Report No. 38107-LK Sri Lanka Strengthening Social Protection January 16, 2007 Human Development Unit South Asia Region Document of the World Bank CONTENTS EXECUTIVE SUMMARY ...........................................................................................................................................i 1. POVERTY AND VULNERABILITY ....................................................... ......................................................... 1 1 , I GROWTH,INEQUALITY AND POVERTY 1 .2 SOURCES OF RISK ............................... Aggregate Shocks........................................................................................................................................................ 4 Individual-Specific Risks ............................................................................................................................................ 5 1.3 POVERTYPROFILE Children........................ 2. ENHANCING EMPLOYMENT OPPORTUNITIES AND EMPLOYABILITY ........................................ 14 2.1 KEYLABOR MARKET TRENDS AND OUTCOMES ............................................................................................ 15 Civil Sector HiringPolicies............... 2.3 ENHANCINGLABOR SUPPLY BY INCREAS ............................... 24 2.4 POLICY OPTIONS. Reforming Labor Market Institutions: Moving from Job to Worker Protection 3. IMPROVING SOCIAL INSURANCE .............................................................................................................. 34 3.1 FORMALSOCIAL INSURANCE SCHEMES. Old Age Income Support through Pensio Disability and Survivors Insurance............................................................................................................................ 37 Unemployment Insurance.................... 3.2 SCHEMES FOR THE INFORMAL SECTOR Pensions....................................................... Disability and Survivors Insurance.............. Unemploynent .................................... ...................40 Cross Cutting Issues ............................ 3.3 POLICY O i v o N s ............................................................................................................................................ 41 Improving the Coverage and Efficiency of'Forma1Sector Schemes ......................................................................... 41 Developing a coherent approach to the risks of death, disability, and ill health 44 Building Blocks of' Unemployment Insurance System Adapted to Sri Lanka........................................................... 44 Improving Coverage and Efficiency of Informal Sector Schemes............................................................................. 46 4. COPING WITH RISK: SOCIAL SAFETY NETS ......................................................................................... 49 4.1 OVERVIEW OF SOCIAL SAFETY NETINTERVENTIONS .................................. ................................. 49 4.2 INCOME PROGRAMS TRANSFER The Sainurdhi Program .... Income Support for Disabled .................................................................................................................. 57 4.3 COPING WITH LARGE-SCALE DISASTERS. DROUGHTS. AND DISPLACEMENT ................................................. 59 4.4 REFORMINGSAFETYNETS ............................... (a) Reformingthe SamurdhiSystem: Addressin .................................... 61 (b) Options for Helping the Poor EscapePoverty.................................................................. (c) IHelping the Poor Cope with Vulnerability .... .......................................................................... 67 5. CONCLUSIONS AND RECOMMENDATIONS ........................................................................................... 71 5 .I SUMMARY OF AKALYTICAL FINDINGS ................................................................................. 71 5.2 GAPSIN KNOWLEDGE ................................................................. ..... ....... .75 5.3 POLICY DIRECTIONS ................................................................... 75 BIBLIOGRAPHY ..................................................................................................................................................... 80 ANNEXES .................................................................................................................................................................. 88 Figure A3.1: Generosity of TEWA orders and compensationindex. 2002-03....................................... 91 Box A5 : Microfinance for Social Protection in the North and East . ................................................................... 98 Boxes Box 1.1 Definition of poverty and vulnerability............................................................................................. 8 Box 1.2. Life cycle effects of child malnutrition........................................................................................... 11 Box 2.1: Helpingjob-seekers to be placed in the private sector.................................................................... 24 30 Box 2.3: A proactive approach to small- and medium-size enterprise support ............................................. Box 2.2: Training for the informal sector: The Jua Kali experience, Kenya................................................. 31 Box 2.4: Bangladesh: underprivilegedchildren's education program (UCEP)............................................. 33 Box 3.1 : Examples of informal social security schemes in Sri Lanka........................................................... 40 Box 4.1: The voluntary sector.... .......................................................................................................... 51 Box 4.3: Developing a proxy means test formula......................................................................................... Box 4.2: How much money will it take to move people out of poverty through income transfers? ............. 56 64 Box 4.4: Community participation in targeting welfare benefits-Sri Lanka's experience with Janasaviya.. 65 Box 4.5: Conditional cash transfers as a way to improve school attendance, health and nutrition of the poor 66 Box 4.6: Combining public works with financial services............................................................................ ....................................................................................................................................................................... 67 Box 4.7: Key design elements of a successful workfare program................................................................. 69 Box 4.8: Emergency labor redeployment project .......................................................................................... 70 Figures Figure 1.1: Growth of GDP and GDP per capita. 1991-2003 (in percent) ...................................................... 1 Figure 1.3: Index of real GDP change by province, 1996-2003(in percent) .................................................. Figure 1.2:Change in consumption expenditure by quintile, 1980-2002 (per adult equivalent, in %)........... 2 3 Figure 1.4: Incidence of poverty by sector, 1990/91-2002.............................................................................. 3 Figure 1.5:Demographic ageing, Sri Lanka and world regional averages(percentage of population over 60 years old, 2000-2050) ...................................................................................................................................... 5 Figure 1.6: Employment losses in tsunami-affected districts ......................................................................... 6 Figure 1.7. Income and expenditure by province ............................................................................................ 9 11 Figure 1.9. Formal and informal employment, 1997-2003 (in million) ........................................................ Figure 1.8. Children with low weight for age (percent) ................................................................................ 12 Figure 1.10: Share of unemployment by income decile, 2002 ...................................................................... 13 Figure 2.1: Labor force participation rate, total and by gender, 1994-2003................................................. 15 Figure 2.2: Employment by type ofjob, 1992-2003 (in million) .................................................................. 15 Figure 2.3: Unemployment rate by age group ............................................................................................... 15 16 Figure 2.5: Job creation andjob destruction flows ........................................................................................ Figure 2.4: Unemployment rate by education................................................................................................ 19 Figure 2.6: Number o f firms by size, 2000-03 .............................................................................................. Figure 2.7: Employment growth o f firms (deviations from the common growth trend), 1995-2003............19 19 Figure 2.8: Coverage o f TEWA by income quintiles (in percent, 2000) ....................................................... 20 Figure 2.9: Education by sector o f employment............................................................................................ 25 Figure 2.10: Reasons for not attending school, in percent (for those who have dropped out) ...................... 25 Figure 2.11: Children in communities with poor school resources, by expenditure quintile (percent o f total) Figure 2.12: Education and training received by disabled, 2003 (by type o f disability, in percent) ............. 25 ....................................................................................................................................................................... 27 Figure 4.1: Expenditures o f the Ministry o f Samurdhi and the Ministry o f Social Welfare (percent o f GDP) Figure 4.2: International comparison o f social assistance expenditure (percent o f GDP), 1996-2003 .......... 53 ....................................................................................................................................................................... 52 Figure 4.3: Growing outreach o f t i e Samurdhi program............................................................................... 55 Figure 4.4: District allocations for Samurdlii income transfers ..................................................................... 58 Tables Table 1.I: Vulnerability to shocks by members o f microfinance institutions. 2004 ...................................... 6 Table 1.2: Reproductive health indicators by socio-economic groups, 2000 ................................................. 