for example, has undertaken several such deconcentration initiatives over the years, leading most recently to authorizing district assemblies to assume control over primary and sec- ondary schools. Guinea is another example of a country in which teacher manage- ment is gradually being shifted to the regional level, and nonsalary budgets are gradually being shifted to the prefectorial level. Although deconcentration may improve the efficiency of MOE operations and may provide a more hospitable envi- ronment for more serious forms of decentralization, it often does very little to increase community participation, enhance parental voice, and improve account- ability. For those reasons, deconcentration experiences are not examined in any depth here. 332 Education Decentralization in Africa: A Review of Recent Policy and Practice Table 10.3. Country Examples of Education Decentralization in Africa Type of education decentralization Country examples Devolution to regions Ethiopia, Nigeria, South Africa Devolution to localities Tanzania, Uganda Explicit delegation to schools Burkina Faso, Mali, Senegal Source: Authors. what was deemed to be the dominant or focal aspect of the reform, recog- nizing that some countries have adopted mixed approaches. Decentralization to Regional Governments This section describes the experiences of three countries that have strong regional governments that bear important responsibilities for the finance and provision of primary and secondary education: Ethiopia, Nigeria, and South Africa. The distribution of key education decisionmaking responsibil- ities in these countries is summarized in table 10.4. As detailed below, each country has had a distinctive experience. Even so, the following conclusions seem to apply broadly to this group: • Without decentralization of budget and personnel powers, the reform is unlikely to take hold in any meaningful way. The center must maintain the responsibility to provide the evaluation and other data necessary for subnational entities to operate effectively. • Regionalization often preserves many of the problems and inefficien- cies of centralization, because often regional governments are large enough to act as central authorities themselves. Thus good incentives for regional governments to decentralize further (for example, to schools and communities) should be put in place centrally. Often only lip service is paid to the further transfers of powers to local juris- dictions and actors. • Regional authorities may be better at picking subregional actors with capacity to take on core decentralized responsibilities, but they may not be any more attuned to local demands for service and local preferences, thus thwarting many of the potential efficiency gains driving the reform. • Countries often spend a great deal of time developing the legal framework for regionalization with very little implementation. • Clear interjurisdictional assignment of functions is critical, and in fact subnational entities prove quite responsive to the incentives they are given. Thus perverse incentives lead to perverse outcomes. Alec Ian Gershberg and Donal R. Winkler 333 Table 10.4. Regional Decentralization in African Education Function Ethiopia Nigeria South Africa Teacher Negotiated and funded Negotiated by the fed- Negotiated between compensation centrally, administered eral government—with national government regionally some input from the and three major unions states Teacher Regional governments State government Provincial with some recruitment under central school input guidelines Principal Regional government State government Provinicial with some recruitment school council input Regional Regional governments 10 to 15 percent of Mostly from central school finance create budgets from funds block grant with some centrally transferred regional additions from general resources and central revenue sharing revenue sharing Local school Funded centrally with Funds subtracted auto- Mostly from central finance some regional matically from federal block grant with some additions and school revenue sharing funds regional additions from feeds to localities. Although a central revenue sharing few local governments support capital pro- jects, they generally do not provide additional own-source revenues Allocation of Schools are “not autho- Localities with some Heavily influenced by budget rized to procure inputs input from schools, national and provincial. under any circum- though in fact dictated Schools have some lee- stances” (Girishankar, in large part by center way and ultimate Alemayehu, and responsibility, however. Ahmad 2001, p. 8) although both woredas and schools demon- strated capacity to monitor budgets, man- age personnel, and undertake monitoring and evaluation table continues on following page 334 Education Decentralization in Africa: A Review of Recent Policy and Practice Table 10.4. (continued) Function Ethiopia Nigeria South Africa School Funded centrally, Localities officially Provincial with central construction administered regionally responsible for school and some community with some input from construction. Both the funding schools federal and the state levels claim they have this function, however, and the localities only rarely receive capital investment funds. Source: Authors. • The all-too-common combination of a lack of a legal framework for the reform with poor delineation of finance and service responsibility leads to duplication of efforts at best and chaos at worst. Countries often end up with a dizzying array of boards, ministries (central and regional), commissions, and authorities—none of which has a clear mandate or the power to implement change. • Overall government size can get bigger, not smaller, as a result of regionalization. This consequence can be unintended and thus nega- tive, or it can be a necessary and planned strategy, although the for- mer is more common. • Regionalization is not likely to be any more successful for education than the overall country strategy for decentralization. That is, the big, cross-sectoral picture has to be in order for the smaller picture to take shape well. ETHIOPIA. Until the end of its civil war in 1991, Ethiopia’s Derg Marxist regime kept education and most other government services highly central- ized. The primary impetus behind efforts to decentralize after the war was to legitimize the state and give voice and power to the country’s largest ethnic groups. Because the ethnic groups were located by regions, a regional devo- lution of government was a natural fit for reform of the state. Regional gov- ernments are funded from central government block capitation grants, and regional councils create social sector budgets, with compensatory support for poor regions supported by the World Bank and other donors. Education decentralization, then, took place more as part of a wider governmental decentralization than as a sector specific reform aimed at improving school performance. Alec Ian Gershberg and Donal R. Winkler 335 Ethiopia has four levels of subnational government: regions, zones, woredas (similar to municipalities or school districts) and community coun- cils. Each level of government has a department of education that reports to a council, which then reports up to the next level of government. The regional bureaus report to the Ministry of Finance and the Ministry of Edu- cation. The regional bureaus are responsible for curriculum development— including the provision of textbooks in ethnic dialects. Regions, too, are responsible for teacher training and certification under recruitment guide- lines issues by the Ministry of Education (Moulton and others 2001, p. 50). Despite the apparent regional emphasis of its reform, in 1994 the govern- ment released an Education Sector Strategy that emphasized the role of the community in the school: The administration of elementary and secondary education and train- ing shall be decentralized in line with the ongoing regionalization process. Schools will be strongly linked with the community which will take responsibility in its well-being and upkeep. They will be made responsive to the local needs and requirements and shall act as centers for all educational activities of the community. The management of each school will be democratized and run with the participation of the community, the teachers, the students, and the relevant government institutions. In as much as possible educational institutions will be encouraged to run on an autonomous basis (AED 1996, p. 3). Implementation of the strategy has been slow. A United Nations Educa- tional, Cultural, and Scientific Organization review of 130 documents con- cluded that even though decentralization was clearly one of seven main education reform themes, “[T]here are no clear and detailed outlines regard- ing the relationship between the Ministry of Education and the regional, zonal, and wereda [woreda] educational bureaus” (Workineh and others 1999). By most accounts the regions often act as 11 centralized regimes. A study of one region stated that since “decision-making authority over even relatively small concerns [was] centered in the hands of the few, Regional and Zonal officials were simply overloaded with work. Meanwhile, school and woreda officials often waited for months for responses to even routine requests” (AED 1996, p. 22). School governance remained largely unchanged. A detailed World Bank study of one woreda (Girishankar, Alemayehu, and Ahmad 2001) confirmed much of this assertion. The woreda’s “educa- tional system was one of almost exclusive reliance on top-down or hierar- chical controls during allocation and implementation” (p. i). Schools are “not authorized to procure inputs under any circumstances,” although both woredas and schools “demonstrated capacity to monitor budgets, manage 336 Education Decentralization in Africa: A Review of Recent Policy and Practice personnel, and undertake monitoring and evaluation” (p. 8). In fact, the Bank study documents just how resourceful school and local officials can be, even within relatively centralized constraints—for instance, accelerating teacher promotions to offset restrictions on augmenting teacher salaries. Schools are required to create school management committees, which raise fees, help construct and maintain buildings, and advise on school improve- ment plans, but the Bank study documents that they were often “unable to strike an appropriate balance between hierarchy and participation, between control and flexibility in education planning” (Girishankar, Alemayehu, and Ahmad 2001, p. 5). The fact that school management committees are often co-opted by local elites suggests the education client has very little voice in Ethiopia. NIGERIA. Nigeria was reestablished as a democracy in 1998 after two decades of military rule. A federation with 36 states and 774 localities, Nigeria has a history of significant involvement of both state and local gov- ernment in the provision of primary education. A set of 1976 federal guide- lines and the 1979 Constitution established primary education as a local government responsibility and ended significant direct federal support for the sector. These reforms in the 1970s took place in the context of an effort to increase revenue sharing funds to state and local governments. During the 1980s and 1990s, however, the central military government sought to re-centralize and otherwise control a system that was officially decentral- ized, in part because of declining oil revenues, which led in turn to declin- ing revenue sharing funds. The overall process has created a highly unstable policy environment: The response of the Federal government [to the declining revenues] in 1988 was to establish the National Primary Education Commis- sion (NPEC) to coordinate and supervise the development of pri- mary education across the country, and to contribute 65 percent of the estimated total cost of primary teachers’ salaries. The intention was that the local governments would contribute a further 20 per- cent with the state governments providing the rest. At the same time, the Federal Government’s share of the Federation Account was reduced from 55 to 50 percent and that of local government raised from 10 to 15 percent. While overall the states were appreciative in principle of the increased funding, resentment developed over the powers of the NPEC and . . . the manner in which the Federal funds were distributed across states. . . . The reaction of the Federal gov- ernment took the states by surprise. In 1991, full responsibility for Alec Ian Gershberg and Donal R. Winkler 337 primary schooling was transferred to the local governments and their share of the Federation Account was increased to 20 percent and that of the states reduced to 25 percent, NPEC was abolished, and Federal financial support withdrawn. . . . [In 1993] NPEC was re- established and the actual cost of teacher salaries began to be deducted at source from the Federation Account allocation to each local government. This arrangement remains in place but changes are being debated (Hinchliffe 2002, p. 4). Federal education spending represents less than 10 percent of its budget and less than 1 percent of gross national product, a remarkably low level of effort. Primary enrollment rates and measures of efficiency (as well as other available achievement measures) are low, even for Africa. In addition, both inputs and outcomes vary considerably across regions and localities. In Nigeria expenditure and policy responsibilities are not clearly defined and hence have gravitated (whatever the original intention) to those agen- cies and levels of government that have the most political power. The situa- tion for primary education appears worse than that of secondary education. Primary schooling is financed mainly from the local governments’ revenue allocation, with some modest state and, largely ad hoc, federal government support, but it is managed by state primary education boards (SPEBs). In addition, although the SPEBs, which are viewed as deconcentrated arms of the federal Ministry of Education,3 and state Ministries of Education (SMOEs) officially manage primary schooling as dictated by the Constitu- tion, the local government funds pass through the federal Universal Basic Education Commission. The Universal Basic Education Commission has created a larger role for the federal government in the sector in the past few years. Officially the Universal Basic Education Commission is supposed to provide policy direc- tion, norms, and standards only, but clearly it has gone further and is imple- menting policy and providing services parallel to those of the SPEBs and SMOEs. In addition, local government education authorities (LGEAs) have input in the process, but their role and powers are not clear, although in principle LGEAs have the right to participate in primary education and run the schools. Both the federal and state governments, however, dictate the nature of that participation and have retained all significant functions such as personnel, budget, and most policy development. For example, a previ- ous education law made LGEAs responsible for school construction, but 3. This dynamic appears to be true even though SPEB chairpersons and perma- nent members are appointed by state governors. 