Report No. 22535-CE Sri Lanka Poverty Assessment June 26, 2002 Poverty Reduction and Economic Management Sector Unit South Asia Region Document of the World Bank CURRENCY EQUIVALENTS Sri Lanka Rupee (SLR) US $1.00=SLR 96.17 (June 25, 2002) FISCAL YEAR (FY) January 1-December 31 ACRONYMS AND ABBREVIATIONS ADB - Asian Development Bank BOI - Board of Investments CARE - Cooperative for Assistance and Relief Everywhere CFS - Consumer Finance Survey CGES - Commissioner General for Essential Services CWE - Cooperative Wholesale Establishment DCS - Department of Census and Statistics EPF - Employment Provident Fund EPZ - Export Processing Zone ETF - Employees Trust Fund FGT - Foster, Green, Thorbecke measures of income poverty GDP - Gross Domestic Product GNP - Gross National Product GTZ - Gesellschaft fuer Technische Zusammenarbeit (German Aid agency) HIES - Household Income and Expenditure Survey IRDP - Integrated Rural Development Program LFSS - Labor Force and Socio-economic Survey LTTE - Liberation Tigers of Tamil Eelam NCRCS - New Comprehensive Rural Credit Scheme NEIAP - North East Irrigated Agriculture Project NGO - Non-Government Organization NHS - Nutrition and Health Survey PA - People's Alliance PPP - Purchasing Power Parity PSPF - Public Sector Provident Fund SANASA - Samupakara Nayadena Samithi (Thrift and Credit Co-operative Societies) SASANA - Samurdhi Development Credit Scheme SAVANA - Samurdhi Enterprise Credit Scheme SBS - Samurdhi Banking Societies SDCP - Surathura Diriya Credit Programme SEEDS - Sarvodaya Economic Enterprises Development Service SLIS - Sri Lanka Integrated Survey TEWA - Termination of Employment and Workmen Act UN - United Nations UNDP - United Nations Development Program UNHCR - United Nations High Commission for Refugees UNP - United National Party WFP - World Food Program Vice President: Ms. Mieko Nishimizu Country Director: Ms. Mariana Todorova Chief Economist & Sector Director: Mr. Sadiq Ahmed Task Manager: Ms. Manuela Ferro SRI LANKA POVERTY ASSESSMENT CONTENTS Preface E xecutive S u m m ary .................................................................................................................................................. i Chapter 1: The Evolution and Nature of Poverty in Sri Lanka........................................................................ Poverty Trends in Sri Lanka: Slow Poverty Reduction and Increasing Regional Divergences ..................... 5 Poverty and Hardship in the Northeast .......................................................................................................... 23 Chapter 2: Anti-Poverty Programs and Safety Nets........................................................................................ 28 The Government Safety Net in the Late 1990s.............................................................................................. 29 Sam urdhi: A n Evaluation ............................................................................................................................... 30 Assistance to Displaced Persons and to Populations Affected by W ar ......................................................... 41 Other Government Programs ....................................................................................................orm.......-.42..... 42 Private, Semi-formal, and Informal Safety Nets............................................................................................ 46 Programs Sponsored by Nongovernmental Organizations and International Organizations......................... 54 Chapter 3: Elements of A Strategy to Accelerate Poverty Reduction............................................................ 55 Creating a Policy Environment that Facilitates Poverty Reduction............................................................... 55 Providing Effective Core Public Services That Reach the Poor....................................................................58 Achieving Peace and Social Harmony........................................................................................................... 61 A n n ex A6...............................................................- ................................................................................................. 62 A nn ex B ......................................................... .....-............ -.............................................................................Ann x 7B4 R eferen ces.................................................... ........................................................................................................R.77 List of Tables Table 1.1 Incidence of Poverty and Per Capita GNP in Sri Lanka and East Asian C om parator C ountries.........................................................................................2 Table 1.2. Per Capita Energy Consumption and Communications, Sri Lanka and C om parator C ountries.........................................................................................4 Table 1.3 Contribution to Poverty by Province, (reference and higher poverty lines) 1985-96 ........................................................................................9 Table 1.4 Poverty Reduction Rates, by Province (percent) 1985-86 to 1995/96 ........................................................................................................9 Table 1.5 Poverty by Province, 1995/96........................................................................... 10 Table 1.6 Contribution to Poverty by Occupation of Household Head, 1985/86 to 1995/96 ....................................................................................... 13 Table 1.7 Gini Coefficient of Consumption Inequality, all Provinces, 1995/96 and 1999/2000 .................................................................................14 Table 1.8 Contribution to Poverty by Ethnicity of Household Head, 1985/86 to 1995/96 ..........................................................................................15 Table 1.9 Gini Coefficient of Consumption Inequality in Sri Lanka, by Ethnicity, '1995-96 and 1999-2000 .................................................................................... 15 Table 1.10 Poverty by Gender of Household Head, 1985/86 to 1995/96 ..........................16 Table 1.11 Contribution to Poverty by Employment of Principal Income Earner, 1985/86 to 1995/96 ........................................................................... 16 Table 1.12 Average Consumption and Access to Basic Infrastructure Services by Consumption Quintile, Sector, and Ethnicity, 1999/2000................................17 Table 1.13 Access to Economic and Social Facilities by Consumption Quintile, Sector, and Ethnicity, 1999/2000 (travel time in minutes)...............................18 Table 1.14 Contribution to Poverty by Education of Household Head, 1995/6................. 19 Table 1.15 Contribution to Poverty by Literacy Rate of Household Head, 1985-96 .........20 Table 1.16 Poverty Incidence, Literacy, and Language......................................................21 Table 1.17 Health Indicators by Expenditure Quintile, 1999/2000....................................22 Table 1.18 Health Indicators by Sector, 1999/2000 ...........................................................22 Table 1.19 Health Indicators by Province 1999/2000 ........................................................ 22 Table 1.20 Households Reported having a Member Killed as a Result of the Civil Conflict, by Expenditure Quintile and Province........................26 Table 1.21 Reasons for Household Move by Area of Residence.......................................27 Table 2.1 Social Expenditures, Sri Lanka and Comparator Countries.............................28 Table 2.2 Social Transfers and Social Expenditure, 1999................................................ 30 Table 2.3 Samurdhi Expenditures, 1999........................................................................... 32 Table 2.4 Sam urdhi Eligibility Criteria ............................................................................ 33 Table 2.5 Sam urdhi Food Stam ps.....................................................................................36 Table 2.6 Food Stamp Distribution by Population Quintile .............................................37 Table 2.7 Model I-Probability of Receiving a Samurdhi Grant (Survey Probit Regression)............................................................................38 Table 2.8 Food Stamps Issued to Districts and Welfare Indicators............. Table 2.9 Distribution of Samurdhi Consumption Grants across Province-Specific Quintiles of Per Capita Expenditure Distribution and Welfare Indicators ....41 Table 2.10 Incidence and Distribution of Public Transfers, 1999-2000.............................44 Table 2.11 Incidence of Private Transfers and Migrant Labor Income..............................47 Table 2.12 Distribution of Private Transfers and Migrant Labor Income by Quintile.......47 Table 2.13 Patterns in Household Saving and Borrowing.................................................50 T able 2.14 U ses of L oan ..................................................................................................... 51 Table 2.15 Distribution of Number of Loans by Source ....................................................53 Table 2.16 Deposits and Advances Selected Semiformal Institutions, at end 1999...........53 Table 3.1 Urban Population, Sri Lanka and Comparative Countries...............................58 List of Boxes, Box 1.1 Measuring Income Poverty in Sri Lanka: Methodology and Data Sources .......6 Box 1.2 Why Do the Poor Lack Assets? .................................................. 18 Box 1.3 Ethnic Diversity and Violence in Sri Lanka.....................................................23 Box 1.4 Social Fragmentation and Conflict...................................................................24 Box 1.5. Ethnic D ivisions and Conflict...........................................................................25 Box 1.6 The Sithamparapuram Welfare Center.............................................................26 B ox 2 .1 W elfare .............................................................................................................3 1 Box 2.2 Politicization of Sam urdhi................................................................................39 Box 2.3 Nutrition Indicators in Sri Lanka................................. Box 2.4 The Role of the Middleman in the Rural Economy..........................................49 List of Figures Figure 1.1 Per Capita Gross National Product (GNP), Atlas Method Sri Lanka and East A sian Com parator Countries ..............................................................................2 Figure 1.2 Incidence of Poverty and Consumption Inequality in Sri Lanka, 1985/86-1995/96 .................................................................................................5 Figure 1.3 Incidence of Poverty in South Asia (latest estimates W ith national poverty lines)............................................................................ 7 Figure 1.4 Per Capita Income (GNP) at Purchasing Power Parity (PPP), 1999..................7 Figure 1.5 Workforce Participation Rates of Children in South Asian Countries............... 8 Figure 1.6 Trends in the Incidence of Poverty in Sri Lanka, by Province, 1985/86-1995/96 ................................................................................................. 8 Figure 1.7 Poverty Maps Incidence of Poverty and Contribution to total poverty, by Province, 1995-96 (percentage) ................................................................ 10 Figure 1.8 Poverty Maps Incidence of Poverty and Contribution to total poverty, by Province, 1995-96 (percentage)............................................................... 10 Figures 1.9a Trends in the Incidence of Poverty, by Sector, 1985/86 to 1995/96 ................11 1.9b & 1.9c Figures 1.10a Sector Contribution to Overall Poverty, 1985/86 and 1995/96........................11 & 1.10b Figure 1.11 Growth in Agriculture, Manufacturing and Services, 1985-96........................12 Figure 1.12 Growth in Agriculture, Manufacturing and Services, 1985-96........................12 Figure 1.13 Proportion of Students Mastering Basic Language and N um erical Skills ............................................................................................21 PREFACE Over the past three years, the Government of Sri Lanka has made an extraordinary effort to deepen its understanding of the nature and root causes of poverty in Sri Lanka, with a view to reassessing and reformulating its policy framework for reducing poverty. This report is a contribution to the development of this new poverty reduction strategy for Sri Lanka. It also informs the World Bank's upcoming Country Assistance Strategy for Sri Lanka. The report, which builds on the 1995 World Bank Poverty Assessment, is the product of close collaboration between the Government of Sri Lanka, the World Bank, several development partners, and Sri Lankan academics and researchers. The concept note was shared and discussed with the Government, and some of the early analytical work that underpins the analysis was carried out in collaboration with and was utilized by the Government in its 2000 Framework for Poverty Reduction paper. In addition, findings of this report were discussed broadly with the government, the research community, and NGOs at three workshops held in Sri Lanka in 2000 and 2001. This report was initiated in March 2000 and was prepared by a multi-sectoral World Bank team led by Manuela Ferro, Senior Economist, under the guidance of Eric Bell and Tercan Baysan, Lead Economists. The report draws on contributions from: Elena Glinskaya (Samurdhi), Eric Bell, Rapti Goonesekere and Salvatore Schiavo Campo (public sector management), Sryiani Hulugalle and Iffath Sharif (consultant, microfinance), Valerie Kozel (design of the Sri Lanka Integrated Survey, SLIS), Uwe Deichman (poverty and service delivery maps), Daya Samarasinghe (health), Jean-Claude Balcet, Terrence Abeysekera, and Geeta Sethi (agricultural policy), Harsha Aturupane (education and design of SLIS), Naresha Duraiswami (private sector development), Princess Ventura (research assistance and conflict issues), Barbara Parker (consultant, qualitative analysis of poverty), Beatriz Godoy and Xun Wu (consultants, SLIS), Sasanka Perera (formerly with the World Bank, currently U. Colombo, conflict), and Shunalini Sarkar (final editing and formatting). The report benefited greatly from collaborative work with Sri Lankan institutions and researchers, including: Dileni Gunewardena (University of Peradenyia, analysis of 1990/91 and 1995/96 Department of Census and Statistics data), Nisha Arunatilake (Institute of Policy Studies, health outcomes), K. Tudor Silva (University of Peradenyia, qualitative analysis), Upul Sonnadara (University of Colombo, SLIS processing), and the staff of the Sri Lanka Business Development Center, which carried out collection of the Sri Lanka- Integrated Survey. The peer reviewers were: Gaurav Datt and Gunnar Eskeland. The team also received- useful comments on the analysis of poverty programs and safety nets from Margaret Grosh and on the analysis of microfinance programs from Gerard Caprio. Sadiq Ahmed, Edgardo Favaro, Emmanuel Jimenez, Tercan Baysan, Mark Sundberg, Tara Vishwanath, Dina Umali-Deininger, Gajan Pathmanathan, Miriam Witana, Hugo Diaz-Etchevehere, Deborah Bateman, Kenichi Ohashi, Mariana Todorova, and Valerie Kozel, who led the early stages of the work, provided useful input at different stages of the report. EXECUTIVE SUMMARY Sri Lanka has led developing nations in recognizing the importance of health and education for development. Sri Lanka now boasts social indicators more in line with those of a developed nation than those of a country in which per capita income is just US$800. Successive governments have maintained the country's emphasis on universal health and education and built on the high level of human capital inherited at independence. Universal and free health and education services, and focus on gender equality and the education of women have been permanent features of Sri Lanka's social policies over several decades. These remarkable achievements have meant higher levels of well-being for all Sri Lankans, and have considerably reduced non income poverty. Less clear is Sri Lanka's success in reducing income poverty. In analyzing the nature and causes of income poverty in Sri Lanka, three questions come to mind: * Why has Sri Lanka's considerable economic advance over Korea, Thailand, and Malaysia in the 1960s turned negative in more recent times, and significantly so? * Why has poverty in Colombo and the rest of the Western Province fallen rapidly and to a relatively low level, whereas it remained high in other parts of the country? * Have the large resources and prominence Sri Lanka has given to redistributive programs really help reduce poverty? A simple answer to these questions is Sri Lanka's hesitant embrace of progressive economic and social policies. Such policies would have lead to: (a) withdrawal of the state from all areas where the private sector can be more efficient and removal of regulations hindering the effective functioning of markets, with a view to evolving a policy framework that supports creation of opportunities; (b) focusing the state on the provision of much needed infrastructure and social services that the private sector cannot provide; (c) social policies accommodating diversity - ethnic, linguistic, political, religious - and understanding among different groups, providing fair and transparent access to resources and opportunities, and providing smaller and more transparent anti-poverty programs, focused on.the poorest. Instead, on the economic policy front, notwithstanding over two decades of gradual but persistent liberalization of the economy, the economy continues to be more protected than countries in Latin America and East Asia, which started liberalization much after Sri Lanka. The regulatory environment, particularly restrictions on labor and land markets, and less than fully competitive financial markets, make the investment climate much less friendly in Sri Lanka than in its competitors to the East. Hence, while reforms started in 1977 helped accelerate growth in the 80s and 90s (from stagnation in the 60s and 70s), Sri Lanka has not been able to grow at the double digit rates of its neighbors to the East. The ethnic conflict has obviously also been a complicating factor, which has set back poverty reduction and growth by diverting a large share of public resources to defense, discouraging tourism, and investment. Sri Lanka ended the twentieth century with per capita income less than half that of Thailand, one-fourth that of Malaysia, and one just one-tenth that of the Republic of Korea, countries at comparable stages of development just a few decades ago. Regarding the role of the state in the economy, notwithstanding significant advances in privatization (plantations, telecommunications, national airline, a variety of industrial plants), the - ii - state continues to own a large share of the banking system, two large insurance companies, power utilities, all water utilities, railways, food marketing companies, and all higher education institutions, among others. This has both prevented the private sector from investing and diverted public resources away from much needed investments in infrastructure, such as roads, that the private sector cannot undertake. For example, now as 15 years ago, it takes nearly 4 hours of travel to cover just over 100 km to Kandy, the country's second city, and about the same to cover nearly the same distance to the country's second largest port of Galle. Regions only slightly more distant, such as Uva and the North Central Province, are even more isolated. As a result, while the liberalization and associated growth brought benefits to the Western Province with good infrastructure, it failed to spill over the rest of the country. Colombo and neighboring areas experience a low incidence of poverty, but it is several times higher in the rest of the country. It is not only the poor that suffer from inadequate infrastructure. One of Sri Lanka's surprising features, is that wealthier households too face significant shortfalls in the provision of. such basic infrastructure. Indicators of per capita consumption of energy and number of phone lines, for instance, are significantly lower in Sri Lanka than in countries with comparable per capita income levels, such as Bolivia and China. These shortfalls affect the welfare of households directly and also indirectly, by limiting economic growth in the poorer areas of Sri Lanka and contributes to keeping the returns to education and health low. Poorer regions of Sri Lanka and poorer households face the most severe shortfalls. The poor have less access to basic services, such as clean drinking water, sanitation, electricity, safe cooking fuel, and communications, than wealthier households. Provision of basic infrastructure, such as power, water, and sanitation, is deficient especially outside Colombo, where most of the poor live. In terms of social policies, Sri Lanka has a long tradition of protecting acquired rights, and encouraging patronage-rather than stimulating market-based creation of opportunities. Job banks, access to public sector employment, distribution of public lands, language policy, and anti poverty programs, have all in their own different ways sought to bring benefits to targeted poor, more often chosen on grounds other than objective need. As documented in academic work and official reports, this has eroded trust in public institutions, and some see it as an important reason for the eruptions of youth violence. The report of the 1991 Presidential Commission on Youth, which had been appointed to examine the causes of youth discontent and unrest that led to the Marxist rebellion of 1987-89, reported that there was a strong consensus within the country that politicization and abuse of power and injustice were the main causes of youth unrest. In particular, the Commission felt that the politicization of employment, an extreme manifestation of which is the "chit system", was deemed by youth as incompatible with the basic norms of fairness, equity, and merit. The Samurdhi anti-poverty program is one more recent example. Samurdhi, which costs about 1 percent of Sri Lanka's GDP and covers nearly half of all households, suffers from design weaknesses that have resulted in the exclusion of nearly 40 percent of the poorest from the program, many of them members of ethnic minorities, and the transfer of resources to many of the wealthy. There are also indications that like Janasaviya (the program in place during the early 1990s), Samurdhi may have also become politicized, with support granted on the basis of political affiliation. The slower pace and unbalanced composition of economic growth in Sri Lanka relative to its East Asian neighbors have in turn translated into slower and geographically uneven progress in reducing overall income poverty. According to the Government's 2000 Framework for Poverty Reduction paper, between one fourth and one fifth of Sri Lankans lived below the national poverty line in the mid 1990s, corresponding to nearly 7 percent living on less than - 111 - US$1 a day, and over 45 percent on less than US$2 a day. Although it may have declined further since then, the incidence of poverty is still about twice that in Malaysia, the Republic of Korea, and Thailand, for largely comparable national poverty lines. Poverty is also higher in Sri Lanka for internationally comparable poverty lines. In addition, obstacles to socio-economic mobility and integration, physical (such as poor roads and communications) and regulatory (such as the rigid labor and land regulations) contribute not only to keep growth below potential, but also to relatively high price and wage differentials, and ultimately to increasing regional disparities in poverty. Within Sri Lanka, wherever growth has been stronger -- in manufacturing and service activities, concentrated in the Western Province - poverty has declined rapidly and is now relatively low. Where growth has stagnated or has been slow -- in agriculture, the main activity in rural areas - poverty has declined slowly and remains high. Who are the poor? Sri Lanka's achievements in providing universal and free education and health care to the general population are reflected also in good social indicators of the poor. The poor fare considerably better in some key non-income dimensions of poverty in Sri Lanka than in most other developing countries. Over 90 percent of the poor in Sri Lanka have completed at least primary education (although the education provided to the poor may be of poor quality) and around half have at least secondary education. Other social policies appear to have been successful. The incidence of poverty is identical among female and male headed households. Less than 2 percent of the poor are unemployed, and although child labor remains a problem, Sri Lanka has been more successful than other developing countries at curbing the practice. In other respects, the poor in Sri Lanka are like the poor in most other nations. Relative to wealthier households, they own fewer assets and are engaged in activities (small-scale agriculture) that earn these assets a lower rate of return. More than 90 percent of Sri Lanka's poor live in the rural and estate (or plantation) sectors; for half of them, small-scale agriculture and wage labor on farms and plantations provides the main sources of income. The poorest work in paddy production or in plantations. There is poverty in urban areas, but poverty and isolation is worst and most widespread in remote, rural areas, where the depth and the severity of poverty are nearly twice that of urban areas. In Moneragala district, Uva Province, for example, nearly half of all households live in absolute poverty. In the Western Province, where the capital city and most manufacturing and services are located, the incidence of poverty is relatively low. In the Colombo district, for instance, 1 in 10 households is poor. - The incidence of poverty varies little across the major ethnic groups (Sinhalese, Sri Lankan Tamils, Indian Tamils, and Muslims or Moors) with the exception of Indian Tamils, most of whom are poor. Indian Tamils are among the poorest people in Sri Lanka. Living largely on tea, rubber, and other plantation estates, they also suffer the highest degree of social and economic isolation. They are often deprived of basic infrastructure and receive the lowest quantity and quality of other publicly provided services, such as education and health services. Because of the civil conflict and the often remote location of estates, their mobility is also restricted. Language barriers (many speak only Tamil) and, for some, lack of identity cards also severely curtail their ability to integrate themselves into the economy and society. As a result, while many benefited from the restructuring in the plantation sector, others were unable to relocate. In the northeast and areas bordering it, which are directly affected by the civil conflict, the extent of material deprivation and human suffering are highest. Poverty estimates are unavailable for the northeast, due to the impossibility of carrying out a representative household - iv - survey in a war area. Data from the 1999/2000 Sri Lanka Integrated Survey, which attempted to cover the North East to the extent possible, reveal that household incomes would be amongst the lowest in the country, were it not for remittances received from abroad. Foreign and domestic remittances are an important source of income for households throughout Sri Lanka, but remittances from abroad are the main safety net of households in the northeast. Compared to other Sri Lankan households, households in the northeast and border villages are the most disadvantaged in terms of their ability to obtain a secure livelihood. Virtually every household in these areas experiences the impact of the war, either directly, through loss of life, assets, or displacement, or indirectly, as a result of the deterioration of infrastructure that has resulted from the conflict. Among the most severely affected groups are households that have been displaced, sometimes repeatedly, as a result of conflict in or threats to their home villages. Displaced families have lost productive assets (agricultural equipment, livestock, shops, mills), including land they had cultivated. Most of the displaced remain in refugee camps, often for years. Death, injury, and displacement statistics can only approximate the extent of human suffering in the North East. According to SLIS data, individuals in the North East are far more likely to suffer personal injury than households anywhere else in the country, in particular in the Western Province. One out of every twelve households in the North East reported that a family member was killed as a result of the civil conflict. Among the poorest households, in the North East the rate is even higher with one out of every seven poor households reporting a member killed. Several studies reveal very high levels of wasting and stunting among children and very poor nutritional status among women, especially among the Tamil and in "uncleared" (under LTTE control) areas. Crafting a Strategy for Accelerating Poverty Reduction Over the last three years, the Government of Sri Lanka has made an extraordinary effort to deepen its understanding of the nature and root causes of poverty in Sri Lanka, including in the areas directly affected by the conflict, with a view to reassessing and reformulating its policy framework for reducing poverty. The Government's Framework for Poverty Reduction paper presented to donors at the Development Forum in December 2000 already signals a welcome shift in the government's poverty reduction paradigm. The paper, a product of recent analytical work (part of which in collaboration with the World Bank and the UNDP) and of a broad consultation process within Sri Lanka, indicates the need to rely more on economic opportunities and growth than on redistribution to reduce poverty; to correct inefficiencies and inequities in the social protection system, in particularly in the Samurdhi program; to empower the poor through improved governance; and to achieve a peaceful resolution to the civil conflict. More recently, and on the basis of President Kumaratunga's Vision for the 21st Century (August 1999) and the Budget Policy Statement of 2001, the Government initiated preparation of a "Vision 2010" to be discussed with civil society, donors and the private sector for determining the country's long- term development goals and strategic priorities. In addition, an island-wide consultative process was launched to collect the views of key stakeholders on how the Government, civil society, and the donor community could work together on improving relief and rehabilitation in war-affected areas, develop a code of operation, coordination and institution building, and promote reconciliation and peace building. The primary goals of the Government's strategy are to secure peace and to achieve significant transformation in the quality of life of the people, especially in rural areas, through increased incomes, economic modernization and invigoration of the rural economy. Tackling poverty in Sri Lanka will not be achieved by adding a government program here or there. -v- Reducing poverty will require