7 Table 1.3: Probit estimates o f the probability that a household is poor, 2002............................................... 0 Table 2.1: Job flow dynamics, 2001102 and 2002103 .................................................................................... 8 Table 2.2: Estimation o f earnings function (per hour take-home and total pay), 2000 ................................. 22 Table 2.3 : Analyses on gender-wagelearnings differentials .......................................................................... 23 Table 2.4: Analyses on ethnicity-based wage/earnings differentials ............................................................. 23 Table 2.5: International comparison o f intergenerational schooling correlations.......................................... 27 Table 2.6: Summary o f impact evaluations o f active labor market programs ............................................... 29 Table 3.1: Nominal, actual and potential coverage o f formal old age schemes, 2002................................... 36 Table 3.2: Options for old age income support schemes by redistribution, required pre-funding, and risk allocation........... ............................................................................................................................. 43 Table 4.1: Social safety nets expenditures, 2000-04 (Rs. Million)................................................................ 50 Table 4.2: Samurdhi eligibility criteria.......................................................................................................... 54 Table 4.3: Samurdlii recipients by category o f allowance, and o f 2002 ........................................................ 55 Table 4.4: Samurdlii program, deductions o f compulsory payments (Rs.).................................................... 56 Table 4.5: Public works organized in tsunami-affected areas, 2005 ............................................................. 61 Table 5.1 : Summary o f social protection programs, key issues, and proposed policies................................ 73 ACKNOWLEDGEMENTS This report was prepared by tlie team led by Milan Vodopivec, Senior Economist (SASHD), based on background papers or contributions by Nisha Arunatilake (statistics on labor market outcomes and trends, analysis o f tlie informal sector); Harsha Aturupane (inputs on sections on education and training); Ramani Gunatilaka (chapter on social safety nets, and sections on microfinance and analysis o f wages); Dileni Gunewardena (background paper on o f risks, vulnerability, and social safety nets), Rasmus Heltberg (conceptualization o f tlie approach, analysis o f wages), Siryani Hullugalle (microfinance), Ambar Narayan (analysis o f poverty), Kumari Navaratne (reproductive health indicators), Robert Palacios (chapter on social insurance), Rasika Ranasinghe (intergenerational schooling mobility, analysis o f growth, poverty and vulnerability), D.G. Harendra de Silva (child malnutrition), and Princess Ventura (civil service employment). The report was prepared under the overall guidance o f Julian Schweitzer (Sector Director, SASHD), Mansoora Rashid (Sector Manager, Social Protection, SASHD), and Peter Harrold (Country Director, Sri Lanka). Tlie team benefited from comments at the concept note stage from Carlo del Ninno (HDNSP) and Stefan0 Scarpetta (HDNSP) and information obtained from the Ministry o f Labor and Foreign Employnient, as well as from valuable comments by Ilisan Ajwad, Christine Allison, Rocio Castro, Nina Kataya, Esperanza Lasagabaster, Philip O'Keefe, Kalanidhi Subbarao, and Tara Vishwanath. Tlie team would also like to thank the Central Bank o f Sri Lanka for providing data on contributors to Employees' Provident Fund, to the Ministry of Labor and Foreign Employment for data on TEWA cases, and to Fei Gao and Jakob Tomse for excellent research assistance. EXECUTIVESUMMARY "... given the near universality of market failures in poor countries, it should be possible, with a combination of careful research and good thinking, to identify opportunities for redirecting resources to poorer people who are in aposition to make very good use of them. WorldDevelopment Report 2006 1. Sri Lanka's accomplishments in the area of social policy surpass most developingcountries. Sri Lanka has made stellar progress on the millennium development goals: the country has already attained universal primary education and completion and has achieved gender parity at primary and secondary education levels. Sri Lanka's child and infant mortality rates are unusually low relative to other developing countries, and also compare favorably with some middle income countries. While challenges remain-particularly in improving the quality of education services, tackling persistentchild malnutrition, and addressingregional disparities in social outcomes-Sri Lanka's accomplishments are remarkable for a developingcountry, particularly one that has endured a 20 year civil conflict (World Bank 2005a). 2. Sri Lanka's performance on poverty reduction has been more modest. Almost a quarter of Sri Lankan population is poor, with poverty concentrated in rural areas. Despite average annual economic growth of 4.5 percent in the last decade (1994-2003), the decline in poverty has been modest, dampened by growing inequality.' If inequality had not increased, poverty reduction would have been more than 5 fold higher. That said, given its much higher level of income, poverty rates in Sri Lanka are still much lower than for most South Asian countries (World Bank 2005a) and while inequality is much higher than most countries in the region, it is far lower than for many Latin American countries (World Bank 2006). 3. Aside from poverty (low income), Sri Lankans also face considerable vulnerability to income shocks. The poor, with few assets and limited access to markets, are often the hardest hit, but income shocks can also drive many non-poor into poverty. Results from a recent survey, consistent with results from other South Asian countries, show that the major individual risks faced by Sri Lankan households are sickness, disability and death o f a family member and unemployment, and the main comnzunity-wide (aggregate) shocks include drought, crop failure, and other natural disasters (the most recent and disastrous being the tsunami).2 Moreover, Sri Lanka has one of the most rapidly aging populations in South Asia. Over the next 25 years, the share o f the population over 60 will double from about 10 to 20 percent and this demographic trend will have an aggregate impact on the economy, 'Gini Index o f per capita consumption expenditures, a measure of inequality, increased from 0.32 to 0.40 over the period 1990/91 to 2000. 2The microfinance survey found that over a twelve month period, over a half o f households experienced significant income shocks, with 21 percent of households being exposed to sickness, injury and death, 16 percent to unemployment, 23 percent to natural calamities, and 18 percent to crop failures n Sri Lanka is amongst the highest in the region, it is lower than found in many countriesof Latin America. 1 potentially changing patterns o f labor force participation and the composition o f health care, and imposing a strain on traditional and formal income support system^.^ 4. There are several poor and vulnerable groups deserving attention. Among children, these groups include those living in large families (the poverty incidence o f households with 3 children is 57 percent above the national average and o f households with 4 or more children i s more than double the national average), disabled children and child workerdstreet children, children affected by the conflict (child soldiers), children who drop out from school (though their numbers are very small relative to other South Asian countries), and malnourished children. Among working age adults, the most poor and vulnerable are informal sector workers living in remote and infrastructure poor areas (particularly uneducated, casual and own-account workers), unemployed youth from low income families, displaced due to the civil conflict or natural disasters, and the disabled. Among the elderly, only the oldest face above-average poverty rates (the incidence of poverty among households with member over 80 years old is I O percent above the national average), but rapid aging o f the population may lead to strains in the traditional and formal income support systems and contribute to future vulnerability among the elderly. 5. Sri Lanka is committed to promoting equitable growth and has instituted several social protection programs. A fairly extensive social protection system comprises three main components: (i) e/iip/oy/iient yrorecrion and promotion: labor legislation, unions, collective bargaining and related instit11tions, and training/retraining of workers; (ii) social security/insurance programs: pensions, disability, survivor insurance (coupled with universal health coverage); and (iii)safely nets, or protection o f last resort: mainly cash transfers and social welfare and care services. Informal networks, including those relying on migration and remittances, continue to have an important role in helping people weather chronic poverty and income shocks, but the viability o f such networks given urbanization and aging is in question; and evidence suggests that informal arrangements are not able to provide protection against community wide shock^.^ 6. This report reviews Sri Lanka's social protection programs and proposes strategic options for enhancing their role in promoting growth with equity. Well designed social protection (SP) systems can help address poverty and inequality through redistribution, and mitigate risks and facilitate employment opportunities, thus contributing to both growth and equity goals.5 The report first identifies the poor and vulnerable (chapter I)and then reviews and evaluates employment protection and promotion policies and programs (chapter 2), social security/insurance schemes (chapter 3), and safety net programs (chapter 4) and proposes policy options The concluding chapter summarizes key analytical findings and presents a unified policy framework to improve social protection. The report relies mainly on extensive existing literature from Sri Lankan and international researchers. The remaining part o f this section -'While more evidence on income shocks is merited, risks facing Sri Lankan households are not dissimilar to those affecting other households in South Asian region. Recent evidence on income risks in such diverse countries as Afghanistan, India (Andhra Pradesh) and Maldives suggest that health and disability shocks are amongst the most important income shocks (in terms o f frequency and lost income) faced by households in these countries (World Bank 2005b, 2005c, 2005d). And South Asian countries are highly prone to natural disasters, as the most recent earthquake in Pakistanbears witness. See Townsend (1994), Morduch (1995, 2002), Ravallion and Chauduri (1997), and Munshi and Rosenzweig (2005) for evidence on informal support networks in promoting consumption smoothing. 