338 Education Decentralization in Africa: A Review of Recent Policy and Practice both the federally linked SPEBs and the SMOEs claim they have this func- tion and the LGEAs only rarely receive capital investment funds. Further- more LGEAs report directly to the SPEBs, bypassing the SMOEs. Thus the states do not really manage primary education. Capacity, authority, and budget control at the state and local level is low, in part because of the federal re-centralization. The roles of the three levels of government overlap tremendously and create duplications of functions. The multiplicity of organizations in primary schooling suggests an abundance of administrative and technical staff. The decentralization–re-centralization process has resulted in a larger, not smaller, public sector. The local governments’ role is unusual. Although officially they fund primary education, the funds are in fact subtracted from their federal trans- fers before those transfers are made and with no input whatsoever from the local governments. Additional uncertainty is brought into the budget process, because the federal transfers are highly dependent on world oil prices from year to year. Indeed, because of rapidly increasing teacher salaries, some local governments receive no transfer once the school funds are subtracted; overall, the average local government loses about 42 percent of its budgeted federal funding before it even receives it. The federal gov- ernment—with some input from the states—negotiates salaries (the pri- mary use of those funds) and dictates class-size targets, accreditation, and other standards, and curriculum.4 States provide some additional salary incentives and some criteria for teacher training eligibility. This arrange- ment provides a perverse incentive for local governments: if enrollments go up, local discretionary income is likely to go down. Aside from a few local governments that support capital projects, local governments do not provide additional own-source revenues to the educa- tion sector. Officially, local governments provide about 85 percent of the funds for primary education, state governments 10–15 percent, and the fed- eral government the small remainder.5 In reality, the federal government appears to control most spending decisions. For the most part, however, LGEA monitoring of schools is limited to physical inspections, not student performance, although some monitoring does help identify effective teach- ers to receive additional training. We found some examples, however, of 4. Note that the obligation to provide teachers’ paychecks has migrated several times between federal, state, and local entities, and one result has been inconsistent delivery of teacher pay leading to strikes. 5. With all these budget figures it is important to note that complete expendi- ture figures have not been estimated since 1962. Through a case study of 10 states, Hinchliffe (2002) provides most of the rough estimates we present. Alec Ian Gershberg and Donal R. Winkler 339 LGEAs and elected local governments attempting to manage teachers and improve accountability. Secondary schooling is a bit more straightforward, funded primarily by state governments—between 70 and 100 percent with the remainder com- ing from the federal government. The average state share is about 80 per- cent, although these data are less reliable because states are not required to report education expenditures, and thus they do not. States manage sec- ondary school teachers and pay them. Finally, private household and teacher expenditures on schooling are sig- nificant. Average fees range from a few dollars per student annually for pri- mary schools to more significant sums for secondary schools, and parents and students pay similar costs for materials. Funds raised by the community, the parent-teacher association (PTA), and the school management committee pro- vide a large share of basic operations and maintenance monies. In addition, in some schools PTAs employ directly as many as half the teachers (Government of Nigeria 2001). In turn, teachers generally pay for their own inservice and preservice training. Schools do bear some of the burden of inservice training, because they must continue to pay teacher salaries while they are on leave. In principle, the federal government requires school management com- mittees and supports PTAs and nongovernmental organizations (NGOs) as means to improve local accountability. The capacity of the school councils and LGEAs needs support, however, if this process is going to be effective. This backing is particularly important given the propensity of donors to support school autonomy and provide funding that goes directly to the schools or their communities or both. The World Bank, the United King- dom’s Department for International Development, and the U.S. Agency for International Development all have projects to provide support to some aspect of community participation to build civil society and local govern- ment and to foster NGO and community-based organization involvement. Many of Nigeria’s accountability problems stem from the fact that the 1999 Constitution does not spell out the division of functions and responsi- bilities between the three levels of government and the myriad agencies that have some role in funding and providing basic education. Although some analysts have argued that putting the legislative horse before the reform cart is not always necessary, (see, for example, Gerschberg 1999) Nigeria appears to provide an example where the reform cart has failed. In part this failure derives from the complex set of institutional and intergovernmental relations that have been set up to provide basic education, leaving no government or agency clearly accountable for results. The federal government has used low state and local capacity to justify its role, but has done little to increase that capacity. In sum, the center has had a hard time letting go of any real power, 340 Education Decentralization in Africa: A Review of Recent Policy and Practice and one group of analysts categorized Nigeria’s education decentralization as “de-responsibility” (De Muro, Salvatici, and Conforti 1998, p. 13). SOUTH AFRICA. The historic end of apartheid in South Africa and the election of a new government brought great hope that intergovernmental fiscal and administrative relations would change for the better. The country established a Finance and Fiscal Commission and began an overall transfer of responsibilities for social service provision to the nine provinces. After the democratic elections in 1994, a great deal of attention was focused on transferring government functions in South Africa to its nine newly created provinces and, to a lesser extent, to its 284 municipalities. Education expenditure responsibilities were transferred to the provinces, each of which now has its own line ministry. Financing is still provided centrally. First, provinces get an equitable shares revenue sharing allocation to fund their general budget (some of which they may allocate to education). In addition, the central government funds schooling directly using a funding formula based largely on the num- ber of students. Some additional weight is given to poor and rural provinces. The central ministry sets curriculum and evaluation guidelines and credentialing standards for teachers and has (increasingly since the ini- tial decentralization) attempted to target central funds to priority areas of concern. For instance, immediately following the transfer of expenditure responsibility to the provinces, personnel spending soared while real non- personnel expenditures fell (Taylor 2001). The central education ministry and the Ministry of Finance have responded by earmarking central formula transfers for textbook purchases and other nonpersonnel expenditures. Provinces are required to rank schools by a poverty index and allocate funds for nonpersonnel expenditures according to a formula using that ranking. Personnel expenditures represent over 90 percent of the spending, and although the provinces appoint, assign, and transfer teachers and head- masters, the labor regulations negotiated with the national teachers unions do not leave much room to maneuver. Even the number of teachers a school gets is dictated by a centrally regulated postprovisioning model. The government has faced many challenges relating to accountability and management of schools. As Taylor (2002) says, “[T]he new government inherited a system of education in which the authority of the state [not to mention line managers] had been steadily eroded over a period of two decades” (p. 7). An additional challenge that arose with the provincial con- trol over education and other social sector spending was the apparent over- spending identified in 1997. The basic cause of the overspending was the requirement that provinces fulfill centrally established criteria such as Alec Ian Gershberg and Donal R. Winkler 341 maximum pupil–teacher ratios and nationally negotiated teacher salaries (Momoniat 2001). Thus the official idea expressed by one minister of edu- cation that “the whole of the national budgeting system was reorganized to allow provinces to determine their allocations based on their own priorities and the needs of their people” was neither entirely accurate nor (appar- ently) effective (Minister of Education 2002, p. 2). Since 1994, the propor- tion of the central budget allocated to education through the provinces has risen sharply, although as mentioned, this increase was absorbed entirely by hiring new teachers and paying them more. The 1996 South African Schools Act established, at each school, elected school-site councils composed of school staff, parents, and students, build- ing on a tradition of parent-teacher-student associations. These councils can make decisions about curriculum, personnel, budgeting, finance, and the school calendar. All decisions are subject to national and provincial guide- lines, which in many of the most important areas means the councils effec- tively have little power. For instance, personnel decisions are subject to nationally negotiated salary, hiring, and promotion standards. Principal candidates are interviewed and recommended by the school-site councils but approved by the provincial ministry. Schools can hire additional teach- ers out of their own-source revenues, but this step is usually taken only by the wealthiest schools. The focus appears to have been on giving principals training and some power to make decisions especially about financial man- agement. Schools deemed more capable by the ministry are given more free- dom. The councils do not yet have the evaluation information or capacity to hold principals accountable for school and student performance. The school-site councils can set fees and school calendars, recommend teachers and school staff for appointment, and recommend language and religious policies. “School governing bodies which qualify, may also apply and be allocated additional functions such as the ability to maintain school property, determine curriculum, buy textbooks and equipment, and pay for services to the school using the non-personnel allocation for that school,” (Minister of Education 2002, p. 4). Decentralization to Local Governments Unlike other regions of the world (for example, Eastern Europe), Africa has relatively few examples of local governments that have significant educa- tion responsibilities. Although a number of countries have plans to devolve more financing and provision decisions to municipalities and districts, few have advanced far. Mali is an example of comprehensive planning, but little implementation. It has plans to transfer to municipalities construction and 342 Education Decentralization in Africa: A Review of Recent Policy and Practice management of primary schools, recruitment and payment of primary school teachers, and preparation of school development programs. This section focuses on the two best-known examples in Africa of decen- tralization to local governments: Tanzania and Uganda. Some key proposi- tions that emerge from these two cases include the following: • The boundary between true devolution to local government and deconcentration of central ministry functions is blurred. Even for these two countries, debate continues about which best characterizes their reforms. • As in other country cases from around the world, decentralization to local governments alone seems to be insufficient to assure credible downward accountability of schools. The largest gains in this area appear to have come from complementary reforms that empower parent committees. • Reforms that do not transfer real power over budget and personnel (among others functions) tend to be weak and ineffective. Central bureaucrats are loath to give up their old roles and responsibilities. National negotiations over teacher contracts in nearly every country pose great obstacles to true decentralization. • Local