all policies, expenditures, and programs to be evaluated for their impact on the livelihoods of the average household and on the poorest Sri Lankans. Three broad areas of action are crucial to reducing poverty and raising per capita incomes: creating a policy environment that is conducive to raising the incomes of the poor, providing effective public services that reach the poor in the core areas of government responsibility, and finding a peaceful resolution to the conflict. Creating a Policy Environment that Facilitates Poverty Reduction Households and individuals tend to escape poverty and reduce vulnerability to downward shocks through their own initiative and efforts, which public policies can either encourage and facilitate or discourage and hinder. A policy framework that helps accelerate growth, including in agriculture; increases real wages and employment; facilitates the mobility of labor and other factors to rapidly growing sectors; and limits inflation, will create an environment in which poverty can be reduced. Prudent fiscal management, financial sector restructuring, outward- oriented trade and investment policies, and a transparent climate for small, medium, and large- scale investors, are all critical features of this environment. Fiscal policy. High fiscal deficits put upward pressure on interest rates, thereby restricting growth. Despite efforts to contain expenditures, increasing defense and interest outlays caused the government to run budget high deficits throughout the 1990s. Defense expenditures also crowd out development spending, namely on much needed investment in, and maintenance of infrastructure. Furthermore, for the public budget to benefit the poor, the government needs to review the pattern of spending across and within sectors, look hard at what tax concessions and subsidies can be eliminated, and design a pattern of expenditures that favors the poor. Deregulation and privatization. Accelerating growth also requires that the pace of structural reform speed up. A considerable reform agenda of privatization, deregulation, and removal of distortions in markets and prices has been highlighted by the President, and remains unfinished. There has been some reduction in state control in recent years through the divestiture of state-owned enterprises and fiscal retrenchment, but the state still intervenes in many production and trading activities, crowding out more efficient private initiative. The business environment is constrained by the inadequate public institutions, public policies, and the regulatory framework is not in tune with the needs of the market economy. As a result, the private sector has not realized its full potential, especially in rural areas. Policies for agricultural growth. The sources of sustained growth and poverty reduction will continue to be manufacturing and services and facilitating the growth of these sectors should continue to be a priority. In the short and medium run, however, many of the poor will continue to remain engaged in agriculture. The policy framework for agriculture, including the trade regime for agricultural inputs, technology, and outputs, is currently a serious constraint to reducing poverty. Increasing agricultural incomes and productivity requires improving access, use, and tradability of land and reducing government involvement in agricultural marketing. Land market reform is a key ingredient for high agricultural growth rates, as experience in Indonesia and Thailand has shown. Past land policies in Sri Lanka, which aimed at equitably distributing land among small-scale farmers, may have fostered smallholder subsistence agriculture. The land market in Sri Lanka is constrained by the state's role as dominant landlord, by restrictions on land sale and use in some areas, and by inadequate land registry. State control - vi - over land, along with ill-defined property rights and tight restrictions on the transferability and use of land, restrict farmers' ability to obtain credit, move, or shift into higher-value crops. Trade policy for agricultural outputs and inputs must be made more consistent and transparent. Sri Lanka followed a relatively open international trade regime in the 1990s, but trade policy on agricultural goods and foodstuffs is unpredictable, distorting production-decisions and contributing to the volatility of farmer incomes. Import tariffs for agricultural produce are often changed on an ad hoc basis, and interventions by the state-owned Cooperative Wholesale Establishment have contributed to sharp variations in output prices in the past. Erratic trade policy has led to large and unpredictable variations in the quantity of imports of major food commodities and thus their farm gate prices. The import regime also discourages the private sector from investing in.food commodity processing and storage, leaving the activity solely to the government. A dynamic private sector could perform all of these functions more effectively. Labor market flexibility. For growth to be effective in reducing poverty, the poor need to be able to increase the returns on their assets and outputs. Policies that facilitate rather than discourage mobility and reallocation in the markets for labor, capital, land, and natural resources are key to increasing the incomes of the poor. Currently, excessive restrictions prevent labor, capital, and land from being allocated to more productive uses. Labor laws, for example, restrict the scope and range of labor contracts, and the state remains an active employer, paying above- market wages, encouraging queuing. The state owns about 60 percent of all land and holds a quasi-monopoly in education (about 95 percent of schools are government run). These distortions constrain growth and contribute to the persistence of poverty. To encourage private sector growth and job creation, labor regulations should be made more flexible. Sri Lanka's Termination of Employment and Workmen Act is among the most restrictive in the world. Under this law, termination of employment is possible only with prior written consent of the employee or the Commissioner of Labor. The process leading to the consent of the Commissioner of Labor is nontransparent, with dispute resolution taking several months and sometimes years, during which time firms must continue paying the salaries of redundant workers. The government has attempted to ease these restrictions by creating Export Processing Zones and providing special concessions through the Board of Investment, but such measures provide relief only to large enterprises located near Colombo. For smaller enterprises and those located away from the capital, inflexible labor laws continue to impose high costs, discouraging employment creation in rural areas. Providing Effective Core Public Services That Reach the Poor Infrastructure for connectivity and integration. Regional disparities in the incidence of poverty in Sri Lanka reflect, at least partly, the lack of a coherent strategy to integrate and develop different regions of Sri Lanka. Inproved provision roads, energy, water and sanitation, and other types of infrastructure is needed to support the development of manufacturing and services in rural areas. One of the most significant problems is the absence of a good road network connecting major supply areas to demand centers, which limits not only non-farm growth but also limits the ability of farmers to switch to higher value-added crops or expand nonfarm activities. Improved infrastructure is needed, but more investment in infrastructure alone is not the answer. New infrastructure and maintenance programs must be designed to support welfare through both income creation and improved living standards. Funds are wasted when programs - Vii - fail to provide the services people need, infrastructure is unreliable because of poor materials or workmanship, or the intended beneficiaries are incapable of using or maintaining the goods and services provided-all outcomes that have resulted from top-down approaches. Improved governance, including a leaner, better organized, and more accountable public administration; a more effective budget process; and a stronger local administration, can help improve the level and quality of basic services to the poor. Quality public services: education. Improving the quality of other publicly provided services, such as education and health, will help the poor acquire knowledge and skills valued by the labor market. For the poor, quality of education may be low, as reflected in lower learning outcomes for poorer regions of Sri Lanka. Class sizes in poor communities, most rural, are often more than twice as large as in wealthier ones. Findings from the Sri Lanka Integrated Survey also reveal a growing shortage of teachers in English, mathematics, and science. The lack of teachers in these subjects, which are considered vital to employability, means that instruction is either inadequate or unavailable in many rural areas. Schools in war-affected areas and border areas, as well as villages with poor road access and transportation limitations, find it particularly difficult to recruit qualified teachers. The deficient provision of education for the children of poor households today will result in the persistence or increase in poverty and inequality tomorrow. Effective and affordable safety nets for the most vulnerable. For the majority of the population, including many of the poor, government policy can be most effective by helping the poor help themselves. Untargeted subsidies and large income support programs, such as Samurdhi, are costly and unnecessary. In Sri Lanka, basic assistance and income support through well targeted safety net programs would still be required for four groups of vulnerable people. These are: (i) a small group of people likely to remain chronically poor (orphans, the physically disabled, and elderly and widows without family and kinship ties) because they cannot take advantage of existing opportunities; (ii) people "trapped" in pockets of poverty because they face constraints to mobility and integration in the short run, such as estate, irrigation and resettlement colony populations; (iii) people affected directly by the war, through death of breadwinner, injury, displacement, and loss of assets; (iv) people affected by catastrophic risk, such as floods or severe droughts, for whom the best public policy is the promotion of a combination of risk prevention and coping support. A larger group of people who experience intermittent periods of ill health or unemployment may require a temporary cushion that enables and encourages them to return to productive activity. An effective safety net should help prevent a very low standard of living by reducing and mitigating downside socioeconomic risk or coping with its occurrences; it should not create dependence or discourage individual effort. In its current form, the Samurdhi program fails to meet this standard. Furthermore, economic theory as well as cross-country evidence from developing countries indicate that transfer programs like Samurdhi cannot achieve sustained reductions in poverty. Inferring need through means testing alone, as the Samurdhi program tries to do, is always difficult. In Sri Lanka, it has lead to leakages, thus becoming costly, and become subject to capture by ethnic or political groups, thus undermining social cohesion. A more effective and affordable set of safety net programs could use a combination of methods of identification, including categorical and means testing criteria (for the chronically poor), geographical criteria (for those in pockets of poverty or affected by catastrophic risk), and self-targeting (for all groups). Revamping the safety net program in this way would serve the needs of the poor who - viii - cannot participate in the growth process and reduce their vulnerability to negative shocks. > ,M Geographical transfers to estate populations and others should not, however, replace efforts to, remove the obstacles to social and economic integration suffered by these communities. Public administration. Improvements in the delivery of physical and social infrastructure as well as social protection will require a more effective and accountable public administration. Traditionally, public sector management in Sri Lanka was considered superior to that in many other developing countries. There is a perception that the quality of governance has declined, however. Sri Lanka's move to an open market economy, however hesitant, has not been accompanied by similar transformations in public administration. The public sector is large, costly, and suffers from low effectiveness and efficiency. With 12 percent of its labor force employed in government in 1999, Sri Lanka has the largest bureaucracy per capita in Asia. This bureaucracy is also expensive to maintain, as public sector workers with the lowest levels of education earn almost twice as much as their counterparts in the nonpublic sector. Sri Lanka has also failed to realize many of the potential benefits of decentralization. The design of decentralized government structures and the ways in which implementation have been executed in Sri Lanka have inhibited high-quality, demand-driven service delivery to all segments of the population. Inadequate attention has been devoted to economic incentives associated with the fiscal aspects of decentralization, and the central government has not withdrawn from the devolved functions. As a result, Sri Lanka has developed a highly complex framework of decentralization, which has created confusion and ambiguity regarding roles and responsibilities of both elected and administrative arms of government. The result has been reduced accountability and transparency, which constrain the growth and maturation of decentralized government institutions. Achieving Peace and Social Harmony Part of Sri Lanka's difficulty in raising incomes and reducing poverty can be traced to the 17-year old ethnic conflict. Reducing poverty and human suffering in Sri Lanka, including the Northeast, requires an end to the civil conflict, the consequences of which are felt in every sphere of social and economic life. A study conducted by the Institute of Policy Studies estimates the present value cost of the conflict during 1984-96 at nearly 170 percent of Sri Lanka's 1996 GDP. The 1999 annual report of the central bank of Sri Lanka estimates that the conflict has reduced Sri Lanka's economic growth by about 2-3 percentage points a year. The conflict has also contributed to a widespread feeling of despair and hopelessness among youth, an increase in violence, and brain drain. The costs of the war have fallen disproportionately on the poor. Faced with fewer opportunities to earn a living, the rural poor serve in the armed forces on both sides. Expenditures on the war effort, which have accounted for about 5 percent of GDP in recent years, crowd out a vast range of pro-growth and pro-poor public expenditures. The instability brought about by the war also reduces investment and job creation. And the protracted conflict has fueled ethnic polarization and discrimination and desensitized communities to the effects of violence. Households in the northeast and border villages are the most disadvantaged in terms of their ability to obtain a secure livelihood. Virtually every household in these areas experiences the impact of the war, either directly, through loss of life, assets, or displacement, or indirectly, as a result of the deterioration of infrastructure that has resulted from the conflict. Among the most severely affected groups are households that have been displaced, sometimes repeatedly, as a result of conflict in or threats to their home villages. Displaced families have lost productive - 1x - assets (agricultural equipment, livestock, shops, mills), including land they had cultivated. Most of the displaced remain in refugee camps, often for years. Policies that build social harmony and economic integration could be an important element of Sri Lanka's poverty strategy. Education and language policy, for instance, could help reduce social divisions and improve the employment and educational prospects of the poor. Education in Sri Lanka is currently provided in either Sinhala or Tamil, based on the ethnicity of the parents. There is no requirement for Sinhalese to learn Tamil or Tamils to learn Sinhala. English is used for instruction in only a few private schools, most located in Colombo. Mastery of English, the language of the private sector, could be a powerful asset for the poor. English is also the language in which many of the university degrees are in high demand, such as medicine and engineering, are taught. Teaching children from poor households to speak English could help them obtain better-paid jobs in the growing areas of the economy. Chapter 1 The Evolution and Nature of Poverty in Sri Lanka 1.1 The case of Sri Lanka has often been cited as an example of a "support-led" strategy for improving basic capabilities (Sen 1981; Dr&ze and Sen 1989; Isenman, 1980, 1987). This approach is based on the belief that economic growth occurs too slowly to provide substantial benefits to the poor within a reasonable period of time and that the government must therefore act to ensure a minimum standard of living. In Sri Lanka, a large share of public expenditures, between around 7-10 percent of Gross Domestic Product (GDP), has been redistributed to households over the years in the form of free education and health services, as well as food subsidies, food stamps, and subsidized credit to improve living standards and ensure minimum consumption levels of households perceived to be in need. 1.2 Sri Lanka has led developing nations in recognizing the importance of health and education for development. As a result, it now boasts social indicators more in line with those of a developed nation than those of a country in which per capita income is just over US$800. Successive governments have maintained the country's emphasis on universal health and education and built on the high level of human capital inherited at independence. Universal and free health and education services, and focus on gender equality and the education of women have been permanent features of Sri Lanka's social policies over several decades. Many of these programs have benefited from strong external donor support and/or Non Governmental Organization (NGO) involvement. These remarkable achievements have meant higher levels of well-being for all Sri Lankans, and have considerably reduced non income poverty. 1.3 Sri Lanka's success in reducing income poverty is less noteworthy, especially when contrasted with that of East Asian countries that were at comparable levels of development only a few decades ago (Table 1.1 and Figure 1.1). In the 1960s, Sri Lanka's income per capita was comparable to that of Malaysia, the Republic of Korea, and Thailand, and prospects for balanced growth and development were brighter. By the end of the twentieth century, income in Sri Lanka was less than half that of Thailand, one-fourth that of Malaysia, and just one-tenth that of the Republic of Korea. Although consumption poverty in Sri Lanka declined to about 25 percent (excluding the war-affected North East) in 1995/96 and may have fallen further since then, this is still about twice that in Malaysia, the Republic of Korea, and Thailand, for largely comparable national poverty lines. Poverty is also higher in Sri Lanka than in Thailand and Korea for internationally comparable poverty lines.1 Poverty might have declined since the mid-nineties, in particular because 1996 was a drought year, yet this remains an unacceptably high level of poverty for most Sri Lankans. Furthermore, poverty indicators are likely to underestimate poverty in the country. Because of the conflict, suitable household surveys could not have been launched in the war-affected areas of the northeast since the mid-1980s. However, it is incontestable that poverty and certainly human suffering have increased in this region with the escalation of the war. Other--more difficult to measure--forms of poverty, such as personal insecurity and vulnerability also appear to be on the rise throughout the country, partly as spillovers of the conflict. Lack of comparable data makes it impossible to contrast the pace of poverty reduction. -2- Table 1.1: Incidence of Poverty and per Capita GNP in Sri Lanka and East Asian Comparator Countries Poverty incidence Income (% of population living below poverty line) (US$) International poverty lines a National poverty line Population below Population below $2 Per capita GNP $1aday aday Country 1990s 1960s 1999 Indonesia 15 66 20 55b 580 Rep. of Korea less than 2 less than 2 - 94 8,490 Malaysia - - 15 338 3,400 Sri Lanka 7 45 25 151 820 Thailand less than 2 28 13 149 1,960 - Not available. Notes: Poverty incidence estimates exclude the North and East Poverty estimates are based on a local (reference) consumption poverty threshold of 2,500 calories and 53 grams of protein per day and male adult (20-39 years) equivalent costing 792 rupees per person a month at 1995-96 prices. a. Latest estimates, 1990s. b. 1967 figure. Source: Word Bank, 2000a, Datt and Gunewardena 1997, 1995/96: Gunewardena 2000. Figure 1.1: Per Capita Gross National Product (GNP), Atlas Method Sri Lanka and East Asian Comparator Countries $11,000. $10,000 $9,000 $8,000 - --Indonesia r $7,000 0 -Korea $6,000 - ---Malaysia $5,000 - 1 ----67- Sri Lanka $4,000 - 0Thailand $3,000 $2,000 $1 000 , SO 1964 1970 1975 1980 1985 1990 1995 1999 Source: World Bank. 1.4 An explanation for Sri Lanka's relatively slower progress in reducing poverty and increasing average incomes is the country's hesitant embrace of modern economic and social policies. Such policies would have lead to: (a) withdrawal of the state from all areas where the private sector can be more efficient and removal of regulations hindering the effective functioning of markets, with a view to evolving a policy framework that supports creation of opportunities; (b) focusing the state on the provision of much needed infrastructure and social services that the private sector cannot provide; (c) social policies accommodating diversity - ethnic, linguistic, political, religious - and understanding among different groups, providing fair and transparent access to resources and opportunities, and providing smaller and more transparent anti-poverty programs, focused on the poorest. -3- 1.5 Instead, on the economic policy front, notwithstanding over two decades of gradual but persistent liberalization of the economy, the economy continues to be more protected than countries in Latin America and East Asia, which started liberalization much after Sri Lanka. The regulatory environment, particularly restrictions on labor and land markets, and less than fully competitive financial markets, make the investment climate much less friendly in Sri Lanka than in its competitors to the East. Hence, while the reforms started in 1977 helped accelerate growth in the 80s and 90s (from stagnation in the 60s and 70s), Sri Lanka has not been able to grow at the double digit rates of its neighbors to the East. The ethnic conflict has obviously been a complicated factor that diverted large share of public resources to defense, and discouraged tourism, and investment. 1.6 Regarding the role of the state in the economy, notwithstanding remarkable advances in privatization (plantations, telecommunications, national airline, a variety of industrial plants), the state continues to own a large share of the banking system, two large insurance companies, power utilities, all water utilities, railways, food marketing companies, and all higher education institutions, among others. This has both prevented the private sector from investing and diverted public resources from much needed investments in infrastructure, such as roads, that the private sector cannot undertake. For example, now as 15 years ago, it takes around 5 hours of travel to cover just over 100 km to Kandy, the country's second city, and around 6 hours to cover nearly the same distance to the country's second largest port of Galle. Slightly more distant areas, such as Uva and the North Central Province are considerably more isolated. As a result, while the liberalization and associated growth brought benefits to the Western Province with good infrastructure, it failed to spill over the rest of the country. Colombo and neighboring areas experience a low incidence of poverty, but it is several times higher in the rest of the country. 1.7 It is not only the poor that suffer from inadequate infrastructure. One of Sri Lanka's surprising features, is that wealthier households too face significant shortfalls in the provision of such basic infrastructure. Indicators of per capita consumption of energy and number of phone lines, for instance, are significantly lower in Sri Lanka than even in countries with comparable per capita income levels, such as Bolivia and China (Table 1.2). These shortfalls affect the welfare of households directly and also indirectly, by limiting economic growth in the poorer areas of Sri Lanka and contributes to keeping the returns to education and health low. Poorer regions of Sri Lanka and poorer households face the most severe shortfalls. The poor have less access to basic services, such as clean drinking water, sanitation, electricity, safe cooking fuel, and communications, than wealthier households. Provision of basic infrastructure, such as power, water, and sanitation, is deficient especially outside Colombo and in plantations, where most of the poor live. -4- Table 1.2. Per capita Energy Consumption and Communications, Sri Lanka and comparator countries Per capita GNP (1999) Per capita Energy Telephones (1997) Atlas Method Measured at PPP Consumption (1998) Main lines Mobiles (US$) (Kilowatt-hours) (n. per 1,000 people) Albania 870 2892 851 31 1 Bolivia 1010 2193 391 69 27 China 780 3291 714 70 19 Indonesia 580 2439 329 27 5 Sri Lanka 820 3056 227 28 9 Notes: PPP - Purchasing Power Parity; because nominal exchange rates do not always reflect international differences in relative prices, GNP can be converted to international dollars by the PPP exchange rate. At the PPP rate, one intemational dollar has the same purchasing power over domestic GNP that the US dollar has over US GNP. Source: World Bank, 2000a. 1.8 In terms of social policies, Sri Lanka has a long tradition of protecting acquired rights, and encouraging patronage-rather than stimulating market based creation of opportunities. Job banks, access to public sector employment, distribution of public lands, language policy, and anti poverty programs, have all in their own different ways sought to bring benefits to targeted poor, more often chosen on grounds other than objective need. As documented in academic work and official reports, this has eroded trust in public institutions, and some see it as an important reason for the eruptions of youth violence. The report of the 1991 Presidential Commission on Youth, which had been appointed to examine the causes of youth discontent and unrest that led to the Marxist rebellion of 1987-89, reported the there was a strong consensus within the country that politicization and abuse of power and injustice were the main causes of youth unrest. In particular, the Commission felt that the politicization of employment, an extreme manifestation of which is the "chit system" (the practice of receiving a letter from one's Member of Parliament in order to find public sector employment), was deemed by youth as incompatible with the basic norms of fairness, equity, and merit. The Samurdhi Program, analyzed in Part II of this report, is one recent example. Samurdhi, which costs about 1 percent of GDP, suffers from design weaknesses, which have resulted in the exclusion of nearly 40 percent of the poor, many of them members of ethnic minorities, and the transfer of resources to many of the wealthy. There are also indications that like Janasaviya (the anti-poverty program in place during the early 1990s), Samurdhi may have also become politicized, with support granted on the basis of political affiliation. 1.9 The slower pace and unbalanced composition of economic growth in Sri Lanka have in turn translated into slower and geographically uneven progress in reducing overall income poverty. In addition, obstacles to socio-economic mobility and integration, physical (such as poor roads and communications) and regulatory (such as the rigid labor and land regulations) contribute not only to keep growth below potential, but also to relatively high price and wage differentials, and ultimately to increasing regional disparities in poverty. Within Sri Lanka, wherever growth has been stronger -- in manufacturing and service activities, concentrated in the Western Province - poverty has declined rapidly and is poverty lower. Where growth has stagnated or has been slow -- in agriculture, the main activity in rural areas - poverty has declined slowly and remains high. -5- Poverty Trends in Sri Lanka: Slow Poverty Reduction and Increasing Regional Divergences 1.10 Overall, poverty in Sri Lanka has declined since independence. Both the 1995 Poverty Assessment and other studies revealed a significant decline in the incidence of poverty in Sri Lanka between 1953 and 1985.2 Analysis of the three most recent household data sets indicates that the declining trend in poverty has continued since the mid-1980s but that poverty reduction slowed during the early 1990s (Box 1.1 and Figure 1.2). This trend is apparent regardless of the income level at which the poverty threshold is set (see Annex A, Figure Al).3 A similar pattern occurs in inequality. Figure 1.2: Incidence of Poverty and Consumption Inequality in Sri Lanka, 1985186-1995/96 40 0.5 35 31 30 -- 25 0.45 25 - 20 20 - - 0.4 Gini 15 10 0.35 0 0.3 1985/86 1990/91 1995/96 i Incidence -4-- Gini Coeff. (Expenditure) Note: Poverty incidence estimates are based on DCS datasets that exclude the North and East. Estimates are based on a local (reference) consumption poverty threshold of 2,500 calories and 53 grams of protein per day and male adult (20-39 years) equivalent, costing 792 rupees per person a month at 1995-96 prices. Source: 1985/86-1990/91: Datt and Gunewardena 1997; 1995/96: Gunewardena 2000. 1.11 The increase in the incidence of poverty in rural areas between 1990/91 and 1995/96 is likely to at least partly reflect the impact of the 1996 drought, suggesting a high degree of household vulnerability to shocks.4 The years 1990 and 1996, during which household surveys 2 Because of data comparability problems, the analysis compared data on private per capita consumption with data on the distribution of household income. A decomposition of changes in the Foster, Greer, and Thorbecke (FGT) (1984) measures of poverty according to the methodology set out in Datt and Ravallion (1992) reveals that between 1985/86 and 1990/91 national poverty fell because of consumption growth and favorable redistribution. Between 1990/91 and 1995/96, the lack of consumption growth explains the rise in poverty. The pattem of redistribution was more complex: favorable redistribution offset the lack of consumption growth in the urban sector. In the rural and estate sectors, both growth and redistribution effects were adverse, contributing to the significant rise in poverty (Gunewardena 2000). 