5 See World Bank (2001) for a description of how social protection issues are addressed by the social risk management framework, and World Bank (2005e) for a discussion o f how equity contributes to growth. Note that social risk management approachcategorizes mechanisms into three categories: (i)those that reduce social risks (for example, reduce the probability of becoming unemployed or sick); (ii)those that mitigate social risk (for example, reduce the impact of unemployment or sickness); and (iii)those applied in response to the undesirable event- coping mechanisms (for example, social assistance). ii summarizes the main issues and policy options. (The main results o f the poverty and vulnerability chapter have been detailed in paragraphs 2-4 above.) I.SriLanka'sSocialProtectionSystem 7. As in most countries, Sri Lanka's social protection system has three main elements: employment protection and promotion; social security/insurance, and safety nets. Its coverage is much more extensive than in other South Asian countries, with up to a third o f the workforce covered in case o f employment protection and social security, and about 40 percent o f households covered in the case o f the largest safety net (cash benefit) program. Social protection expenditures amounted to approximately 3 percent o f GDP in 2004 (2 percent for pensions and 1 percent for safety netsebut the large implicit debt o f the pension system (estimated at 60 percent o f GDP) means that the above costs underestimate the full fiscal costs o f the social protection system. 1. EmploymentProtectionand Promotion 8. Sri Lanka is to be commended for providing basic protection of core labor standards- albeit to formal sector workers. Sri Lanka has ratified eight ILO conventions on core labor standards, including the right to collective bargaining and freedom of association. Tripartite mechanisms for social dialogue have been in place for a long time and trade unions, employers and the government interact collectively to resolve critical issues in industrial relations, e.g., wages and other disputes. While the problem of child labor (street children and child soldiers) remains, strong child focused policies (e.g., inclusive education system and employment protection) have led to one of the lowest rates o f out of school children (and incidence of child labor) in all o f South Asia. Gender discrimination in wages is present-women earn less than men, as a result o f productivity differences, but also because o f other non- economic factors-but there is no evidence o f ethnicity-based differences in earnings. Key Issues 9. However, Sri Lanka's labor market institutions provide excessive job protectionfor formal sector workers. Sri Lanka's severance pay (TEWA) system is one o f the most restrictive severance pay systems in the world (for example, a Sri Lankan worker with 20 years o f experience is awarded by a severance pay o f 39 monthly wages vs. average o f 16 in other Asian countries and 6 in OECD countries), and the process o f separation requires prior approval by the Labor Commissioner and is thus non- transparent and arbitrary. Wage setting institutions have also considerably raised the returns o f public sector workers and `protected' private sector workers (covered under the TEWA) relative to informal sector workers. Adding to perverse labor market incentives (and contributing to government over- staffing) are patronage-based appointments to the civil service. 10. Excessivejob security leads to lower productivity and exclusion of vulnerable workers from formal sector jobs. Labor market institutions that encourage worker voice can help reduce discriminatory practices, enhance work place health, safety, and training-improving equity and productivity o f covered workers. However, international evidence suggests that excessive protection to incumbent workers ('or insiders') can reduce the reallocation of labor required to enhance economic productivity. It is perhaps not surprising, then, that job flow rates in Sri Lanka are far lower than those found in developed, transition and other developing countries (in early 2000s, Sri Lanka's job creation rate was 8 percent and job destruction rate was 4 percent, vastly lagging behind the average job creation 6 For an extensive survey on the literature of the economic and employment impact of unions and collective bargaining see Aidt and Tzannatos (2000), based on OECD experience, and World Bank (2005) and Nickell and Layard (1997), based on experience ofother countries. ... 111 rate of 14 percent and job destruction rate of 11 percent in selected 17 developed, transition and developing countries, see chapter 2). Moreover, international evidence shows that restrictive employment protection legislation, like the TEWA, reduces job prospects of vulnerable groups (or `outsiders') e.g., women, youth, and informal workers, as observed in Sri LankaS7Thus excessive protection o f workers has not only efficiency costs, but also results in exclusion of vulnerable groups from the labor market. 11. Wage setting institutionsincreasewages of formal sector workers relative to informalsector workers. There is compelling evidence that wage setting institutions (unions and collective bargaining, tripartite pay commissions) have created an artificial gap betweenbetter-payingjobs in the public and the "protected" private sector, and low-paying jobs in the "unprotected" private sector. Empirical analysis shows that public and formal private sector jobs command an important wage premium which cannot be explained by the productive characteristics of the workers. Other things equal, returns of public sector workers are 14 percent more, and returns of the `protected' private sector workers (covered under the TEWA) are 11-12 percent more than returns of informal sector workers. In addition, public sector workers enjoy other benefits such as tax exemption, job security and non-contributory pensions. Unions are also more likely to be active in this sector, compelling greater compliance with protective labor legislation and engaging in collective bargaining. However, women earn less than men, as a result of productivity differences but also because of other factors, such as discrimination. On the positive side, there is no evidence of ethnicity-based differences in earnings (with few exceptions, studies show statistically insignificant differences in earnings, once differences in observable characteristics of individuals are accounted for). 12. Recent evidence supports that high unemploymentis linked to the queuing and institutional hypotheses for labor market rigidities.' The civil service wage premium attractsjob-seekers to queue and thus generates unemployment. Similarly, the wage premium of TEWA-coveredjobs increases costs and reduces labor demand.Note that TEWA-coveredjobs also carry non-wage costs-the expectedcosts of layoffs (if explicitly acknowledged at the time of the contract, the expected costs of layoff should reduce wages, and so the amount of the wage premium of the TEWA jobs underestimatesthe additional costs borne by employers). Adding to perverse labor market incentives (and contributing to government over-staffing) are patronage-based appointments to the civil service as a way of obtaining political support or fulfill promises, with insufficient private job creation often used as a rationale (in the 2005 Budget announcement, for example, 30,000 graduates were recruited and 40,000 temporary staff were made permanent). 13. Informal sector workers have lower education levels which reduce their labor market prospects-a result, in part, of high dropouts among children from poor families. Many school dropouts can only qualify for unskilled occupations, the occupations which are paid the lowest and which are associated with the highest poverty incidence. In 1999/2000, 7 percent of 5-14 year olds did not attend school. Dropouts occur both because poor families cannot afford schooling costs and because the expected benefits of schooling for childreii from poor families are low. Direct and opportunity costs of schooling of poor families are relatively higher because by sending children to school, the sacrifice in terms of foregone work at home and direct costs of schooling are higher than the sacrifice of non-poor families. Expected returns of schooling for poor children are lower because their schooling is of lower quality and becausethey lack the social networks which would enable them accessing higher payingjobs. 14. Sri Lanka has few passive or active labor market programs, with scope for piloting new approaches. Publicly administered and provided active labor market programs (job counseling, 7See, for example, OECD (1999)and Elmeskov, Martinand Scarpetta (1999) for the impact of strict limitationson terminations of employees. We are referring to the findings o f Raina(2003), World Bank (1999), and Heltbergand Vodopivec (2004). iv information, training, wage subsidies) exist on a very small scale in Sri Lanka, and their effectiveness is not known. Indeed, this is quite positive as Sri Lanka can design and develop programs based on an evaluation o f both domestic interventions and international experience. Evidence from OECD countries suggests that publicly provided programs are often very costly and ineffective, with job counseling/search and job information being the exception. Well targeted programs to vulnerable groups can also be effective, but their cost needs to be carefully controlled. Evidence from developing countries suggests that privately provided, but publicly regulated training programs work best to link job seekers to jobs. The Government is considering the introduction o f a passive labor market program: a formal unemployment insurance (UI) system, modeled largely on an OECD model. However, international evidence suggests that unemployment benefits need to be much more simply designed if implemented in low income countries (World Bank 2003). Sri Lanka is also addressing unemployment among youth by preparing a national action youth plan for implementation in the coming year. Policy Options 15. Strong economic growth, a sound investment climate, and high quality education systems are critical for improving employment, productivity and wages for all workers. However, well functioning labor market institutions that balance job and worker protection also importantly contribute to favorable labor market outcomes. 