officials appear to be more in touch than their regional and cen- tral counterparts with the demands of their constituencies and thus are better able to respond to the with well-matched service provision. • As with regionalization, the central government maintains a number of important roles—most often evaluation—but the need to maintain such controls is easily perverted to pure power politics on the part of the central government. • Publishing and publicizing the amount and purpose of funds trans- ferred to local entities can be a simple, effective, and nearly costless accountability enhancer. • Capture by local elites is a concern. UGANDA. After decades of civil war and dictatorship, the National Resis- tance Movement began to bring some stability to Uganda in 1986. This sta- bility included an overall effort to decentralize government legislated primarily through the 1995 Constitution and the 1997 Local Government Act. Uganda is divided into 45 districts with elected councils and chairs and over 800 subcounties. Recently, the government of Uganda has increased its effort in the educa- tion sector, raising spending from 2.6 percent of gross national product in 1996 (with only 43 percent allocated to primary schooling) to 4 percent in Alec Ian Gershberg and Donal R. Winkler 343 2000, or nearly a third of its discretionary recurrent budget. This increase was necessitated by the much celebrated “big bang” approach the government took to universal primary education in 1997, abolishing all fees for primary schooling and fully assuming the responsibility for financing the sector. Up to that point, household contributions represented about 60 percent of funds for primary schools. As a consequence, enrollments skyrocketed, and pupil–teacher ratios increased. By contrast, the government provides only 30 percent of funds for secondary schooling, and enrollment rates are low. Despite annual economic growth of 7 percent during the early 1990s, social services hardly improved in many respects. A now famous Public Expenditure Tracking Survey found that as little as one quarter of primary education grant monies actually reached schools and furthermore that schools operated under perverse incentives to misreport enrollment and fee data (Ablo and Reinikka 1998; Reinikka 2000). Since 1995, the government of Uganda has sought to redress these problems, namely, [T]o improve the flow of information, and make budget transfers transparent by: i) publishing amounts transferred to the districts in newspapers and radio broadcasts; ii) requiring schools to maintain public notice boards to post monthly transfer of funds; iii) legally providing for accountability and information dissemination in the 1997 Local Governance Act; and iv) requiring districts to deposit all grants to schools in their own accounts, and delegating authority for procurement from the center to the schools” (World Bank n.d.). Officially, the districts are responsible for providing primary and sec- ondary schooling but are supposed to devolve primary education to the subcounties and other local governments (villages and parishes) and schools, but the division of powers under the Local Government Act is not entirely transparent (Azfar and others 2000) (see table 10.5.) Districts recruit teachers, but teacher pay is both determined and provided by the central government. Lang (2000) captures the recent progress in this area nicely: When Uganda introduced [universal primary education], it also intro- duced a capitation grant system, which provides about $4 per child per year for children in grades one through three and $6 per child per year for children in the next four years. The government pays teachers salaries and textbooks, but the grants are used to fund other school needs. Uganda’s grant system is calculated centrally and released as a conditional block grant to districts, which in turn, release all funds to schools on the basis of enrollment. The ministry has also released guidelines to schools for allocation of funds, for example, 50% for 344 Education Decentralization in Africa: A Review of Recent Policy and Practice Table 10.5. Education Devolution in Uganda. Function Description Teacher compensation Set nationally, administered regionally (district level) through conditional grants Teacher recruitment Regional level Principal recruitment Regional level Allocation of budget Central transfer of funds to regions, subcounties, and schools; some school-level budget responsibility School construction Funded centrally, administered and overseen regionally, and implemented largely by schools Source: Authors. scholastic materials, 5% for administration, and so on. The School Management Committee manages the money at school level. The amounts received from the district office are posted publicly in the school. Some schools publicly display expenditures, but anyone can ask to see the records of how the money is spent. There have been regu- lar audits that show increasing evidence that the funds do reach the schools and are utilized for the purposes intended (p. 1). Regarding the role of the school management committees, Azfar and others (2000) explains: The School Management Committee, which is distinct from but often associated (or overlapping) with the PTA, now appears to be the most important governance mechanism dealing with education locally. These committees are empowered to sign checks for the headmaster, oversee the schools, and investigate problems. . . . The committees also oversee school construction and improvements (p. 9). The Ugandan decentralization experience has won significant interna- tional praise, although naturally it has had its pitfalls.6 Transparency of budgetary allocation, largely in response to the quite negative outcomes of the Public Expenditure Tracking Survey, has played a large role in its early 6. “In Uganda, decentralization had various positive developments: increased participation, increased transparency and accountability, improvements in capacity building, etc. Nevertheless, the central government transfers still are insufficient, and the local governments are neither involved, nor even consulted on the national budget” (De Muro, Salvatici, and Conforti 1998, p. 11). Alec Ian Gershberg and Donal R. Winkler 345 successful aspects. For instance, information on the conditional grants to districts are published in the national press and provided to schools. Schools and subcounties, in turn, must publicize their budgets and sources of funds. In addition, a rare, detailed analysis of the Ugandan reforms7 shows that in fact some of the touted theoretical benefits of decentraliza- tion can occur in practice. Specifically, subcounty government officials are well aware of the preferences of parents even if institutional rigidities pre- vent them from matching those preferences well. Unfortunately, devolu- tion may be reproducing centralization at the regional level, in part because the assignment of, and expectations for, subregional responsibili- ties is not well articulated in the decentralization legislation. This result is particularly negative because regional officials were found to be less in touch with citizen preferences than either subcounty or national officials. TANZANIA. Tanzania exhibits both poor performance and low levels of educational funding. Only 83 percent of the age cohort enters grade 1, and of those only 70 percent graduate. Public education spending is just 2.3 per- cent of gross domestic product. The low level of education spending is largely a function of low levels of government spending overall: about 20 percent of government spending is on education, and the pupil teacher ratio of 40:1 is on the low side for Sub-Saharan Africa. Sixty-two percent of gov- ernment education spending is on primary education. Cost recovery, fees, and household spending on education are thus high, representing over half of all primary school resources, not including in-kind donations to school projects and construction. Fees are considered an obstacle to enrollment. Tanzania has 20 regional units (regional secretariats) and 113 districts or local government authorities (LGAs). Subdistrict units include wards, villages (about 11,000), subvillages (about 70,000), and schools. LGAs in their current form were created in 1984, but only in the late 1990s were they solidified as locally elected, democratic bodies. Educational decentralization in Tanzania is part of a more general government decentralization often called the Local Government Reform Program in which various service responsibilities have been transferred to districts and LGAs through the President’s Office-Regional Administration and Local Government. Therkildsen (1998) summarizes: Primary education is the most important responsibility of local gov- ernments in Tanzania. Half of all their funds is spent on this activity 7. Azfar and others (2000) interviewed or surveyed 137 subcounty education officials, 18 district education officials, 145 school principals, over a thousand house- holds as well as key national officials. 346 Education Decentralization in Africa: A Review of Recent Policy and Practice (although most funds are provided by the central government), and two-thirds of all local government employees are teachers. Local gov- ernment decision-making is vested in the district council. A majority of its members are directly elected at the ward level. At the lower level, the village council has much the same functions as the district council. At the school level the school committee—in which parents are represented—is supposed to oversee the running of the school. Unfortunately, as implemented, education reforms in Tanzania often suf- fer from poor relations and coordination between the Ministry of Education and Culture; the president’s decentralization coordination unit, the Presi- dent’s Office-Regional Administration and Local Government; and the LGAs. The Ministry of Education and Culture has had difficulty letting go of direct service responsibility and reorienting itself toward policy forma- tion, monitoring, and evaluation.8 According to Therkildsen (1998): [T]he local governments (and the minister responsible) have little real influence on educational issues proper: curriculum; examinations; the relative weight given to academic and practical activities in the schools; the length of class room instruction; and so on. . . . Even the earlier permission for local authorities to adjust school terms to the agricultural practices in the has been removed. Nevertheless, the LGAs, too, are now expected to play an increasingly sup- portive role for village and subvillage councils and school committees. Some observers assert that Tanzania exemplifies deconcentration rather than devolution because the autonomy of the districts has been more formal than real: autonomy of local political authorities is severely limited, and a state of deep crisis exists (DAC 1997). As a consequence, some analysts argue that Tanzania’s system is still a deconcentrated one in which political, fiscal, and administrative controls remain firmly centralized (McLean 1997). The central government continues to provide most of the funds for teachers’ salaries, which are administered through the district level. Schools are now receiving some funds that they control directly, however. The cen- tral government uses capitation grants to schools of approximately US$10 per student for textbooks, teaching materials, teacher training, and other school projects. School committees help manage these funds and must develop an annual school improvement plan, which includes mobilizing 8. The Ministry of Education and Culture is still responsible for the provision of secondary schooling. Alec Ian Gershberg and Donal R. Winkler 347 additional contributions in cash and kind from the community. Other important central grants support school construction and rehabilitation, which are also supposed to be managed by the school committees with sup- port from the villages, especially regarding connections to community- based organizations and NGOs involved supporting the program. The World Bank has supported grants as well as school and subdistrict level capacity building necessary to manage such funds. Officially, districts have significant personnel control. In practice, how- ever, assigning teachers to schools most in need has proven to be difficult, especially because their compensation is negotiated at the national level (see table 10.6). Tanzania provides an example of a top-down decentralization reform in which the center has retained most of the decisionmaking power. The fail- ure to decentralize further has been hampered by the lack of clarity concern- ing the role of local governments at the outset and by the fact that the motivation for the decentralization was more political than educational. Nevertheless, in recent years, the Tanzanian reforms have begun to show some of the promise of improved service delivery from locally driven and controlled governance. Delegation to Schools and School Councils The empowerment of schools and school councils falls under several rubrics—community schools, school-based management, and community participation. Although these terms are often used interchangeably, they Table 10.6. Education Devolution, Tanzania Function Description Teacher compensation Set nationally, administered regionally, augmented at the school and community level through fees Teacher recruitment Regional level Principal recruitment Regional level Allocation of budget Central transfer of funds to regions, LGAs, and schools; some school-level budget responsibility School construction Funded centrally, administered and overseen regionally, and implemented by LGAs and school Source: Authors. 