4 The incidence of poverty in any one year is best characterized as a random variable that can follow a trend but is susceptible to random fluctuations. These fluctuations are deviations from the mean that would occur in a large set of repeated drawings conditional on government policies and exogenously determined variables in a given year. were conducted by the Department of Census and Statistics, were exceptional years. In 1990 the recovery in agriculture allowed the economy to grow 6 percent, up from 1 percent in 1988-89. In 1996 the economy was hit by several shocks, including a prolonged drought, the bombing of the central bank, and the escalation of the conflict, which negatively affected business confidence, tourism, and foreign investment. Economic growth in 1996 was just 3.8 percent, with agriculture. falling 5 percent and paddy production declining 27 percent. The economy rebounded to grow 6.4 percent in 1997. Box 1.1: Measuring Income Poverty in Sri Lanka: Methodology and,data sources. The Bank's first Poverty Assessment for Sr Lanka was completed in 1995? Itfocused oovertyoutcomesiupto.1990 and based its poverty estimates ons the analysis f household data collected dunng-thefirstthree rounds of the 199/91 survey andthecorresponding. months of the 1985/86 survey.This reportanalyzsthe ful985/861990/91data ts.as\well:asthe latest(1995/96)datafrom which comparable poverty lines can be estimated. it uses the concept of consumption poverty (per capita consumption expenditure, excluding expenditure on durables) as the main indicator of standard of living To allow for a companson between earlier findings and those emerging from more recent data, this report follows the same approach (See Deaton (1995) for a discussion of the relative merits of using per capita consumption as the individual welfare indicator for developing countnes See Ravallion (1998) for a discussion of how the concept of capabilities (Sen. 1985. 1987) can be related to the more conventional approaches to welfare measurement found in economics) Poverty lines. Internationally comparable poverty lines are useful for producing comparable aggregates of poverty. They test for the ability to purchase a basket of commodities that is similar across the world These lines are less suitable for the analysis of poverty within a country. The national poverty line for Sn Lanka used in this report is based on a reference food poverty line of Rs 200 in 1985/86, denved from Nanayakkara and Premaratne (1987), whose food poverty line is Rs 203 at 1985/86 prices, corresponding to a normative threshold of 2500 calories and 53 grams of protein per adult (age 20-39) male equivalent This poverty line was updated using a temporal food price index derived from unit values and expenditure weights from the 1985/86 LFSS Regional food poverty lines were denved for ten regional areas, using spatial food pnces for 1995/96 For further details, see Gunewardena (2000). Poverty measures. A poverty profile typically answers the question 'If an individual exhibits a particular.characteristic (e.g lives in a particular region) what is the likelihood of this individual being poor' In other words, what proportion of individuals with a-given characteristic are poor'? This measure is known as the headcount index or incidence of poverty, which is part of a larger family of measures known as the FGT (Foster. Greer. Thorbecke, 1984) measures of consumption poverty on which this report relies to quantify poverty in Sn Lanka. The FGT definitions of poverty measures used in this report include, * Headcount index. H, (the incidence of poverty) measures the percentage of individuals in a given population whose standard of living lies below the poverty line * Poverty Gap index, PG, (the depth of poverty) measures the average shortfall between an individuals level of consumption and the poverty line. where the shortfall for all the individuals whose consumption is above the poverty line is zero. * Scuared Poverty Gap index, PG2, (the seventy of poverty) measures the squared shortfall between an individual's level of consumption and the poverty line, thereby placing a greater weight on poorer individuals Data sources. The analysis of trends in poverty levels is based on three broadly comparable data sources: the 1985/86 Labor Force and Socio-economic Survey (LFSS), and the 1990/91 and 1995196 Household Income and Expenditure Survey (HIES), all camed out by Sri Lanka's Department of Census and Statistics (DCS) An additional household survey. the 1997/97 Consumer Finance Survey collected by the Central Bank, has become available for public use only recently. The poverty ggfile is based on both the 1995/96 HIES and three quarters of Sn Lanka Integrated Survey (SLIS) data for 1999/2000. The SLIS, with a multistage stratified random sample. was designed as part of the joint Government of Sri Lanka-VVorld Bank-UNDP work towards the development of a new poverty strategy for Sri Lanka. This report is based on three quarters of the sample (approximately 6,600 households) collected between October 1999 and July 2000 across all provinces of the country Three features of the SLIS are noteworthy First, because it includes a detailed module on various public services and poverty programs it is possible to analyze the effectiveness of these public services in reaching the poor Second, because it has national coverage (including households in the North-East), it is possible to identify and analyze the distinguishing features of poverty and welfare in war-affected areas Third, because it includes a qualitative assessment of poverty complementary to the quantitative data set, the SLIS can be used to include the perspectives of the poor themselves It is expected that future household surveys will Integrate these useful features Because the SLIS survey questionnaire is less detailed regarding the food consumption basket than is required for the calculation of a poverty line, the profile is based instead on the characteristics of households in the lowest expenditure quintile, i.e the poorest 20 percent of households. -7- 1.12 Sri Lanka compares favorably with other South Asian countries in terms of poverty and average incomes. (Figures 1.3 and 1.4). It also compares favorably in terms of nonincome dimensions of poverty, such as education and health, and the prevalence of child labor (Figure 1.5). Figure 1.3: Incidence of Poverty In South Asia (latest estimates with national poverty lines) 45 40 35 - 030 - IMI E 25- a 20I 15 - 10- 5 0 IM IM CL L6 4jo0) M - CY 2000 C C0 - 1 ) C- Source: World Bank, 2001. Figure 1.4 Per Capita Income (GNP) at Purchasing, Power Parity (PPP), 1999 5000- 4000- S3000- IL 2000- 1000 0 4) C L E E C C COci 0 :5 0) -iOV Source: Wodld Bank, 2001. -8- Figurel.5: Workforce Participation Rates of Children in South Asian Countries 30 25 * 20O Boys S20- M Girls e 15 - M-Total 10 5- Bangladesh India Nepal Pakistan Sri Lanka Source: World Bank, 2001. 1.13 Regional Disparities. Poverty indicators reveals a complex and changing landscape within Sri Lanka. Poverty has declined rapidly and to low levels in the Western Province (Figure 6). Poverty studies based on earlier data (World Bank 1995; Datt and Gunewardena 1997) indicated limited regional variation in poverty. Analysis of the 1995/96 Household Income and Expenditure Survey and the 1999/2000 Sri Lanka Integrated Survey data indicates large and growing disparities in the incidence, depth, and severity of poverty across regions (Figure 1.6 and Tables 1.3 and 1.4).. Figure 1.6: Trends in the Incidence of Poverty in Sri Lanka, by Province, 1985186-1995/96 40 35- 25 20 10 Western Central Southern N.Western N.Central Uva Sabaragarnua E 1985/8%1990/9101995/96 Note: Poverty incidence estimates are based on DCS datasets that exclude the North and East. Source: 1990/91: Datt and Gunewardena 1997; 1995/96: Gunewardena 2000, based on Department of Census and Statistics (DCS) data. -9- Table 1.3: Contribution to Poverty by Province, (reference and higher poverty lines) 1985-96 (percent) Poverty line Rs792 per person per Poverty line Rs. 950 per person per month month Province/measure of poverty 1985/86 1990/91 1995/96 1985/86 1990/91 1995/96 Western Headcount 19 23 17 21 24 19 Poverty gap 19 24 16 19 24 17 Squared poverty gap 18 24 14 19 24 16 Central Headcount 15 18 17 16 17 17 Poverty gap 14 19 18 14 18 17 Squared poverty gap 13 20 19 14 19 18 Southern Headcount 19 18 16 18 18 16 Poverty gap 21 19 16 20 18 16 Squared poverty gap 23 19 16 22 18 16 North Western Headcount 15 13 18 15 13 18 Poverty gap 14 11 17 14 12 17 Squared poverty gap 13 10 16 14 11 17 North Central Headcount 7 6 8 8 7 7 Poverty gap 7 5 7 7 6 8 Squared poverty gap 6 4 7 7 5 7 Uva Headcount 9 9 11 9 9 10 Poverty gap 11 9 13 10 9 11 Squared poverty gap 12 8 15 11 8 13 Sabaragamuwa Headcount 15 13 14 14 12 14 Poverty gap 15 14 14 15 13 14 Squared poverty gap 15 15 14 15 14 14 Note: Poverty incidence estmates are based on DCS datasets that exclude the North and East. Source: Gunewardena 2000. Table 1.4: Poverty Reduction Rates, by Province (percent) 1985/86 to 1995/96 1985-1990 1990-1995 1985-1995 Western 22 11 30.3 Central 22 -19 7.3 Southern 40 -12 32.4 North-Western 47 -88 -0.3 North-Central 45 -71 5.7 Uva 41 -56 8.6 Sabaragamuwa 44 -37 22.9 Note: positive rate Implies a decline in poverty over the period. Source: World Bank calculations based on Gunewardena 2000. 1.14 There are sharp regional disparities in economic welfare and living standards across Sri Lanka (Figures 1.7 and 1.8, Table 1.5, Annex A, Table A2, and Annex A, Figure A4). In the Western Province, where the capital city, 75 percent of manufacturing, and most services are - 10 - located, the incidence of poverty is relatively low. In the Colombo district, for example, just 10 percent of households are poor. In the northeast and the areas bordering it, which are directly affected by the civil conflict, the extent of material deprivation and human suffering is highest. In the rest of Sri Lanka, most of which is rural and coastal, small-scale or plantation agriculture and fisheries remain the main economic activities, and poverty and isolation are high. In the Moneragala district, Uva province, for example, nearly half of all households live in absolute poverty. Figures 1.7 and Figure 1.8: Poverty Maps Incidence of Poverty and Contribution to total poverty, by Province, 1995-96 (percentage) 14 -8 26-28 11-14 31-32 16-17 Northem 34-37 Norhn 18 no data no data .North central North \ Eastern North North North Eastern Western Western y Central U- -JUva we stern West4Uv Sabaragomuwa SOURCE o WORLD BANK, BASED ON GUNEWARDENA (2000). Table 1.5: Poverty by Province, 1995196 Province Incidence of Poverty Depth of Poverty Severity of Poverty Index Contribution Index Contribution Index Contribution Poverty line=792 rupees per person per month Western 14 17 3 17 1 14 Central 28 17 6 18 2 19 Southern 27 16 6 16 2 16 N.Westem 34 18 7 17 2 16 North Central 31 8 6 7 2 7 Uva 37 11 10 13 4 15 Sabaragamuwa 32 14 7 14 2 14 Note: Poverty incidence estimates are based on DCS datasets that exclude the North and East Source: Gunewardena 2000. - 11 - 1.15 Poverty is lower and has unambiguously declined in the urban sector, where the most dynamic activities - manufacturing and services - are concentrated (Figures 1.9a, 1.9b, 1.9c).5 With the opening of the economy and better infrastructure in Colombo, manufacturing and services have expanded rapidly, at over 6 percent a year on average. 90 percent of Sri Lanka's poor, however, live in rural and estate areas and in those poverty remains high (Figures 1.1 Oa. and 1.1Ob). Individuals in households whose principal income earner was in agriculture experienced a 46 percent fall in the incidence of poverty in the 1985-90 period, but a 62 percent increase between 1990 and 1996. In 1995/96 they comprised 43 percent of the poor. Figures 1.9a, 1.9b, 1.9c: Trends in the Incidence of Poverty, by Sector, 1985/86 to 1995/96 Urban Sector Rural Sector Estate Sector 40 40 40 35 35 35 330 3 30 25 25 25 27 20 20 20 25 22 21 15 18 15 15 10 is 15 10 10 12 5 5 5 0 0 0 1985/86 1990/91 1995/96 1985/86 1990/91 1995/96 1985/86 1990/91 1995/96 Figures 1.10a. and 1.10b. Sector Contribution to Overall Poverty, 1985/86 and 1995/96 Estate Urban Estate Urban 5% 12% 4% 8% Rural Rural 83% 88% Note: Poverty incidence estimates are based on DCS datasets that exclude the North and East Sources' Graph data from: Daft and Gunewardena, 1997 for 1990/91; Gunewardena, 2000 for 1995/96, based on DCS data. 1.16 More than 90 percent of all poor people in Sri Lanka live in rural or estate areas, however, where poverty remains high. Estate households, although accounting for only 4 percent of all poor households, are among the poorest in Sri Lanka. The civil conflict and the often 5 There are no reliable regionally-disaggregated valued added statistics for Sri Lanka. A series for the period 1990-95 was produced for the 1998 UNDP Human Development Report for Sri Lanka and indicated high regional and inter-year variability in growth rates. However these statistics are not produced on a regular basis. Sector-wise and some regional disaggregation for manufacturing and agriculture are available and were used in this report. It is important that regional value added estimates be produced and disseminated in the future. - 12 - remote location of estates restrict the geographic mobility of estate workers. Language barriers (most estate workers speak only Tamil), discrimination, and, for some, lack of citizenship cards further curtail their ability to integrate themselves into the Sri Lankan economy and society by finding jobs outside the estate sector.6 These households also receive the poorest quantity and quality of publicly provided services. 1.17 More than 40 percent of principal income earners in poor households are employed in agriculture, and the poorest households rely on agriculture for more than half of their income (earned either as small-scale farmers or wage laborers) (Table 1.6). Households in which the principal income earner worked in agriculture experienced a 46 percent decline in the incidence of poverty in the 1985-90 period but a 62 percent increase between 1990 and 1996. The performance of the agricultural sector thus has a strong impact on poverty. Agriculture has expanded slowly in Sri Lanka, and performance has been volatile (Figures 1.11 and 1.12). Official statistics on agricultural value-added show an annual average growth rate of about 2 percent between 1990 and 1999 and just 1 percent during 1990-98. High levels of profitability were restored in the plantation/estate sector following privatization, but social and geographical constraints to the mobility of estate populations-which have reportedly tightened since the mid- 1980s, with the intensification of the conflict-prevented surplus labor.from moving into other activities. In nonplantation agriculture, low-technology, inflexible land markets, and distorted and unstable output prices have continued to keep returns to traditional agricultural crops low, contributing to the persistence of poverty in rural areas. Figures 1.11 and Figure 1.12: Growth in Agriculture, Manufacturing and Services, 1985-96 30 90 7 -30 Aru Agricult 10.0 Sre S10 --Pdy 5.01 0 -Rubber 0.0' ~ I.90 I~9' iI~9'~ I~94~T9 ~I'b~ I 1 8598 1988 198919901991 1992199 9495 6The closed nature of the labor market in the estate sector is discussed in Korale (1988). Korale also describes government programs aimed at transferring people from densely populated wet zones to more sparsely populated dry regions through agricultural development schemes, colonization, and village expansion schemes. Agricultural productivity, defined as agricultural value-added per agricultural worker, is low and has increased little over the past decade in Sri Lanka (see Annex A, Figures A2 and A3). - 13 - Table 1.6: Contribution to Poverty by Occupation of Household Head, 1985/86 to 1995/96 (percent) Poverty line = Rs. 792 per person Poverty line = Rs.950 per person per month per month Occupation of household 1985'86 1990/91 1995196 1985'86 1990191 1995196 head Agriculture Headcount 46 38 41 45 39 40 Poverty gap 46 35 41 45 37 41 Squared poverty gap 46 33 41 46 35 41 Mining and quarrying Headcount 1 1 2 1 1 2 Povertygap 1 1 2 1 1 2 Squared poverty gap 1 1 2 1 1 2 Manufacturing Headcount 7 6 6 8 7 7 Povertygap 7 7 6 7 7 6 Squared poverty gap 6 7 5 7 7 6 Construction Headcount 5 4 5 5 4 5 Poverty gap 4 4 4 5 4 5 Squared poverty gap 4 3 4 4 4 4 Wholesale and retail trade Headcount 5 7 6 5 7 7 Povertygap 4 7 6 4 7 6 Squared poverty gap 3 7 6 4 7 6 Transportation Headcount 2 2 2 2 3 2 Poverty gap 2 2 2 2 2 2 Squared poverty gap 1 2 2 2 2 2 Finance Headcount 0 0 0 0 0 0 Povertygap 0 0 0 0 0 0 Squared poverty gap 0 0 0 0 0 0 Communications Headcount 4 8 6 5 7 6 Povertygap 4 8 6 4 8 6 Squared poverty gap 3 8 6 4 8 6 Unclassitied Headcount 8 10 10 7 9 9 Povertygap 9 11 11 8 10 10 Squared poverty gap 10 11 11 9 11 10 Unemployed/labor force nonparticipants Headcount 23 24 21 23 23 22 Poverty gap 24 25 22 23 24 22 Squared poverty gap 25 27 23 24 25 22 Note: Poverty incidence estimates are based on DCS datasets that exclude the North and East Source: Gunewardena 2000. -14- 1.18 Most of the poor live in provinces that are generally poor, namely, the North Western, North Central, Uva, and Sabaragamuwa provinces, where they earn a living from agriculture and wage labor, often on a day basis. Average incomes and wages in rural and estate areas are much lower than in Colombo. 1999/2000 data from the Sri Lanka Integrated Survey confirms that economic welfare, measured in terms of per capita consumption expenditure, varies sharply across consumption quintiles and provinces (see Annex A, Table A5). The poor, defined as those with per capita expenditure falling in the lowest quintile of the distribution, consumed less than half the average for the country and only two-thirds the level enjoyed by the next highest quintile. Overall income inequality in Sri Lanka is relatively low (Table 1.7). Table 1.7: Gini Coefficient of Consumption Inequality, all Provinces, 1995/96 and 199912000 Gini coefficient of consumption inequality Province 1995/96 1999/2000 Western .339 .358 Central .297 .349 Southern .294 .325 North Western .268 .292 North Central .284 .264 Uva .287 .297 Sabaragamuwa .280 .331 North East - .235 Sri Lanka .326 .35 Notes: Gini coefficients not strictly comparable due to differences in the definition of expenditures in the 1995/96 and 1999/2000 surveys. - Not available Sources: 1995/96:Gunewardena 2000, based on 1995/96 Household Income and Expenditure Survey; 1999/2000: World Bank, based on 1999/2000 Sri Lanka Integrated Survey. 1.19 Ethnicity. With the exception of Indian Tamils (often referred to as estate Tamils), the majority of whom are poor, the incidence of poverty varies little across ethnic groups. Although aggregate consumption inequality in Sri Lanka is not high (Gini coefficient of .32), inequality across provinces and particularly across ethnic groups is low (Tables 1.8 and 1.9). Aggregate consumption is lowest among Tamils living on estates, where inequality is low, indicating that the majority of households are homogeneously poor. - 15 - Table 1.8. Contribution to Poverty by Ethnicity of Household Head, 1985/86 to 1995/96 (percent) Poverty line = Rs. 792 per person Poverty line = Rs. 950 per person per month per month Ethnicity of household 1985/86 1990/91 1995/96 1985/86 1990/91. 1995/96 head Sinhalese Headcount 87 84 84 86 83 84 Poverty gap 89 83 83 88 83 84 Squared poverty gap 89 82 83 88 83 83 Lankan Tamil Headcount 4 5 5 4 5 5 Poverty gap 3 6 6 3 5 5 Squared poverty gap 3 7 6 3 6 6 Indian Tamil Headcount 3 2 4 3 3 5 Poverty gap 2 2 3 3 2 4 Squared poverty gap 2 1 2 2 2 3 Lankan Moors Headcount 6 8 7 6 8 6 Poverty gap 6 8 8 6 8 7 Squared poverty gap 6 9 8 6 8 8 Note: Poverty incidence estimates are based on DCS datasets that exclude the North and East. Source: Gunewardena (2000). Table 1.9: Gini Coefficient of Consumption Inequality in Sri Lanka, by Ethnicity, 1995/96 and 199912000 Ethnic group 1995/96 1999/2000 Sinhalese .327 .325 Lankan Tamil .344 .265 Indian Tamil .219 .249 Moor .312 .291 Source: 1995/96:Gunewardena 2000, based on 1995/96 Household Income and Expenditure Survey; 1999/2000: World Bank, based on 1999/2000 Sri Lanka Integrated Survey. 1.20 Gender and Unemployment. Individuals in households headed by women are about as likely to be poor as those living in households headed by men, regardless of the level at which the poverty line is set (Table 1.10). -16- Table 1.10: Poverty by Gender of Household Head, 1985/86 to 1995/96 (percent) Poverty line = Rs792 per person Poverty line = Rs.950 per person per month per month Gender 1985/86 1990/91 1995/96 1985186 1990/91 1995/96 Male Headcount 31 20 26 44 33 39 Poverty gap 7 4 5 12 8 10 Squared poverty gap 3 1 2 5 3 4 Female Headcount 32 21 24 45 33 38 Poverty gap 8 5 5 13 8 10 Squared poverty gap 3 1 2 5 3 3 Note: Poverty incidence estimates are based on DCS datasets that exclude the North and East. Source: Gunewardena 2000. 1.21 Unemployment does not contribute significantly to poverty in Sri Lanka. As previous research has indicated, few of the poor are unemployed, with households whose head was unemployed making up just 2 percent of poor households in 1995/96 (Table 1. 11). Results from the Sri Lanka Integrated Survey indicate similar figures for households in the lowest expenditure quintile. Table 1.11: Contribution to Poverty by Employment of Principal Income Earner, 1985/86 to 1995/96 (percent) Poverty line = Rs. 792 per person Poverty line = Rs. 950 per person per month per month Employment status 1985/86 1990/91 1995196 1985/86 1990/91 1995/96 Employed Headcount 91 92 95 91 93 95 Poverty gap 91 91 95 91 92 95 Squared poverty gap 90 91 94 90 92 95 Unemployed Headcount 2 1 1 2 1 1 Poverty gap 2 1 1 2 1 1 Squared poverty gap 2 1 1 2 1 1 Labor force nonparticipants Headcount 7 7 5 7 6 5 Poverty gap 7 8 5 7 7 5 Squared poverty gap 8 9 5 8 8 5 Note: Poverty incidence estimates are based on DCS datasets that exclude the North and East Source: Gunewardena 2000. 1.22 Lack of Access to Infrastructure. The poor, especially those living in rural areas and estates, have significantly less access to clean drinking water, sanitation, electricity, safe cooking fuel, and communications than wealthier households. Only a fraction of poor families in 1999/2000 had access to safe cooking fuel (2 percent), electricity (38 percent), safe sanitation (55 percent) and clean drinking water (61 percent) (Table 1.12). Particularly striking is the fact that only an estimated 60 percent of the SLIS sample had access to electricity. Limited access to these resources reduces their living standards both directly and indirectly, by limiting economic growth in poor regions. - 17 - Table 1.12: Average Consumption and Access to Basic Infrastructure Services by Consumption Quintile, Sector, and Ethnicity, 199912000 (percent, except where otherwise indicated) Monthly average Access to consumption per capita Safe Latrine Safe Safe cooking Electricity (SL Rupees) drinking sanitation fuel Item water Consumption quintile Poorest 821 61 84 55 2 38 Second 1,211 74 85 67 5 49 Third 1,537 78 89 75 8 60 Fourth 1,986 82 90 82 22 67 Richest 3,860 89 94 89 51 82 Sector Urban 2,809 97 94 91 51 84 Rural 1,816 74 88 72 14 57 Estate 1,449 72 76 60 3 43 Ethnicity Sinhalese 1,925 76 92 75 19 61 Tamil 1,920 75 69 63 11 45 Estate Tamil 1,396 76 81 73 4 43 Moor 2,061 91 81 79 23 78 Note: A household has access to 'safe drinking water if it obtains its drinking water from protected well, public tap, tube well, tap within unit, and tap outside unit). A household has access to 'safe sanitation' if the type of latrine it uses is either water seal or flush toilet A household has access to 'safe cooking fuer if it uses either gas or electricity for cooking. Source: World Bank, based on 1999/2000 Sri Lanka Integrated Survey. 1.23 In contrast, access to community level economic and social facilities, such as main roads, post offices, banks, markets, bus stops, and local government offices, measured in terms of distance and travel time, is similar for poor and non-poor households (Table 1.13). On average, people in Sri Lanka can reach these facilities in 15-30 minutes, with little variation across consumption quintiles. - 18 - Table 1.13: Access to Economic and Social Facilities by Consumption Quintile, Sector, and Ethnicity, 199912000 (travel time in minutes) Grama Item Main road Post office Bank Market Bus stop Sevaka Consumption quintile Poorest 22 22 31 30 28 26 Second 20 22 31 29 24 20 Third 21 20 30 28 25 21 Fourth 20 19 27 26 23 22 Richest 16 15 22 22 20 19 Sector Urban 11 10 12 13 13 8 Rural 19 18 28 26 24 19 Estate 38 42 50 53 44 32 Ethnicity Sinhalese 20 20 28 26 24 20 Tamil 21 22 32 28 22 25 Estate Tamil 36 34 50 63 50 45 Moor 13 9 18 21 23 3 Source: World Bank, based on Sri Lanka Integrated Survey 199912000. 1.24 Outside of the conflict areas, economic welfare and living Box t2: Why Dothe Poor Lac -Assets? conditions are worst for households .. living on estates, most of whom are Lacking assets is both'a cause.and an outcome of poverty Poor Tamils. Estate Tamils are health. deficient! kills: limited access to basic services, and the humiliation of soc4al'exclusion refldct deprivations in personal, descendents of workers brought over public, and social assets Humii, physical, and natural assets from South India by British also lie at the core .whether an individual, househ6ld, or group plantation companies during the lives in poverty. or .escapes it. These assets interact with market nineteenth and early twentieth and social opportuhiti6s to genertte. income, a better quality of centuries to work on the hill country life,. and a' sense of py hlogicil-well-being Assets are also . plantations. They have the lowest central to coping with shoks and reducing the vulnerability that consumption levels in Sri Lanka and is a constant feature of.poverty. the worst basic living conditions. Per capita consumption in the estate Poor people lack assets because,they-live.in poor countries or in sector is only about half that in the poor regions within communities. Inequality in the distuibution 6f wealth and the benefits of Fliblic'action also- depnve the poor of assets In Sri Lanka, for example:, 82 percent of the wealthiest that in the rural sector. Access to households and just 38 percent of the poorest households have safe cooking fuel (3 percent), access to electricity. electricity (43 percent), and sanitation facilities (60 percent) are Source: World Bank, 2000-2001 World Development Repon and World well below levels in other sectors. Bank estimates 6ased 6n 1999/2000 Sri Lanka Integratd Survey Access to public facilities is also much more limited, with travel time to main roads, banks, post offices and markets about twice that in rural areas (Table 1.13). Limited access to public facilities reflects the remoteness of communities located deep inside large estates and the absence of regular bus transportation. - 19 - 1.25 Economic welfare is low and living conditions difficult in other rural areas. Rural consumption is about two-thirds urban consumption, and access to clean sanitation, good drinking water, electricity, and safe cooking fuel is much more limited. Travel time to facilities such as main roads, banks, post offices, and markets is about twice that of urban areas, due chiefly to longer travel distances and more limited transportation. 1.26 Education and Health Status of the Poor. Sri Lanka's achievements in education and health have been remarkable. Less than 10 percent of the population is illiterate, and 100 percent of children are enrolled in school. These strong social indicators have improved the standard of living, even of households below the poverty line. While poverty is highest among individuals whose principal income earner has no schooling, such households make up a small percentage of the poor, a percentage only slightly higher than their share in ,the total population (5 percent) (Gunewardena 2000). Less than 12 percent of the poor have no schooling (Tables 1.14 and 1.15), a percentage similar to the literacy rates in the general population. Table 1.14: Contribution to Poverty by Education of Household Head, 1995196(percent) Poverty line = Rs792 per person Poverty line = Rs.950 per per month person per month Level of education No schooling Headcount 12 11 Poverty gap 13 12 Squared poverty gap 14 13 Primary school Headcount 43 41 Poverty gap 44 43 Squared poverty gap 45 44 Lower secondary school Headcount 25 26 Poverty gap 25 25 Squared poverty gap 25 25 Upper secondary school Headcount 13 14 Poverty gap 12 13 Squared poverty gap 11 12 GCE O/L Headcount 6 6 Poverty gap 5 6 Squared poverty gap 5 5 GCE A/L Headcount 1 1 Poverty gap 1 1 Squared poverty gap 1 1 Graduate and above Headcount 0 0 Poverty gap 0 0 Squared poverty gap 0 0 Note: Poverty incidence estimates are based on DCS datasets that exclude the North and East Source: Gunewardena 2000. - 20 - Table 1.15: Contribution to Poverty by Literacy Rate of Household Head, 1985-96(percent) Poverty line = Rs792 per person Poverty line = Rs950 per person per month per month Literacy status 1985186 1990/91 1995/96 1985/86 1990/91 1995196 Literate Headcount 77 76 81 78 78 81 Poverty gap 74 74 .79 76 76 80 Squared poverty gap 73 74 78 74 75 79 Illiterate Headcount 23 24 19 22 22 19 Poverty gap 26 26 21 24 24 20 Squared poverty gap 27 26 22 26 25 21 Note: Poverty incidence estimates are based on DCS datasets that exclude the North and East. Source: Gunewardena 2000. 1.27 Nearly half of the poor live in households whose principal income earner has some secondary education. These low returns to education of the poor may reflect the low quality of their education, the fact that they are unable to make the best use of their education by moving to activities or areas where returns are higher, or low returns to education in Sri Lanka (as compared to other countries). Although education is free, the quality of education provided to the poor is inferior to that provided to richer households, limiting the ability of children of poor parents to take advantage of the growing economy. Province-level primary learning outcomes do suggest that here may be significant differences in the quality of education acquired by members of poor and nonpoor households (Figure 1. 