16. Reducing excessive job security for formal sector workers will not only enhance productivity, but also improve job prospects for vulnerable groups. The costly protection o f formal sector jobs via excessive termination benefits could be gradually reduced to international norms. In particular, it would be important for policy makers to reconsider maintaining the exceedingly high costs o f separation and the discretionary power o f the Labor Commissioner for granting permission for layoffs. Avoiding direct government interference in wage setting (apart from setting o f minimum wages) would also reduce labor costs. Ad-hoc recruitment o f unemployed graduates should also be reconsidered, as it reduces incentives for youth to take non-government jobs. Reducing excessive protection for `insiders' could, in tandem with other economic reforms, promote job flows, and hence higher economic productivity. Increasing access to formal sector jobs (with greater access to basic workers rights and social security-see next section) for `outsiders' or less educated and marginalized workers would also promote equity and contribute to a reduction in labor market duality. I7. Replacing `job' protection with `worker' social protection programs can promote market efficiency and helps allay the political costs of reform. Reducing job protection for formal sector workers is politically difficult, to say the least, and often meets with stiff resistance from `insiders' because it increases the prospects o f layoffs and dismissals from previously protected jobs. Yet, many countries have taken this difficult step. Promoting social dialogue and negotiating solutions acceptable to all stakeholders have been critically important for fostering change. Ensuringprotection o f basic workers rights is also essential. Moreover, the introduction o f well designed, cost-effective active and passive labor market programs that provide more efficient protection to workers in the long run (in lieu o f protection o fjobs) and that help allaying the short term political costs o f reforms has proved instrumental in achieving this policy shift (for example, in Eastern European co~ntries).~Once in place, active and During the early transition o f Eastern European countries, reductions in job protection were accompaniedby the active and passive programs to ease restructuringilayoffs. While the programs were initially generous and increased fiscal costs, their generosity was significantly reduced over time, as the goals of the restructuring agenda were met, though reducingentitlements was difficult (see Vodopivec, Worgotter and Raju 2005). That said, it should be noted that too generous worker protection programs can also impede effective functioning of the labor market, and these programsalso need to be cost effective in design (World Bank 2000). V passive programs also facilitate restructuring efforts and thus promote dynamic efficiency. Sri Lanka is already taking steps to create a more explicit `worker' protection system to support less stringent protection of jobs, but a key challenge will be to adapt these programs to the fiscal and administrative capacity o fthe country. In this respect: The unemployment insuranceprogram will need to be adapted to Sri Lankan context. This program could potentially lay the foundation for a more efficient protection of formal workers and ease the political costs of reducingjob protection, but its design, costs, and administration will need to be modified from a standard OECD model and adapted to the Sri Lankan context (see social insurance section below). Active measures to enhance skills/employabilityof lower incomeworkers should be piloted, prior to scaling up. Strengtheningcost-effective programs(e.g.job counseling/information) and providing informal sector workers better access to well designed and affordable training through demand driven, private sector based approaches could bring greater equity and productivity gains. Programs specifically targeted to women, youth, disabled, and the poor living in remote rural areas could also be piloted (for youth and disabled these pilots should be based on the recoininendations o f nationaI action pIans). 2. Social insurance 18. Sri Lanka provides the most extensive social security (pensions, disability, survivor and health) coverage in South Asia. The provision of social security is largely tied to participation in the formal sector, although since tlie 1990s, a new public sponsored scheme reaches a significant number of farmers and fishermen, despite its voluntary nature. Together, these programs cover a third o f the population, much more than any other South Asian country. Aside from pensions, these workers are offered disability and survivor insurance as part o f their benefit package. Coverage for medical expenseddrugs to cope against catastrophic health shocks is provided through a universal health insurance scheme that is not tied to scheme membership, although formal sector workers have added benefits. Key Issues 19. Nevertheless, the coverage of current schemes is limited and their administration and financial sustainability could be considerably strengthened. First, there are large gaps in coverage, with tlie vast majority of informal sector workers lacking coverage and considerable evasionamongthose in the formal sector. Two thirds o f the labor force remains without insurance for life's major risks in the face of dramatic demographic ageing trends. Second, pensions provide some consumption smoothingover the life cycle only for civil servants, but not for workers participating in other retirement schemes. The Employees' Provident Fund provides insufficient benefits due to low investment returns, a low pensionable age, and the lump sum payout that does not offer longevity insurance (for outliving one's savings). Third,the fiscal sustainability of these schemes is questionable.At 2 percent of GDP, Sri Lanka has the highest pension spending on civil service pensions in the region, and preliminary estimates indicate that the pension system faces an unfunded liability on the order 60 percent of GDP. Although the EPF is a defined contribution scheme which is self-financing by definition, it, too, depends on government financial capacity because practically the entire EPF portfolio consists of government vi paper." Andfourth, pension programs have weak administration and regulation, and limited portability o f benefits (pension rights that are not transferable acrossjobs) constrains labor mobility. 20. Two publicly sponsored programs offer social security for informal sector workers, but face similar and even more complex challenges. Farmers, fishermen and self-employedare the main groups served by these programs. Because these workers do not earn regular wages, their contributions and benefits are set in flat nominal terms and both are steadily eroded by inflation. Minimum pensions have been increased by fiat. 111the long run, the relationship between the assets and the liabilities in these two schemes may lead to broken promises, a massive bailout, or both. Administrative costs are high, eating into members' contributions, and investments are once again concentrated in government paper. 21. Informal sector workers also participate in quasi-formal group arrangements for death benefits. There exist a large number o f self-help associations that have traditionally provided funeral assistance or death benevolence benefits, and they have recently beenjoined by a few non-governmental organizations. However, the benefit levels o f these schemes are set ad hoc and are not indexed to inflation, the financial sustainability of schemes is not known, and their regulation and administration are likely to be weak. 22. As noted above, a formal unemployment insurance (UI) system is being considered and workfare programs-a safety net cum unemployment insurance program for informal sector workers-are largely ad hoc. The government has formulated several OECD type proposals for unemployment insurance programs, but this scheme will be geared to formal sector workers. To provide income support for the workers in the informal sector, workfare schemes exist on an ad-hoc basis, but Sri Lanka does not have a permanent workfare scheme (offering low wages to target the most vulnerable able bodied workers) that can be scaled up to help informal sector workers earn benefits in times o f unemployment caused by layoffs or natural disasters, and scaled down when these emergencies are passed (a well-known example of such a scheme is provided in Argentina). Policy Options 23. Social insurance institutions could substantially improve their services, as well as expand their coverage and scope, and financial solvency. Inthe short run, important gradual reforms include: 0 Improvements in the functioning of old age income support schemes, and consideration of possibilities to expand coverage to the informal sector. In the case o f the civil service scheme, these measures include gradually reducingtarget replacement rates (pension benefits as a share o f individual's wages) and introducing employees' contributions, increasing retirement ages in Employees' Provident Fund and approved retirement schemes, and improving the linkage o f benefits to contributions (or introducing fully contributory schemes) rather than specifying nominal benefits and contributions in schemes covering informal sector workers. Moreover, with regard to death and disability insurance, public policy objectives should be clarified with a view to reducing arbitrary differences and inequities between workers. Once again, political economy considerations will be important, especially in case when benefits are affected. However, many countries have addressed this issue by introduced changes very gradually, with several options, including grandfathering, available to workers impacted by the change (e.g., India, Russia). Adaptation of the proposed unemployment insurance scheme and a consideration of workfare to address unemployment shocks. The proposed UI system will need to be adapted to I OIn the defined contribution scheme, the stream of contributions and the returns on ensuing investments determine pension benefits; in a defined benefit scheme, pension benefits are determined by a specific, predeterminedformula (and thus benefits are usually only loosely linked to contributions). vii Sri Lanka's fiscal and administrative capacity. For example, the elimination o f personalized monitoring o f continuing eligibility conditions and a much simplified benefit and administration structure would render it more appropriate for a low income country. For informal sector workers, particularly the poor, a workfare program might be considered which can be self-targeted to the able bodied poor. However, its costs will need to be assessed and administrative arrangements developed to ensure self-targeting and the program piloted, prior to scaling up. For all schemes, information systems (record keeping, forecasting) need to be upgraded and supervision improved. Funded schemes also need to strengthen fund management and regulation, and coiisider diversification o f their investment options. 