348 Education Decentralization in Africa: A Review of Recent Policy and Practice have quite different meanings when arrayed along a continuum of voice and authority. • Community schools tend to have strong parental voice and high authority, in which parents select the governing board. The govern- ing board in turn selects the school director and other personnel and, along with the school director, has a high degree of authority to make decisions. • School-based management is a term typically used to describe schools in which a high degree of authority has been delegated to the school principal, but parents may have limited voice in terms of assigning the director and other key personnel, in terms of selecting the govern- ing board (if there is one), and in terms of making important person- nel and budget decisions. • Community participation refers to the voluntary participation of par- ents and other citizens in school councils. Typically these school councils are advisory bodies rather than decisionmaking bodies and, typically, they fall apart if they are not granted serious decisionmak- ing responsibilities.9 Community schools, the focus of this section, are the most common form of educational decentralization in Sub-Saharan Africa as well as the most studied and evaluated (for example, see Miller-Grandvaux and Yoder 2002). In some countries they represent a significant percentage of all schools. For example, they represent 20 percent of primary schools in Togo, 32 percent of primary schools in Mali, 50 percent of secondary schools in Tanzania, and most primary schools in Ghana. Community schools are found at both the primary and secondary levels. Most are in rural (often remote) areas and rely upon inexperienced, uncertified teachers (often high school dropouts from the local community). They provide examples of some of the most successful efforts to date in the sector, and they also show the possible drawbacks. Community schools have deep historical roots, both as a response by communities to the failure of the state to provide access to schools—espe- cially, in remote rural areas—and as a result of efforts by churches to create autonomous, religious schools.10 Increasingly, however, governments are 9. Schaeffer (1994) distinguishes usefully between participation and involve- ment, with the former being the stronger of the two, involving a real transfer of power and responsibilities of important decisions. 10. Governments also limit access to public schools through the use of enroll- ment fees and dress requirements, which may put the cost of public education beyond the reach of the poor. Alec Ian Gershberg and Donal R. Winkler 349 stimulating their creation and growth, with the objective of rapidly expand- ing coverage at relatively low cost. Community schools now often fit squarely into larger reforms to decentralize school systems and govern- ments. Table 10.7 presents a simple typology for classifying the emerging variety of community schools. Each of the elements of the typology is explored further in the following paragraphs. The financing of community schools usually includes school fees, but a number of national and international NGOs—CARE, World Learning, World Education, Action Aid, World Care, Save the Children, the Associa- tion for the Development of Education and Training in Africa, Aide et Edu- cation, and the United Nations Children’s Fund—provide partial or full funding. Increasingly, these NGOs act as conduits for government, bilateral and multilateral grants and loans. A more recent phenomenon is govern- ment assistance. For example, Burkina Faso, with a net primary enrollment rate of only 32 percent, has begun paying the salaries of newly recruited teachers contracted by community school councils with the dual objective of rapidly improving access and doing so at much lower cost than would be possible through the traditional public school system with its highly paid teachers. Senegal, also, is moving to local recruitment of contract teachers. Guinea provides assistance to community schools in the form of training, materials, and teacher salary supplements. Government financial support need not come from the central government; in Ethiopia the local govern- ments match revenues from fees at the school level. One risk of decentralization to community schools with government assistance is the subsequent removal of government financial support. In Zambia, for example, after initially encouraging community schools, the Table 10.7. Typology of Community Schools Characteristics Definitions Motivation Community initiative versus government initiative Finance Self-finance through fees, government subsidies, and nongovernmental organization subsidies Governance Selection and composition of school council Council powers School construction versus school governance and management Regulation and supervision Benign neglect, encouragement, discouragement, or co-optation Source: Authors. 350 Education Decentralization in Africa: A Review of Recent Policy and Practice government subsequently reduced their funding, thus forcing schools to charge student fees to meet minimum costs. Currently, Zambian households spend almost as much on a pupil’s education (US$17 annually) as does the government (US$22 annually) in Zambia (UNICEF 2001). The governance of community schools varies across and even within countries, but it usually includes participation by teachers, parents, and community members. When school directors serve on the governing coun- cil, as in Guinea, they can come to dominate the other members, many of whom may be illiterate. Although the powers of community schools vary across and even within countries, they nearly always include community involvement in school construction and maintenance (for example, Malawi and Senegal). The pro- vision—including finance and construction—of a school by the community is often a condition for the government to place a teacher in the community. In many cases, communities provide all the financing, either in cash or in- kind, for construction. In other cases, communities may provide only a small share of matching financing (usually 10 percent), but also assume the responsibility for procuring and managing construction activities. Of course, by definition participation requires significant time, effort, and other noncash resources as inputs from parents and community members. Although such participation is generally touted as an unequivocal benefit, participation is itself a scare resource and any true evaluation of cost effec- tiveness would need to take account of the cost side of the participation ledger (Gershberg 1999). An important distinction is between community school councils that construct and maintain schools and those that actually govern and manage schools. With respect to the latter, almost all set the school calendar and daily schedule to fit parental needs, and almost all set their own fee levels. In Mali, Zambia, and several other countries, community school councils can hire and fire teachers. In Togo, however, the central government selects the school directors. Furthermore, community school councils with real powers—either in construction and maintenance or in governance and management—must be distinguished from and those with illusory powers. Ghana, for example, has legislatively mandated the creation of school man- agement councils, but has given them only advisory powers. Similarly, in Guinea school councils have a largely advisory role to play in terms of teacher management. In general, community schools are lightly regulated, if not in legal terms, at least in practice. In most countries, the central government regulates the core curriculum and sometimes textbooks, although no country but Zambia Alec Ian Gershberg and Donal R. Winkler 351 has gone so as far as to create a separate central entity to regulate and coor- dinate community schools services and activities. Even in Zambia, however, technical assistance to and supervision of community schools is rare, and the result can be inconsistent educational quality and a lack of qualified teachers and supplies (Miller-Grandvaux and Yoder 2002). Although benign neglect in regulation may be viewed as a benefit by the schools themselves, it is also a legal risk. Many community schools lack the firm legal basis to ensure their long-term viability. The legal and political risks to community schools are likely to increase as community schools become viewed as threats by labor unions and other education stakeholders. Mounting evidence indicates that community schools are educationally successful, especially in terms of reducing student and teacher absenteeism. Furthermore other evidence suggests learning has improved in some cases.11 Of particular note is the careful empirical study by Dowd (2001) showing that, in Malawi, the enhancement of community support for local accountability has had an important independent impact on student learn- ing. More broadly, anecdotal and qualitative evidence reveal that parental and citizen participation in the governance of community schools has not only improved over time, but also has spilled over into other realms of civil society.12 The studies that provide this evidence must be interpreted with caution, because many are self-studies or studies commissioned by the programs being evaluated, and few have the necessary baseline data and experimen- tal controls to provide statistically reliable results. A need remains for sys- tematic evidence of the impact of various community schools’ experiences on educational quality and other outcomes. Nevertheless, the impressions, however anecdotal, are positive enough to warrant guarded optimism. Conclusions and Lessons Learned Africa is on the path to decentralizing educational decisions from central government ministry offices in the capital city to regional and local adminis- trative units—sometimes of the MOE itself and sometimes of elected regional and local governments—and to local school communities. This 11. For example, see Odonkor’s (2000) evaluation of CARE’s School and Com- munity Oriented Education Project in Ghana and Hyde and others’ (1997) evalua- tion of community schools run by Save the Children in Malawi. 12. Odonkor (2000) reports that community members have become more vocal and have even begun making complaints at district education offices in Ghana. 352 Education Decentralization in Africa: A Review of Recent Policy and Practice experience is too recent to know its full effects or even to know which decentralization policies and strategies work best. To date the best evidence is consistent with international experience in showing that moving respon- sibilities to schools governed by elected school councils can improve accountability and performance, whereas decentralization to regional or local government provides mixed results highly dependent on (a) the true decentralization of budget and personnel authority, (b) the incentives cre- ated by intergovernmental fiscal transfers and mandates, and (c) the politi- cal power struggle inevitable in decentralization reforms. Unlike many regions of the world, where decentralization policies have been almost exclusively designed and implemented from the top down, much education decentralization in Africa is by and large a grass-roots phe- nomenon. Community schools have taken root in most countries of the region, and governments are increasingly coming to view them as effective and cost-effective options for increasing access and quality. Numerous questions remain unanswered, and numerous issues remain unresolved concerning education decentralization in Africa. One important issue is the need for education ministries to begin actively supporting com- munity schools, not just with finance but also with technical assistance and guidance. A second important issue is how best to implement education decentralization. Most efforts in Africa (for example, Guinea), including those supported by the World Bank, are incremental in nature, first decon- centrating to regional ministry offices, then introducing advisory school councils, and eventually, according to plans, moving decisions to the level of local communities or schools. Many hazards have arisen from the age-old failure to spell out clearly the assignment of responsibilities across different levels of government and schools. Other African efforts (for example, Mali) have produced beautiful plans for the decentralized system of the future, but with little in the way of operational strategies to bring those plans to fruition. Based on the country cases and on the received literature on African edu- cation decentralization, table 10.8 summarizes African experience with respect to several of the lessons learned from international experience.13 In short, Africa does relatively well in terms of informal or formal parental par- ticipation, does reasonably well in terms of the design of financial transfers 13. The grades represent our summary judgments based on our reading of the literature and our discussions with experts. They do not represent any systematic measures as such as those attempted by Ndegwa (2002). Alec Ian Gershberg and Donal R. Winkler 353 Table 10.8. Assessing African Experience with Education Decentralization International African experience lessons learned (graded 1–5, 1 = best) Comments Accountability is critical 5—Weak formal account- Informal accountability for results. ability mechanisms mechanisms work well in community schools. Assignment of functions 5—Role of local govern- Significant divergence and responsibilities ments poorly defined or exists between legal state- must be clear and not overlapping ments of roles and reality. overlapping. Parental participation 2—Parental participation Tradition of community and empowerment are in school councils often schools contributes to essential to good encouraged parental involvement. governance. Well-trained principals 4—Role and capacity Very little evidence of are crucial for well- of principals not well- serious attention to the managed schools. developed issue is available. Design of financial trans- 3—Very mixed Capitation grants to sub- fers determines equity experience—some good, national governments and efficiency. some bad and schools are used increasingly. Ministries of education 4—Few examples of Failure to restructure and must be restructured to restructuring to provide reorient ministries is support the decentraliza- information, technical causing them to fight to tion process. assistance, and other sup- retain their traditional port services role. Source: Authors. to schools and local governments, and does quite poorly in terms of clearly assigning roles and responsibilities to local governments and in providing the mechanisms and information required for accountability. The next steps in education decentralization in Africa would appear to be obvious: build on what already exists, provide financial and other stimuli to increase the size and number of community schools, provide monitoring and technical assistance to help resolve local problems before they become crises, and foster the communication and exchange of successful experi- ences to shorten the feedback loop and improve community schools. 