13). Although this effect is not captured by the existing household surveys, it is possible to identify measurable proxies for education quality, such as class size or language of instruction. Class sizes in rural areas and estates are considerably higher than in urban areas. As for the language of instruction, a primary education in Tamil, for instance, is likely to yield lower market returns than one in Sinhalese or English. A basic analysis of district-level poverty incidence and English literacy reveals a strong correlation (Table 1.16). Further research on the relationship between language of instruction and returns to education could help clarify these effects further. -21- Figure 1.13: Proportion of Students Mastering Basic Language and Numerical Skills 45 40 35 a30 - 1 - |11iLiteracy ElNumeracy Source: National Institute of Education 1995. Table 1.16: Poverty incidence, literacy, and language Province District Poverty incidence Literate in Illiterate 1995/96 Sinhala Tamil English Western Colombo 10 82 18 39 6 Gampaha 11 93 5 25 5 Kalutara 26 86 11 22 8 Central Kandy 30 76 22 23 10 Matale 35 75 19 16 14 Nuwara Eliya 21 40 43 12 22 Southern Galle 25 89 5 22 9 Matara 28 87 4 19 11 Hambantohta 27 87 1 12 13 North Western Kurunegala 34 89 7 15 9 Puttalam 33 80 18 16 7 North Central Anuradhapura 33 87 6 11 10 Polonnaruwa 27 87 5 16 10 Uva Badulla 30 65 22 13 18 Moneragala 49 82 4 10 16 Sabaragamuwa Ratnapura 37 81 10 16 13 Kegalle 25 84 12 17 9 Note: Correlation coefficient between district-level poverty incidence and literacy in English is -0.713. Poverty incidence estimates are based on DCS datasets that exclude the North and East Sources Poverty estimates: Gunewardena 2000; self-assessed literacy rates by language from Demographic Survey 1994. - 22 - 1.28 Similarly to education outcomes,'there seems to be no consistent differences in the long term (stunting) or short-term (wasting) nutritional indicators of children in Sri Lanka across income groups or across provinces according to data from the Sri Lanka Integrated Survey. Nutritional indicators of children across expenditure quintiles and provinces vary little (see Tables 1.17 through 1.19). There are significant differences, however, across ethnic groups and sectors of residence, with nutritional outcomes of estate children (most of whom are ethnically Indian Tamils) being inferior to those of other ethnic groups and population in the rest of the country. Table 1.17: Health Indicators by Expenditure Quintile, 199912000 Quintile of per-capita Proportion of children Proportion of children Proportion of children wasted expenditure underweight* stunted* distribution 1 (lowest) 25 29 30 2 24 29 27 3 23 32 25 4 19 25 18 5 (highest) 27 27 32 Total 24 29 27 * underweight is defined as weight for age below 2 standard deviations from the age and gender specific mean "stunting is defined as height for age below 2 standard deviations from the age and gender specific mean **wasting is defined as weight for height below 2 standard deviations from the age and gender specific mean Source: Worid.Bank, based on 1999/2000 Sr Lanka Integrated Survey; calculations from three quarters. Table 1.18: Health Indicators by Sector, 199912000 Sector Proportion of children Proportion of children Proportion of children underweight* stunted" wasted Urban 16 26 20 Rural 24 29 27 Estate 42 37 37 Total 24 29 27 * underweight is defined as weight for age below 2 standard deviations from the age and gender specific mean "stunting is defined as height for age below 2 standard deviations from the age and gender specific mean "'wasting is defined as weight for height below 2 standard deviations from the age and gender specific mean Source: World Bank, based on 1999/2000 Sri Lanka Integrated Survey; calculations from three quarters. Table 1.19: Health Indicators by Province, 1999/2000 Province Proportion of children Proportion of children Proportion of children wasted underweight* stunted* Western 20 27 25 Central 21 25 33 Southern 26 24 28 N.Eastern 20 30 20 N.Western 25 36 21 North Central 36 28 33 Uva 30 34 27 Sabaragamuwa 27 33 32 Total 24 29 27 * underweight is defined as weight for age below 2 standard deviations from the age and gender specific mean "stunting is defined as height for age below 2 standard deviations from the age and gender specific mean *"wasting is defined as weight for height below 2 standard deviations from the age and gender specific mean Source: World Bank, based on 1999/2000 Sri Lanka Integrated Survey; calculations from three quarters. - 23 - Poverty and Hardship in the Northeast 1.29 Part of Sri Lanka's difficulties in raising incomesand reducing poverty can be traced to the civil war, in which an estimated eBox 1.3: Ethnic Diversity and Violence in Sri Lanka 60,000 people have lost their lives. i i Sri Lanka has been immersed in war Sri Lanka's ethnically diverse population com-irise an ethnic for the last 17 years. The majority, theSinhAlese, and three rinty groups, the Sfi humanitarian, social and economic Lankan Tamils, the Indian Tamils, and the Moors The. main impact of the war is felt most distinguishing characteristic of these groups is language. The directly by populations in the North mostly Buddhist Sinhalese. who make up 74 percent of the and East and the areas bordering it. population. speak a language that belongs to the Indo-Aryan But the effects of the conflict have familN. The largely Hindu Sri Lankan Tamil minority, which far reaching economic, social, and represents about 12 percent of the population. speaks Tamil, a Dravidian language. The Sri Lankan Tamils have long psychological repercussions, which claimed that the, Sinhalese discriminate against them. extend beyond the theatre of battle. Another 6 percent of the population are Indian Tamils, who Some of the effects of the conflict live apart from the rest of Sri Lankan society, largely on include: loss of civilian lives and estates or plantations. -The Muslims, who represent around 7 psychological trauma, damage to percent of the population, are predominatly traders and infrastructure and homes, maintain cordial ties with both the Sinhalese and Tamils. displacement, restricted mobility in some areas in the country, Simmering ethnic tensions intensified in 1956, after Sinhala disruption of local economies, ~ was declared the onl official language, and in 1971, when the rural Sinhalese youth-based mo%ement. the Janatha disruption of community and %'imukthi Peamuna (the . Peoples' Liberation Front), institutional networks, disruption of espousing an. ideology combining leftist populism and children's education, high Sinhalese nationalism, organized an armed uprising. The dependence on relief, deterioration uprising was violently crushed by the government Violence in the health status of the erupted again in 1983, when anti-Tamil violence in the south population, and widespread led to a mass exodus of Tamils and to support for an armed vulnerability and insecurity among struggle for a separate Tamil state (comprising the Northern the population. At the root of the and Eastern provinces of the country), led by the Liberarion problem has been Sri Lanka's Tigers of Tamil Eelam (LTTE). inability to forge a multicultural _ identity (Boxes 1.3, 1.4 and 1.5). 1.30 Expenditures on the war effort, which have consumed about 5 percent of GDP in recent years, crowd out a vast range of pro-growth and pro-poor public services. The instability brought about by the war also reduces investment and job creation. And the protracted conflict fuels ethnic polarization and discrimination and desensitizes communities to the effects of violence. 1.31 'Direct costs of the war include military expenditures borne by the government (and the. Liberation Tigers of Tamil Eelam) and the costs associated with the destruction of or damage to physical and social infrastructure. The government's military expenditures increased from 1.3 percent of GDP in 1982, before the outbreak of war, to about 5-6 percent of GDP in 2000. Arunatilake and others (2000) estimated .the economic cost of government's expenditures at about 41 percent and those of the Liberation Tigers of Tamil Eelam at 4 percent of Sri Lanka's 1996 GDP. Adding the costs of providing for refugees and the costs of lost assets yields*a total direct opportunity cost of 61 percent of Sri Lanka's 1996 GDP. The indirect costs of the 'war are at least as high. These costs represent the income lost as a result of forgone domestic investment, reduced tourism, the decline in foreign direct investment, the human capital loss associated with death, injury, and displacement. - 24 - Box 1.4: Social Fragmentation and Conflict Group differentiation by such characteristics as ethnicity, race, religion, and- language can sometimes result in. social fragmentation, with groups perceiving themselves as having distinct interests even though they may have similar socioeconomic status. Ethnicity - a multidimensioral phenomenon and a controversial notion - is based on perceived cultural differences between groups in a society, differences that form-a powerful source of identity and a base for political mobilization (Horowitz, 1999). Common ethnic affiliations can be a basis for bonding social capital, providing community members with a range of benefits. (credit, employment, etc) while imposing significant obligations and commitments (financial7 support, conformity). Membership in -an. ethnic community can, also generate negative externalities, as with conflict between ethnic groups. Such divisions can be obstacles to collective action. Ethnicity can become a basis for competition for political power and for access to material resources (Turton, 1997). Unless institutions-of the state and civil-society offer forums for mediating intergroup rivalries and forging cross-cutting ties -among diverse ethnic groups, these ethnic cleavages can lead to conflicts, tearing a society and economy. apart, leaving everyone vulnerable to poverty. Ethnic cleavages affect development outcomes in many ways. They,influence the internal organization ot govemment. and the allocation of public spending, leading to unequal distribution of public goods and -services. They encourage rent seeking, reducing the efficiency of public spending (Alesina et al. 1998). Further economic distortions enter-when ,powerful ethnic groups-use their political power to increase their incomes relative to those of others. The-extent to which social fragmentation leads to conflict depends largely on administrative and political institutions. To- create a functioning society, a -whole,range of social and political institutions must work together. By contrast,. breakdowns in:govemance and in the delivery of public goods and related social services create conditions for social unrest and conflict-as do breakdowns in the institutions of conflict mediation, such as represeritative politics and the rule of law; Sources: Alesina 1998, Horowitz 1999, Turton 1997, and World Bank 2000 1.32 The human and economic costs of the war are felt most directly by populations in the North East. According to Ministry of Planning estimates, the size of the overall economy of the Northern Province shrank from US$ 350 to US$ 250 million between only 1990 and 1995, corresponding to a negative annual average GDP growth of 6.2 percent per year. Poverty estimates are unavailable for the North East, where conditions make it too difficult to administer a representative household survey. The 1999/2000 Sri Lanka Integrated Survey (SLIS) attempted to cover the North East, to the extent it was possible to do so without jeopardizing the lives of interviewers. However, the difficulties encountered in administering a household survey in war areas were many. The findings based on the SLIS for the North East will, therefore, not fully reflect the gravity of the situation in the area due to the following reasons: (i) sampling in the Northeast is not random (as it was in the rest of the country) due to the very difficult and fluid security situation and the frequent displacement of entire villages covered during the survey period; (ii) the survey was unable to cover uncleared areas (areas under LTTE control) due to security reasons -- and there is reason to believe that the situation in the uncleared areas may be worse than in the rest of the country; and (iii) the survey results do not capture the situation in refugee camps. 1.33 Taking into account these caveats, data from the Sri Lanka Integrated Survey suggest that household incomes in this region would be among the lowest in Sri Lanka were it not for the safety net provided by remittances from family members living abroad. However, contrary to expectations, SLIS data has not shown income or consumption levels in the North East to be lower than in the rest of the country. Monthly average consumption per capita, adjusted by a regional prices index to take into account the higher price level in the North and East, is the second highest in this area, next to consumption in the Western Province (Annex A, Table A5). Foreign remittances and labor income from abroad are responsible for ensuring minimum consumption levels. - 25 - 1.34 Other sources reveal high levels of wasting and stunting among children and very poor nutritional status of women, especially among Tamils living in uncleared areas. Although the civil conflict in Sri Lanka is confined to the North East, violence has become an islandwide and endemic feature of Sri Lankan society. Violence has become,a means to attain legitimacy, wealth, and protection (Goodhand 2000). The much higher incidence of violent crime in Sri Lanka compared with other South Asian countries may reflect the impact of the conflict. So,Et ie r natnd Hfne 1its mosn 1xtremeforan ude ndon of97 Lecnm depivaion can desed Bnk. 200. 1.35ic Thi poonfic bensifediproportiont hare of the costury of the war.ernced wofith shfeeromwr oppotuenintios to earnfalitvwin ttes ural o nfserv in bthue a d csonsequie of oorecooi conflict im ar most of the con i loss of lifeainury datiobstan t m a. t Houeholds in e Norut of ph ysiclages oRderingcia apithe ostrsverely affecal ind humAn thaility dirton of are liveih ofod economuaranoal, 19 . irtuall m resoues n theebaeasha expeditur edaa from productiv of producaivityehrouighos c ivites; m agration or igysplaementworks asna raers of he d ation aboad. Thesercotr at as coute foear. 1.36 Themog theawrsaffectedporoate sare househeostso thhe been diaced, soetes reportutie s a earn a living, th raso the home flles. Anb s odted the conflict and bear most of the burden in terms of loss of life, injury, dislocation, and trauma. Households in the North East and villages bordering it are the most severely affected in terms of their ability to obtain a secure livelihood (Shanmugaratnam, 1999). Virtually every household in these areas has experienced the impact of the war, through loss of life, assets, or displacement or as a result of the deterioration of infrastructure that has resulted from the war. 1.36 Among the worst-affected groups are households that have been displaced, sometimes repeatedly, as a result of conflict in or threats to their home villages. A survey conducted in the eastern area of Trincomalee indicated that 40 percent of Muslim households and 27 percent of Tamil households had been displaced and that almost two-thirds of household heads in uncleared areas were unable to read and Write.8 Displaced families have lost productive assets (agricultural equipment, livestock, shops, mills), including in some cases lands they had cultivated before being displaced. Most of the displaced remain in refugee camps, often for years (Box 1.6). Those who eventually return to their home villages sometimes find that their lands have been appropriated by other households. Reinhard and Kraemer (1999). Baseline Survey on Health and Nutrition. Ministry of Plan Implementation and Parliamentary Affairs, Integrated Food Security Programme, Trincomalee. - 26 - Box 1.6 The Sithamparapuram Welfare Center Sithamparapuram Welfar&CeCteris the largest welfare center in the Vavuniya District This was established in, 1992 to accommodate the flow of displaed people from the war-affected areas including cleated and uncleared areas of the Vavuniya district; such as Podivikulamand Cheddikulam, Mullativu, Kiinochchi, Manner and Jaffna'- isitricts. As of April 1999, the total population i this welfare center was 7,341 persons. Mbstof the occupants in. Sithamparapuram are poor, They have lost whatever modest assets they had due to displacement or attacks. Their more affluent relatives and neighbors had managed to move to Colombo, Vavuniya, Trincomalee and other urban centers outside of the war ione including foreign destinations as they had enough cash to travel to such locAtions; secure passes and other identity documents necessary for such a move. and settle in such new locations. In effect, the poorest of displaced people are the ones who remain in welfare centers for long periods of timeL ,;60ly a -small -proportion of .ocupants in the: Sithampar autam welfare center live in permanent buildings and temporary.sheds, while the majority live In thatched huts. Most of these permanent buildings were either original -"school buildings or those built by the administration for purposes such as storage and subseuenfly been distributed among the newcorners to the center. A vast majority of the people live in thatched hutments with mud walls and cadjan roofs. While, temporary sheds are mainly occupied by a floating population from nearbyvillages who move into the center when there is potential danger from fighting and return to their villages once danger subsides Sov:Silva 2000.1 1.37 Death, injury, and displacement statistics can only approximate the extent of human suffering in the North East. Individuals in the North East are far more likely to suffer personal injury than households anywhere else in the country, in particular in the Western Province. One out of every twelve households in the North East reported that a family member was killed as a result of the civil conflict (Table 1.20). Among the poorest households, in the North East the rate is even higher with one out of every seven poor households reporting a member killed. Table 1.20: Households Reported Having a Member Killed as a Result of the Civil Conflict, by expenditure quintile and province (percentage) 1 (lowest) 2 3 4 5(highest) Total Western Province 0 0 0 1 1 0 North and East 15 10 6. 8 7 8 Rest of Sri Lanka 1 1 1 1 1 1 Souce: 1999-200 Sri Lanka Integrated Survey calculatons from three quarters. 1.38 Displacement as a direct consequence of the war is a striking feature of life in the North East. Data from the 1999/2000 Sri Lanka Integrated Survey indicates that one of the most common reasons for changing the place of residence in the North East is the war, and around 10 percent of all households that experienced an episode of mobility in the last 20 years moved . because of the war. Not only the households iresiding in the North East are more likely to change their residence, they are more likely than households in other parts of the country to move more often. Among those residents of the North East province who have moved in the past 20 years, 17 percent moved 3 or more times. In comparison, respectively 4 and 2 percent of households in similar circumstances residing in the Western and other provinces of Sri Lanka had moved more than 3 times. - 27 - 1.39 Nearly all households in the North East (97 percent) that moved due to the war suffered loss of property. Compensation for these losses was rarely paid, and when it was paid well-off households were more likely to receive compensation for their losses than the poor households: 7 percent of all households who suffered income losses reported receiving compensation, while none of the households in the two lowest quintiles did. Table 1.21: Reasons for Household Move by Area of Residence (percentage) Reasons: Work Available Marriage Other War Resettlement Other (as reported by the land family household heads who were reasons not born in the place of current residence) Western 26 13 41 7 1 3 9 North and East 10 56 17 3 12 2 1 Rest of Sri Lanka 18 30 31 9 1 4 6 -28- Chapter 2 Anti-Poverty Programs and Safety Nets 2.1 Sri Lanka's success in improving non-income dimensions of poverty is due to a great extent to successive government's efforts to provide universal and free education and health care to the population. The efforts of successive governments, involving expenditures of between 4 and 5 percent of GDP (4 percent in 1997) appear to have paid off, not only for the wealthy, but also for the poor (Table 2.1). The poor fare considerably better in some key non-income dimensions of poverty in Sri Lanka than in most other developing countries. Around 90 percent of the poor have completed at least primary education. Although health indicators for the poor are not readily available, indicators such as infant and maternal mortality rates and life expectancy at birth above 70 years of age, are comparable to those of middle income countries. Although there is a perception that the quality of public education and health services have been declining, a rigorous analysis of which is beyond the scope of this study, Sri Lanka's achievements are nonetheless remarkable. 2.2 As discussed in Chapter 1 of this report, the effectiveness of Sri Lanka's strategy in reducing income poverty and vulnerability has been far less remarkable. A key element of this strategy has been the transfer of public resources to households through untargeted subsidies and more recently large welfare programs. Welfare programs in Sri Lanka have traditionally covered a large percentage of the population, including those in upper income strata. This may have contributed to larger public support for these programs.9 As a result, government expenditures in social security and welfare programs have been traditionally high, amounting to nearly the education and health expenditures combined in 1997 (Table 2.1). Table 2.1: Social expenditures, Sri Lanka and Comparator Countries % of government expenditure % of GDP Education Health Social Total Education Health Social Total Security and Security and Welfare Welfare US$800-1,500 per capita income countries (1999) Sri Lanka (1997) 9.8 5.3 14.0 29.1 2.5 1.4 3.6 7.5 Egypt (1997) 14.8 3.3 0.5 18.5 4.5 1.0 0.1 5.7 Philippines (1997) 20.3 3.2 2.5 26 3.9 0.6 0.5 5.0 Bolivia (1998) 19.6 3.4 25.9 48.9 4.3 0.7 5.7 10.8 Bulgaria (1998) 29.3 25.8 9.2 64.3 2.3 2.0 0.7 5.0 US$2,000-30,000 per capita income countries (1999) Thailand 1998) 23.1 9.2 4.1 36.4 4.3 1.7 0.8 6.7 Malaysia (1997) 22.8 6.3 7.2 36.3 4.5 1.2 1.4 7.1 Singapore 1997) 18.8 6.7 1.8 27.3 3.2 1.1 0.3 4.5 Source: IMF 1999. 9 For political economy arguments for nontargeted transfers, see Gelbach and Pritchett (1997). It is debatable whether there is a need in Sri Lanka to gain popular support for poverty programs, given the country's socialist tradition and concern for social welfare. - 29 - 2.3 In general, governments maintain safety nets to prevent very low standards of living by means of reducing and mitigating downside socio-economic risk or coping with its occurrences. 1o The poor in Sri Lanka have access to a variety of mechanisms to cope with shocks, including government, private and non-governmental arrangements. There are four groups of vulnerable people in Sri Lanka who could be covered by government safety nets, including: (i) a core group of people that cannot take advantage of existing opportunities (handicapped and orphans, for instance); (ii) people "trapped" in pockets of poverty because they face constraints to mobility and integration, such as the estate or Indian Tamils; (iii) people affected directly by the war, through death of breadwinner, injury, and displacement; (iv) people affected by catastrophic risk, such as drought or floods. Instead of focusing only on these vulnerable groups, government support programs in Sri Lanka have provided income support to a larger share of the population over extended period of time. The Government Safety Net in the Late 1990s 2.4 The Sri Lankan Government currently has several social assistance programs and subsidy schemes. Samurdhi became the largest government social assistance program in 1995, when it replaced the Janasavyia program, and claims about $170 million per year or 1 percent of GDP (Table 2.2)." By 1999, two million families-about 50 percent of all families in the country- were receiving benefits from Samurdhi.12 Assistance to the displaced and war-affected population was the second-largest item in the social transfer budget, claiming $63 million-0.3 percent of GDP. There are several other programs with implicit or explicit safety net functions. Nutrition intervention programs, Public Assistance, free textbooks and school uniforms, and a school transport subsidy continued to be provided, as did producer and import subsidies intended to 1o Policies promoting growth and increasing access to health and education services improve the average well being of the poor; policies that create opportunities for households to manage risk better, thus reducing the impact of negative shocks, fall under the purview of social protection. It is important to recognize, therefore, that since poverty and vulnerability are closely linked and reinforce each other, the policy responses to them overlap considerably and also have important synergies (Risk Management in South Asia, 2001). 1 An act of parliament i 1995 stipulated that the main functions of the Samurdhi National Program were "to improve the economic and social conditions of youth, women and disadvantaged groups of the society by (a) broadening their opportunities for income enhancement and employment; (b) integrating them into economic and social development activities; (c) linking family level economic activities with community development projects at village, district, divisional and provincial levels; (d) mobilizing their participation in the planning and management of projects and schemes for their upliftment; (e) fostering cooperation among them, promoting savings amongst them and assisting them to obtain credit facilities; (f) facilitating the delivery of inputs and services of government departments, public corporations, local authorities, private sector organizations and nongovernmental organizations to beneficiaries of the program, and to implement the program so formulated and other programs of the government for poverty alleviation." In "Samurdhi National Programme for Poverty Alleviation 1999" the Samurdhi Authority of Sri Lanka restated that "Samurdhi is the national programme launched by the present government for poverty alleviation. The main objectives of this program are to provide a solution for the problem of unemployment which is the most pressing problem among the youth, eradication of poverty from the country at large, and obtaining the direct participation of the youth in development." 12 The 1999 Central Bank of Sri Lanka Annual Report indicates that Samurdhi covers around 50 percent of all families in Sri Lanka, whereas the 1999-2000 Sri Lanka Integrated Survey results show a slightly lower percentage Samurdhi covering 40 percent of all households. The discrepancy may be due to the fact that the SLIS covers the Northeast, where Samurdhi is less active, or to definitional issues. -30- increase household incomes, including a fertilizer subsidy ($25 million in 1999). In addition, programs with transfer elements, such as housing development, IRDP, and microfinance programs, as well as various schemes to create jobs for unemployed educated youth were administered. Table 2.2: Social Transfers and Social Expenditure, 1999 Percent of Millions of Percent of total social Sri Lankan rupees US$ millions social transfers expenditures Social transfers Samurdhi total 10,007 170 56 14 Of which Samurdhi food stamps 8,191 139 46 11 Refugees (dry rations, welfare camps, Unified Assistance Scheme) 3,700 63 21 5 Of which dry rations 2,661 45 15 4 Public Assistance 2,149 36 12 3 Poor relief nutrition program 97 2 1 0.1 Triposha 320 5 .2 0.4 Fertilizer subsidy 1,451 25 8 2 Subtotal 17,724 300 100 24.5 Other social subsidies School uniforms 500 8 0.7 Textbooks 700 12 1 Season tickets 300 5 0.4 Other social expenditures Pensions 20,723 351 28 Education 22,231 376 31 Health 10,651 180 15 Total 72,829 1,234 100 Sources: Central Bank of Sri Lanka, Ministry of Plan implementation, IMF. Samurdhi: An Evaluation 2.5 Samurdhi's effectiveness in reaching the poor is weak. Estimates from the 1999-2000 Sri Lanka Integrated Survey (SLIS) indicate that although the transfer component of Samurdhi reaches 40 percent of Sri Lankan households, it misses 36 percent of households in the poorest per capita expenditure quintile, those who would be most in need.3 More than 40 percent of Samurdhi transfers in the sample were made to the wealthiest 60 percent of the population. 2.6 More importantly, there are substantial and significant differences in the probability of being a Samurdhi recipient by ethnic status. The patterns of targeting errors are consistent with a finding of bias against ethnic minorities. Relative to Sinhalese households, Sri Lankan Tamils, Indian Tamils, and Moors are less likely to receive Samurdhi transfers, controlling for income 13 Quantitative and qualitative information from the 1999-2000 Sri Lanka Integrated Survey was used for evaluation of the distributional impact of government programs and to discern patterns of household behavior. -31 - status and other characteristics that might influence program eligibility. Estate populations, including poor households, are virtually left out. 2.7 Samurdhi is not an efficient poverty alleviation program. It is too large and unaffordable. It aims to reduce poverty, but also to create white-collar employment to unemployed youth. The targeting efficiency of the program is compromised by design flaws that allow for political and ethnic biases, thus discouraging social cohesion. In addition, the emphasis on large income support programs such as Samurdhi, have also created a culture of entitlement, which discourages individual efforts. Other government-run safety net programs-such as the Public Assistance program that provides assistance to widows, orphans, and the disabled-appear to work better, but are under-funded. Box 2.1: Welfare "The government has become the mother and father of the people." In th6 better-off areas, we found that the Samurdhi grant-is perpetuating a-dependency culture bred by the populist policies-of successive governments. Most beneficiaries now expect welfare measures as'their right. People are constatly;!ookingto the'goverhmerittoprovide their basic necessities. In the better-off villages in Haldummulla; for-example; the operation of several govemment and NGO programs has encouraged people to become lazy: they-either feel that they do not need to do anything to improve their-living standards, or they feel that they cannot do anything for themselves.-Therefore, the oerriment has to doitfor them. As the -Samurdhi'Managers in Haldummullapgutit, 4if th people were given hives4t- rear bees; they would ask the government to catch aid give them the:bees as well.! Source: Gunatilaka et al. 1997 2.8 Generating white-collar employment for program administrators is also one of the program's explicit goals; nearly 10 percent of the overall program budget is allocated to pay administrators' salaries. Development officers and animators are explicitly considered to be beneficiaries of Samurdhi program, whether or not they are poor. At the end of 1999, 21,700 young men and women were serving as development officers and animators; their number increased to 22,700 by the end of 2000. There were 1,780 Samurdhi managers in December 2000. These officials are selected from among unemployed local youths having at least an O-level education. Gunatilaka et al. (1997) report that appointments of these officials are made upon recommendations from local politicians. Samurdhi is also not a sustainable employment generation program. In addition to not being conducive to the creation of an effective public administration, the policy of hiring unemployed youth into the public sector is also unaffordable, in particular given Sri Lanka's fiscal stress. It is also not justifiable on poverty reduction grounds: as shown in Part I of this report, only 2 percent of the unemployed are poor, indicating that unemployment is not a key factor in determining poverty. 2.9 Structure and Administration. The Samurdhi program has three main components. The first, which provides a consumption grant transfer (food stamps) to eligible households, is the largest, claiming 80 percent of the program's budget. The second is a savings and credit program operated through so-called Samurdhi banks, and loans meant for entrepreneurial and business development (described later in this chapter). The third component is rehabilitation and development of community infrastructure through workfare and social or human development programs. Samurdhi also operates social insurance, funded through compulsory contributions of beneficiaries of the highest consumption grants. 