24. A longer-term strategy is to create a seamless, unified framework for formal and informal sector workers, covering old age, disability, survivor and unemployment risks. Social security (old age, death, disability) should be based on a defined contribution scheme (or with contributions strongly linked to benefits) with parameters set so as to produce a target replacement rate. This would allow for complete portability o f benefits (individuals would not lose accumulated benefit when they move jobs, easing labor mobility) and increase economies o f scale in administration (e.g., record keeping) and avoid the creation o f new liabilities. Death and disability insurance, fairly priced based on actuarial results, could be purchased for members o f this scheme. Annuitization o f benefit (the conversion o f a lump-sum benefit into a stream o f benefits paid in regular intervals) after retirement could also be made possible. Similarly, a funded severance portion o f the system could be added and integrated into the unemployment insurance (UI) program, with UI benefits being funded by withdrawals from both individual saving accounts as well as from the solidarity fund, the latter being financed by pooled contributions o f workers and their employers. This approach offers advantages not only in terms extension o f coverage, but also greater labor mobility and financial solvency. A unlJied system would need to be designed carefully, and would only be implemented gradually once improvements in regulation and administration of current programs are achieved. Increased formalization of the economy would complement expansion in coverage. 3. Social Safety Nets 25. Sri Lanka has put in place an extensive social safety net system to address chronic poverty. The safety net system comprises Sainurdhi, the main program cash transfer program to address chronic poverty, and disability payments. I n addition, the system provides limited social welfare and care services; and disaster relief to all affected persons on the basis of the impact and injuries suffered. Relief programs are also in place to assist families cope with income loss associated with drought. The Government and private response to the recent Tsunami was exceptional, with special cash transfer and other programs implemented to help affected families cope with the disaster. 26. The safety system could be better designed to serve its objectives. Specific areas where the safety net system could be strengthened include: The targeting, the level of benefit and work incentives of the Samurdhi program could be improved. The program is poorly targeted, with a low level o f benefits, and limited exit provisions. Since its introduction in 1995, the income transfer component o f the program utilizes the bulk o f resources and has the widest outreach. At the end o f 2002, a total o f 1.9 million families or 41 percent o f the population received the income transfer (Central Bank o f Sri Lanka 2003). While the allocation o f funds at the central level is relatively equitable, much o f the mistargeting occurs due to the lack of explicit targeting criteria at the local level. Moreover, reflecting a decrease o f beneficiaries and an erosion o f nominally fixed benefit levels, program expenditures declined from 0.9 percent o f GDP in 2001 to 0.4 percent o f GDP in 2004. Its ... Vlll welfare impact is minimal-for example, in 2000 on average it amountedto less than 15 percent of total household food expenditure. Thus, as currently designed, the program provides inadequate assistance to the poor. It also does not adequately help the poor escape poverty (e.g. through investinent in their h11man capital), The disability program does not reach all poor disabled people. The second largest income transfer program, cash conipensationfor disability, is primarily focused on the disabled soldiers injured in the conflict, leaving other disabled groups under-covered(survey data show that only 4 1 percent of the disabled receive any sort of income transfer, including the disability allowance, from the government). Program expenditures have increased steadily over the years, reaching approximately 0.2 percent of GDP in 2004. The combined spending on the two main cash transfer programs is thus about 0.6 percent of GDP-generally consistent, or even higherthan countries in the region. It has to be stressed that disability payments are a limited way of supporting the disabled poor, and that a more comprehensive approach to disability is needed that addresses preventionand enables the disabledto integratebetter into society. Social welfare and care services exist, but provide limited coverage. Social welfare and care services are an essential complement to cash programs, addressing social risks (e.g. alcohol and drug abuse, orphan hood, protection o f legal rights), particularly for the poor. Currently services are provided publicly and through non-governmentalorganizations, but the coverage is small and their impact is not well known. Relief programs could be strengthened in their ability to support affected populations. Displaced persons are also eligible for rehabilitation assistance for permanent settlement but fiscal constraints limit program coverage. The relief for the displaced (in the form of food assistance) during drought conditions is insufficient to meet basic nutrition needs. Po[icy 011tiotis 27. The equity objectives of the safety net should be strengthened, but the program could also benefit by focusing on alleviating vulnerabilities and promoting growth. Policy options include: Revamping existing social assistancekashtransfer programs to target the poor. The ongoing reform-aiming to increase the amount o f support within existing budget through better targeting-is a an important initiative, and similar reforms have been successfully introduced in many countries around the world (e.g., Armenia, Mexico). Introducing a proxy means testing targeting formula will improve objectivity and transparency. Eligibility could also be validated via community participation. Because inflation has strongly eroded the real value o f benefits, current benefits should at least be maintained in real terms and continually adjusted for inflation." But higher benefits may create work disincentives (poverty traps), so suitable exit/incentive clauses from the program need to be designed (e.g., workfare to improve community infrastructure, which can act as a work test for able bodied poor), and continual eligibility frequently monitored. Once disability is included as a criterion for receiving Samurdhi benefits, it would be possible to consider the option of folding disability transfers into Samurdhi, or at least ensure consistency across the two main cash transfer programs as a first step. Similar coordination on benefit and eligibility could be developed for other cash transfer programs, so that over time a consolidated cash transfer system could be developed, allowing an increase in benefits (within fiscal envelope and with attention to work incentives) and enhancing adininistrative and targeting efficiency . IIPerhapsthe original 1996 valuescould be restored,which would roughly double2004 benefit levels. ix Strengthening incentives to help the poor escape poverty. The safety net program could be designed to help the poor participate in human development and income earning programs, and improve their access to skill development. The Samurdhi reform could be strengthenedby using cash transfers providing incentives to improve school enrollment and reduce dropouts, and also to help the poorest participate in nutrition based programs, thus confronting child malnutrition. Helping the poor access micro-finance and training will also assist poor people to exit the safety net (successful examples include Mexico, Brazil, and Turkey). Upgrading programs to address individual vulnerabilities. Strengthening existing social welfare and care services for vulnerable groups are critical for addressing individual vulnerabilities. While these services can be outsourced to private providers/NGOs, the government agencies need to shape policy, set standards of care, license providers, and regulate the provision of services. Strengtheningand mainstreamingdisability is also important to address the special needs o f this vulnerable group. The implementation of the national disability action plan will require considerable coordinated efforts across a number of agencies, both public and private, to raise awarenessireduce stigma of disability, strengthen prevention of disability, and promote the inclusion of disabled in socio-economic activities. And community wide vulnerabilities. As noted above, to address disasteddrought relief, the government could develop a package of interventions: cash/livelihood grants, workfare, and social welfare and care services, ready to be scaled up in case of disasters. Relief for conflict njjiicted groups and areas could be strengthened by helping demobilized soldiers to reintegrate into civil society, by attending the special needs of children and youths affected by the conflict, and by rehabilitating schools, roads and essential infrastructurethrough public works. 11.Main Conclusions 28. Sri Lanka's social protection system provides extensive coverage, but could still be strengthened to better support growth and equity. The wide range of social protection programs indicates the Government's commitment to promote growth with equity. However, the system could be much more effective. Many vulnerable groups are not covered, and conversely, some non-vulnerable groups are included. Aside from equity issues, the programs also create inefficiencies that can constrain growth. For example, pension programs are fiscally unaffordable and limit labor mobility, employment protection institutions impede job flows, and safety nets are not linked to improving training and job prospects for the poor. 29. The report outlines strategic directions for reforming social protection programs to improve equity and facilitate economic growth. Examples of such "double dividend" policies are discussed for all three areas o f social protection: By opening "good" jobs to underprivileged informal sector workers, improved labor market policies would not only level the playing field for these workers but also facilitate the fulfillment o f their potential, thereby enhancingthe productive potential o f the whole economy. By enabling widespread risk pooling, social insurance not only helps preserve the wellbeing of individuals but also avoids the use of costly self-insurance such as accumulating assets as a form o f precautionary savings, and allows individuals to engage in more productive and higher riskireturn production. By linking cash transfers to investment in humandevelopmentof children and enhancing income opportunities for the poor, social safety nets not only reduce poverty and vulnerability but also X enhance opportunities o f the poor, thus helping the poor and making the economy more productive. 