354 Education Decentralization in Africa: A Review of Recent Policy and Practice References The word “processed” denotes informally produced works that may not be available through libraries. Ablo, E., and R. Reinikka. 1998. “Do Budgets Really Matter?” Working Paper no. 1926. World Bank, Washington, D.C. AED (Academy for Educational Development). 1996. Decentralizing Education: The BESO/Tigray Case Study, A Summary. SARA Project. Washington, D.C. Azfar, Omar, Satu Kähkönen, Jeffrey Livingston, Patrick Meagher, and Diana Rutherford. 2000. Making Decentralization Work: An Empirical Investigation of Gov- ernance and Public Services in Uganda. College Park: University of Maryland, Cen- ter for Institutional Reform and the Informal Sector. DAC (Development Assistance Committee). 1997. Evaluation of Programs Promoting Participatory Development and Good Governance. Synthesis report. Paris: Organisa- tion for Economic Co-operation and Development. De Muro, Pasquale, Luca Salvatici, and Piero Conforti. 1998. An Overview of Decen- tralization in Sub-Saharan Africa. Washington, D.C.: World Bank. Dowd, Amy Jo. 2001. “The Impact of Community Support for Education on Stan- dard 2 Mathematics Learning in Malawi.” Paper presented at the 2001 Annual Conference of the Comparative and International Education Society, Washing- ton, D.C. Gershberg, Alec Ian. 1999. “Decentralization, Citizen Participation, and the Role of the State: The Autonomous Schools Program in Nicaragua.” Latin American Per- spectives 26(4): 8–38. Girishankar, Navin, Abebaw Alemayehu, and Yusuf Ahmad. 2001. “Handling Hier- archy in Decentralized Settings: Governance Underpinnings of School Perfor- mance.” In Tikur Inchini, ed., West Shewa Zone, Oromia Region. Africa Region Working Paper no. 21. World Bank, Washington, D.C. Government of Nigeria. 2001. Education for All (EFA) in Nigeria. Lagos: Federal Republic of Nigeria. Hinchliffe, Keith. 2002. Nigeria: Public Expenditures on Education: Issues, Estimates, and Some Implications. Washington, D.C.: World Bank. Hyde, Karin, Esme Kadzamira, Juliet Sichinga, Mike Chibwana, and Ronald Ridker. 1997. “Village Based Schools in Mangochi, Malawi: An Evaluation.” Working Paper Series: on the Determinants of Educational Achievement and Attainment in Africa. Institute for Policy Reform, Washington, D.C.. Lang, Grace. 2000. School-Based Budgeting: Q&A for the Web/Knowledge Nugget. World Bank: Washington, D.C. http://www1.worldbank.org/education/global educationreform/pdf/SBbudgetingQ&A.pdf. McLean, Keith. 1997. Characterizing Decentralization in Tanzania. World Bank, Washington, D.C. Processed. Alec Ian Gershberg and Donal R. Winkler 355 Miller-Grandvaux, Yolande, and Karla Yoder. 2002. A Literature Review of Community Schools in Africa. SARA Project, Office of Sustainable Development. Washington, D.C.: U.S. Agency for International Development. Minister of Education (South Africa). 2002. Speech to the Gauteng Department of Education Conference on Education Decentralization, June 11. Momoniat, Ismail. 2001. Fiscal Decent in South Africa: A Practitioner’s Perspective. Intergovernmental Fiscal Relations and Local Financial Management Program. Washington, D.C.: World Bank. Moulton, Jeanne, K. Mundy, M. Welmond, and J. Williams. 2001. Paradigm Lost? The Implementation of Basic Education Reforms in Sub-Saharan Africa. Sustainable Development Publication Series, Office of Sustainable Development Bureau of Africa Technical Paper no. 109. Washington, D.C.: U.S. Agency for International Development. Ndegwa, Stephen. 2002. “Decentralization in Africa: A Stocktaking Survey.” Africa Region Working Paper Series no. 40. World Bank, Washington, D.C. http://www. worldbank.org/afr/wps/wp40.pdf. Odonkor, Martina. 2000. Final Evaluation of Care International’s School and Community Oriented Education Project. Frontier Analysis Consulting Associates. OECD (Organisation for Economic Co-operation and Development). N.d. Education at a Glance. Paris. http://www1.oecd.org/els/education/ei/eag/. Reinikka, R. 2000. Recovery in Service Delivery: Evidence from Schools and Clinics. Wash- ington, D.C.: World Bank. Rondinelli, Dennis A. 1981. “Government Decentralization in Comparative Theory and Practice in Developing Countries.” International Review of Administrative Sci- ence 47(2):133–47. Rondinelli, Dennis A., and John R. Nellis. 1986. “Assessing Decentralization Policies in Developing Countries: The Case for Cautious Optimism.” Development Policy Review 4(1): 3–23. Rondinelli, Dennis A., James S. McCullough, and Ronald W. Johnson. 1989. “Analyz- ing Decentralization Policies in Developing Countries: A Political-Economy Framework.” Development and Change 20(1): 57–87. Shaeffer, Sheldon. 1994. Participation for Educational Change: A Synthesis of Experience. Paris: United Nations Educational, Scientific, and Cultural Organization, Inter- national Institute for Educational Planning. Taylor, Nick. 2001. “Outcomes, Effort, and Values in Schooling.” Presentation to the conference on New Millennium Business, May 15, Wits University Business School, South Africa. _____. 2002. “Accountability and Support: Improving Public Schooling in South Africa.” Paper presented to the National Consultation on School Development, January 29, Department of Education, South Africa. 356 Education Decentralization in Africa: A Review of Recent Policy and Practice Therkildsen, Ole. 1998. “Local Government and Households in Primary Education in Tanzania: Some Lessons for Reform.” Working Paper no. 98.6. Centre for Development Research, Copenhagen. http://www.cdr.dk/working_papers/ wp-98-6.htm. USAID (U.S. Agency for International Development). 1997. Education Decentraliza- tion in Africa. Washington, D.C.: Africa Bureau. UNICEF (United Nations Children’s Fund). 2001. “Educating Children Out of the System: The Community Schools Movement in Zambia.” New York: Columbia University, School of International and Public Affairs for UNICEF’s Zambia office. Winkler, Donald. 1989. Decentralization in Education: An Economic Perspective. Popula- tion and Human Resources Department Paper. Washington, D.C.: World Bank. Workineh, Tilahun, Tirussew Teferra, Ayalew Shibeshi, and Malcolm Mercer. 1999. Studies of Education in Ethiopia: An Inventory and Overview of Education Sector Stud- ies in Ethiopia 1994–1997. Paris: Association for the Development of Education in Africa and United Nations Educational, Cultural, and Scientific Organization, Working Group on Education Sector Analysis. http://www.adeanet.org/ wgesa/en/doc/ethiopia/chapter_5.htm. World Bank. n.d. Uganda: Tracking Public Expenditures (PETS). Washington, D.C.. http://www.worldbank.org/wbi/urban/docs/KampalaCaseDecentralisation% 20Mwesigye.pdf . 11 Reflections Dele Olowu, Joel D. Barkan, and Njuguna Ng’ethe Nine of the chapters in this book were presented as papers and discussed at a workshop sponsored by the World Bank Institute, in Johannesburg, South Africa, June 10–11, 2003. Chapter 10, “Education Decentralization in Africa” by Alec Ian Gershberg and Donald R. Winkler, was not presented at the workshop. In addition, time limitations prevented a PowerPoint presenta- tion on experience with rapid decentralization in Ethiopia, South Africa, and Uganda from being converted into a chapter for this book. Aside from the author-presenters, the approximately 60 participants at the workshop consisted of senior government officials and practitioners based in Africa (largely from anglophone countries) plus eminent academic analysts of the African scene. At the conclusion of the workshop, three members of the last group were invited to reflect on the proceedings. This chapter presents summary versions of their contributions. Professor Dele Olowu I begin these reflections by thanking the organizers of this meeting on behalf of other Africans, both participants and nonparticipants, for the excellent arrangements made to have a real dialog in a genuine partnership mode with a focus on lessons learned. This type of interaction on the policy processes related to African public sector reforms was often lacking in the past. Because the menu of discussions was so large, touching on all that I have heard, which emerges as important, is not practical. Rather, I will highlight the significant messages that have been shared in this meeting in three areas while also raising additional issues that might require further attention in 357 358 Reflections each area. These areas are (a) building and sustaining meritocracy in Africa’s poor economies, (b) decentralization, and (c) accountability and anticorruption efforts. Building and Sustaining Meritocracy On the first topic, building and sustaining meritocracy, I begin by noting that merit systems require a fairly long time to build. At least one century was required in many western European countries and two centuries in some of them, for example, in the United States the law on civil service examinations was passed in 1883, but not until 1993 did this law capture 90 percent of all federal public servants (for details of the struggle for an Amer- ican merit bureaucracy see Mosher 1968, chapter 3; for France see Rouban 1998; for the United Kingdom see Young 1961). To effectively restore the initiative on efforts to build meritocracy, begun shortly before independence in many African countries, this workshop has suggested that the old idea of having servicewide applications of meritoc- racy is clearly unsustainable. First, many states cannot afford it, and second, many people in the public service already should not have been there in the first instance. Retrenching this redundant work force (mostly in the lower cadres) is a very difficult and costly proposition. Some of the most effective strategies include—selected accelerated salary enhancement schemes, conversion into an agency structure, devolu- tion, and privatization. Of these several options, I have found the first par- ticularly fascinating because it forces countries to identify their policy priorities, and these priority sectors then drive the reforms. A number of lessons were also learned on how management development institutions are becoming effective postexperience training institutions. That some of them (with some assistance from their governments and donors) have transformed themselves from being supply-driven to demand-driven insti- tutions is a welcome development. The lesson from the positive experience of the Ghana Institute of Management and Public Administration under- scores the point that some of these institutions can be weaned off state sub- vention. Conversely, they could become so successful that they cater more to the needs of fee-paying private and nongovernmental organizations than those of public service organizations. Two important missing elements in this discourse on civil service reforms relate to the politics and economics of reforms. On the political front, although analysts generally recognize that the politics of public sector reform is important and must be well managed, the role of professional associations as significant major players in the civic society who might help Dele Olowu, Joel D. Barkan, and Njuguna Ng’ethe 359 to promote the cause of reform has not been highlighted. Yet, such associa- tions played critical roles in the experience of many western countries. On economics, analysts seem to have paid much more attention to the expendi- ture side of the equation than the revenue side. Yet reforms are not sustain- able in the medium to long term without domestically generated revenues. Therefore much more attention needs to be given to raising domestic rev- enue sources in the medium to long term, even though in the short term, reliance on grants and loans from bilateral and multilateral organizations continues. Some analysts have argued that Africans are already overtaxed, but many private sector organizations operating in these countries do not pay any or sufficient taxes, yet they are the organizations that require the services provided by the public sector. Decentralization With respect to decentralization, four main messages have been conveyed. First, the need was canvassed for refining analysis on process, actors, and outcomes before proceeding with policy intervention. Second, conference participants made a strong plea for closer links between public sector and local government reforms, especially because country choices with respect to decentralization are dependent on the nature of politics at the center. Third, discussions also called attention to the potential contribution of a variety of organizational possibilities at the local level (for example, field administrations of the central government, community-based organiza- tions, project enclaves such as the Malawi Social Action Fund, and single- purpose authorities), rather than being fixated on one organizational mechanism: multipurpose local governments. To the extent that African countries have both weak field administrations and weak local govern- ments, this contribution is welcome. It will also enable countries to be less ambiguous on what they refer to as decentralization. Finally, the point was made that capacity building under rapid decentralization is an intergovern- mental design challenge. Three themes related to decentralization need more emphasis than they received. First is the importance of building on available social capital—in the form of local organizations and resource bases