2.10 The Samurdhi National Program of social assistance was launched by an act of parliament in 1995. By the end of the 1990s, Samurdhi was fully functioning in 21 of the - 32 - country's 25 districts.14 Its main functions are to improve "the economic and social conditions of youth, women and disadvantaged groups of the society."1s Originally administered by the Ministry of Samurdhi, Youth Affairs and Sports, since fall 2000 it has been administered by the Ministry of Samurdhi and Rural Development and Parliamentary Affairs. Three departments within the ministry coordinate various Samurdhi functions: the Department of Poor Relief, the Department of the Commissioner General of Samurdhi, and the Samurdhi Authority. There are district, divisional, and zonal-level Samurdhi authorities. Samurdhi expenditures amount to nearly 1 percent of GDP; more than 80 percent of expenditures are for the food stamp program (Table 2.3). Table 2.3: Samurdhi Expenditures, 1999 Program component Expenditure Percent (millions of SL rupees) of total Samurdhi relief program (food stamps) 8,190.6 81.9 Monitoring and evaluation 2.5 0.0 Financial Management (recurrent) 828.5 8.3 Subtotal/administration 831.0 8.0 Rural Development program 616.8 6.2 Agricultural Development 7.1 0.1 Human Resources and Institutional Development 116.6 1.2 Markedng 10.0 0.1 Banking and financial activities 201.8 2.0 Animal Husbandry and Fishery 3.2 0.0 Industrial and vocational skills development 8.0 0.1 Financial Management (capital) 17.5 0.2 Social Development (human development) 4.8 0.1 Subtotal capital and credit projects 985.9 9.9 Total 10,007.4 100.0 Source: Ministry of Samurdhi, Youth Affairs, and Sports, Summary of Financial Progress, 1999. 2.11 Program monitoring and evaluation are done at the national level by the Ministry of Plan Implementation. Divisional secretaries and district assistant Samurdhi commissioners are supposed to conduct monthly progress review meetings. There is no external (third-party or NGO) monitoring of progress or any random checks of beneficiary selections. 2.12 Four features of Samurdhi should be noted at the outset. First, program administrators are considered among the beneficiaries of the program, even though they are not necessarily poor themselves; 8 percent of the total Samurdhi budget is allocated for their salaries. Second, beneficiaries of the highest consumption grants pay a premium to finance a social insurance program available to all eligible households. Third, the same beneficiaries contribute 20 percent of the value of their stamps in lieu of "forced savings." These "savings" are unavailable for 14 In late 2000, the program was still not in full operation in some districts of the Northeast (Jaffna, Mannar, Killinochi, Mullaitivu). 15 Parliament of the Democratic Socialist Republic of Sri Lanka. "Samurdhi Authority of Sri Lanka Act, No. 30 of 1995." -33 - withdrawal until after four years. Fourth, multipurpose cooperative stores are the only places where stamps can be exchanged for goods. These stores belong to a de facto government- managed distribution network administered by the Ministry of Co-operatives and Development. 2.13 Development officers, "animators" (niyamakas in Sinhalese or ukkuvipalar in Tamil), and Samurdhi managers administer the program. Initially, the majority of animators were hired as temporary workers, with monthly salaries of Rs. 2,000. In 1999, almost all of them were promoted to permanent development officers, and their salaries doubled before being increased again to Rs. 5,000 in August 2000. Animators and development officers are responsible for identifying families that deserve assistance (according to instructions from the Commissioner of Poor Relief) and for distributing that assistance. They are expected to convene meetings of Samurdhi Societies (Samurdhi recipients and poor families) and Samurdhi Force (local youth groups). The Samurdhi Task Force is in charge of identifying infrastructure projects and other activities. 2.14 Development officers trained in banking procedures conduct the activities of Samurdhi banks. About 5,000 persons were trained in banking operations as of April 2000. Development officers and animators are also expected to supervise poor families who are saving in small groups-"like a teacher and thereafter like a friend." Program officers are expected to maintain a list of pregnant anemic women and malnourished children. Most commonly, development officers and animators receive a three-day training course in general operations, social mobilization, and leadership skills. Those involved in banking receive more extensive training in the form of three one-week courses. Training is carried out by the staff of the divisional secretaries, Samurdhi authorities, and consultants. 2.15 Eligibility and Use of Food Stamps. Households with a combined monthly income of less than Rs: 1,000 are eligible for consumption grants. Program development officers and animators screen potential beneficiaries using household questionnaires about income sources, living conditions, possession of durable goods, and Janasaviya and food stamp receipts (Table 2.4). Table 2.4: Samurdhi Eligibility Criteria Monthly household income in rupees and household size Grant size (rupees per month) Before August 2000 After August 2000 Less than Rs. 500, 4 or more members 1,000 1,000 Rs. 500-999, 3 or more members 500 700 Less than Rs. 1,000, 2 members 200 350 Less than Rs. 1,000, 1 member 100 250 Former Janasaviya recipients 250 400 Graduating Samurdhi membersa 125 *Once household income exceeds Rs. 2,000 Samurdhi members must graduate, but they still can receive Rs. 125 to cover their social insurance contribution (Rs. 25) and compulsory savings (Rs.100). Source: Ministry of Samurdhi. 2.16 Samurdhi officials widely admit that since income is not easily verifiable, means testing alone does not identify the poor. Moreover, the officially established threshold is artificially low. Animators are thus advised to use other characteristics of welfare in conjunction with means testing to identify beneficiaries. After one-time identification, a list of potential beneficiaries was -34 - compiled and displayed in public places. The grievances were considered by the Samurdhi Task Force. 2.17 Households identified as beneficiaries receive a supply of stamps every six months. The stamps are validated for use at multipurpose cooperative stores at the beginning of each month. There are multiple reports of inferior quality in these stores and inconvenient hours of operation. Contributing labor is a mandatory requirement for Samurdhi beneficiaries, but it is not always enforced. 2.18 When a household's income exceeds Rs. 2,000 and remains at that level for more than six months, or when a household member finds employment, the household must exit the program. Family members of military personnel are exempted from this rule. If a household changes its place of residence, it must be tested and enrolled at a new location, and there is reluctance to enroll new beneficiaries. Eligible extended families residing in one dwelling usually receive only one grant. 2.19 Effectiveness and Outcomes. The following analysis of Samurdhi focuses on evaluating distributional outcomes of the program's food stamp component.16 Rather than being a safety net for a group of households in need, Samurdhi more closely resembles an income support program. Using a representative household survey (Sri Lanka Integrated Survey), distributions from the program are measured across households with different welfare ranks. Welfare ranking of households is determined on the basis of their position in the per capita expenditure distribution before receiving the transfer. All households in the Sri Lanka Integrated Survey sample were divided into five groups (quintiles), each with the same number of individuals. Expenditure measures were adjusted by province-specific price indices that represented differences in the cost of living of the low-income populations. 2.20 Households with low relative welfare are more likely to receive Samurdhi assistance; that is, the program's incidence is progressive. However, transfers miss 36 percent of households ranked in the lowest expenditure quintile, while a substantial number of households with higher relative welfare receive Samurdhi consumption grants and other forms of Samurdhi assistance. Around 40 percent of the total Samurdhi transfer budget is spent on households from the third, fourth, and fifth quintiles-those whose members are relatively well off. These targeting outcomes are inferior to the outcomes of targeted programs in other countries. 2.21 Food grants cover more than 20 percent of the food costs of low-income households receiving Samurdhi but, as expected, Samurdhi transfers have considerably lower impact on the welfare of the better-off households. Forty percent of the total Samurdhi envelope is spent on about 2 percent of the food expenditures of households in the richest 60 percent of the population. 16 This evaluation does not assess the behavioral response of households to receiving Samurdhi income. However, theoretical considerations indicate that a household will change its behavior with respect to labor supply, investment, and consumption decisions, and well as to its efforts to seek private transfers within and across households or time, in response to government grants. Empirical evidence of increased demand for leisure was found by Sahn and Alderman (1996) in Sri Lanka. There is a widespread perception in Sri Lanka that successive transfer programs including Samurdhi do have the effect of reducing effort. Large government transfers may also displace private transfers (Cox and Jimenez, 1992 and 1995). -35- 2.22 Analysis of the probability of receiving Samurdhi grants in a multifactorial framework indicates that factors that generally correspond to the notion of low welfare-large household size, low income, disability, landlessness, no access to electricity or latrines-are associated with a higher probability of Samurdhi receipt. After controlling for income, place of residence, and other factors, there are still substantial and significant differences in the probability of being a Samurdhi recipient by ethnic status. Compared to Sinhalese households, Sri Lankan Tamils, Indian Tamils, and Moors are less likely to receive Samurdhi. 2.23 Even though the goals of Samurdhi are to improve the economic and social conditions of youth, women, and disadvantaged groups, the program's resource allocation and design reveal it as an income maintenance program that also provides public employment. 2.24 As an income-maintenance program Samurdhi is not effective. Features of Samurdhi- substantial errors of exclusion and inclusion and the non-randomness of these errors-reflect flaws in the design of the program that allow for the possibility of deliberate omission of certain groups of vulnerable individuals. The types of errors indicate that identifying the poor is not an informational problem, and suggest that unless the incentives faced by program administrators are fundamentally changed or the administrators are replaced, improvements in directing resources to the poor are unlikely to occur. 2.25 In addition, there are efficiency losses in the process of exchanging stamps for goods in the public multipurpose cooperative stores. Second-best arguments can be made in favor of using in-kind transfers to encourage certain types of behavior (for example, providing school lunches so that parents will bring children to school) or using low-quality goods as an element of self- targeting.17 But since there is no particular behavior to encourage and the costs of administration are high, there seems to be no good reason to require that purchases be made from particular stores. 2.26 Empirical Evaluation of Food Transfer Distribution. Households with low relative welfare are more likely to receive Samurdhi assistance; that is, the program's incidence is progressive (Table 2.5). However, Samurdhi misses 36 percent of households ranked in the lowest expenditure quintile, while 30 and 14 percent of households from two highest expenditure quintiles receive Samurdhi consumption grants. The size of the transfer is slightly larger among the households in the lowest quintile, but in per capita terms, the grant is larger for the better-off households. The average Samurdhi transfer is Rs. 403, which translates into Rs. 88 per person in per capita terms. 17 The best policy would be to provide information on the benefits of the desired behaviors and let households decide. -36- Table 2.5: Samurdhi Food Stamps 1 5 Quintile (lowest) 2 3 4 (highest) Total N (sample size) 463 514 510 560 618 6,534 Benefit incidence and average amount of transfer (conditional on receiving transfer) Incidence (percent) 64 53 42 30 14 40 Rupees, per household 403 390 378 362 303 381 Rupees, per capita 88 96 93 105 107 95 After compulsory deductions Rupees, per household 322 315 307 301 260 310 Rupees, per capita 71 78 77 88 94 78 Excluding Janasavia households Incidence 53 41 32 21 11 31 Rupees, per household 359 334 320 291 251 326 Rupees, per capita 79 83 79 84 82 81 Samurdhi food transfer as share of household food expenditure Conditional on being a recipient 21 12 10 9 8 14 Unconditional 14 7 4 3 1 6 After compulsory deductions Conditional on being a recipient 18 10 8 7 7 11 Unconditional 11 5 3 2 1 5 Source: Calculations from 3 quarters 1999/2000 Sri Lanka Integrated Survey. 2.27 After applying compulsory deductions for savings and social insurance, the amount of the average transfer falls to Rs. 322 per household and Rs. 71 per person. Excluding former Janasavia recipients-who receive grants of Rs. 250 (Rs. 400 after August 2000) due to the government's legal obligations under the previous social assistance program-does not change the relative distribution of transfers. 2.28 Among households that receive Samurdhi, an average of 14 percent of total household food expenditure is supplied by the Samurdhi grant." The impact on the welfare of households in the lowest quintiles is considerably larger than on higher-income households. Around 21 percent of total food expenditures of the lowest quintile households are financed from the Samurdhi food stamp grant, whereas in households in the second lowest quintile only 12 percent of total food expenditures comes from Samurdhi. The program's impact on consumption of the richest 60 percent of the population is even smaller, ranging from 1 to 8 percent of total food consumption. Forty percent of the total Samurdhi transfer budget (based on the sample) goes to households in the third, fourth, and fifth quintiles-that is, to those who are relatively well off (Table 2.6). '8Total food consumption includes home-produced and home-consumed goods. -37- Table 2.6: Food Stamp Distribution by Population Quintile 1 5 Quintile (lowest) 2 3 4 (highest) Total Reported receipts 33 27 20 14 .6 100 After compulsory deductions 32 27 20 14 7 100 Excluding Janasaviya households 34 27 20 13 6 100 Source: 1999-2000 Sri Lanka Integrated Survey; calculations from three quarters. 2.29 How do these targeting outcomes in Sri Lanka compare to programs elsewhere? The targeting outcomes of Samurdhi are inferior to the outcomes of targeted programs in other countries; in fact, they are comparable with the outcomes of untargeted systems (such as primary health care or primary education). The practical difficulties of administering targeted programs are well known, and the way to judge whether these outcomes are successful or not, is to compare them with the outcomes of similar programs in other countries. Grosh (1994) reviewed the design and targeting outcomes of 47 government systems in Latin America. Across the considered countries, a median targeted program delivered more than 70 percent of all benefits to the poorest 40 percent of the households. A median untargeted program had outcomes close to Samurdhi, transferring about 60 percent of the total budget to the poorest 40 percent of households. Examples of such programs are primary health care and primary education. The administrative cost of a median program using individual assessment as a targeting method is reported at about 9 percent of the total program cost, which is comparable with the administrative cost of Samurdhi (8 percent). 2.30 Who does and does not receive Samurdhi food stamps? We used a multifactorial framework and a series of models that relate the characteristics of households to their probability of being selected as a Samurdhi beneficiary to investigate the determinants of the probability of being a Samurdhi recipient. We begin with a model relating the probability of being a Samurdhi recipient to the exogenous demographic and geographical characteristics of households, as well as their pretransfer income (model I).'9 A second model-model fl-uses with the same outcome measures and a larger set of factors normally associated with low welfare (possession of land and durable goods, type of energy used, access to latrines, presence of chronically sick members (see Annex A,Table A8). Estimates cover the entire country except for the North East province, since Samurdhi was introduced there much later and may still not be functioning fully. 19 Pretransfer income is not an exogenous characteristic and might be affected by unobservables, which would in turn affect the probability of selection. Other behavioral responses of participation also might lead to changes in income. These considerations were not addressed in the above. -38- Table 2.7: Model I-Probability of Receiving a Samurdhi Grant (Survey Probit Regression) Scaled Variable Coefficient Standard error T P>|t| coefficient Number of children 0-7 years old -0.036 0.041 -0.863 0.389 -0.535 Number of children 8-17 years old 0.016 0:027 0.571 0.568 0.369 Number of elderly -0.014 0.037 -0.366 0.715 -0.055 1, if no men in the household 0.174 0.094 1.861 0.064 1.041 Household size 0.107 0.020 5.241 0.000 1.000 Type of settlement, traditional village omitted 1, if irrigation colony -0:289 0.145 -1.996 0.047 -5.425 1, if expansion colony 0.251 0.173 1.446 0.149 0.626 1, if settlement scheme -0.008 0.124 -0.064 0.949 -1.776 1, if plantation or estate -0.649 0.223 -2.909 0 004 -7.531 1, if urban low income neighborhood 0:143 0.178 -0.800 0.424 0.587 1, if urban mid.income neighborhood -0462 0.112 -4.115 0.000 -6.230 Province, Westem omitted 1, if Central 0.384 0.112 3.427 0:001 3.864 1, if Southem 0.449 0.100 4.511 0.000 3.873 1, if North Western 0.535 0.112 4.760 0.000 4.083 1, if North Central 0.490 0.119 4.124 0.000 4.755 1, if Uva 0.347 0.130 2.667 0.008 2.829 1, if Sabaragamuwa 0.545 0.111 4.916 0.000 4.193 Ethnicity of the household head, Sinhalese omitted 1, if ethnicity is Sri Lankan Tamil -0.490 0.121 -4.046 0.000 -3.018 1, if ethnicity Indian Tamil -0.741 0.197 -3.761 0.000 -4.269 1, if ethnicity is Moor -40.358 0.125 -2.858 0.005 -3.803 1, if ethnicity is other -0.514 0:376 -1.369 0.172 -5.729 Log household income, less Samurdhi -0.291 0.032 -9.088 0.000 -2.535 Constant 1.505 0.268 5.604 0.000 13.106 Number of observations= 4762; Number of strata= 2; Number of PSUs = 325; F(22, 302) = 13.50; Prob > F = 0.0000 Note: Shaded if (jointly) significant at 5 percent level; absolute coefficient values in probit model cannot be interpreted as marginal effects; scaled coefficients measure the effect of particular characteristics relative to the effect of household size. Source: 1999-2000 Sri Lanka Integrated Survey; calculations from three quarters. 2.31 Findings from model I show large disparities across provinces and types of settlement (Table 2.7). Conditional on their real income level, households residing in urban middle-income neighborhoods, irrigation colonies, and estates are least likely to receive Samurdhi grants. Results also show an effect of ethnicity: Controlling for income and area of residence, Sri Lankan Tamils, Indian Tamils, and Moors are all less likely to receive Samurdhi than are Sinhalese households. Large households are more likely to receive Samurdhi food stamps, as are households without adult males. Household income is an important determinant of eligibility. 2.32 Estimations from model II confirm the findings of model I with respect to the ethnic differences in benefit receipt and attest to the notion that households with characteristics of low - 39 - welfare are more likely to be Samurdhi recipients. Factors that generally correspond to the notion of low welfare-large household size, low income, disability, landlessness, no access to electricity or latrines-are associated with a higher probability of Samurdhi receipt. Specifically, households with disabled or chronically sick members are more likely to receive Samurdhi assistance. Households owning livestock and other assets are less likely to receive assistance. Land ownership is the most significant determinant of assistance grants. 2.33 The probabilities of grant receipt differ significantly by'the occupation of the household head. Households headed by farm laborers and households with members in the military are more likely to receive assistance. Housing conditions, too, appear to be associated with the probability of receipts. Households with no access to latrines or to communal latrines are more likely to receive assistance than households with private latrines. Energy source also seems to be an important correlate of receiving Samurdhi. Households that use kerosene for lighting or have no sources of lighting are more likely to receive Samurdhi than are households with access to electricity. 2.34 In addition to allowing for ethnic and location biases, Samurdhi may also have become politicized. Qualitative work, from the Sri Lanka Integrated Survey and other sources such as Gunatilaka et al. 1997 and Asian Development Bank 2001, suggests that this may already be the case (Box 2.2), with party affiliation and voting preference reportedly influencing allocation of Samurdhi consumption grants. This happens partly because there are no strict rules for program eligibility and not enough checks and balances to prevent Samurdhi officers from acting on the demands of politicians. Box 2.2 Politicization of Samurdhi. There i 'a widespread percepfion that Samurdhi is politicized,.. much like its predecessor, Janasaviya. A'study bythe Institute of Policy Studies reports that "political-affiliaioh determined the eligibility of some to receive the consumption grant.... This problem was:confirmed by several Samurdhi Managersiahd nyamakas we spoke to.The hiyamakas said that they are under'pressure from area:politicians:and village-level party brganizations togive 'the Samurdhigrant to PA fanmilies and:to denythe'nto UNP supporters... Rublic sect6r employees who pasted poster for politcians during the local govermrit elections successfully, petitionbd and got themselves selected for Samurdhi in the second round." Source:.Gunatilaka et al, 1997. As-tudy 'by Parker and Silva (2000) also suggests perceptions of unfairriess in the selection of both the Samurdhi administrators and the beneficiaries f' Samurdhi transfers. "Although a number of points of criticism were raised, unfaimess. in the:selection of both Samurdhi animators (niyamakas) and beneficiaries was'the complaint hesrd: most often' about the Samurdhi program The prmary abuse appears to be the inclusion of large numbers f'ineligible households 'as areward for political loyalty. to the governing party. In many of the 'villages visited; half or more. of the village.households were found to bereceiving Samurdhi benefits. 'Reports* ereheard of party organizers attractihg voters. to the party with the promise of Samurdhi benefits. Less commonly, Samurdhi has been utilized to threaten:or punish low-itcome supporters of the opposition party. ln some' villages eligible Samurdhi, recipients: who are not supporters of the ruling party were told that if they ventured to vote in an upcoing election, they would be removed from' 'the Samuirdhi lists and their benefits would be terminated. Although errors or abuses of exclusion appear to be far less common than errors:of inclusion, they nevertheless exist." Sburce: Parker ard Silva 2000. 2.35 Allocations of Samurdhi Transfers Between and Within Provinces. The allocation of benefits across provinces corresponds to poverty rates, that is, poorer provinces and districts are allocated a higher amount of Samurdhi transfers in per capita terms. Identifying the poor on the -40 - ground is less successful. Within the province, provinces with higher poverty rates are less effective in reaching their poor than are richer provinces. Correlation between funds allocated to each district and district-specific poverty measures are presented in Table 2.8. The correlation between per capita district-specific Samurdhi funds and poverty rate (depth) is .814 (.711). This high positive correlation attests that allocation of Samurdhi assistance at the central level is pro poor. It also suggests that most targeting errors are due to misallocation at the local level. Poorer provinces allocate smaller shares of their Samurdhi funds to their poor (Table 2.9). Specifically, the correlation between the share of province-specific Samurdhi funds transferred to the poorest strata and province-specific poverty indicators ranges between -.454 and L-.789. Table 2.8: Food Stamps Issued to Districts and Welfare Indicators 1999 total Stamps per Poverty and welfare indicators (millions of Population person rupees) (thousands) (rupees) Average Poverty expenditure Gini District Poverty rate depth (rupees) Colombo 375 2,057 182 10 1.77 2038 0.35 Gampaha 710 1,708 416 11 1.8 1720 0.31 Kalutara 384 938 410 26 5.78 1392 0.33 Kandy 557 1,221 456 30 7.04 1321 0.34 Matale 290 423 686 35 8.53 1096 0.28 Nuwara Eliya 181 671 270 21 3.61 1145 0.22 Galle 462 955 483 25 5.62 1323 0.31 Matara 448 754 594 28 5.84 1214 0.28 Hambantohta 338 516 655 27 5.4 1234 0.29 Kurunegala 986 1,378 715 34 7.06 1093 0.27 Puttalam 396 601 660 33 6.69 1105 0.26 Anuradhapura 461 676 681 33 6.15 1143 0.28 Polonnaruwa 191 332 576 28 6.45 1240 0.29 Badulla 321 750 428 30 6.44 1139 0.27 Moneragala 269 365 737 49 14.47 927 0.30 Ratnapura 664 917 725 37 8.26 10921 0.28 Kegalle 389 760 512 25 5.07 1230 0.28 Trincomalee 146 .. .. .. Ampara 332 .. .. .. Batticaloa 263 .. .. .. Vavuniya 27 .. .. .. Pearson correlation: Stamps per person and column 1 0.7109 -1 -0.311 Source: Samurdhi Authority and Gunewardena 2000. -41- Table 2.9: Distribution of Samurdhi Consumption Grants across Province-Specific Quintiles of Per Capita Expenditure Distribution and Welfare Indicators (percent) No access No access 1 5 Poverty Poverty to safe to safe Province (lowest) 2 3 4 (highest) rate depth water sanitation Western 33.0 32.9 18.8 12.3 2.5 14 2.6 18 11 Central 27.0 24.3 16.6 22.7 9.4 28 6.2 26 24 Southern 31.8 21.6 25.5 15.5 5.7 26 5.6 35 20 North Western 27.4 21.9 23.1 19.5 8.1 34 7 35 30 North Central 22.3 20.7 27.8 15.7 13.6 31 6.3 48 32 Uva 27.4 28.3 22.8 14.0 7.6 37 9.5 45 34 Sabaragamuwa 32.1 21.2 23.0 13.4 10.2 32 6.8 32 23 Pearson correlation: share accrued to the first quintile and column -0.534 -0.454 -0.694 -0.789 Pearson correlation: share accrued to the first two quindles and column -0.667 -0.486 -0.712 -0.738 Note: Province-specific quintiles of per capita expenditure are used in the investigation of Samurdhi distribution within provinces. Source: Calculations from three quarters of Sr Lanka Integrated Survey and National Human Development Report. Assistance to Displaced Persons and to Populations Affected by War 2.36 The government operates assistance programs for 800,000 internally displaced persons, 200,000 of whom live in government welfare centers. Government initiatives include providing (i) dry food rations for displaced persons, both inside and outside of welfare camps, (ii) operation of welfare camps, and (iii) resettlement and relocation grants, along with other types of compensation, under the Unified Scheme of Assistance. The government also provides transfers under several smaller schemes. 20,21 Total government expenditures under these schemes amounted to Rs. 3.7 billion in 1999 (0.3 percent of GDP). Despite these considerable efforts, the war-affected population remains vulnerable. Restrictions on the mobility of internally displaced persons and people living in welfare camps, as well as restrictions on flows of goods and services to the war-affected areas, exacerbate the problems of this population. 2.37 Dry rations. Internally displaced persons are eligible for dry rations provided as stamps redeemable in retail outlets of multipurpose cooperatives and as in-kind food items. The most recent estimates indicated that 730,000 persons were receiving dry rations in late 2000. Most recipients live in districts of the North East in "cleared" and "uncleared" areas, but displaced persons staying with friends and relatives in other provinces also received dry rations. 2.38 The amount of the monthly benefit ranges from Rs. 336 for a family of one to Rs.1,260 for a family of five. Those amounts are equivalent to 0.25 and 0.95 of the median per capita income in Sri Lanka in 1999. Government funds are channeled through the Commissioner General for Essential Services (CGES), and identification of beneficiaries is devolved to the local administration. In 1999 the CGES spent Rs. 2,331.3 million on dry rations. 20 These smaller schemes include economic assistance to fishermen affected by the ban on fishing and assistance to the "poorest of the poor" in lieu of the Samurdhi in Jaffna. Little is available about actual incidence of those programs. 21 UNCHR estimates; government estimates for the same time period (730,000 persons in total and 171,000 persons in welfare camps) are lower, and are based on the number actually receiving dry rations. -42 - 2.39 Welfare Camps and Resettlement and Relocation Grants. According to the government estimates, CGES houses 171,000 people in 348 welfare centers in 14 districts. Maintenance expenses amounted to Rs. 72 million in 1999. Dry rations to households in welfare camps are provided by the World Food Program (WFP) through the Department of Social Services (Ministry of Social Services and Housing Development). Welfare camps vary from abandoned buildings, schools, and hospitals to huts and communal shelters. The Unified Scheme of Assistance provides compensation to internally displaced persons who suffered injuries or property damage, and to families of war casualties. Households in welfare camps are also eligible for certain grants. The program is administered by the Resettlement and Rehabilitation Authority of the North. 2.40 These are substantial efforts. But the population on both sides of the conflict remains greatly affected. Loss of personal property and employment, loss of a breadwinner, risk of disability, and risk of recurrent displacement are elements of everyday life. As their ties with communities are broken and their assets are lost, the displaced lose access to traditional risk- coping mechanisms. Civilian populations living in the conflict areas have limited access to basic services, and their ability to earn income is compromised by restrictions on the flow of goods. They are often caught in the crossfire. People living in welfare camps also often lack access to services, informal networks, and income opportunities. 