30. Moving forward in this direction will require attention to the following areas: the financing of social protection, program administration and garnering broad based support for reforms. Ensuring adequate and fiscally sustainable funding for the social protection programs. Social protection spending in Sri Lanka is skewed towards higher income groups both in the case o f safety net programs (e.g., Samurdhi) and in social insurance (e.g., pensions). The proposed Samurhdi reforms, as well as social insurance proposals o f this report, would not only help to correct these imbalances, but they would also help keeping public expenditures on these programs in check. In particular, improved targeting under the proposed Samurdhi reform will create ample room for increasing benefit per recipient -the level o f which has been strongly eroded because the amount per beneficiary has been nominally fixed -without necessarily increasing the overall spendinglevel. Over the long term, to ensure fiscal sustainability the country will need to reduce spending on pension schemes for civil servants (and the scheme's large implicit debt burden) and link pension benefits strongly to contributions. Moreover, public expenditures may also be re-directed towards (potentially subsidized) insurance coverage o f the poor and towards adequately funded and efficiently targeted safety net programs. Overall spending for safety nets in developed countries peaks at about 2 percent o f GDP (World Bank 2005e) and could serve as upper bound target expenditure for safety net prograins for Sri Lanka, including cash transfers and social welfare services. Some flexibility and set aside in safety net spending may be needed to respond to social protection needs stemming from emergencies, such as natural disasters. Strengthening capacity to administer programs is critical for program implementation. Although social protection reforms may bring strong efficiency gains and may promise to improve welfare o f the poor, such reforms are technically demanding and, therefore, the capacity o f the country to design and implement such reforms need to be developed. Social protection reforms are difficult to design because they require meticulous attention to adequacy, incentives, fiscal costs, and monitoring and evaluation, and because they often have to be coordinated across different areas (ranging from social protection to financial markets, fiscal management, education, and health), and because local circumstances need to be carefully accounted for. Appropriately accounting for political economy considerations is vital for ensuring successful adoption of proposed changes. Social protection reforms are politically difficult because many groups prefer the status quo. For example, current beneficiaries o f strong employment protection- incumbent workers and their unions-may oppose any reduction o f job protection as they consider alternatives, such as iinemployment insurance, inappropriate; existing beneficiaries o f Sainurdhi may oppose reforms because they do not wish to lose eligibility if a more transparent method o f selection is adopted. And because potential beneficiaries are often without adequate political representation (informal sector workers, or the poor who are excluded from safety net programs, for example), and because the benefits o f reforms often accrue only in the long run (such as in reforms o f old age income support), social protection reforms need to be coupled with adequate attention to political economy o f reform. xi Summary of Social Protection Policy Options Create a more equitable and efficient employment protection program by reducing excessivejob Employment Protection protection (reducing termination benefits and discretionary powers for layoffs, avoiding and Promotion excessive interference in wage setting mechanisms, phasing out ad hoc civil service recruitment practices)and strengtheningprotection of basic rights and social protection of workers Consider cost effective active labor programs, e.g.job counseling and information, and demand driven training programs for vulnerable, disadvantaged workers (incorporating employment options for youth and disabled from respective national action plans) II.SOCIAL INSURANCE: Prom ting Security Improve the adequacy and fiscal sustainability of formal social security schemes (e.g. gradually Social security: old age, disability reducing target benefits, increasing retirement age, increasing employee contributions and and survivor strengthening their links to benefits, improving record keepingiforecasting, strengthening fund management, and developing sound investment options) Explore possibilities to expand the coverage o f formal social security schemes to the informal sector, including potential subsidization of the very poor Introduce a modified version o f the UI system by eliminating personalized monitoring of unenlployment beneficiaries and simplifying benefit and administrative procedures, consistent with Sri Lanka's insuranceiworkfare income level, labor market conditions, and administrative capacity Strengthen self-targetedworkfare programsto help individuals cope with unemployment that can be scaled up during seasonal downturns, droughts, and other natural disasters Over the long term, under appropriate pre- conditions, explore options for creating an integrated All social securityiinsurance social security system, based on a defined contribution scheme (or with contributions strongly linked to benefits) with parameters set so as to produce a target replacement rate, to allow portability of benefit, and to increase economies of scale in administration (e.g. record keeping), and to avoid the creation of new liabilities. Ill. SOCIAL SAFETY NETS: EI ancing Equity Samurdhi program, civil-conflict ( 1) Address chronic poverty: relatcd programs, flood and Improve targeting o f the program by introducing proxy means testing, and avoid poverty traps drought relief. other safety net by introducing exit /incentive policies (e.g. workfare) programs Increase benefit level, without compromising work incentives and fiscal sustainability Improve fiscal, administrative and targeting efficiency by gradual consolidation of disability and other cash payments under the Samurdhi program (2) Help the poor escapepoverty: * Consider conditional cash transfers to reduce school non-attendanceand child malnutrition Promote access of to poor through skill development andaccess to micro-finance ( 3 ) Address individual vulnerabilities: Strengthen delivery social welfare and care services through public private partnerships Implement national disability policy focusing on raising awarenessireducing stigma, preventing disability, and including disabled people in socio-economic activity) (4) Address aggregate vulnerability: Develop a package of interventions: cashilivelihood grants, workfare, and social welfare and care services, to be scaled up in case of disasters Strengthen relief for conflict affected groups and areas by helping demobilized soldiers to reintegrate into civil society, by attending the special needs of children and youths affected by the conflict, and by rehabilitating schools, roads and essential infrastructure through public works. xii 1. POVERTYAND VULNERABILITY "Icannot think of earning even the minimumfor myfamily to survive ifI do not go in search of work like thisJFomplace toplace. " P.H. Sirisena, a 41 year old casual laborer in Kegalle district 1 . As a natural starting point for designing an effective social protection system, this chapter reviews growth, inequality and poverty trends, identifies key sources o f risk, and profiles the most important poor and vulnerable groups.`* The chapter finds that a large proportion o f households in Sri Lanka live in poverty (with low levels o f income), and these and many other individuals are also vulnerable to income shocks (or variance in income) resulting from individual specific shocks, such as sickness, loss o f employment, or more community wide shocks, such as natural disasters and the civil conflict. Across the life cycle, the chapter uses a mix o f poverty estimates and more qualitative evidence to identify the poorest and most vulnerable groups as children: child workers, children living in large fainilies and disabled children; working age adults: informal sector workers (particularly uneducated, casual and own-accoiint workers), unemployed youth, farmers and others affected by natural disasters, internally displaced persons, and the disabled; and the oldest among the elderly. While precise estimates o f vulnerability are not undertaken in this report, the poor, those without any human and physical assets, are likely to be the most vulnerable and hardest hit by income shocks. 2. This chapter begins by describing the macro picture in Sri Lanka in terms o f trends in economic growth, inequality and poverty. Then the chapter describes the main risks and identifies chief poor and vulnerable groups in society. The following chapters o f the report draw on the findings o f this chapter to discuss formal and informal social protection programs aimed at reducing poverty and vulnerability in Sri Lanka, and evaluate their ability to protect the poor. 1.1 Growth, Inequality and Poverty Figure 1.1: Growth of GDP and GDP per capita, 1991-2003 (in percent) 3. Sri Lanka's economy has been steadily growing over the last few decades. The average yearly GDP growth in the last decade (1994-2003) averaged at 4.5 percent (with a large drop in 2001-Figure 1.1). Trend in GDP per capita has a similar pattern and grew steadily since 1990s. Compared to other countries in the region, Sri Laiika's GDP growth was average-it equaled the rate o f growth o f Nepal, lagged behind India's ` - - ~ G D P 1, GDP percapita average growth rate of 5.8 percent, and exceeded the Pakistan's a average growth rate o f 3.6 percent Source: Central Bank of Sri Lanka, Annual Report (the comparison refers to 1990-2003). (various issues), I 2Guidelinesabout building social protection systems are provided by World Bank (2000) and World Bank (2002a). 