already developed by the people themselves. Reformers must seek to integrate these informal organi- zational forms with the more formal structures, as ways of enhancing local capacity for developmental action. Decentralization programs that leave out these steps have tended to be less than productive in terms of empower- ment and improved service delivery. Related to this issue is a second issue: the need to improve opportunities for local taxation. This is particularly 360 Reflections important in urban areas, where substantial opportunities exist for urban property taxation. A third issue is the need to give time and space to discus- sions on local level capacity building. For decentralization to be effective, capacity building is needed in three critical sites—the central government, local government, and communities—and we need to focus on all of them, especially in sustaining monitoring and accountability arrangements. Accountability and Anticorruption Efforts On accountability, we now know what works and what does not work in fighting corruption. More laws, anticorruption agencies, and cultural or ide- ological explanations do not contribute much to our understanding or abil- ity to confront corrupt practices. What seems to work include effective legislatures, meritocratic public services, and independent judicial organs— in other words all of the architecture of good governance discussed at this meeting. Research on legislatures (presented at the meeting) has high- lighted the important factors that can make this institution effective in over- seeing the executive. More country cases will help us to test some of the hypotheses suggested by the present rather limited studies. More informa- tion on party financing legislation would also be helpful. Conclusions These thoughts lead to three cross-cutting conclusions. First, the diverse discussions are closely interconnected. Meritocracy, stronger parliaments, judiciaries, and local governments improve public accountability and respon- siveness. Second, we still need more information and hence more policy- related research and collaboration on these subjects between researchers within and outside the World Bank, within and outside Africa, and with African governments. Finally, sharing experiences in a spirit of partnership, as has taken place in this workshop, is important for all participants—country practitioners, policymakers, donors, and academics. Professor Joel D. Barkan The implications of interdependence among political institutions for efforts to strengthen governance and public management has emerged as a major theme at this workshop. We are unlikely to succeed by focusing only on technical fixes with respect to the bureaucracy and bureaucratic actors. We must also consider how political institutions function, particularly the top political leadership within cabinet (both formally and informally) and the Dele Olowu, Joel D. Barkan, and Njuguna Ng’ethe 361 legislature (the home of second echelon political actors) and their relation- ships to the bureaucracy. The relationships between these three institutions must be robust, but the extent and quality of these relationships, and espe- cially the extent to which these relationships have been institutionalized, varies greatly—both within countries and between. These three sets of institutions (cabinet, bureaucracy, and legislature) and the relationships between them exist within a broader societal context over which technical reformers (that is, reform-oriented bureaucrats) and donors have little control. Moreover, the broader societal context is often an impediment to reform. Societal pressures include the following. • The nature of the political process. In most African countries, the politi- cal process is reflective of African society (typically agrarian and plural), and so emphasizes representation of geographically specific communities and the distribution of state resources, including social service delivery, across these communities. • The persistence of neopatrimonial, that is, clientelist, politics in the context of liberalized political systems. • Weak political parties and weak and distorted civil society, that is, civil society that is located largely in the urban areas. • Cultural impediments that inhibit consultation, dialog, and negotiation— ethnicity or so-called “constructed ethnicity” the basis of political mobi- lization (of which we have heard no mention in this workshop). These factors often undermine attempts to strengthen both the formal institutions of the state (both political and bureaucratic) and the relation- ships between them. Put differently, they make attempts at reform difficult. We need to remember that although under authoritarian rule little or no reform took place, political liberalization and democratization make gov- ernment more rather than less susceptible to societal pressures of the types just described. Indeed, these pressures are the same ones that weakened the African state in the decades immediately following independence. Public demands outstripped state capacity to manage and channel politics (see the basic argument in Huntington 1968). If these observations are valid, then they invite several hard questions, both technical and political, that must be answered with candor if we are to advance the enterprise of public sector reform: • Is it realistic to expect that key political and bureaucratic institutions, and the relationships between them, can be strengthened? If so, how, and under what conditions? In all countries (even though conditions vary greatly)? Or in different countries in different ways—one size 362 Reflections decidedly does not fit all? Can, for example, public sector reform be pursued equally effectively in the context of uncompetitive pluralism, competitive pluralism, or uncompetitive monopolistic systems? If not, what lesson is implied for donor agencies including the World Bank? • If political will and commitment on the part of the top leadership, both political and bureaucratic, is the key to success, how might that leadership be replicated and sustained? How might one create such leadership where it does not exist or is in short supply? Is this sim- ply beyond the capacity or appropriate role of donor agencies? And if it is indeed beyond their capacity, what does it suggest for their operations—continued repetition of past approaches, on the grounds that this capacity is only what their mandate permits? Greater selec- tivity in terms of which partners one assists? • If best practice is not replicable everywhere, how do we approach the less successful reformers? By abandoning them in the short-run on the grounds that they will eventually realize that they cannot keep doing business as usual? Or, alternatively, by engaging them even though reform never really becomes self-sustaining? Professor Njuguna Ng’ethe The theme of this workshop—governance and economic development—has been timely for two main reasons: one having to do with the evolution of development thinking in general and one having to do with recent events and thoughts on the subject of development in Africa. For quite some time now, development thinking has taken for granted that governance and economic development are somehow linked. The proposition is a very general one, and some consensus seems to have emerged in the workshop around the suggestion that the best way to handle this broad issue is, perhaps ironically, to be quite specific on the elements of governance and economic development that one is seeking to conjoin. Fur- thermore, one must be conscious of both place and time as complicating fac- tors of the general proposition. A further refreshing aspect of the papers presented and discussed at the conference is that everyone made a con- scious attempt to bring different disciplines together not simply at the level of verbal discourse, which has been the norm, but also at the more meaning- ful level of conducting research jointly, thus allowing different disciplines to inform each other. Although our understanding is still a long way from where we want it to be, this kind of multidisciplinary research is likely to lead the way toward our desired goal. Dele Olowu, Joel D. Barkan, and Njuguna Ng’ethe 363 Within Africa, the big and recent shift in development thinking has been the acceptance, by the politicians in particular, of the importance of good governance in fostering economic development. This new and refreshing “ideology” has been given the official seal of approval by the New Partner- ship for African Development, through the members’ commitment to good governance and to collective self-monitoring to make sure that practice is commensurate with the publicly declared intentions. The workshop was able to start generating key areas that could serve as starting points in moni- toring governance—either by the New Partnership for African Development or by other interested parties, such as academics and development agencies. The importance of this effort should not be underestimated for few issues are as controversial as choosing the governance areas to spotlight, let alone the specific variables to address. At the end of the conference, one got the feeling that a consensus was beginning to emerge on this highly emotional issue, perhaps not on the specifics but certainly on the general trend of things. The issue is no longer the necessity of reform. Rather, as most partici- pants pointed out, it is the still troublesome one of pace and sequencing. Wrong pace and wrong sequencing could easily result in an array of unwanted and disruptive consequences, including severe budgetary con- straints, social dislocation, and loss of regime legitimacy, all likely to have a negative impact on economic development. To minimize the probability of negative consequences, public sector reform must indeed be public—in the new sense that it must be part of a much wider agenda whose objective is to widen public participatory space. As part of a wider democratization agenda, public sector reform is more likely to incorporate the views of all those with some interest in the subject, including public servants, trade unions, and politicians. That way, the reform process is also likely to become more sustainable. Fortunately, a new sense of partnership is beginning to emerge. Now not even the most orthodox development agencies or the most authoritarian regimes seriously believe that reform can be decreed. If reform is to take place at all, it must be a partnership between development agencies, rele- vant civil society, and the state. Each one of the partners has the potential to derail, if not to sabotage, the reform process. In theory, therefore, a de facto consociational arrangement should exist in which each partner enjoys a potential veto over the reform process. In practice, the new partnership is evolving on the basis of trust, even though civil society is slow, if not reluc- tant in some instances, to trust its two partners. Going forward, four major challenges can serve as beacons. A first chal- lenge is to take into account Africa’s past experience with reforms. This 364 Reflections consideration is especially important with regard to civil service reform. Often forgotten is the historical fact that the first generation of civil service reform—Africanization—took place soon after independence. This first generation of reform was often in the hands of young public servants, most of them barely over 30 years old and having little previous experience. In a remarkable number of cases, the reforms worked well. What can the sec- ond generation of reforms learn from the first? The starting point is to admit that the past may offer something to be learned. A second challenge is the need to be innovative in the reform process. One is quite often struck by the formulaic nature of the reforms, leading one to wonder: Does the reform business have any room for innovation? In the- ory, room for innovation should be ample, in view of the fact that innova- tion is normally the result of creative tension. Somehow, however, even though the process of public sector reform is, self-evidently, replete with tension, allowing room for innovation does not seem to have come with cre- ativity. The challenge facing reformers, therefore, is to explore the windows of opportunity in which tension and room for creativity might coexist, thus paving the way for innovation. Such windows must surely exist! A third challenge where an innovative approach might be required is how to incorporate the local development elites in the reform process and thus harness their energies for economic development. A new local develop- ment elite is emerging as a result of the various forms of decentralization and, perhaps even more important, as a result of highly skilled and articulate public sector employees retiring into their localities. This new elite is begin- ning to change the configuration of both local development and politics. The challenge lies in how to treat this group as part of the modern pacesetters rather than as part of the so-called traditional and backward rural society. A fourth challenge is that of harnessing political will as a partner in the reform process. Political will is the result of individual leadership character- istics interacting with institutional characteristics to produce a collective outcome, in this case reform. With or without external stimuli, reform has, historically, never been undisruptive. For this reason, it has always required some will, especially on the part of those manning the state, and exists in the context of the fact that some of them are likely to become the biggest losers from the reforms. How to gather this will, especially when one stands to lose from the reforms, requires some thinking and strategizing. A com- monly suggested way out of the problem is to follow the path of minimum loss for each of the would-be losers. Finally, a common observation by many African policymakers is that too many little things are going on all over the public sector and they cannot Dele Olowu, Joel D. Barkan, and Njuguna Ng’ethe 365 keep track of what is going on. This confusion points to the need to provide a framework for the different components of public sector reform—easier said than done, of course. One way of beginning to address the challenge is to look into history, with a view to persuading all participants that the reform process should be viewed as an internal continuation of the unfin- ished business of state construction and not as an external control imposi- tion. We can hope that all have an interest in internal state construction. References Huntington, Samuel. 