2.41 The major challenges for internally displaced persons are the long-term nature of the conflict, the recurrent nature of the displacement, and diminishing opportunities for people to return to their villages of origin (UNCHR and Consortium of Humanitarian Agencies 2000). Many internally displaced persons have lived in welfare centers for six or seven years. Their relocation is imperative. However, those who are resettled sometimes show slow recovery. Relations with the host communities are sometimes tense and newcomers often lack property rights in their newly acquired homesteads. Restrictions on freedom of movement and the pass system prevent inhabitants of welfare camps from reaching schools and hospitals and from seeking income opportunities. The most persistent consequences of the war will be felt by children and adolescents. Children affected by the war exhibit higher rates of malnutrition, lack of access to education, and reportedly suffer physical abuse and sexual harassment, experiences that will have profound effects on their productivity in the future. Other Government Programs 2.42 Nutrition Programs. The incidence of low birth weight, undernutrition among young children, and anemia among pregnant women in Sri Lanka is fairly high (Box 2.3). The government has been addressing the nutritional status of young children for at least two decades with direct in-kind transfers and other programs. The largest nutrition program, Triposha, was started by CARE in 1973 and provided a fortified food supplement to malnourished mothers and preschool children. In 1987 CARE turned over school feeding to the Sri Lankan government. The World Bank's 1995 Sri Lanka Poverty Assessment concluded on the basis of an independent evaluation (Marga Institute 1990) that Triposha in its present form was not up to the task of eliminating the problem." The program is still in place without any significant modifications. -43 - Box 2.3 Nutrition Indicators in Sri Lanka Nearly ialf of all children under five yearsold and non-pregnant womeri in thecodntry are estimated tolbe anernic. Chrdnicmalhutition-(stuntihg) affects 24 percent,of all,children under five, ahd 16 percent sufferacute undernutrition (wastilig): 38 percent. all children in this age group are underweight(Health Ministry 1998).ro'deficieindy anemia and'InAliutritioh are most pervasive in rain-fed dry zones of the Uva, North Central, ndfNoit i Western pr6vinces. Recent studies in the northeast region suggest that the highest rates of ma`nutrition:may no6W'be.found in the-areas affecteddirect,b t,conflic Source: Tudawe aid Wickreinanayake 1998. The.1994 Nutrition and Health SUrvey (NHS4) reveals that a third of.all nonpreghagi iiothers in Sri Lafiikaiarb undemourished There is-large variation.in the incidence of matermal malnutritior acrossisectors ahd income groups, but maternal'undemourishment represents alproblem even for'the-relativ6lY elllff:Populations it the estate sector show the-highest incidence.of maternalundemourishment. Undernourished Mothers by Sector and lncome (percent) Sector Income quartile Urban 21.1 Lowest 45.7 Rural433 Second 41.1 Estate 58.6 Third 31.4 Highest 18.2 Overall33;4 Souice:Rarnanujam and Nestel 1997. 2.43 The Poor Relief Nutrition Program distributes monthly nutrition cards worth Rs. 100 to lactating mothers. A total of 80,718 cards costing Rs. 97 million were distributed in 1999. Beneficiaries are identified by Samurdhi development officers. Nutrition cards can be exchanged for food in multipurpose cooperative stores. 2.44 Several other programs have components to improve nutrition outcomes. The Nutrition Intervention Program undertakes a set of activities including vitamin supplement and nutrition awareness programs. The Plan of Action for Children has the goal of reducing iron and iodine deficiency disorders. Both programs are administered by the Ministry of Health and Indigenous Medicine and by provincial ministries of health. 2.45 Old Age Security. Sri Lanka has entered a demographic transition, with the proportion of elderly expected to double by 2025. In the near future, the country will have to address safety net policies for those elderly who are poor and lack other means of support. At present, however, increase in support to the elderly is not expected to have a significant effect on poverty. Overall, the elderly are not over-represented among the poor, according to the Integrated Survey. However, there is a large regional variation in the income of the rural elderly. The elderly living in urban areas have a relatively higher standard of living. In the Uva, Sabaragamuwa, and Central regions, by contrast, the elderly are over-represented among the relatively worse off. The elderly living alone are over-represented among the households in the very bottom and very top expenditure quintiles. -44- 2.46 Currently, government pensions are more likely to be received by the relatively well off, since access to public or formal employment provides a relatively higher standard of living. Analysis of the Sri Lanka Integrated Survey shows that whereas 6 percent of all households receive government pensions and 2 percent receive EPF/ETF payments, 14 percent of households in the top quintile of spending received pensions and only 2 percent in each of the two lowest quintiles benefited from government pensions (Table 2.10). Among the households from the top expenditure quintile with members over 60 years of age, 26 percent receive government pensions. The distribution of pension funds across provinces shows that 50 percent of the total pension envelope observed in the Sri Lanka Integrated Survey sample accrues to households in Western province. Central province is the second-largest recipient of pensions, accounting for 14 percent of the total envelope. Uva and North-Central provinces received the lowest proportions of the total-2 percent each. Table 2.10: Incidence and Distribution of Public Transfers, 1999-2000 (percent) 1 5 Quintile (lowest) 2 3 4 (highest) Total Incidence of public transfers overpopulation quintiles Government pensions 1.6 2.1 3.9 6.6 14.3 5.9 Widow and orphan pension funds 1.1 1.6 1.4 1.5 2.9 1.7 Other pensions 0.0 0.1 0.4 0.0 0.2 0.2 Assistance from Social Service Department. 5.0 1.5 1.3 1.0 1.0 1.9 Employer Provident Fund (EPFIETF) 1.7 1.5 2.0 2.6 3.8 2.4 Distribution of public transfers overpopulation quintiles Government pensions 4 6 10 22 58 100 Widow and orphan pension funds 6 9 15 10 61 100 Other pensions 0.5 0.5 89 0 10 100 Assistance From Social Service Department 50 15 13 9 13 100 Employer Provident Fund (EPF/ETF) 7 8 13 21 51 100 Source: 1999-2000 Si Lanka Integrated Survey; calculations from three quarters. 2.47 The pension system is a responsibility of the formal sector and consists of pension schemes and savings funds. The Public Sector Provident Fund (PSPF) is mandatory for government employees who are not entitled to a civil service pension. All non-government employees in the formal sector are required to participate in the government-administered Employees Trust Fund (ETF) and the Employment Provident Fund (EPF) or another approved provident fund. Outside the mandatory schemes, several government-administered voluntary pension schemes offer retirement benefits to farmers, fishermen, and the self-employed.22 * Widows and orphans of participants in EPF receive pensions. The distribution of these benefits is 22 A provident fund is a saving plan designed to provide an individual with retirement income (often paid in a lump-sum) by collecting and investing a regular stream of contributions during the years of employment. -45 - regressive, according to the Sri Lanka Integrated Survey, with over 70 percent of payments in the SLIS sample going to households in the top two quintiles of expenditure distribution. 2.48 "Public Assistance". Sri Lanka's Public Assistance is a relatively well targeted set of social assistance programs. Intended for the very poorest Sri. Lankans, its outcomes confirm that program funds are in fact directed to this population. Several Public Assistance programs are operated by the Social Service Department, Ministry of Social Services and Housing Development. Programs provide assistance to persons with disabilities, the elderly poor (those without sources of income), widows, foster parents of orphans, and persons with long-term illnesses or paralysis. Single person households receive Rs. 100 per month; households with dependants receive Rs. 300. A scheme of assistance to persons affected by tuberculosis, leprosy, and cancer was recently introduced. 2.49 Half of the total transfer envelope of this set of programs benefits the poorest 20 percent of Sri Lankans, according to the Integrated Survey; 65 percent benefits the poorest 40 percent (Table 2.10). Low-income households with disabled members as well as the low-income elderly are more likely to receive funds from Public Assistance programs, according to the Integrated Survey. The benefit size and coverage of Public Assistance programs, however, are low. The benefit is equivalent to 8-23 percent of the per capita median income in Sri Lanka in 1999. 2.50 Administration of these programs is devolved to the provincial councils. Although the Department of Social Services is still nominally responsible for evaluating them, few details are available on their implementation. The 1999 expenditures of the Social Services Department totaled Rs. 2,149 million (0.2 percent of GDP), with Rs. 1,740 million (81 percent of the total) spent on actual transfers.24 2.51 Unemployment Programs. Two distinct groups in Sri Lanka are at high risk of unemployment. The first is unskilled rural laborers, who face seasonal fluctuations in labor demand; the second is so-called educated youth. The unemployment rate among young, educated individuals-those with 0- and A-level university degrees-stood at 12.9 and 17.4 percent in the first quarter of 1999 (Central Bank Annual Report, 2000). In the past the government has attempted to deal with the youth unemployment problem by creating civil service jobs. This has led to an increase in reservation wages of these unemployed, a proliferation of public sector jobs, and grievances on the part of those who felt left out. A recent study of the Sri Lanka labor market demonstrated that high reservation wages and queuing for lucrative jobs in the public sector are the main reasons for high unemployment rates among educated youth.25 In Sri Lanka, less than 2 percent of poor households are unemployed. 2 The Social Services Department also assists victims of natural disasters. 24 Net of World Food Program (WFP) funds. The Department of Social Services and the CGES also provide dry ration assistance (up to six months) to the low-income victims of natural disasters such as drought, floods, earth slides, pests, wild elephant attacks, and outbreaks of malaria, as well as to people in welfare camps. Property damage (up to Rs. 10,000) resulting from natural disasters is also covered by grants from the Department of Social Services. Disaggregated figures on the amount of these transfers are not available. 25 Based on findings of World Bank (1999) "Sri Lanka - A Fresh Look at Unemployment" and Rama (1994) "Flexibility in Sri Lanka's Labor Market." -46 - 2.52 The government's latest response to unemployment has been to create export-processing zones (EPZs) and boards of investment (BOI), which were expected to generate jobs in response to tax concessions and to the shield protecting them from Sri Lanka's rigid labor regulations. The large number of vacancies in these industries implies that tiese jobs are considered undesirable by the unemployed, in particular educated youth. White-collar public sector jobs-teaching, Samurdhi administration-are more popular among graduates. 2.53 Other Programs. Several other government-operated transfer and subsidy programs have certain safety net functions. Uniforms and textbooks are provided free of charge to school children by the Ministry of Education and Higher Education. These programs are meant to encourage school participation, but their operational efficiency could be clearly improved. In case of schools uniforms, for example, distributing vouchers, rather than a uniform, could reduce administrative overheads. 2.54 The fertilizer subsidy is considered in Sri Lanka to be a social transfer because it is intended to benefit small-scale farmers. Sri Lanka produces very little chemical fertilizer domestically (9 percent of the country's total consumption in 1997), and the subsidy is paid to importers of chemical fertilizer. Fertilizer subsidies were in place for several decades after independence, but in the beginning of 1990 they were discontinued for several years, reappearing in 1994. Between 1994 and 1997 subsidies were given to a variety of fertilizers. Since 1997 only the urea subsidy has remained in effect, with the purpose of "channeling the bulk of the benefits of the fertilizer subsidy to the small farmers." The amount of the subsidy depends on the difference between the purchase price and the fixed retail price envisaged for the Sri Lanka domestic market. 2.55 The efficiency of the fertilizer subsidy is questionable on at least two grounds. First, 28 percent of all subsidized urea was used in 1999 in tea, rubber, and coconut production-to the benefit of large producers. Second, in rice farming, the urea subsidy encourages overuse of this fertilizer, which soon reduces yields (the so-called illusive green effect) and causes environmental damage. Private, Semi-formal, and Informal Safety Nets 2.56 Remittances. In Sri Lanka as in other countries households make their own arrangements to secure their consumption flows. Foreign employment is an importance source of income for many households. An estimated 5 percent of GDP is net income from abroad. Other types of arrangements include remittances from friends ands relatives, use of village middlemen, and savings. In the Sri Lanka Integrated Survey sample, slightly less then 5 percent of all households reported receiving income from family members working abroad. Households from all income strata receive labor income from abroad (Table 2.11). Variation in use of these methods stems from geographic area of residence and ethnic origin. Moors are the most likely to report income earned abroad (11 percent of households) followed by Sinhalese and Sri Lankan Tamil households (8 percent each). Indian Tamil households have the fewest opportunities to earn income abroad: Only 2 percent of households have members remitting income. Households in the North Eastern and North Western provinces are the most likely to receive labor income from abroad; households in Uva and Sabaragamuwa are the least likely. 2.57 Analysis of the 1999/2000 Sri Lanka Integrated Survey shows that well-off households in Sri Lanka are more likely to receive transfers from abroad, whereas domestic remittances are received by households from all strata (Tables 2.11 and 2.12). There is wide geographic variation -47 - in receipt of remittances. Foreign remittances are more likely to be received by households in the Western and North Eastern regions, while households in Uva and Sabaragamuwa depend on domestic remittances more than households in the other provinces: 14 and 20 percent of all households in these two poorest provinces reported receiving domestic remittances. Table 2.11: Incidence of Private Transfers and Migrant Labor Income (percent) 1 5 Quintile (lowest) 2 3 4 (highest) Total Labor income from abroad 4.7 5.7 5.4 3.8 4.1 4.7 Friends and relatives remittances from abroad 1.6 1.7 2.9 3.0 6.4 3.2 Domestic remittances from relatives and friends 9.4 9.6 9.9 10.6 10.6 10.0 Source: 1999-2000 Sri Lanka Integrated Survey; calculations from 3 quarters. Table 2.12: Distribution of Private Transfers and Migrant Labor Income by Quintile 1 5 Quintile (lowest) 2 3 4 (highest) Total Labor income from abroad 14 19 19 18 29 100 Friends and relatives remittances from abroad 2 5 6 11 76 100 Domestic remittances from relatives and friends 12 12 20 30 26 100 Source: 1999-2000 Sri Lanka Integrated Survey; calculations from 3 quarters. 2.58 Microfinance. An efficient financial market can help households deal with risk. The welfare of the poor can be improved by including them in the financial system so that they have access to high marginal returns on their investments, insurance to deal with the risks implicit in their economic activities, and liquidity for consumption smoothing. Poor people save, but they do not necessarily have access to deposit services or the opportunity to save enough to engage in investment projects. 2.59 Studies have sought empirical evidence on the impact of microfinance services on poverty by measuring whether access to microfinance increases levels of consumption or decreases fluctuations in levels of consumption at the household level (Khandker 1998; Morduch 26 1998; Montgomery and others 1996). Increased investment via access to credit is expected to raise household income (and consumption) and thereby reduce poverty as long as the investment returns are in excess of the loan installment repayments. A temporary reduction in poverty is expected to occur if credit or savings are used to smooth consumption by repaying existing debt, improving housing, or addressing social obligations, thereby reducing household vulnerability to 26 The definition of microfinance and microfmance providers needs to be clarified to avoid confusion. Generally, microfinance services comprise basic financial services that are specifically tailored for poor clients. They have the following characteristics: small loans, typically for start-up business or working capital; informal and quick appraisal of borrowers; collateral-free loans and use of collateral substitutes such as group guarantees or compulsory savings; access to larger repeat loans based on sound repayment performance; frequent monitoring of investment projects and repayment of installments; and small amounts of savings. -48 - income shocks. There is growing evidence in the literature on the income-smoothing (vulnerability-reducing) role of microfinance (Morduch 1998; Zaman 1999). 2.60 Formal microfinance providers are those chartered by the government or subject to banking regulation and supervision. In Sri Lanka, they include public and private banks engaged in commercial and development banking. Semiformal institutions are usually licensed by different government ministries and include cooperatives and NGOs, as well as the credit component of the government Samurdhi program. Informal financial institutions operate outside the boundaries of any government regulation or supervision; they include self-help groups, moneylenders, shopkeepers, and friends and family.27 2.61 The formal microfinance market is largely supply driven, as evidenced by various government interest subsidies or refinancing schemes offered to the two state banks for rural sector lending. The same situation exists in the semiformal market, where the government supports the fast-growing credit component of the Samurdhi program (introduced earlier in this chapter) known as Samurdhi Banking Societies (SBS). Most of the subsidized rural credit programs have resulted in high arrears and have generated losses for the financial institutions administering the programs and for the government. Evidence from other developing countries show that government associated credit programs like Samurdhi also face the risk of failure due to the potential for political capture of these programs. 2.62 Government subsidies and direct state involvement in microfinance through Samurdhi crowd out other well-performing microfinance institutions that do not have access government subsidies. Many NGOs use borrowed funds at market interest rates to on-lend to their clients but have had to restrict their expansion and clientele to remain competitive. Heavy state involvement in the microfinance sector also undermines credit discipline among borrowers, thereby imposing additional credit risks onto these NGOs. 2.63 According to the Sri Lanka Integrated Survey, more than 70 percent of the country's poor have access to some form of savings service, and about 26 percent borrow from various sources. Similar distributions in saving and borrowing are also observed among relatively wealthier households, thus indicating no systematic lack of access to financial services by poor people. Among households that borrow, loan sources are distributed almost equally across all the quintiles. Formal sources of loans are the largest among all groups, followed by informal sources and then the semiformal institutions. Table 2.13 disaggregates household savings and borrowing behavior across all households by per capita monthly expenditure quintiles. 2.64 Sri Lanka has unusually high levels of formal sector lending to the poor compared to most developing countries. About 41 and 25 percent of loans to poor households are made by formal and semiformal microfinance providers, respectively. The saving and borrowing patterns of the population as a whole do not suggest that poor households are systematically excluded from participating in the formal or semiformal financial market. 2.65 The level of poor people's participation in the formal financial sector distinguishes Sri Lanka from most developing countries, where the informal sector is the primary source of credit for poor people. The informal sector in Sri Lanka does provide access to financial services, in 27 The term "microfinance institution" is used when referring to an organization that claims to be providing microfinance services. It is hard to establish, based on the data available, which institutions are strictly providers of microfinance services and which are not. -49 - particular short-term consumption loans (Box 2.4). But the role of the formal sector is also important. Not only do poor households in Sri Lanka have access to credit and mortgage facilities from various institutions, but they can also chose to draw down their savings to meet income shocks. Poor households are therefore better prepared to manage their risks than their counterparts in other developing countries. Box,2.4 The Role of the Middleman in the Rral Econoiny The village middleman has a wide;presence in the village.economy and is said to trap' poor households in acircleof debt. In some cases, the client-patron relationship allows a.village lender to:establish a monopolisti6opresence and earn corresponding profits. A closer exani'ation reveals that middlemen serve a series of essential purposes: A government wpuld be short-sighted ifit'tried to break this relationship by regulating village lenders or-by providing subsidized credit. Regulatiois do not usually work inthe informal econoriy, and the risk assessmerit, monitoring,:and enforcement capabilities of a state-agent are by:far_infeiorto those of avillage lender. Public.policy action thatcoIldinfluence this patronlient relationship ghould ainiat bringing changes to the structure of.the village market,for example, by f6sterifig the emergenceof competitors with a stable presence liprovemerts in;r6ads and other types of communications infrastructure are;examples of such types,of policies. In additibr strengthehing NGO-based initiatives in which peer!groups.perform monitorihg and enforcement roles should improve the competitive structure:of the village credit markets. The middleman (mudatali) in the Sri Lankan'rural economy fills multiple roles and is'generally,viewed with ambivalence,by the rural population. The mudalali offers a supply of consumer goods and inputs on credit fnd purchases produce at the farmh gate; His fubctinsincludethe supplyrof agricultural inputs, retail trade in consumer go6ds, provisiah oftransport services and ofcredit; and purchase-of produce, particularly paddy crops. The ,control :of:this wide range-of essehtial furnctions by a single a'ht has all6wed the mudalalito consolidate!an advantageous position visrb-vis the small farmer. Sinal(farmers particula in isolated ar6as; depend heavily on one or avery few.traders:for inputj ciedit, an uninterrupted supply of corisnier goodsand sale of farm produce. Dunng lean seasons, particularly before harvests; many are delendenton credit from thelocal mudalali shpkeeper for basic staple food' items or, earlier inthe season, for agricultural inputs such as 'fertilizer nd 'pesticides .Under the terms ofa commoitunwritten agreeient, small farmers are expiected to sell their produce to the shopkeep.rd wilihasextended them credit As a result 'they,are oftei compelled to;piay above-market prices for inputs;and other commnodities while,receiving below-market prices for their farm produce; experiencing highly unfavorable terms 6f tfade in the process. Farmers finidit difficult to shopfor better offersldcally, since options arefew and they will be refused credit during the next lean season 'if they sell to any retailer other than their:customary creditor. The smlll numberof traders and the femoteness of some areas results in'a near monopoly on'the'servibes traders offer. Efforts to substitute govemment-sponsored programs and irnstitutions for the mudalali'have generally beeri unsuccessful. Many- farmers indicated that they are aware that the middlemen play an important role, that inudalalis are' more efficient than institutional lenders in;temisof providing credit and services when needed;' and that their profits reflect the- risks, they take; At best sniall producers sometimes viewthe middlemen as a reliable source of help arid-protection. Efforts by the govemrient to 'eliminate middlemen. in the rural economy. are unrealistic and of questionable Wisdom, given the'varied functi6ns'they perform in production' and marketing. Source' Based on Parker and Silva (2000). -50- Table 2.13: Patterns in Household Saving and Borrowing (percent) Quintile 1 2 3 4 5 Total (lowest) (highest) Households that save with formal and 71 74 77 81 88 79 semiformal institutions Households that saved but did not 71 66 61 58 53 60 withdraw from formal and semiformal institutions over the last 12 months Households that have loans outstanding 25 23 28 30 30 28 Borrower households that borrow from 24 26 26 21 17 23 semiformal institutions Borrower households that borrow from 418 41 39 45 57 45 formal institutions Borrower households that borrow from 35 34 34 34 27 32 informal sources Households that mortgaged assets over 13 16 20 17 14 16 the last 12 months Note: Expenditure quintiles are based on per capita spending. As a result the number of households is different in each column. Source: 1999/2000 Sri Lanka Integrated Survey, calculations from three quarters. 2.66 Poor people in Sri Lanka tend not use their savings for liquidity management purposes. According to the Sri Lanka Integrated Survey, more than 70 percent of poor households that saved did not withdraw any of their savings. During extensive discussions, many poor households from Kalpitiya in Puttlam district who were members of Isuru Societies-organized as part of the central bank-sponsored Small Farmers and Landless Credit Project-explained their preference to use loans rather than their savings for household cashflow management. Savings are meant for old age security and for their children's education. This notion of saving for the future is also captured by patterns of loan use by the poor. 2.67 Data on the use of loans suggests that poor households that borrow use most of their loans to meet household consumption needs rather than for investment activity (Table 2.14). The second most popular use of credit by the poor is to improve housing conditions. This confirms the bias expressed by the households in Kalpitya toward using loans to smooth household consumption. Household saving and borrowing patterns indicate that poor people are forward- looking users of microfinance. As part of their risk management strategy poor households save for the future and use credit to smooth current household expenditure. 2.68 Nothing in the patterns of household saving and borrowing suggests that poor households are systematically excluded from participating in the financial market. Poor households share the same distribution of formal, semiformal, and informal loans as all other households. There appears to be no substance to the opinion expressed by many officials in interviews that poor people need to be "taught how to save." The government will be short-sighted if it does not recognize the inherent ability of poor people to manage their household cash flow through various means, saving being one of them. The government's focus should, therefore, be to safeguard poor people's savings and to ensure that they have access to them in case of need. -51- Table 2.14: Uses of Loan (percent) . 1 2 3 4 5 Total Quintile (lowest) (highest) Business or farm use Purchase of raw materials 9 9 8 9 12 10 Purchase of equipment 2 4 4 7 6 5 Purchase of land 0 0 0 3 2 1 Purchase of animals 4 3 2 1 1 2 Building improvement 2 5 5 4 8 5 Other business/farm use 10 15 14 13 14 13 To pay workers 0 0 0 0 0 0 Personal use households consumption 31 26 29 19 14 23 Purchaselimprovement of dwelling 27 28 22 29 29 27 Marriagelfamily event 1 2 2 2 3 2 Consumer durables 1 2 4 3 4 3 To settle another loan 1 1 1 2 5 1 Other personal use 13 4 9 7 1 8 Total 100 100 100 100 100 100 Source: 1999/2000 Si Lanka Integrated Survey, calculations from three quarters. 2.69 To what extent are government directed credit programs useful and a good use of public resources? There are several shortcomings in some of the larger targeted credit programs. Despite interest and refinance facilities, recovery rates of two of the largest programs implemented by the Bank of Ceylon-the New Comprehensive Rural Credit Scheme (NCRCS) and the Surathura Diriya Credit Programme (SDCP)-were 76 and 67 percent. By contrast, the recovery rate of the Bank of Ceylon's Small Enterprise Development Programe, which used the Bank's own funds, 28 had a recovery rate of 81 percent despite charging a higher interest rate 2.70 NCRCS provides short-term loans for farmers to cultivate paddy crops and other subsidiary food crops. At the end of 1999, this program cost the government Rs. 68 million in interest subsidy paid to the implementing banks, and Rs. 117 million in expenses incurred by the central bank under the provisions of its Credit Guarantee Scheme. The cumulative loss incurred by the government and the central bank under the program through the end of 1999 was Rs.343 million (Central Bank Annual Report 1999). 2.71 The Surathura program is targeted toward promoting self-employment through microenterprise among unemployed youth. Preliminary findings of a survey of defaulters reveal that poor identification of borrowers was a key reason for low recovery rates. Even loans offered to Samurdhi welfare recipients under the SASANA and SAVANA programs via the Bank of 28 Most targeted program offer an on-lending rate of 10 percent per annum while the SED program charged 13.5 percent per annum (Central Bank Annual Report 1999; Bank of Ceylon Report on Development Credit Schemes 2000). -52.- Ceylon and Peoples' Bank had recovery rates of 74 and 73 percent. Targeted credit programs in Sri Lanka, therefore, appear to be no exception to the general experience of similar programs elsewhere-low recovery rates, poor identification of borrowers, low incentives for bank officials to recover loans, and low incentives for borrowers to repay. Although loans from state banks are a key source of credit for poor households (see Table 2.15), the performance of these programs suggests that they are mostly supply driven and impose substantial costs on the government. 