1 4. Economic growth raised incomes, but inequality increased. The per capita consumption of richer segments o f the population grew much faster than iiicome o f the poorer segments (Figure 1.2). A recent study (Gunatilal=4 l_l__ 1,000 Household income < 1,000 and number o f members >=3 500 Household income < 1,000 and number o f members 2 200 --Source: Household income < 1,000 and number of members 1 100 Former Janasaviya r*ients 250 Ministry of Samurdhi, Youth Affairs and Sports. "Progress 1999, Programs2000" 131. Coverage nncl expenditures. Samurdhi coverage is extensive, with 41 percent of the population receiving benefit in 2002, with the coverage and expenditures declining since the late 1990s. At the end o f 2002, a total o f 1.9 million families or 41 percent o f the population received the income transfer (Central Bank o f Sri Lanka 2003). The number o f recipient families increased from about 1.5 million families at the start o f the program to 2.3 million families by the late 1990s and remained at these levels until 2001, with budget cuts bringing a reduction o f the number o f beneficiaries in 2002 and thereafter (see Figure 4.3). 64 Program expenditures rose from Rs. 8 billion in 1995 to a high o f Rs. 13.9 billion in 2001, and have declined since. 65 132. Adequacy and work incentives, The level of benefit is low, covering a quarter of the food bill of the poorest households, and not adjusted for inflation. The average size o f the total Samurdhi transfer, according to SLIS 1999/2000 data, is Rs. 381, which translates into Rs. 95 per person in per capita terms (the breakdown o f the number o f families by the size o f the grant in 2002 is presented in the Table 4.3). In 1999/2000, an average o f 14 percent o f total household food expenditure (and 21 percent for the recipients in the lowest income decile) could be met by the Samurdhi grant-roughly translated, five to seven days o f food per month. As with previous programs, Samurdhi benefits are not adjusted for inflation and so their real value has eroded over time (Glinskaya 2000). 133. Beneficiaries receive benefit payments that are much smaller than the stipulated grant. As noted above, the Samurdhi recipients are eligible for food stamps, and participate in savings/micro credit, social security, and other schemes. For this reason, compulsory savings stamp, social security stamp (discussed in chapter 3), and a housing lottery stamp are deducted from the transfer at the time o f receipt. Table 4.4 sets out the amounts deducted for each program component for each category o f grant. SLIS data revealed that after deducting forced savings and social insurance, the average size o f the net transfer is Rs. 3 I O per household and Rs. 78 per person. 64 As a result of these measures, the total number o f households receiving the supplement fell by 24 percentto 1,488,493 by end 2002. However, to encourage ineligible families to exit the program, departing families were paid Rs.140 supplement per month to cover compulsory deposits.This brought the total number of families coveredto 1,886,737 as at end 2002. 65 Note that "transfers" as defined by Samurdhi includes consumption grants to households as well as salaries o f SamLIrdhi officers. 54 Figure 4.3: Growing outreach of the Samurdhi program __ ~ - No. of families Rs.mn. 2,500,000 15,000 2,000,000 10,000 1,500,000 5,000 1,000,000 95 96 97 98 99 00 01 02 - ~- Number of families +Total value ~~~ ~ ~~ - ~~ Source. Central Bank Annual Report (various years). Table 4.3: Samurdhi recipients by categoryof allowance,and of 2002 Cateeorv of Allowance (Rs.) No of families YO IO00 3,954 0.21 600 589,460 3 1.24 400 320,759 17.00 350 238,593 12.65 250 335727 17.79 140 398,244 21.11 Total 1,886,737 100 Source; Samurdhi Ministry Progress report, 2002-2003, cited in CEPA 2004, Table 4.3. 134. The income transfer at best increased, by 7 percent, food consumption of the average household in lowest 40 percent of the income distribution. Glinskaya (2000) also reports that conditional on being a Samurdhi recipient, the size o f the transfer is slightly larger among households in the lowest quintile, but in per capita terms, the grant is larger for better-off households. Nevertheless, there is also qiialitative and quantitative evidence that the Samurdhi income transfer is an important safety net, particularly for households with unstable income (Centre for Poverty Analysis 2004). 55 Table 4.4: Samurdhi program, deductions of compulsorypayments (Rs.) Commodities Encashment Compulsory savings Social Security Housing lottery ..Category stamp ...... .._....._.__..._..I.~...- E?._mP... ..__I"___.______ stamp stamp fund stamp 1000 400 360 200 30 10 600 250 210 100 30 10 400 360 0 0 30 10 350 340 0 0 0 10 250 240 0 0 0 10 140 0 0 100 30 10 Source: Samurdhi Ministry Progress report, 2002-2003, cited in CEPA 2004, Table 4.4. 1 3 5 . Benefit is small when compared with the actual transfer needed to move households out of poverty. In 2002, the average income shortfall needed to move a person over the poverty line, was four times the size o f tlie actual grant. Iftargeted perfectly, Samurdhi could have moved 60 percent o f the poor out of poverty with a budget one and a half times as much as was actually spent that year. Box 4.2 indicates that tlie average income shortfall needed to move a person over the poverty line in 2002 was Rs. 320, four times the size o f the actual grant. Box 4.2: How much moneywill it take to move peopleout of poverty through income transfers? The average poverty gap or consumption shortfall of the average poor person in 2002 was Rs. 320 per person per month.' Twenty three percent of Sri Lanka's population outside the North and East were estimated to be poor in this year, and thus the total income shortfall was Rs. 1,227.6 million. Assuming a further 10 percent for the North and East (allowing for a higher poverty incidence in these provinces), the total budget that was required to reduce poverty in Sri Lanka through income transfers (without allowing for leakages and administration costs) is Rs. 1,350 million per month, or 16,204.5 million annually. This is 1.6 times the budget for Samurdhi benefits in 2002, and 4 percent o f total government expenditure and 1 percent o f GDP in this year. This implies that with perfect targeting, the budget for Samurdhi could have eradicated poverty among 60 percent of the poor. In fact, spending on Samurdhi was substantially reduced in 2002-in 2001, at its peak, spending was 0.9 percent o f GDP, or 3.3 percent o f total government expenditure. Source: Gunewardena (2005). 136. Despite the small size of the income transfer, labor-leisure decisions may be affected. Although the size o f the benefits as discussed above is relatively small, Glinskaya (2000) suggests that the possibility o f a household to lose eligibility upon earning above Rs. 2000 can distort incentives for taking a job (for reference, the statutory miniinurn daily wage rate in 2005 for workers in the tea, rubber and cocontit sectors was Rs. 121, Employers' Federation o f Ceylon 2005). In practice, this criterion has been universally ignored, thus making work disincentives less plausible. While no rigorous analysis has been done about tlie incentive effects o f the Sainurdhi program to date, it should be noted that adverse incentive effects were found for other welfare prograins in Sri Lanka. For example, participation in the food stamps program in Sri Lanka (in tlie early 1980s) was found to have reduced labor supply by two to three days per month (Salin and Alderman 1996). 137. Targeting. Samurdhi program suffers from weak targeting of benefits. There is considerable evidence that the program suffers from both types o f targeting errors, leakage to the nonpoor and undercoverage o f tlie poor: There is considerable leakage to the nonpoor. Twenty seven percent o f the non-poor received Samurdlii transfers in 1995196 (Gunewardena 2000), and 44 percent o f households in the top 56 tliree quintiles received them in 1999/2000 (Glinskaya 2000). Gaminiratne (2004) notes the leakage of Samurdlii to tlie non poor and undercoverage o fthe poor elderly. The poor, and the elderly poor in particular, are undercovered. Only 66 percent o f households in the bottom decile received Samurdhi assistance in 1995/96 (Gunewardena 2000) and 60 percent o f households in tlie bottom quintile received it in 1999/2000 (Glinskaya 2000). Undercoverage is even more pronounced for the elderly poor. Gaminiratne (2004) reports that only 23 percent o f the poorest quintile o f the elderly and 27 percent o f the second poorest quintile received Samurdlii assistance. Targeting outcomes are similar to those o f untargeted systems such as primary health care or primary education. 138. Targeting errors are systematic, as some groups more likely to receive Samurdhi income transfer than others. Poorer households in irrigation colonies, urban middle income neighborhoods and estates were less likely to receive Samurdhi than those in a traditional village (but no difference was observed for expansion colonies). The probability o f being a Samurdhi recipient differed significantly according to ethnic statiis (controlling for income and area o f residence, Sri Lankan Tamils, Indian Tamils and Moors are less likely than Sinhalese to receive Samurdlii benefits). Migration status was not related to the probability o f Samurdhi receipt, but not having a national identity card reduced the probability o f receipt (Glinskaya 2000). Party political affiliation also plays a significant role in determining one's chances o f getting the transfer. For example, families identified as traditional UNP voters were less likely to have been granted Sainurdlii benefits (ADB-PIMU 2000, Glinskaya 2000, Gunatilaka et al. 1997). 1 39. Nevertheless, household characteristics associated with poverty also increase the chances of receiving support. Households with disabled or chronically sick members were more likely to receive Samiirdlii benefits, while houses owning livestock, and other assets, including land and possessing land title were less likely to receive Saniurdlii benefits. Housing conditions were associated with the receipt o f Samurdlii (households with no access to latrines, or communal latrines and households with no lighting or kerosene for lighting were more likely to receive benefits). The probability o f receiving Samurdlii differed significantly by occupation o f household head (farm laborers and members in the military were more likely to receive, government and private salaried employees and retirees less likely to receive benefits) (Glinsltaya 2000). 