1968. Political Order in Changing Societies. New Haven, CT: Yale University Press. Mosher, F. C. 1968. Democracy and the Public Service. New York: Oxford University Press. Rouban, Luc. 1998. The French Civil Service. Paris: La Documentation Français. Young, M. 1961. The Rise of Meritocracy. Baltimore, MD: Penguin Books. Index (Page numbers in italics indicate material in tables or figures.) Accountability. See also Expenditure Anti-Corruption Bureau (ACB) (Malawi), accountability 271–272 in anticorruption measures, 360 Anti-Corruption Commission (Sierra Leone), downward, 22–25 270–271 in education decentralization, 329, 340 focus on, 11–12 Anti-Corruption Commission (Zambia), horizontal, 18–22, 211–254 (See also 276–277 Legislatures; Parliaments) Anticorruption agencies. See also Corruption improvements in, 26 characteristics of successful, 266–268 outward, 18 as complement to state institutions, in public service reform, 77–79 268–270 Adjustment conclusions about, 279 economic performance and, 37–40 development of, 265–266 economic policy and, 6–7 function of, 261, 262 as phony populism, 270–273 Administrative capacity building, 12–17 role of, 268 Africa. See also specific countries as short-term substitute, 273–279 background of economic development Anticorruption measures. See also Corruption strategies in, 1–2 accountability for, 360 building state capacity in, 11–29 (See also conclusions about, 279–281 State capacity building) local or citizen-based, 259–260 governance and economic performance national, 259 in, 2–11, 29–42 (See also Economic populist, 260 development; Governance) process interventions as, 258–264 legislatures in 1990s in, 214–219 structure-process interactions for, 260–264 neopatrimonial downward spiral of, 5–6 Audit Service Law of 1999 (Ghana), 201 African Parliamentary Network Against Corruption, 265 Australia, 177, 266 Africanization, 44, 364 Banda, Hastings, 306 Agricultural value added, 39 Benin Amin, Idi, 295, 303 administrative reform and, 14 Anti-Corruption Act of 2000 (Sierra Leone), 270 community schools in, 29 367 368 Index legislature in, 19, 213–215, 221, 224–228, parliament oversight and, 203–204 231, 233, 238–242, 244, 246–249, 251, prerequisites for, 198–200 252, 254 supreme audit institutions and, 198, pay reform in, 111–113, 115–117, 122, 200–202 128–131, 133–137, 139–143, 145, 146 Buganda, 295 political organization in, 136–137 political system in, 135, 215–216 Bureaucracy regime ideology in, 139 as agents of political principals, 25–26 regime legitimacy in, 137 governance diamond and, 3, 4 trade unions in, 140, 141 neopatrimonialism and, 5 societal pressures on, 361 Bolivia, 23n Burkina Faso Botswana administrative reform and, 14 administrative reform in, 14 auditing in, 201 anticorruption measures in, 20, community schools in, 29, 349 268–270, 272 expenditure accountability in, 203–204 pay reform in, 114, 116, 117, 122, 123, pay reform in, 111–113, 115–117, 122, 128–138, 140, 141, 143–147 128–141, 143–147 political organization in, 136–137 political organization in, 136–137 political system in, 135 political system in, 135 regime ideology in, 138 public institutional development project regime legitimacy in, 137 in, 101–102, 106 trade unions in, 140, 141 regime ideology in, 138–139 Botswana Democratic Party, 131, 136 regime legitimacy in, 137 trade unions in, 140, 141 Budget Act of 2001 (Uganda), 164 Budget execution Cabinets explanation of, 16, 192–194 budget process role of, 150–153, financial management information 174–177 systems and, 194–195 societal pressures on, 361 internal management control and, 195–196 in South Africa, 156–159 lessons from experience with, 196–198 in Uganda, 163, 166–168 Budget formulation in Zambia, 160, 162, 163 in African countries, 16, 187–190 Cameroon, 201 characteristics of effective, 185–187 lessons from experience with, 190–192 Canada, 176, 177 Budget process Cape Verde, 102, 106–107 conditions for, 153 Capital Development Fund (United Nations), legislature and, 249–252 310 linking PRSPs to, 149, 150, 173–174 in Mozambique, 168–171 Capital Fund for Local Collectivities in Organization for Economic Co- (Senegal), 291–292 operation and Development Chad, 208 countries, 174–177 overview of, 15–16, 152 Chama Cha Mapinduzi (Tanzania), 50, 138 role of cabinet in, 150–153, 174–177 Chiluba, Frederick, 50, 51, 130, 136, 140, 142, in South Africa, 154, 156–160 276–278 in Uganda, 163–168 variations in, 172, 173 Civil Service Department (CSD) (Tanzania), in Zambia, 160–163 55, 70, 71, 77, 78, 83 Budget reporting Civil Service Performance Improvement lessons from experience with, 204–206 Program (Ghana), 70 overview of, 16–17, 21 Civil United Front (Tanzania), 138 Index 369 Community activism sustainability of, 287–289, 302–305 decentralization and, 302–304, 313–315 tax issues and, 359–360 in school councils, 348 in Uganda, 22–23, 284, 289–296, 298–300, 303, 304, 309, 316–319 Community schools authority in, 348 Decentralization and Local Governance background of, 348–349 Support Program (U.S. Agency for decentralization of, 349–350 International Development), 301n effectiveness of, 351, 352 Delegation financing of, 349 de facto, 326 overview of, 24–25, 29 implicit, 326 regulation of, 350–351 to schools and school councils, 347–351 Compaore, Blaise, 130 Democratization Constituency Road Fund (Kenya), 232 during early 1990s, 6, 214, 215, 219, 257 explanation of, 214n, 253 Control systems, 72–79 Third Wave of, 224 Corrupt Practices Act no. 18 of 1995 Devolution, education, 325–327, 344 (Malawi), 271–272 DevPol II (Malawi), 310 Corrupt Practices Investigation Bureau (CPIB) (Singapore), 266 Directorate for Corruption and Economic Crimes (DCEC) (Botswana), 268–270 Corruption. See also Anticorruption agencies; Anticorruption measures Directorate of Ethics and Integrity detection of, 261 (Uganda), 273 efforts to combat, 9–10, 20–21, 258 Directorate of Public Prosecution focus on, 257–258 (Uganda), 273 prevention of, 261 District Assemblies Common Fund Côte d’Ivoire, 285 (Ghana), 232 District Focus for Rural Development De facto delegation, 326 (Malawi), 310, 311 Decentralization Downward accountability, 22–25 administrative, 292–294 big bang vs. incremental, 286 design of, 318–319 Economic Commission of West Africa detailing, 287, 288, 298–302 States, 116 education, 23–24, 323–353 (See also Economic Crimes Law (Kenya), 265 Education decentralization) empowerment and, 319–320 Economic development enabling, 287, 288, 294–298, 310–313, adjustment and, 6–7 317–318 links between governance and, 2–11, fiscal, 290–292, 308–309 29–42, 362–363 local level capacity building and, 360 overview of, 1–2 in Malawi, 22–23, 284, 285, 305–316, public service reform and, 52–53 318, 319 Economic policy reform, 1 overview of, 22–23, 283–285 participatory communities and, 313–316 Economy, 3, 4 politics and, 285–290 Education decentralization. See also progress in, 289–294 Decentralization propositions for nurturing, 317 assessment of, 324–328 realities of, 316–317 conclusions about, 351–353 in Senegal, 22–23, 284, 285, 289–305, 309, delegation to schools and school 316, 318, 319 councils, 347–351 social capital and, 359 370 Index in Ethiopia, 333–336 civil service reform in, 46 international experience with, 328–331 control and accountability in, 73, 75, 77, to local governments, 341–347 79–82 in Nigeria, 333–334, 336–340 decentralization in, 285 overview of, 23–24, 323–324 economic growth in, 37 to regional governments, 331–333 legislature in, 19, 204, 213–215, 224–228, in South Africa, 333–334, 340–341 230–234, 238–242, 245, 248–252, 254 pay reform in, 57–59, 111–118, 122, Education deconcentration, 325 128–143, 145, 146, 148 Education delegation, 326 performance management in, 69, 70 political environment in, 51, 53 Education devolution political organization in, 136–137 explanation of, 325–326 political system in, 135, 215–217 in Uganda, 344 public financial management reform Education for All initiative, 323 program in, 104–105 public sector management reform Empowerment, 319–320 project in, 104 Ethics and Code of Conduct Law (Kenya), 265 public service reforms in, 43, 48, 49, 51, 53–55, 57–59, 62–64, 66, 69, 70, 73, 75, Ethiopia 77, 79–85 auditing in, 201 public service training in, 80–82 decentralization in, 285 regime ideology in, 138 educational decentralization in, 24, 285, regime legitimacy in, 137 333–336 state restructuring in, 62–64 Executive agencies (EAs) subvented agencies in, 67 explanation of, 47, 65 trade unions in, 140, 141 risks of, 65–66 World Bank public sector reform project in Tanzania, 66–68 in, 208 Expenditure accountability. See also Ghana Education Service, 67 Accountability Ghana Institute of Public Administration, budget execution and, 192–198 81–82 budget formulation and, 185–192 budget reporting and, 198–206 Goldenberg’s scandal, 264 conclusions about, 206–208 Governance. See also Public sector HIPC expenditure tracking process and, governance (PSG) 180–185 indicators of progress in, 96–100 overview of, 16–17, 179 interdependence among political institutions to strengthen, 360–362 Financial management information systems links between economic performance (FMIS), 194–195, 197 and, 2–11, 29–42, 362–363 sources of change in, 40–42 Finland, 175–177 trends and challenges, 7–11 Formal political institutions, 3, 4 Governance Baseline Survey, 78 France, 177 Governance diamond Fulton Commission, 44 cause-effect relationships in, 30–37 explanation of, 2–4 interactions in, 4–5 Gambia, 46, 201 Governance polarized countries General Confederation of Labor (Burkina average annual real growth rates in GDP Faso), 140 for, 7 Ghana corruption control in, 9–10 administrative reform in, 12–14 economic performance in, 37–38 auditing in, 201, 202 list of, 6 Index 371 relationships between private firms and expenditure accountability in, 201, government in, 8–9 202, 204 institutional development and civil Guinea service reform project in, 102, 107 auditing in, 202 legislature in, 19, 29, 213–215, 222, capacity building for service delivery 224–228, 230–233, 239–242, 245, project in, 105 248–252, 254, 265 civil service reform in, 46 political system in, 215, 217–218 economic policy in, 37 public sector management technical assistance project in, 105 Harambee development project, 230, 232 public service reforms in, 48 Heavily Indebted Poor Countries (HIPC) Kenya African National Union (KANU), 217, Initiative, 149, 179 218, 265 budget execution and, 193, 194 Kenya Anti-Corruption Authority budget formulation and, 185–187, 190 (KACA), 265 budget reporting and, 198–200 expenditure accountability and, Kenya Anti-Corruption Commission, 265 180–185, 201, 206–208 Kenya National Assembly, 214, 217, 218, 222, HIPC. See Heavily Indebted Poor Countries 228, 232, 244, 247–249 (HIPC) Initiative Kenya National Union of Teachers, 244 Hong Kong, 266–267 Kerekou, Mathieu, 139, 215–216, 247, 251 Horizontal accountability, 18–22 Keynes, M., 175 Kibaki, Mwai, 262, 264 Implementation completion reports (ICRs), 87–88, 94 Kufuor, John, 217 Implicit delegation, 326 Labor unions. See Trade unions Independent Commission Against Corruption (ICAC), 266–267 Lao-tzu, 29 Indonesia, 285 Later adjusters average annual real growth rates in GDP Industrial value added, 39, 40 for, 7 Inspectorate of Government (Uganda), corruption control in, 9–10 273, 274 economic performance in, 37, 38 list of, 6 Institutional checks and balances (CHECK), relationships between private firms and 88, 96, 100 government in, 8–9 Internal management control systems, Law on Public Administration Reform 195–196 (Burkina Faso), 132 International Development Association, 181 Leadership, 53–54 International Monetary Fund Leadership Code of Conduct (Uganda), 273 budgetmaking and, 16, 52, 162, 163 expenditure accountability and, 181, Legislators 184, 208 characteristics of, 235–237 Kenya and, 264 patronage practices and, 223–224 standards and codes report, 181 public expectations of, 222–223 structural adjustment programs of, 46 roles and functions of, 243–249 salary for, 223, 229–331 Kaunda, Kenneth, 50, 51, 142 Legislatures. See also Parliaments Kenya areas of inquiry for study of, 212–214 administrative reform in, 94–95 authority of, 224–229, 252 anticorruption measures in, 264–265, 268 budgetary process in, 249–252 372 Index composition of, 237–243 Mali determinants of behavior in, 219–220 auditing in, 202 evolution of, 252–254 community schools in, 24, 350 formal rules and, 224–229 decentralization in, 341–342 institutional resources available to, Mauritania, 37 229–239 in 1990s, 214–219 Means of administration and coercion overview of, 211–212 (MAC), 125–126 societal pressures on, 361 Media, 267 societal structure and political system and, 220–224 Medium-term expenditure framework structural variables in, 220–234 (MTEF) budget and, 150, 156, 173, 174, 207 Lesotho, legislature in, 221 explanation of, 71–72, 74–75 Liberal Democratic Party (Kenya), 218n in Mozambique, 169, 171 in South Africa, 154, 156–160 Liberalization in Uganda, 163, 166, 168 explanation of, 214n, 253 in Zambia, 160 during 1990s, 214, 215 in Senegal, 219 Medium-Term Public Service Pay Policy of 1999 (Tanzania), 59 Local Administration Act of 1967 (Uganda), 295 Merit principle, 47, 358 Local Authorities Service Commission Meritocracy, 358–359 (Malawi), 307–308 Millennium Development Goals (United Local Government Act (Malawi), 306 Nations), 180, 323 Local Government Act (Uganda), 290, 292, Mills, Atta, 51 299–300, 342, 343 Ministerial Committee on the Budget (MCB) Local government education authorities (South Africa), 156–159 (LGEAs) (Nigeria), 337–339 Ministries, departments, and agencies (MDAs), Local