2.72 Although they were intended to be a different type of financial institution, the Samurdhi Banking Societies (SBS) cannot be differentiated from the two state banks in the way they operate. Even though the recovery rate of SBS loans is reported to be 104 percent, it is not clear whether that rate is indicative of the quality of loan performance or of other idiosyncrasies of the program. For example, access to Samurdhi food stamps is conditional on loan repayment. Because money is fungible, one could argue that what is being observed is simply a net re- transfer of Samurdhi food stamps in the form of loan repayments. Further, SBS has only been in operation since 1997, but has increased its portfolio size by 600 percent. Such a high level of expansion in itself raises concern about SBS's institutional base and its capacity to maintain the same level of performance, especially in terms of assessing credit risks and monitoring investments. There is no guarantee that over time people, will not come to view Samurdhi loans as government transfers that need not be repaid, therefore inviting complete capture by "willful" defaulters as was the case in India. 2.73 Another cause for concern is SBS's recent acquisition of Pan Asia Ltd., a commercial bank. It is not yet clear what business Samurdhi intends to conduct through this acquisition and what it all means for poor depositors. Should SBS become insolvent under any circumstance, the government will have to foot the bill using billions of rupees that SBS owes to poor savers. As of the end of June 2000, SBS owed its members Rs. 2 billion in voluntary savings and Rs. 5 billion in compulsory savings due to mature within 12 months. 2.74 Because SBS and the two state banks have access to government resources, they may crowd out the more market-oriented semiformal institutions, mainly small NGOs, that operate using funds borrowed on local capital markets. For example, the Arthacharya Foundation, an NGO that provides microcredit services to a client base of 3,000, borrows from various private commercial banks and other sources at an average cost of about 14 percent. It is no surprise that NGOs like the Arthacharya Foundation remain minuscule in terms of their coverage when compared to the state banks and Samurdhi, or that they target both poor and nonpoor households to diversify risks and minimize costs (Table 2.15). -53- Table 2.15: Distribution of Number of Loans by Source (percent) 1 2 3 4 5 Total Quintile (lowest) (highest) Relatives 10 6 6 6 6 7 Friends 9 14 15 14 9 12 Landlord 1 1 0 0 0 0 Employer 2 1 3 4 0 4 Shopkeeper 6 4 4 2 1 3 Money lender 7 8 6 7 3 6 Pawnbroker 1 0 0 1 0 1 NGOs 2 2 4 2 2 2 Samurdhi Bank 10 9 6 5 2 6 SANASA 7 8 8 8 5 7 Community bank 4 5 7 7 5 5 Commercial bank 27 27 26 26 39 30 Other institutions 11 14 8 8 10 8 Source: 1999-2000 Sr Lanka Integrated Survey, calculations from three quarters. 2.75 The larger semiformal microfinance institutions, cooperative rural banks, SANASA, and Sarvodaya Economic Enterprises Development Service (SEEDS) are all based on a federated three-tier cooperative structure. These institutions largely rely on member savings and donor grants for their operations. None lends exclusively to the poor. For instance, SANASA is more of an institution for the "middle class", an assessment endorsed by people rinterviewed for this. report. All of these institutions mobilize substantial amounts of savings from their members. For example, the cooperative rural banks showed the highest ratio of deposits to credits (20.7), followed by SEEDS at 4.62 (Table 2.16). The deposit-to-credit ratio would have been higher for SANASA except that part of its deposit base is used to build the capital base of SANASA Development Bank. It is no surprise that all four institutions have plans to "graduate" into the formal financial sector given their strong liquid bases. Table 2.16: Deposits and Advances Selected Semiformal Institutions, at end 1999 Deposits Loans Deposit-to-credit (Rs. millions) outstanding ratio (Rs. millions) Cooperative rural banks 12915 623 20.7 SEEDSa 573 124 4.62 Thrift and credit cooperative societies 4238 2033 2.08 (SASANA) Samurdhi Banking Societiesb 1957 1685 1.16 Total 43831 4465 9.82 a. As at end September 2000 b. As at end June 2000 Source: Annual reports and statistical tables provided by each institution - 54 - Programs Sponsored by Nongovernmental Organizations and International Organizations 2.76 Many local and international NGOs provide elements of a safety net in Sri Lanka. Sewa Lanka and Sarvodaya-two local NGOs-have a large presence in the districts of the North East. Sewa Lanka works on several projects in the conflict zone, both in "cleared" and "uncleared" areas, including supply of temporary shelters, agricultural relief project for resettling populations, supplementary food for the displaced and resettling, and rehabilitation and development (tree planting and microcredit). Sarvodaya is involved in construction or repair of basic sanitation facilities, schools, and roads and in making in-kind and monetary transfers for the purpose of building income-generating activities. It also conducts health and nutrition training and operates medical clinics for mothers and children and early childhood development centers. 2.77 The United Nations High Commissioner for Refugees (UNHCR) is the main UN coordinating and implementation agency for ensuring protection of the displaced population. UNHCR implements its activities directly and indirectly by providing project assistance.29 This assistance is aimed at facilitating the resettlement and reintegration of the displaced. Many local and international NGOs and UN agencies are partners in this effort. ' UNHCR's Country Strategy for Sri Lanka 2000-01 (UNHCR 2000). 30 The World Bank's own North East Irrigated Agriculture Project (NEIAP) is a five-year, $32 millioi initiative that is expected to assist war-affected communities in the North and East. - 55 - Chapter 3 Elements of A Strategy to Accelerate Poverty Reduction 3.1 Over the last three years, the Government of Sri Lanka has made an extraordinary effort to deepen its understanding of the nature and root causes of poverty in Sri Lanka, including in the areas directly affected by the conflict, with a view to reassessing and reformulating its policy framework for reducing poverty. The Government's Framework for Poverty Reduction paper presented to donors at the Development Forum in December 2000 signals a welcome shift in the government's poverty reduction strategy. The paper, a product of recent analytical work and of a broad consultation process within Sri Lanka, indicates the need to rely more on economic opportunities and growth than on redistribution to reduce poverty; to correct inefficiencies and inequities in the social protection system, in particularly in the Samurdhi program; to empower the poor through improved governance; and to achieve a peaceful resolution to the civil conflict. More recently, and on the basis of President Kumaratunga's Vision for the 21S" Century (August 1999) and the Budget Policy Statement of 2001, the Government initiated preparation of a "Vision 2010" to be discussed with civil society, donors and the private sector for determining the country's long-term development goals and strategic priorities. In addition, an island-wide consultative process was launched to collect the views of key stakeholders on how the Government, civil society, and the donor community could work together on improving relief and rehabilitation in war-affected areas, develop a code of operation, coordination and institution building, and promote reconciliation and peace building. 3.2 Sri Lanka's experience confirms cross-country evidence that a sustainable reduction in poverty can be achieved only through a continued period of rapid growth. Extensive poverty programs can help the poor in the short run-though they are costly and difficult to target-but in the medium and long run, only policies that trigger fast, private-sector-led growth, combined with the removal of restrictions on the mobility of labor and land, can produce significant and sustained reductions in poverty. 3.3 Tackling poverty in Sri Lanka, therefore, will not be achieved by adding a government program here or there. Reducing poverty will require all policies, expenditures, and programs to be evaluated for their impact on the livelihoods of the average household and on the poorest Sri Lankans. Three broad areas of action are crucial to reducing poverty and raising per capita incomes: creating a policy that is conducive to raising the incomes of the poor, providing effective public services that reach the poor in the core areas of government responsibility, and finding a peaceful resolution to the conflict. Creating a Policy Environment that Facilitates Poverty Reduction 3.4 Households and individuals best escape poverty and reduce vulnerability to downward shocks through their own efforts. Public policies can either encourage or discourage those efforts. A policy framework that helps accelerate growth (in agriculture and other areas), increase real wages, expand employment, and limit inflation will create an environment for poverty reduction. Fundamentals of economic management, discussed in reports from the World Bank and others on Sri Lanka, highlight the importance of prudent fiscal management, financial sector restructuring, outward-oriented trade and investment policies, and a transparent climate for investors large and small. Monitoring the impact of reforms on households, in particular the poor, should continue to be a part of the government's agenda. -56- 3.5 Fiscal constraints mean tougher choices. The fiscal situation in Sri Lanka has been a concern throughout the 1990s, with the deficit remaining between 8 and 10 percent of gross domestic product (GDP). Efforts were made during the decade to contain expenditures, despite increasing defense and interest outlays. In 2000 those efforts were frustrated by the escalation of the war, which caused security-related expenditures to reach 5.6 percent of GDP. If the public budget is to benefit the poor, the government needs to examine the pattern of spending across and within sectors, look hard at what tax concessions and subsidies can be eliminated, and design a pattern of expenditures-not only poverty programs-that emphasizes the poor. In Sri Lanka, the fiscal costs of even untargeted direct subsidies-such as the rice rations of the 1950s, the food stamps of the 1970s, and the Janasaviya and Samurdhi programs of the 1990s-are dwarfed by the costs of bank restructuring and investments in loss-making public enterprises. This highlights the need for an overall analysis of public expenditure patterns from a poverty perspective. 3.6 Accelerating Deregulation and Privatization. To increase growth the pace of structural reform must also increase. In Sri Lanka, reforms to privatize and deregulate the economy, and to remove distortions in markets and prices, remain unfinished. This may help explain the low rates of return on investment in Sri Lanka compared to East Asian comparator countries (Ahmed and Ranjan, 1995). Although some reduction in state control has occurred in recent years through the divestiture of state-owned enterprises and fiscal retrenchment, the state still intervenes in many production and trading activities, crowding out more efficient private initiative. The business environment is constrained by the inadequate adjustments over the years of public institutions, public policies, and the regulatory framework to the needs of the market economy. As a result, the private sector, has not developed to its full potential. 3.7 Policies for Agricultural Growth. In the long run, the sources of sustained growth and poverty reduction will continue to be manufacturing and services. Facilitating the growth of and mobility of resources into these sectors should continue to be a priority. In the short and medium run, however, many of the poor will continue to remain engaged in agriculture. Increasing productivity and returns to labor in agriculture is, therefore, key for poverty reduction in the short run. Currently, agricultural policy, including the trade regime of agricultural inputs, technology, and outputs, is a serious constraint to reducing increasing returns to farming. 3.8 Trade policy for agricultural outputs and inputs must be made more consistent and transparent. Although Sri Lanka followed a relatively open international trade regime in the 1990s, trade policy on agricultural goods and foodstuffs has been unpredictable, contributing to the volatility of farmers' incomes and distorting production decisions. Import tariffs for agricultural produce are often changed ad hoc, and the market interventions of the state-owned Cooperative Wholesale Establishment (CWE) further introduce sharp variations in output prices. This inconsistent and uncertain trade policy has led to large and unpredictable variations in the quantity of imports of major food commodities-including rice, onions, chilies, and potatoes- and thus on the prices Sri Lankan farmers can charge for those commodities. The import regime also discourages the private sector from investing in food commodity processing and storage, leaving the activity solely to the government. CWE is a large parastatal engaged in wholesaling, storing, and retailing of a wide range of consumer goods, including major food commodities. In 1999 it accounted for about 40 percent of total rice imports and about 15-20 percent of total imports of other major food commodities to Sri Lanka. A dynamic private sector can perform more effectively the functions that CWE is meant to perform. 3.9 Flexible Land Markets. The experiences of East Asia, especially Indonesia and Thailand, have shown that an efficient land market is a key ingredient for high growth rates in agriculture. Increasing agricultural incomes and productivity in Sri Lanka will require improving access, use, -57- and tradability of land and reducing government involvement in agricultural marketing. The land market in Sri Lanka is constrained by the state's role as dominant landlord, restrictions on land sale and use in some areas, and inadequate land registry. State control over land, along with ill- defined property rights and tight restrictions on transferability and use of land restrict farmers' ability to obtain credit, move, or sometimes shift into higher value crops.31 Land policies aimed at distributing land equitably among the peasant class have fostered small-holder agriculture with a subsistence orientation. 3.10 Successive measures of land tenure, particularly in the paddy sector, have been introduced in Sri Lanka since the early 1950s. These have established strong state control over land but not clearly defined property rights. Land has been distributed to improve food production and as a means by which indentured workers could earn income. 3.11 The state still dominates land ownership in Sri Lanka, owning between 60 and 80 percent of the land in the country. Two-thirds of state-owned land is farmed by private farmers under varying tenure arrangements. Tenancy arrangements vary in their restrictions on land use and transfer, ranging from very restrictive leases to land grants. Even land grants, which convey title permanently, have transfer restrictions-the government must give its permission before a sale can take place. Two Land Reform Acts, passed in 1972 and 1975, vested a total of one million acres of private and plantation lands in the state. About 20 percent of that land was distributed to the landless in quarter-acre to one acre lots. Between 1920 and 1985, an additional 1.8 million acres of state land were distributed to landless families under various land settlement projects. Another 250,000 acres were added under the Mahaweli Program and other irrigated-settlement projects in the form of village expansion and regularization of spontaneous settlements. Arable land in large irrigation systems in the dry zone is owned by the state, which cedes rights to farm the land to settlers under the land development ordinance permit system. The system permits transfers from one generation to the next, but sales and other land transactions are prohibited. As a result, land markets in major irrigation schemes are nonfunctional-a major long-term constraint for intensification of agriculture. 3.12 Other deficiencies in land administration include documentation, transfers, and the use of land as collateral for mortgages. The Colombo Land Registry has very limited capacity and is unable to cope with the rapidly growing volume of transactions. Land market restrictions limit the transferability of leases and freeholdings, cause distortions, encourage illegal rent-seeking, and increase transaction costs. Land administration problems also affect the system of property taxation, limiting municipal revenue and the availability of land required for infrastructure and other development purposes. Another casualty of ill-defined property rights and weak land administration is labor mobility. 3.13 Flexible Labor Markets. For growth to be effective in reducing poverty, the poor need to be able to earn fair returns on their assets and outputs. Excessive restrictions make Sri Lanka's labor and land markets inflexible, preventing labor, capital, and land from being allocated to more productive uses and limiting returns on assets. Labor laws restrict the scope and range of labor contracts, and the state remains an active employer, paying above-market wages, encouraging queuing and often acting as an employer of last resort. The state owns between 60 and 80 percent of the land in the country and holds a quasi-monopoly in education-around 95 percent of 31 The government may also have feared a scenario in which indebted farmers would be compelled to mortgage or sell their land and thereby revert to landlessness. By preventing farmers from having full ownership of their land, however, land policies may have kept returns to farming low. -58- schools are government run. These distortions constrain growth and contribute to the persistence of poverty. 3.14 Sri Lanka's labor regulations should be made more flexible to encourage private sector growth and job creation. The Termination of Employment and Workmen Act is one of the most restrictive job-security regulations in the world. It protects all workers who have job tenure of at least one year in establishments with 15 or more staff. Under the legislation, termination is possible only with prior written consent of the employee or the commissioner of labor. Just as damaging is the fact that the process of obtaining the commissioner's consent is nontransparent, with dispute resolution taking several months and sometimes years, during which time the firm must continue paying the salary of its redundant worker. One of the consequences of the existing labor legislation is that firms in the informal sector prefer to remain small and not graduate to the formal sector. This impedes economies of scale and job creation.32 3.15 The government has attempted to ease the restrictions by creating export processing zones and providing special concessions through the Board of Investment, but these moves have provided only partial solutions, and only to large enterprises located near Colombo. For smaller enterprises and those located away from the capital, the inflexible labor legislation continues to impose high costs. Providing Effective Core Public Services That Reach the Poor 3.16 In the interest of long-term, sustained poverty reduction, the government must continue to be involved in basic health and education, and improve its delivery of basic infrastructure and social assistance-all core public services that form another area of action for poverty reduction. 3.17 Infrastructure for connectivity and integration. Regional disparities in the incidence of poverty in Sri Lanka reflect, at least partly, the lack of a coherent strategy to integrate and develop different regions of Sri Lanka. Improved provision roads, energy, water and sanitation, and other types of infrastructure is needed to support the development of manufacturing and services in rural areas. One of the most significant problems is the absence of a good road network connecting major supply areas to demand centers, which limits not only non-farm growth, but also limits the ability of farmers to switch to higher value-added crops or expand nonfarm activities. 3.18 Compared to countries at a similar Table 3.1: Urban Population, Sri Lanka and stage of development, Sri Lanka is a more Tbe.: r ation,riL a rural economy. Only about one fourth of Comparative Countries the population lives in urban areas (see Per Capita Urban Table 3.1). Better roads, energy, water and ($us) Population (%) sanitation, and other infrastructure are Albania 750 38 needed in rural areas and linking these to Bolivia 950 62 faster growing urban centers to support China 860 32 incomes. Sri Lanka's rising regional Georgia 840 59 disparities in the incidence of poverty Sri Lanka 810 23 reflect at least partly the lack of a coherent Source: World Bank WDR. regional development strategy. The poor quality of infrastructure-including electricity, water, transport, and communications between 32 The Department of Census and Statistics conducted a survey of industries in 1995, which showed that a staggering 97 percent of the 133,000 firms surveyed in the rural sector employed less than 15 workers. -59- supply centers and markets-limits the switch to higher-value-added crops in agriculture and the expansion of non-farm activities. One of the most significant problems is the absence of a good road network connecting major supply areas to demand'centers. Instead, rural roads, many inadequately maintained, absorb budgetary resources already limited by rising defense expenditures. The absence of rapid access to markets increases transport costs, and as a result prices of goods and even unskilled wages vary sharply across regions. 3.19 As in other areas, "more investment in infrastructure" is not the answer, at least not by itself. Infrastructure must be designed to support welfare through income creation and improved living standards. Funds are wasted when infrastructure and related measures to improve mobility and access miss their targets, when things break down due to poor materials or workmanship, and when the beneficiaries are incapable of using or maintaining the goods and services provided. Adequate funding for operations and maintenance of priority infrastructure, increasingly crowded out by defense spending, needs to be ensured. Top-down approaches have often led to such outcomes. Improved governance-including a leaner, better organized, and more accountable public administration, a more effective budget process,,and a stronger local administration--can help improve the level and quality of basic services to the poor. 3.20 Quality public education services. Improving the quality of other publicly provided services, such as education and health, can help the poor acquire knowledge and skills valued by the labor market. At present, there is a perception that the quality of education is declining. There is a growing shortage of teachers in mathematics, the sciences, and other basic disciplines. In many rural areas these subjects-considered vital to employability-are poorly taught or not taught at all. Class sizes in poor communities are often twice those in wealthier ones. Schools in war-affected areas, border villages, and isolated areas find it particularly difficult to obtain qualified teachers. Poor education for the children of poor households today will result in more poverty and inequality tomorrow. 3.21 In the past, education policies may have reinforced social divisions and had powerftil redistributional implications. Appropriately modified, education and language policy could be a powerful force for social cohesion and poverty reduction. Currently, education is provided in either Sinhala or Tamil, the two official languages, according to the ethnicity of the parents. Students do not typically choose the language in which they learn;. Sinhalese are not required to learn Tamil, nor are Tamils required to learn Sinhala. English, a bridge language, is used as a medium of instruction in a few private schools, most located in Colombo. English is the language of the private sector, and students from upper-class homes in which English is spoken are at a premium in the labor market. English is also the language of instruction for many high-demand university degrees, such as medicine and engineering. The country's language policy severely restricts the ability of children of poor households to obtain better-paid jobs in the growing areas of the economy. 3.22 Effective and affordable safety nets for the most vulnerable. For the majority of the population, including many of the poor, government policy can be most effective by helping the poor help themselves. Untargeted subsidies and large income support programs, such as Samurdhi, are costly and ineffective in reducing poverty sustainably. Basic assistance and income support through well-targeted safety net programs would still be required for four groups of vulnerable people. These are: (i) a small group of people likely to remain chronically poor (orphans, the physically disabled, and elderly and widows without family and kinship ties), because they cannot take advantage of existing opportunities; (ii) people "trapped" in pockets of poverty because they face constraints to mobility and integration in the short run, such as estate, irrigation and resettlement colony populations; (iii) people affected directly by the war, through - 60 - death of breadwinner, injury, displacement, and loss of assets; (iv) people affected by catastrophic risk, such as floods or severe droughts, for whom the best public policy is the promotion of a combination of risk prevention and coping support. A larger group of people who experience intermittent periods of ill health or unemployment may require a temporary cushion that enables and encourages them to return to productive activity. 3.23 An effective safety net should help prevent a very low standard of living by reducing and mitigating downside socioeconomic risk or coping with its occurrences; it should not create dependence or discourage individual effort. In its current form, the Samurdhi program fails to meet this standard. Furthermore, economic theory as well as cross-country evidence from developing countries indicate that transfer programs like Samurdhi cannot achieve sustained reductions in poverty. 3.24 For purposes of identifying beneficiaries of government safety net programs, inferring need through means testing alone, as the Samurdhi program attempts to do, is always difficult. In Sri Lanka, it has lead to leakages, thus becoming costly, and become subject to capture by ethnic or political groups, thus undermining social cohesion. A more effective and affordable set of safety net programs could use a combination of methods of identification, including categorical and means testing criteria (for the chronically poor), geographical criteria (for those in pockets of poverty or affected by catastrophic risk), and self-targeting (for all groups). Geographical transfers to estate populations and others should not, however, replace efforts to remove the obstacles to social and economic integration suffered by these communities. Revamping the safety net program in this way would serve the needs of the poor who cannot participate in the growth process and reduce their vulnerability to negative shocks. 3.25 Geographically targeted, in-kind benefits could also help to address problems of low birth weight, child malnutrition, and anemia. Children of households living in estates and resettlement colonies should be targeted first. Part of these health problems are generally attributed not only to low incomes, but also to low access of households living in estates and resettlement colonies to water, sanitation, and other public services. Therefore, geographically targeted transfers need to be complemented with improved delivery of these public services to households in estates and resettlement colonies. 3.26 It is critical to alleviate conditions caused by the war. Government partnerships with United Nations agencies have benefited war-affected populations in many countries, including Sri Lanka. This report reiterates the recommendations of the United Nations High Commissioner for Refugees that resettlement can be a good strategy for securing livelihoods of refugees and displaced persons, and that emergency relief in cash and in-kind care in refugee camps should continue in the short term. Policy actions to remove restrictions on the mobility of people, goods, and services are expected to help populations affected by the war. 3.27 Public administration. Improvements in service delivery will require a more effective and accountable public administration. Sri Lanka has been compared favorably to other developing countries for its progress on social development and its good public sector management. However, the quality of governance appears to have deteriorated in recent years. Sri Lanka's move to an open, market economy has not been accompanied by similar transformations in the public administration. The public sector is large, costly, and inefficient. With 12 percent of the labor force employed in government in 1999 (17 percent if the mixed sector is included), the country has the largest bureaucracy per capita in Asia. Public sector workers with the lowest levels of education on average earn almost twice the wages of their equivalents in the nonpublic sector. -61- The progressive weakening of public administration is due to several factors. First, spending for new investments, maintenance, and operations has been crowded out by the high wage and pension bill and by rising defense expenditures. While the wage bill has declined in recent years, pensions have increased, and the total has remained at a high level: wages and pensions for the central government alone account for about 5 percent of GDP and 19 percent of total central government expenditure. Between 1986 and 1999 defense expenditures increased nearly three fold in real terms. Second, the civil service suffers from severe skill and motivation problems. Third, the machinery of government is antiquated, fragmented, and overly complex. Finally, the Sri Lankan public administration is laboring under an increasing weight of political pressure and influence, which is attributed by many to the abolition of the independent Public Service Commission.33 3.28 Sri Lanka has not yet realized many of the potential benefits of decentralization- permitting local governments to make their own choices about what services to offer and how to provide them. The design of decentralized government structures in Sri Lanka and the ways in which those designs have been implemented inhibit high-quality, demand-driven service delivery to all segments of the population, particularly the poor. This is due to two main problems. First, inadequate attention has been paid to economic incentives associated with the fiscal aspects of decentralization. Second, the central government has not withdrawn from the devolved, functions. As a result, Sri Lanka has developed a highly complex framework of decentralization that has created confusion and ambiguity regarding the roles and responsibilities of the elected and administrative arms of government. This has reduced accountability and transparency, thereby constraining the growth and maturation of decentralized government institutions. Achieving Peace and Social Harmony 3.29 Reducing poverty and human suffering requires an end to the civil conflict. The consequences of the conflict are felt in every sphere of social and economic life. A study conducted by the Institute of Policy Studies estimates the present-value cost of the conflict during 1984-96 at nearly 170 percent of Sri Lanka's total GDP for 1996. The Central Bank of Sri Lanka, in its 1999 annual report, estimates that the conflict has reduced Sri Lanka's economic growth by about 2 to 3 percentage points a year. There is a widespread feeling of despair and hopelessness among youth, and Sri Lankan society has become more violent. 3.30 Socially divisive policies, initiated during colonial times and continued after independence, have led to social fragmentation along ethnic and political lines. Because institutions of the state and civil society have failed to offer forums for mediating intergroup rivalries and forging cross-cutting ties among diverse ethnic groups, the ethnic cleavage led to conflict, tearing Sri Lankan society and economy apart and leaving many vulnerable to poverty. 3.31 Ethnicity-distinguished primarily by language-and political affiliation have become an accepted basis for competition for political power and for access to material resources. This social fragmentation and the perception of group-based resource allocation have become an obstacle to collective action that keeps the country from fulfilling its potential. Accelerating poverty reduction and growth calls for policies that put national interests above narrow political or ethnic interests. Policies that build social cohesion and economic integration should have a key role in this strategy. 33 The Conunission was restored in 1978, but without its original independence. - 62 - ANNEXA -63 - Table Al. Poverty Measures for Selected Poverty Lines, 1985186-1995196 Poverty line 1985186 199091 1995196 (in 1995196 prices) Incidence of Depth of Severity of Incidence of Depth of Severity of Incidence of Depth of Severity of poverty poverty poverty poverty poverty poverty poverty poverty poverty Rs.633 17 4 1 9 2 0 12 2 1 Rs.673 20 4 2 11 2 1 15 3 1 Rs.713 24 5 2 14 3 1 18 4 1 Rs.752 27 6 2 17 3 1 22 4 1 Rs.792 31 8 3 20 4 1 25 5 2 Rs.831 34 9 3 23 5 2 29 6 2 Rs.871 38 10 4 26 6 2 32 8 3 Rs.910 41 11. 4 29 7 2 36 9 3 Rs.950 45 13 5 33 8 3 39 10 4 Note: Reference poverty line is Rs. 792 per person per month at 1995/96 prices. Excludes the North and East Source: Gunewardena 2000. - 64 - Figure Al. Incidence of Poverty in Sri Lanka for Reference Poverty Line and 20 Percent Higher Poverty Line, 1985/86-1995/96 50 - 45 45 b 40 39 335 - > 30 - 3 0 a. 2 5 25 Rs. 791.67 o +URs. 950.00 O 20 - ~15 .S 10 5 0 1985/86 1990/91 1995/96 Note: Excludes the North and East. Source: Gunewardena 2000. - 65 - Table A2. Incidence, Depth, and Severity of Poverty in Sri Lanka, by District, 1995196 (percent) Incidence of poverty Depth of poverty Severity of poverty District Index Contribution Index Contribution Index Contribution Lower poverty line (Rs. 792 per person per month) Anuradhapura 33 6 6 5 2 4 Badulla 30 5 6 5 2 5 Colombo 10 6 2 4 0 4 Galle 25 6 6 7 2 7 Gampaha 11 5 2 4 0 3 Hambantota 27 4 5 4 2 3 Kalutara 26 7 6 7 2 7 Kandy 30 9 7 10 2 11 Kegalle 25 5 5 5 2 5 Kurunegala 34 12 7 12 2 12 Matale 35 4 9 4 3 5 Matara 28 6 6 6 2 6 Moneragala 49 5 14 7 6 9 Nuwara Eliya 21 4 .4 3 1 3 Polonnaruwa 27 2 6 2 2 3 Puttalam 33 5 7 5 2 5 Ratnapura 37 9 8 9 3 9 Higher poverty line (Rs. 950 per person per month) Anuradhapura 50 5 12 5 4 5 Badulla 48 6 12 5 4 5 Colombo 19 6 4 5 1 5 Galle 39 6 10 6 4 7 Gampaha 21 6 4 5 1 4 Hambantota 43 4 10 4 4 3 Kalutara 38 6 10 7 4 7 Kandy 42 8 12 9 5 10 Kegalle 41 5 10 5 3 5 Kurunegala 53 12 13 12 5 12 Matale 51 4 14 4 6 4 Matara 44 6 11 6 4 6 Moneragala 66 5 22 6 10 7 Nuwara Eliya 40 5 8 4 2 3 Polonnaruwa 40 2 11 2 4 2 Puttalam 51 5 13 5 4 5 Ratnapura 52 8 14 9 5 9 Note: Excludes the North and East. Source: Gunewardena 2000. - 66 - Table A3. Poverty in Sri Lanka by Sector, 1985/86 to 1995/96 (percent) Poverty line =Rs. 792 per person per Poverty line =Rs. 950 per person per month month Sector 1985186 1990/91 1995/96 1985186 1990/91 1995/96 Urban Headcount 18 15 15 28 25 25 Povertygap 4 3 3 8 6 6 Squared poverty gap 2 1 1 3 2 2 Rural Headcount 36 22 27 50 36 41 Povertygap 9 5 6 15 9 11 Squared poverty gap 3 1 2 6 3 4 Estate Headcount 21 12 25 35 28 45 Povertygap 4 2 5 8 5 10 Squared poverty gap 1 1 2 3 2 3 Note: Excludes the North and East Source: Gunewardena 2000. Table A4. Contribution to Poverty in Sri Lanka by Sector, 1985186 to 1995/96 (percent) Poverty line Rs. 792 per person per Poverty line Rs.950 per person per month month Sector 1985/86 1990/91 1995/96 1985V86 1990/91 1995/96 Urban Headocunt 12 16 8 13 16 9 Poverty gap 12 18 8 13 16 8 Squared poverty gap 12 19 7 12 17 8 Rural Headcount 83 80 88 81 79 87 Poverty gap 84 79 89 83 79 88 Squared poverty gap 85 78 89 84 79 89 Estate Headcount 5 4 4 5 6 4 Povertygap 4 4 3 4 4 4 Squared poverty gap 3 3 3 4 4 4 Note: Excludes the North and East Source: Gunewardena 2000. - 67 - Figures A2 and A3. Agricultural Productivity in South Asian Countries, 1979-81 and Output per worker 1979-81 and 1996-98 800 700 - 600 - 500 - O 1979-81 400- g 1996-98 0 300 - 200 100- 0 Bangladesh India Pakistan Sri Lanka 7000 --------- 6000 5000 - o 4000 -- 4000O01979-81 00- S3000 - - - 1996-98 2000 1000 - 0 Thailand Malaysia Indonesia Sri Lanka Source: World Bank 2000a. - 68 - Figure A4. Number of Poor in Sri Lanka, by Province Number of poor (thousands) 340 600 Northern 720 óNOrth-Ce'ntral North no data Western - 2000. Noifth Eastern*- westemn Uva Centra Sa aragamuw Southern Source: World Bank, based on Gunewardena, 2000. - 69 - Table A5. Monthly per Capita Income and Expenditure in Sri Lanka, by Province, 199912000 Per capita Income Per capita Expenditure Average Bottom quintile * Average Bottom quintile* Western 2,464 1,126 2,728 875 Central 1,741 1,545 1,783 815 Southern 1,459 1,136 1,680 847 Northeastern 2,026 1,185 2,027 897 Northwestern 1,376 863 1,655 845 North Central 1,375 756 1,533 928 Uva 1,068 636 1,740 955 Sabaragamuwa 2,231 945 2,201 890 Source: Word Bank, based on 1999/2000 Sri Lanka Integrated Survey. Figure A5. Daily Wages for Selected Tasks, by Province, 199912000 Sabaragamuwa Uva North Central O N.Western 2 Northeastern 0. Southern Central Western 0 50 100 150 200 250 300 350 Rupees per day @ prevailing agricultural wage N paddy transplanting O construction Source: World Bank, based on 1999/2000 Sri Lanka Integrated Survey. - 70 - Table A6. Incidence of Poverty by Education of Head of Household, 1985186 to 1995196 (percent) Poverty line = Rs792 per person Poverty line = Rs.950 per person per month per month Level of education 1985/86 1990/91 1995/96 1985/86 1990/91 1995/96 No schooling Headcount 43 29 38 60 46 58 Povertygap 11 6 9 18 12 15 Squared poverty gap 4 2 3 7 4 6 Primary school Headcount 41 26 34 57 41 51 Povertygap 11 5 8 17 10 13 Squared poverty gap 4 2 2 7 4 5 Lower secondary school Headcount 29 19 27 44 33 43 Povertygap 7 4 6 12 7 11 Squared poverty gap 2 1 2 4 3 4 Upper secondary school Headcount 22 15 20 35 26 33 Povertygap 5 3 4 9 6 8 Squared poverty gap 2 1 1 3 2 3 GCE OIL Headcount 8 7 10 15 13 17 Povertygap 2 1 2 3 3 4 Squared poverty gap 1 0 1 1 1 1 GCE A/L Headcount 6 4 4 7 5 8 Povertygap 1 1 1 2 2 2 Squared poverty gap 0 0 0 1 1 1 Graduate and above Headcount 1 2 1 5 3 3 Povertygap 0 0 0 1 1 1 Squared poverty gap 0 0 0 0 0 0 Note: Excludes the North and East. Source: Gunewardena 2000. Table A7. Incidence of Poverty by Literacy Rate of Household Head, 1985/86 to 1995/96 (percent) Poverty line = Rs. 792 per person Poverty line = Rs.950 per person per month per month Literacy status 1985/86 1990/91 1995/96 1985/86 1990/91 1995/96 Literate Headcount 28 18 23 41 31 37 Povertygap 7 4 5 11 7 9 Squared poverty gap 2 1 2 4 2 3 Illiterate Headcount 45 30 38 61 46 56 Povertygap 12 7 9 19 12 15 Squared poverty gap 5 2 3 8 4 6 Note: Excludes the North and East Source: Gunewardena 2000. - 71 - Figures A6 and A7. Access to Safe Water and Sanitation in Sri Lanka Percent of population Percent of population that lacks access to that lacks access to safe water sanitation E318 [¯¯]I1 ] 26 20-24 32-35 - 30-34 Northem 45-48 Northdm no data no data orth North Easth Eastem North North Westem - Westem c- Centiaa- Westem Westem Source: W Butha o Suc:World Bank, based on 1999/2000 Sd Lanka, Integrated Survey. -72- Table A8. Model Il-Probability of Receiving a Samurdhi Grant (Survey Probit Regression) Standard Scaled Variable Coefficient error coefficient Mean Std. Dev. Number of-children 0-7 years old -0.047 0.042 -0.470 0.391 0.647 Numberof children 8-17 years old 0.012 0.030 0.120 0.906 1.079 Number 6f elderly 0.026 0.044 0.260 0.397 0.648 1, if no rhen in the household 0.132. O090 1.320 0 097 0.296 Household size 0.100 0.022 1.000 4.558 1.743 1, if disable household member 0.206 0.055 2.060 0.315 0.465 Type of settlement, traditional village omitted 1, ifirrigation colony -0.419 0.151 -4190 0.052 0.223 1, if expansion colony 0.114 0.190 1.140 0:058 0.235 1, if settlement scheme -0.023 0.118 -0.230 0.065 0.246 1, if plantation or estate -0.686 0.214 -6.860 0.034 0.181 1, ifurban low income neighborhood -0.034 0.237 -0.340 0.012 0.110 1, if urban mid-income neighborhood -0.351 0.137 -3.510 0.074 0.262 Province, Westem omitted 1:, if Central 0.375 0.129 3.750 0.166 0.372 1, if Southern 0.434 0.109 4.340 0.175 0.380 1, if North Western 0.450 0.116 4.500 0.123 0.329 1, if North Central 0.455 0.139 4.550 0;086 0281 1,if Uva 0.408 0.127 4.080 0.092 0.289 1, if:Sabaragamuwa 0.475 0.112 4.750 0.120 0.325 Ethnicity of the household head, Singhalese omitted 1, if.ethnicity:is Sri Lankan Tamil -0.515 0.126 -5.150 0.072 0.259 1, if-ethnicity Indian Tamil -0.633 0.227 -6.330 0:024 0.152 1, if ethnicity is Moor -0.300 0.146 -3.000 0.053 0.224 1, if ethnicity is other -0.352 0.397 -3:520 0.007 0.086 1, if household owns livestock -0.024 0.108 -0.240 0.078 0.268 1, if household owns assets -0.184 0.101 -1.840 0.297 0.457 Type of identification card, "govemment id" omitted . 1, if "estates id" card 0.118 0.158 1.180 0.029 0.168 1, if "other" card -0.109 0.111 -1.090 0.074 0.262 0.000 1, if member in the military 0.348 0.130 3.480 0.047 0.211 Migration status, never moved omitted 1, if moved before 1950 0.186 0.179 1.860 0.027 0.164 1, if moved between 1950-70 0.145 0.105 1.450 0.092 0.289 1, if moved between 1970-93 -0.125 0.160 -1.250 0.013 0.115 1, if moved between 1993-95 -0.035 0.266 -0.350 0.006 0.078 1, if moved between 1995-99 0.172 0.273 1.720 0.013 0.113 continued on the next page - 73 - Table A8. Model Il-Probability of Receiving a Samurdhi Grant (Survey Probit Regression), continued from the previous page Standard Scaled Variable Coefficient error coefficient Mean Std. Dev. 1,if has land title -0.481 0.101 -4.810 0.191 0.393 Land ownership, owned >1 acre omitted 1, if has no land 0.668 0.302 6.680 0.295 0.456 1,;if has <1 acre ownland 0.415 0. 096 4.150 0.179 0.383 1, if uses plot rotation <1 acre -0.378 0.418 -3.780 0.289 0.453 1, if uses-plot rotation >1 acre -0.697 0.724 -6.970 0.002 0.050 1, if uses operator rotation <1 acre -0.306 0.311 -3.060 0.290 0.454 Occupation of the household head, farm laborer omitted 1, if non-farm laborer -0.101 0.156 -1.010 0.168 0.374 1, if government employee -1.308 0.170 -13.080 0.090 0.286 1, if private salaried employee -0.571 0.175 -5.710 0.070 0.255 1, if employed in estates -0.730 0.204 -7.300 0.038 0.192 1, if employed in petty business -0.649 0.154 -6.490 0.107 0.309 1, if employed in domestic service -0.334 0.249 -3.340 0.012 0.108 1, if a farmer -0.372 0.155 -3.720 0.168 0.374 1, ifa fishermen 0 022 0.238 .0220 0.011 0.107 1, employed in other occupations -1.447 0.671 -14.470 0.003 0.057 1, if house duties only -0.345 0.156 -3.450 0.064 0.244 1, if retired -0.730 0163 -7.300 0.158 0.365 1, if disabled -0.166 0.211 -1.660 0.025 0.157 1, if not working -0.539 0.190 -5.390 0.022 0.146 Years of education of the household head -0.058 0.010 -0.580 7.419 3.374 Type of house, own house omitted 1, if attached house -0.152 0.146 -1.520 0.047 0.211 1, if flat -0.249 0.325 -2.490 0.011 0.107 1, if shanty 0.066 0.301 0.660 0.017. 0.129 1, if other type of housing -0.772 0.230 -7.720 0.039 0.194 Type of latrine, own latrine omitted 1, if has no access to latrines 0.238 0.094 2.380 0.094 0292 1, if has access to comm. latrine 0.053 0.151 0.530 0.040 0.196 Type of lighting used, electricity omitted 1, if no light 0.166 0.141 1.660 0.037 0.188 1, if use kerosenefor light 0.524 0.061 5240 0.337 0.473 Log;household income, less Samurdhi -0.201 0:030 -2.010 8.634 1.183752 Constant 1.323 0.285 13.230 Number of obs = 4761; Number of strata = 2; F( 62, 262) = 12.70; Prob > F = 0.0000 Note: Shaded if (jointly) significant at 5 percent level; absolute coefficient values in probit model cannot be interpreted as marginal effects; scaled coefficients measure the effect of particular characteristics relative to the effect of household size. Source: 1999-2000 Sri Lanka Integrated Survey; calculations from three quarters, the North and East excluded. -74- ANNEX В -75- Historical Background on Safety Nets and Anti-Poverty Programs 1. The country's first food aid prograrn-a rice ration-was introduced in 1939, under British rule. After independence, the Sri Lankan government continued to distribute food. In the 1950s, it distributed two kilograms of rice each month, free of charge, to every Sri Lankan. Between 1966 and 1979 expenditures on various food subsidy programs (rice, wheat flour, sug4r) amounted to 3-6 percent of GDP, or 8-17 percent of total government expenditures. In 1978, the government made its first attempt to improve the targeting of food subsidies. Ration books were issued to those in households that declared a monthly income of less than Rs. 300 (then $20) for a family of five. Some 7.7 million individuals (55 percent of the population of the island) received monthly ration books worth Rs. 15-25 per person. 2. In 1979 rationing was replaced by a food stamp program. This too was intended as a targeted program, with eligibility based on means testing, and government officials were instructed to scrutinize declarations of household income. 34 The number of people declared eligible under the food stamp program declined to 7.2 rmillion, even though the estimated number of individuals living in households with incomes of less than Rs. 300 was 3-4 million. Edirisinge (1987) used data from the 1981-82 Consumer Finance and Socioeconomic Survey to report that 50 percent of all households received this targeted assistance in 1981-82." Program outlays accounted for 1-2 percent of GDP between 1979 and 1984. 3. The food stamp program was pro-poor, but suffered from substantial leakages and under- coverage of poor households. Overall, 67 percent of program spending observed in the Consumer Finance and Socioecofiomic Sur-vey sample went to ho ' useholds in the two poorest quintiles in terms of per capita expenditure. The program had significant impact on food consurription by the poor, representing 20 percent of the food spending of recipients in the lowest quintile and only 6 percent of the food spending of better-off recipients. In addition (and as economic theory would predict), the program was shown to induce a reduction in recipients' work effort. Sahn and Alderman (1996) showed that labor supply dropped by an average of 2.3-2.9 days per month in response to the additional income represented by the food subsidy. 4. Tbroughout the 1980s other, smaller social assistance programs were added, and the wheat flour subsidy was reintroduced. Some-the Integrated Rural Development Program (IRDP), Triposha (a supplementary feeding program), Public Assistance (assistance to destitute widows, disabled, elderly), kerosene stamps, an infant milk subsidy-were supplemented with financing from external donors. Mid-day meals and school uniforms were provided to all school- aged children beginning in the late 1980s. 5. Between 1989 and 1995, the Janasaviya program replaced food stamps as the country's main antipoverty program. This targeted, means-tested program aimed to provide Rs. 2,500 in monthly grants to half of all households in the country,for two years. In practice, monthly grants were fixed at Rs. 1,458 per household, comprising a consumption transfer of Rs. 1,000 and a 46savings" component of Rs. 458 intended for investment by the beneficiaries after the two-year program period. In addition, a lump-sum grant of Rs. 25,000 was promised to beneficiaries at the 34 Means testing requires individual assessment of the means (income) of intended beneficiaries. Only a plicants with income below a certain threshold are eligible. 3 Edirisinghe (1987). Data from the Consumer Finances and Socioeconomic Survey 1981-82 collected by Central Bank of Ceylon, The survey is a representative sample of Sri Lankan households comprising of 8,000 households in all provinces of the country. The survey was conducted between October 1981 and September 1982. - 76 - end of two years. To reduce budgetary pressures, the program was implemented in rounds. Only 4 of 11 rounds planned were implemented, and the lump-sum grant was replaced with a monthly payment of Rs. 250. Social mobilization was a main implementation vehicle, and 6,000 social mobilizers (animators) were employed to identify beneficiaries and boost community participation activities. 6. In return for cash payments, beneficiary households had to provide 20 men days of labor per month during the two-year period, which could be spent in training for private investment projects of the beneficiaries themselves or working on community infrastructure projects. In this regard Janasaviya resembled a public works program, but because it lacked a necessary element of well-designed public works programs-self-targeting through lower-than-market wages-it may have displaced other productive activities that beneficiaries might have undertaken. The benefits provided were large in comparison with income levels, and thus may have created disincentives to work and caused pressure on wages. 7. The targeting of Janasaviya was far from perfect. Results from the 1995-96 Household Income and Expenditure Survey reported by the Department of Census and Statistics of Sri Lanka (DCS) show that this transfer program was substantially pro-poor, but that it, too, had substantial leakages and undercoverage.36 In 1995-96, 64 percent of all households in the lowest quintile, and 9 percent in the top quintile received assistance. Prahdan (1999) confirmed the DCS results by using the 1996-97 Consumer Finance Survey to show that 60 percent of program spending accrued to the two lowest quintiles. Janasaviya outlays amounted to 0.7 percent of GDP in 1992. 36,"Household Income and Expenditure Survey 1995-96. Final Report." Department of Census and Statistics, Ministry of Finance and Planning. The survey of 21,000 households in all areas outside of the North East was conducted between November 1995 and October 1996. In 1996 Samurdhi had begun o erations in a few regions, and benefit distributions might reflect both Janasaviya and Samurdhi. Pradhan (2000). The evaluation is based on 1996-97 Consumer Finance Survey, a survey of 9,000 households outside the North East, conducted by the Central Bank of Sri Lanka. The survey instrument combined information about Janasaviya receipts with information about kerosene stamps. -77- References Ahmad, E., Dreze, Hills and Sen. (ed) 1991. Social Security in Developing Countries. Clarendon Press. Ahmed, S. And P. Ranjan, 1995. "Promoting Growth in Sri Lanka: Lessons from East Asia". Policy Research Working Paper 1478, Washington D.C.: World Bank. Alailima, P., 1997. "Poverty and Unemployment in Sri Lanka", Department of National Planning, Ministry of Finance and Planning, Colombo. Alesina, A., R. Baqir, and W. Easterly, 1998. "Public Goods and Ethnic Divisions." Policy Research Working Paper 2108. World Bank, Washington D.C. Arunatilake, N., S. Jayasuriya, and S. Kelegama. 2000. "The Economic Cost of the War in Sri Lanka", Research Studies: Macroeconomic Policy and Planning Series No. 13. Institute of Policy Studies, Sri Lanka. Atkinson, A. B. 1995. Incomes and Welfare State: Essays on Britain and Europe. Cambridge: Cambridge University Press. Aturupane, H. 1997. The Economic Policy Framework and Poverty in Sri Lanka: Efficiency and Equity Implications, University of Colombo, mimeo. Austin, G. 1999. "Background Note on Civil Strife and Poverty in Sub-Saharan Africa." Background paper for World Development Report 2000/2001. London School of Economics and Political Science; and World Bank, Washington D.C. Barr., N. 1994. "On the Design of Social Safety Nets", Working Paper, London School of Economics. Besley , T., and R. Kanbur. 1990. "The Principles of Targeting." Working Paper. Washington, D.C.: World Bank. Besley, T., R. Burgess, and I. Rasul. 2000. "How Much Should Governments Spend on Social Safety Net?" Working paper, Department of Economics, London School of Economics. Bhalla, S., (1987). "Is Sri Lanka an Exception? A Comparative Study of Living Standards", in Rural Poverty in South Asia, T.N. Srinivasan and P. Bardhan (eds), Columbia University Press. Bhalla, S. and Glewwe, P., 1986. "Growth and Equity in Developing Countries: A Reinterpretation of the Sri Lankan Experience.", The World Bank Economic Review, Vol. 1-1. CARE 2000. "Household Livelihood Security Assesment in LTTE-controlled. Areas of the Wanni. Assessment Report". Central Bank of Sri Lanka 2000. 1999 Annual Report, Colombo, Sri Lanka. -78- Collier, P., 1999. "The Political Economy of Ethnicity." In Boris Pleskovic and Joseph E. Stiglitz, eds., Annual World Bank Conference on Development Economics 1998. Washington D.C: World Bank. Collier, P. and A. Hoeffler. 1998. "On the Economic Causes of Civil War." Oxford Economic Papers 50: 563-73. Conning and Kevane. 2000. "Community Based Targeting Mechanisms for Social Safety Nets", World Bank Working Paper. Consortium of Humanitarian Agencies, Sri Lanka, 2000. Yearbook 1999, Sri Lanka, Colombo. Cox, D. and Jimenez, E., 1995. "Private Transfers and the Effectiveness of Public Income Redistribution in the Philippines" in Van de Walle D. and K. Nead (ed.) 1995. Public Spending and the Poor. Theory and Evidence. The Johns Hopkins University Press. Baltimore and London. Cox, D. and Jimenz, E., 1992. Social security and private transfers in developing countries: the case of Peru. World Bank Economic Review; 6:155-69. Datt, Gaurav and D. Gunewardena, 1997. "Some Aspects of Poverty in Sri Lanka:1985-90" Policy Research Working Paper 1738, Washington D.C.: World Bank. de Ferranti, D., et. al. Securing our Future in a Global Economy The World Bank, Washington DC Deaton, A. 1995. The Analysis ofHousehold Surveys: Microeconometric Analys for Development Policy, Princeton University, mimeo. Drze, J. and Sen, A. 1989. Hunger and Public Action, Clarendon Press-Oxford. Edirisinghe, N. 1987. "The Food Stamp Scheme In Sri Lanka: Cost, Benefits, And Options For Modification." Research report 58. Washington DC, International Policy Research Institute. Foster, J., J. Greer, and E. Thorbecke, 1984. "A Class of Decomposable Poverty Measures." Econometrica 52:761-766. Gabriel G. and D. Cornfield, 1995. "Segmentation and Standardization of Income Determination in Developing Nations: An Analysis of Urban-Rural Differences in Income Determinants in Sri Lanka", Economic Development and Cultural Change. Gelbach, J. and L. Pritchett. 1997. More for the Poor is Less for the Poor. The Politics of Targeting. World Bank Policy Research Working Paper 1799. Glewwe, P., 1988. "Economic Liberalization and Income Inequality - Further Evidence on the Sri Lankan Experience", Journal ofDevelopment Economics, 28. Government of Sri Lanka, 2000. Sri Lanka: A Framework for Poverty Reduction. Grosh, M. 1994. Administering Targeted Social Programs in Latin America. From Platitudes to Practice World Bank Regional and Sectoral Studies, Washington, D.C.: World Bank. - 79 - Gunatilaka, R., et al. 1997. "The Samurdhi Programme: A Preliminary Evaluation." Colombo: Institute of Policy Studies. Gunewardena, D., 2000. "Consumption Poverty in Sri Lanka, 1985-1996: a profile of poverty based on household survey data". Background paper, Policy Framework for Poverty Alleviation in Sri Lanka, Ministry of Finance, Colombo. Hazell, P 1992. "The Appropriate Role of Agricultural Insurance in Developing Countries." Journal of International Development 4: 567-581. Holzmann R. and S. Jorgensen. 1999. "Social Protection as Social Risk Management. Conceptual Underpinnings for the Social Protection Sector Strategy Paper". World Bank. Horowitz, D., 1999. "Structure and Strategy in Ethnic Conflict: A Few Steps towards Synthesis." In Boris Pleskovic and Joseph E. Stiglitz, eds., Annual World Bank Conference on Development Economics 1998. Washington D.C: World Bank. Human Development Center, 1999. Human Development in South Asia, 1999: The Crisis of Governance. Oxford University Press. Isenman, P. 1980. "Basic Needs: The Case of Sri Lanka". World Development Report, Vol. 8. Isenman, P., 1987. "A Comment on "Growth and Equity in Developing Countries: A Reinterpretation of the Sri Lankan Experience" by Bhalla and Glewwe". The World Bank Economic Review, Vol. 1, No. 3. Journal of International Development; 11, No. 7:1005-27, November-December 1999. Khandker, S., 1998. "Fighting Poverty with Microcredit " Oxford University Press. Korale, R.B.M. 1988. Employment and the Labor Market in Sri Lanka. Department of Census and Statistics, Ministry of Plan and Implementation. Ledgerwood, Joanna. "Microfinance Handbook: An Institutional and Financial Perspective". The World Bank. 1998 Little, A.W. (1999). Laboring to Learn - Towards a Political Economy ofPlantations, People and Education in Sri Lanka, St. Martin's Press. Luckham, Robin, 1999. "Complex Political Emergencies and the State: Failure and the Fate of the State." Third World Quarterly 20(1):27-50. Marga Institute 1990. Priority Issues and Policy Measures for the Alleviation of Rural Poverty - Sri Lanka Case Study. Mimiograph. Ministry of Plan Implementation, Ethnic Affairs And National Integration. Regional Development Division. Ministry of Samurdhi, Youth Affairs, and Sports. "Progress 1999, Programmes 2000." Report for the Parliament. Colombo. - 80 - Montgomery, R, D. Bhattacharya and D. Hulme. "Credit for the Poor in Bangladesh: The BRAC Rural Development Program and the Government Thana Resource Development and Employment Programme" in Hulme, D and P. Mosley 'Finance against Poverty' vols. 1 and 2, Routledge. London. Morduch, J., 1998. "Does Microfinance Really help the Poor? New evidence from flagship programs in Bangladesh" mimeo. Nanayakkara, A.G.W. and H.A.G. Premaratne, 1987. Food Consumption and Nutritional Levels. Section 2.2 in R.B. Korale (ed) Income Distribution and Poverty in Sri Lanka, Department of Census and Statistics, Colombo. Nayar, R and R. Gunatilaka. 1999. "The Microfinance Sector in Sri Lanka: Environment, Policies and Practices". UNDP Report. Osmani, S.R. 1993. "Is There a Conflict between Growth and Welfarism? The Tale of Sri Lanka." Working Papers N. 109. World Institute for Development Economics Research, The United Nations University. Parker, B. and K. T. Silva. 2000. "People's Perceptions of Poverty: Social Issues in Sri Lankan Poverty", background paper prepared for the Sri Lanka Poverty Assessment. Parliament of the Democratic Socialist Republic of Sri Lanka. Samurdhi Authority of Sri Lanka Act, Act No 30 if 1995. Certified on 1 loDecember 1995. Rama, M.. 1994. "Flexibility in Sri Lanka's Labor Market", Working Paper. World Bank. Ramanujam, P. and P. Nestel. 1997. "Preliminary report on the fourth national nutrition and health survey, July-August, 1995." Ceylon Journal of Medical Science, 40: 13-24. Rannan-Eliya, R. "Ageing and Pensions Study. Final Report" Institute of Policy Studies, health Policy Programme, Colombo Sri Lanka. Ratnayake, R., 1999. "Proposed Entry and Exit Criterion to the Food Stamp Scheme." Colombo: Ministry of Samurdhi, Youth Affairs, and Sports. Colombo, Sri Lanka. Ravallion, M. 1999. "Appraising Workfare", World bank Research Observer, February. Ravallion, M. 1992. Poverty Comparisons: A Guide to Concepts and Methods. World Bank LSMS Working paper n.88. Ravallion, M. 1998. Poverty Lines in Theory and Practice, World Bank LSMS Working Paper n.133. Ravallion, M. and Q. Wodon. 1997. Poor areas or only poor people?, World Bank Policy Research Working Paper 1798. Reinhard and Kraemer. 1999. Baseline Survey on Health and Nutrition. Ministry of Plan Implementation and Parliamentary Affairs, Integrated Food Security Programme, Trincomalee -81 - Rouse, C.E. 1990. A Study of the Poor in Sri Lanka, World Bank Internal Discussion Paper, Asia Regional Series. Sahn, D. and H. Alderman, 1996. "The Effect of Food Subsidies on Labor Supply in Sri Lanka" Economic Development and Cultural Change v45, n1 October 1996, pp 125-45. Samurdhi Authority of Sri Lanka The Vision, Aims And Development Approach of Samurdhi Program Sethsiripaya, Battaramulla Sri Lanka. Sawhill, I. V. 1988. "Poverty in the U.S.: Why Is It So Persistent?", Journal ofEconomic Literature, Vol. XXVI, September. Sedlacek, et. al. 2000. "Targeted Conditional transfer programs in Latin America: An Early Survey", World Bank, Washington DC. Sen, A. 1976. "Poverty, An Ordinal Approach to Measurement", Econometrica 46:437-46. Shanmugaratnam, 1999. "Rural Poverty in Sri Lanka: A Synthesis of Preliminary Qualitative Studies". Mimeo. Shariff, I. 2000. Sri Lanka: From Subsidized to Demand-Driven Microfinance. Background paper for the Sri Lanka Poverty Assessment. Mimeo. Skees, J., P. Hazell and M. Miranda 2000. "New Approaches To Crop Yield Insurance In Developing Countries." Discussion paper No 55, Environment and Production technology Division, International Policy Research Institute, Washington, DC Sri Lanka Department of Census and Statistics, Ministry of Finance and Planning. 2000. Household Income and Expenditure Survey 1995/96 Final Report. Colombo. Stewart, F., F. Humphreys, and Nick Lea, 1997. "Civil Conflict in Developing Countries over the Last Quarter of a Century: An Empirical Overview of Economic and Social Consequences." Oxford Development Studies 25(1):11-41. Subbarao, K. et al. 1997. Safety Net Programs and Poverty Reduction. Lessons from Cross- Country Experience, The World Bank, Washington,-DC Tudawe I., 2000. "Review of Poverty Related Data and Data Sources in Sri Lanka." IPS, May 2000. Turton, D., ed. 1997. War and Ethnicity: Global Connections and Global Violence. Rochester, N.Y.: University of Rochester Press. UNDP, Sri Lanka 1998 National Human Development Report. UNHCR's County Strategy for Sri Lanka 2000-2001. Update as of April 2000. United Nations High Commissioner For Refugees and Consortium of Humanitarian Agencies, 2000. "Internally Displaced Persons In Government Welfare Centers In Sri Lanka. Workshop Results And Proposals For Action." Workshop On The Situation Of Displaced Persons In Welfare Centers Anuradhapura, 14-15 September 2000. - 82 - Van Ginneken, W. ed. 1999. Social security for the excluded majority: case-studies of developing countries. Geneva: International Labour Office, 1999. World Bank 1990. World Development Report. World Bank 1995. Sri Lanka: Poverty Assessment. Washington, D.C.: World Bank. World Bank 1999. Sri Lanka. A Fresh Look at Unemployment. Washington, D.C.: World Bank. World Bank, 2000a. Attacking Poverty. 2000-2001 World Development Report. World Bank, 2000b. Sri Lanka: Recapturing Missed Opportunities. World Bank Country Report. World Bank 2001. Risk Management in South Asia A Poverty Focused Approach. Mimeo. Yaron, J., M. Benjamin, & G. Piprek, 1997. "Rural Finance: Issues, Design, and Best Practices." The World Bank. Zaman, H., 1999 "Assessing the Impact of Microcredit on Poverty and Vulnerability in Bangladesh." Policy Research Working Paper # 2145. IMAGING Report No.: 22535 CE Type: ER