Government Finance Committee 47, 54, 56, 59, 60, 64, 68, 70, 75, 78, 81 (Malawi), 307, 308 Ministry of Private Sector Development Local Government Reform Program (Ghana), 64 (Tanzania), 64, 345 Mkapa, William, 130, 278–279 Moi, Daniel arap, 215, 217, 218, 242, 264 Malawi anticorruption measures in, 20, 268, Monterrey Conference on Financing 271–273 Development, 180 auditing in, 201, 202 Movement for Multi-party Democracy decentralization in, 22–23, 284, 285, (MMD) (Zambia), 50, 51, 135–136, 142 305–316, 318, 319 economic growth in, 37 Mozambique institutional development project in, auditing in, 202 102–103, 107 budget process in, 15, 168–171, 173 parliament in, 203, 204 capacity building-public sector and legal social fund in, 29 institutions development project in, World Bank public sector reform project 103, 107 in, 208 decentralization in, 285 parliament in, 204 Malawi Local Government Association, 308 public service reforms in, 48 Malawi Social Action Fund (MASAF), 23, MTEF. See Medium-term expenditure 306, 312–316, 319, 320 framework (MTEF) Index 373 Muluzi, Bakili, 312 Operational performance, 88 Museveni, Yoweri, 130, 137, 147, 278–279, 295 Operations Evaluations Department (World Bank), 11 Muslims, 297–298, 304, 305 Organization for Economic Co-operation and Mwanawasa, Levy, 136, 277–279 Development (OECD), 99, 172, 174–177 Outward accountability, 18 Namibia, 221 Overseas Development Institute (ODI), 166, National Alliance (Kenya), 218n 167, 171 National Assembly (Benin), 216, 246, 251 National Assembly (Ghana), 233–234, 251 Pakistan, 285 National Assembly (Senegal), 214, 239 Parliamentary Service Commission (Kenya), 232–233, 250 National Assembly (Zambia), 277 Parliaments. See also Legislatures National Democratic Congress (Ghana), effectiveness of, 19–22 51, 217 expenditure accountability and, 203–204 National Institute of Public Administration Parti Démocratique Sénégalais (PDS), 218, (Zambia), 68 219, 236n National Institutional Renewal Program Parti Socialiste (Senegal), 131, 143, 218, 219, (NIRP) (Ghana), 51, 54, 55, 84 236n, 297 National Party (Zambia), 51 Pay. See Public service pay National People’s Party (Ghana), 217, 252 Pay reform. See Public service pay reform National Primary Education Commission Penal Code Act of Uganda, 273 (NPEC), 336, 337 Performance National Rainbow Coalition (Kenya), 218, 265 for capacity building, 100–102 National Resistance Movement (NRM) determinants of, 89–92 (Uganda), 137, 289, 294–296, 298–300, 303, operational, 88 304, 342 in public service reform programs, 68–72 relationship between project design and National School of Administration and the project, 92–96 Judiciary (Senegal), 301 study results of, 101–108 Neopatrimonialism, 5–6 Performance improvement funds (PIFs) Netherlands, 175 function of, 47, 70, 71n in Zambia, 72 New Program for African Development, 180 Peru, 23n New South Wales, Australia, 266, 267 Political commitment, 53 New Zealand, 176 Political environment, 49–52, 83–84 Nigeria auditing in, 201 Political ideology civil service reforms in, 44 origins of, 221 education decentralization in, 24, public service pay reform and, 138–139 333–334, 336–340 Political interests country specific nature of, 26 Oasis Forum, 278 governance diamond and, 3, 4 Obasanjo, Olusegun, 263 Political parties institutionalization of, 136–137, 144 Obote, Milton, 295, 303 strength of, 221 374 Index Political system Public financial management and decentralization and, 285–289 accountability (EXPACC), 88, 92, 93, 95, legislative behavior and, 220–224 96, 100 pay reform and, 124–139 Public Procurement Act (Tanzania), 66 public service pay reform and, 132, 134–136 Public Sector Capacity Building Project (PSCAP) (Zambia), 48, 52, 64, 68, 78, 79, Poverty Alleviation Policy Framework 83, 84 (Malawi), 306, 313–314 Public sector governance (PSG). See also Poverty reduction Governance decentralization and, 310–311 indicators of challenges and progress in, governance and, 96 96–100 Poverty reduction strategy papers (PRSP) operations of, 89–92 application of, 155, 161, 173–174 project design and project performance conditions for, 153 and, 92–96 influence of, 149 study of reform under, 88, 100–101 overview of, 15, 149 Public Sector Management Reform Project Presidents, 228 (PSMRP) (Ghana), 48, 64, 67, 82 Prevention of Corruption Act (Kenya), 265 Public sector reform, 363–364 Prevention of Corruption Act no. 8 of 1970 Public service commissions (PSCs) (Uganda), 273 function of, 47n, 54 power of, 44 Prevention of Corruption Bureau (PCB) (Tanzania), 275–276 Public service pay erosion of, 44, 45 Private sector, 131 results of low, 109 Process interventions trends and structure of, 110–118 anticorruption, 259–264 Public service pay reform linking structural and, 264–265 administrative actors and, 126–130 types of, 258–260 approaches to, 57–62, 145–148 Productivity, 109 civil service institutions and, 127, 131–132 conclusions about, 142–144 PRSPs. See Poverty reduction strategy papers donor community and, 127, 132 (PRSP) institutionalization of political Public administration reform (ADMIN), 88, organization and, 136–137, 143 92, 93, 95, 96, 100 market institutions and, 127, 131 policy objectives of, 109–110 Public expenditure accountability. See political system and stability and, 132, 134 Expenditure accountability political system competitiveness and, Public expenditure management, 15 135–136 politics of, 124–126 “Public Expenditure Management and regime ideology and, 138–139 Financial Accountability Review for regime legitimacy and, 137–138 Zambia” (World Bank), 160–161 representative institutions and, 127, Public Expenditure Review (World Bank), 130–131 169–170 social basis of regime and, 139, 141 tactics and techniques for, 118–124 Public Finance and Accountability Act trade unions and, 140, 142 (Uganda), 164 Public service reform Public Finance Compliance Act (South approaches to, 361–362 Africa), 154 control systems and accountability and, Public Finance Management Act (South 72–79 Africa), 76n early, 44–45 Index 375 economic environment and, 52–53, 359 auditing in, 201, 202 first-generation, 46 public sector management support implementation of, 54–56 project in, 103 leadership and, 53–54 Singapore, 266 lessons learned from, 85 pay and, 44, 45, 57–62, 109–148 Social funds, 23n, 29 performance and, 68–72 Soglo, Nicéphone, 139, 216 politics and, 49–53, 358–359 results of, 82–84 South Africa second-generation, 46–82 budget process in, 15, 16, 154, state restructuring and, 62–68 156–160, 173 training skills and, 79–82 decentralization in, 285 education decentralization in, 24, Public Service Reform Project (Tanzania), 48, 333–334, 340–341 64, 76, 77 legislature in, 221 Public Finance Management Act in, 76n Rawlings, Jerry, 51, 130, 132, 215–217 South African Schools Act of 1996, 341 Rwanda, 201 State capacity building administrative capacity and, 12–17 Sankara, Thomas, 139 downward accountability and, 22–25 horizontal accountability and, 18–22 School-based management, 348 new approach to, 26–29 Schools. See Community schools; Education outward accountability and, 18 decentralization performance studies for, 87–108 (See also Performance) Selected accelerated salary enhancement process of, 1–2, 11–12 (SASE), 56–62 Structural interventions, 264–265 Senegal administrative reform and, 14 Sub-Saharan Africa (SSA). See also specific auditing in, 201 countries bureaucracy in, 300–302 civil service reform in, 44, 46, 68, 76, 82 decentralization in, 22–23, 284, 285, Subvented agencies (SAs), 67 289–291, 293, 294, 296–298, 300–302, 304–305, 309, 316, 318, 319 Supreme audit institutions (SAIs), 198, economic policy in, 37 200–202, 204, 205 fiscal system in, 291 Sustained adjusters legislature in, 19, 213–215, 221, 224–228, average annual real growth rates in GDP 233, 236n, 238–242, 244, 246–250, for, 7 252, 254 corruption control in, 9–10 pay reform in, 111–113, 115–117, 122, economic performance in, 37, 38 128–131, 133–143, 145, 146 list of, 6 political organization in, 136–137 relationships between private firms and political system in, 135, 218–219 government in, 8–9 regime ideology in, 138 regime legitimacy in, 137 Swahili, 240n social conservatism in, 304–305 Sweden, 176 trade unions in, 140, 141 Senghor, Leopold, 218, 297 Tanganyika African National Union, 50 Settlement Law (Senegal), 246 Tanzania Sierra Leone administrative reform in, 12–14 anticorruption measures in, 20, 268, anticorruption measures in, 20, 268, 270–271, 273 275–276 376 Index community schools in, 24 pay reform in, 112–118, 122, 123, control and accountability in, 73, 76–78, 128–137, 139–141, 143–147 80, 81 political organization in, 136–137 decentralization in, 285 political system in, 135 education decentralization in, 345–347 public service reforms in, 48 executive agencies in, 66–68 regime ideology in, 138 financial and legal management regime legitimacy in, 137 upgrading project in, 103–104, 108 trade unions in, 140, 141 pay reform in, 45, 57–61, 112–118, United Democratic Front (Malawi), 312 122–124, 128–138, 140–141, 143–148 performance improvement funds in, 47 United Kingdom performance management in, 69–71 cabinet in, 174, 175, 177 political environment in, 50, 53, 83 Department for International political organization in, 136–137 Development, 339 political system in, 135 EA model in, 64–66, 68 public service reform in, 43, 48–50, 52–55, United National Independence Party 57–64, 66–71, 73, 76–78, 80, 81, 83, 85 (Zambia), 140, 142 public service reform project in, 105–106 public service training in, 80, 81 United Nations Development Programme, regime ideology in, 138 310, 311 regime legitimacy in, 137–138 United Nations Educational, Cultural, and state restructuring in, 62–64 Scientific Organization, 335 trade unions in, 140, 141 United Nations (U.N.) Tanzania Public Service College (TPSC), 81 Capital Development Fund, 310 Ten-Point Program (Uganda), 303 Millennium Development Goals, 180, 323 Third Wave of democratization, 224 Standard International Trade Trade unions Classification, 39 strength and influence of, 140–142 Universal Basic Education Commission in Tanzania, 50, 141, 142 (Nigeria), 337 in Zambia, 51, 140–142 U.S. Agency for International Development, Training, 79–82 233n, 301n, 339 Transparency International, 264 Virtuous cycle, 21 Uganda administrative reform in, 14, 28–29 Wade, Abdoulaye, 218, 219 anticorruption measures in, 20, 268, 273–276 Wolfensohn, James, 257 auditing in, 202 Women, 238–239 budget process in, 15–16, 163–168, 173 bureaucracy in, 298–300 World Bank decentralization in, 22–23, 284, 289–296, budgeting projects and, 16, 52 298–300, 303, 304, 309, 316–319 civil service reform projects and, 11, economic growth in, 37 208, 209 education decentralization in, 342–345 decentralization study and, 283 expenditure accountability in, 203, 204 educational bureaus study and, 335 fiscal system in, 290–292 expenditure accountability operations institutional capacity building project in, and, 29, 179–181, 184, 202, 204 104, 108 governance operations progress legislator salary in, 232 indicators development and, 99 National Resistance Movement in, 137, Malawi Social Action Fund and, 314 289, 294–296, 298–300, 303, 304, 342 Operations Evaluation Department, 88 Index 377 state capacity-building and, 2, 87, 233n, parliament in, 204 234 pay reform in, 57–60, 111–118, 122, structural adjustment programs and, 46 128–130, 133–143, 145, 146 supreme auditing institutions and, 202 performance improvement funds in, 47 performance management in, 69, 72 World Bank Institute political environment in, 50–53, 83 corruption data from, 10 political organization in, 136–137 parliamentary capacity building political system in, 135 program of, 204, 234 public service capacity building project World Development Report 1997: The State in a in, 106 Changing World (World Bank), 7 public service reform in, 43, 48–53, 55, 57–59, 62–64, 68, 69, 72, 73, 75, 76, 78, 80–83, 85 Zambia public service training in, 80, 81 administrative reform in, 12–14 regime ideology in, 138 anticorruption measures in, 20, 268, regime legitimacy in, 137 276–278 state restructuring in, 62–64 auditing in, 202 trade unions in, 51, 140–142 budget process in, 15, 160–163, 173 World Bank public sector reform project community schools in, 349–351 in, 208 control and accountability in, 73, 75, 76, 78, 80, 81 Zambia Congress of Trade Unions, 140, 142 economy in, 37 Zimba, Newsteed, 142 executive agencies in, 68 financial and legal management Zimbabwe, 285 upgrading project in, 104, 108 Zongo, Norbert, 147 legislature in, 222, 238 THE WORLD BANK 1818 H Street, NW Washington, DC 20433 USA Telephone: 202 473-1000 Internet: www.worldbank.org E-mail: feedback@worldbank.org Building the capacity of African states is at the top of the continent’s develop- ment agenda. Reducing poverty and forging productive partnerships with donors and investors depends critically on strong public capacity: M to formulate policies, M to build consensus, M to implement reform, and M to monitor results, learn lessons, and adapt. Learning how to develop country capacity is an ongoing process and a formidable challenge. Efforts by the World Bank and other agencies to help build institu- tional capacity, through for example, civil service reform, have had mixed results. But in recent years, a number of African states have launched new-style programs to strengthen the public sector. Building State Capacity in Africa shares some of the lessons that are emerging from their experiences and identifies some major challenges for the African public sector. How can African states: M be innovative in the reform process, M harness the energies of the local elites, M learn from past reformers, and M harness political will as a partner in the reform process. Building State Capacity in Africa draws on the experiences of public sector reform implementation in more than a dozen African countries to address topics such as the relationship between governance and economic development, public expenditure and accountability, anticorruption reform, decentralization, political structures, and the delivery of public services. ™xHSKIMBy360 02zv":<:!:+:! ISBN 0-8213-6000-0