140. Mistargeting is the result of the lack of explicit targeting criteria at the local level. As shown in Figure 4.4, tlie allocation o f Sainurdlii assistance at the central level corresponds to relative share o f poor households in districts. The correlation o f per capita district-specific Samurdhi funds and the poverty rate is 0.814, and the correlation with the poverty depth is 0.711 (Glinskaya 2000). It follows that most targeting errors occur at the local level.66Glinskaya also found that poorer provinces allocate a smaller share o f their Samurdhi funds to the poor, so their targeting errors are more prevalent. Therefore targeting errors appear to be non-random and tlie consequence o f the lack o f rigorous selection criteria, leavingtoo much discretion to Samurdhi Development Officers, who are in turn influenced by local politicians. These flaws in tlie design o f tlie program thus allow for the deliberate omission o f certain individuals or groups depending on tlie configuration o f local political forces. Income Support for Disabled I 41. Eligibility for disability benefits encompasses various types of disabilities. The Protection o f the Rights o f Persons with Disabilities Act No. 28 o f 1996 defines a person with disability as "any person who, as a result o f any deficiency in his physical or mental capabilities, whether congenital or not, is 66 This contrasts to findings in other countries, e.g., Bangladesh, where pro-poor outcomes of income transfer programswere largely due to the success o f communities in reachingthe poor, rather than central government allocation (Glinskaya 2000). 57 unable by himself to ensure for himself, wholly or partly, the necessitiesof life". This reasonably broad definition encompasses both the medical and socio-economic aspects of disability and forms the basis for government support of the disabled. Accordingly, the Ministry of Social Welfare encompasses people who have visual, speech, hearing, mobility, intellectual, and psychiatric disability and disability arising as a result of epilepsy and other causes in determining eligibility for income support. The Ministry's definition also encompasses multiple disabilities, which are a combination of two or more of these disabilities in a single individual (Ministry of Social Welfare 2003). Figure 4.4: District allocations for Samurdhi income transfers - ~~ ~ - ~ p ~ - ~ - __ _ _ _ ~ ~ OShare of Samurdhi HH 8%of poor HH (DCS) - - - - ___ ~ - ~ ~ p ~ ~ - p p ~ A -- - - -~ - p ~ ~ ~ Source Ministry of Samurdhi, Progress Report2003-2004, CentralBank (2003), DCS Poverty indicators/ CEPA 2004 142. A survey o f tlisablecl persons showed that only 41 percent receive any sort of income transfer (including the disability allowance) from the government (of persons suffering from intellectual, mobility, psychiatric, hearing, or speech disability, with no figures are available about the share of disabled in population-Ministry of Social Welfare 2003). Within the recipients, roughly 60 percent get Samurdhi income transfers and only the rest received income transfers for disabled persons from the Ministry of Social Welfare (the latter transfer can be received in addition to Samurdhibenefits). This split in transfer makes its administration more costly than if it were administered by a single agency. Eligibility for the disability benefit is determinedthrough the Divisional Service Office of the Department of Social Services, and since 2003, it has been fixed at a low level of Rs. 100-300 per month. 143. The main beneficiaries among the disabled are soldiers injured by the conflict, leaving other poor disabled groups undercovered. Income transfers to the vulnerable by the Ministry of Social Welfare are highly skewed towards disability payments for soldiers injured in the conflict and families of soldiers killed in action. This component consumed 92 percentoftransfers by the Ministry in 2003 and 85 percent of its total budget, leaving very little support for other vulnerable disabledgroups. 144. Expenditure on disability payments has increased over time. This increase is explained by the growing burden on the exchequer through mounting loss of lives. The allocations have continued to increase in 2002 and 2003, both years where the ceasefire was in place and casualties among the armed forces were virtually nil. This is because payments have to continue until the year of retirement of a solider who lost hisiher life. In many developed countries, disability payments comprise the most rapidly increasing expenditures of any cash transfer programs. Disability payments suffer from moral hazard or 58 collusion problems. In many countries, individuals who certify disability can collude with individuals to help the latter obtain disability payments. There is often constant pressure on Government to extend the definition o f disability which increases the number o f individuals eligible for disability and therefore disabiIity expenditures. 145. Disability payments are a very limited way of supporting the disabled poor. As noted in chapter 1, information about disability is in short supply, and the disabled suffer from stigma and lack o f acceptance into society. Tlie disabled face limited educational opportunities that increase the likelihood that they drop out or not attend school, and therefore have limited employment prospects. A more comprehensive approach to disability is needed that addresses prevention and enables the disabled to integrate better into society. 4.3 Coping with Large-Scale Disasters, Droughts, and Displacement 146. I n the short-term, disaster relief covers all affected persons on the basis of the impact and injuries suffered (Welfare Benefits Board 2003). Divisional Secretaries in the affected regions select beneficiaries with tlie assistance o f GramaNiladharis and Social Service Officers at the Divisional level. The value o f tlie assistance varies according to the type o f relief and is delivered in the form o f cash grants, food relief, food stamps, etc. Selection o f beneficiaries is carried out on the basis o f the impact caused by tlie disaster and not on income poverty criteria. However, a broad income criterion o f Rs. 3000 per family per month exists for long-term assistance for recipients o f disaster relief. 147. Drought relief is provided to families with income less than less than Rs. 2000 a month from agriculture or related activities. Families become eligible for the relief if the cultivation of seasonal crops have been disrupted or damaged for at least two consecutive seasons due to drought, the crops have not been insured, and the family does not have an alternative source o f income. The drought relief program seems to work effectively. In a case study o f paddy farmers and chena cultivators in the dry zone area o f Anuradhapura, Senaka-Arachchi (1998) found that drought relief by the government constituted the most important source o f income for 85 percent o f afflicted households. However, Samurdhi beneficiaries do not qualify for drought assistance as they receive relief through the Samurdhi program (Senaka-Arachchi 1998). Once again, the payment o f cash through two separate ministries likely increases administration costs o f delivery. 148. Relief for displnceclpersons comprises food assistance but the amount is insufficient to meet basic nutrition needs. The Ministry o f Social Welfare provides food assistance to internally displaced persons (IDPs) if their monthly income less than Rs. 1,500 for 12 months from date o f permanent resettle~iient.~'From September 2004 there has been strict adherence to the eligibility criteria. As a result, tlie number o f eligible families decreased by almost half, with 127,479 families receiving rations in January 2005. The non-eligible IDPs are expected to receive Samurdhi assistance, provided they are eligible for Samurdhi. Tlie quantity o f dry rations issued by the Ministry of Social Services is insufficient to meet nutritional needs (Jayasinghe 2005). 149. Displaced persons are also eligible for rehabilitation assistance for permanent settlement. Rehabilitation assistance is also provided but Permanent settlement is supported by the Unified Assistance Scheme (UAS) coordinated by the Ministry o f Relief, Rehabilitation and Reconstruction. It is "made available only to families who have returned to land/property that they own, or which has been granted by the authorities, or is available to returnees for the purpose o f durable settlement, and where the returnee and his family have demonstrated a clear intention to remain in the location in a permanent 67 There were roughly 600,000 internally displacedpersons (IDPs) in June 2002, of whom 172,000 lived in refugee camps. Since the ceasefire, 345,000 IDPshave returnedto their original homes (World Bank 2004). 59 capacity". The government offers Rs. 25,000 as a grant to returning IDPs to acquire basic tools, inputs and temporary shelter and to restart productive livelihoods. Each permanently resettled refugee/IDP family is also provided with an additional Rs. 250,000. 150. However, fiscal restraints have limited the coverage of this provision and many families await assistance.Currently, over 100,000 families have been paid this assistance, with a further 125,000 families awaiting assistance (of which funds are available only to pay 25,000 families) (Jayasinghe 2005). IDPs in government-controlled and LTTE-controlled areas receive food assistance through the Commissioner General o f Essential Services (CGES). I n 2000, 186,33 1 families comprising 674,150 persons received food assistance including 40,668 families comprising 170,857 persons in welfare centers, I 41,353 families comprising 484,258 persons living with relatives and friends and 4,3 10 economically affected families (Gunewardena 2005). 151. I n late 2004 and early 2005, the government, aided by multilateral agencies, launched several Tsunami relief initiativesto prevent chronic poverty and help restoring livelihood o f the affected population. The main prograins included cash transfers and the provision o f loans and grants for niicroenterprises via microfinance institutions. In addition, numerous public works projects have been successfully carried out by several international agencies and NGOs. It must be stressed that the eiioriiious proportions o f the disaster provoked an unprecedented response from the indigenous voluntary sector immediately after the disaster and even today, it shoulders a large part o f the relief and rehabilitation work. The groups involved included not just NGOs described in Box 4.1, but many others, including private sector firms, university students, schools associations, and other spontaneous groupings o f individuals, revealing a bedrock spirit o f voluntarism, self-organization, resource-mobilization and self-reliance across social strata. 152. A cash transfer program of Rs. 5,000 per affected families was launched in the immediate aftermath of the disaster. As o f August 2005, three rounds of payments were completed. By the second quarter o f 2005, over 245,000 families had received at least one round o f the cash transfer. While no cases o f deserving houses being excluded were reported, there was significant inclusion o f unaffected households in the program confirmed for the first round o f payments. The results will help understand if siich error is preventable in case o f a major disaster. 153. Public works were widely implemented in tsunami-affected communities to help households cope with employment loss. As noted in chapter 3, public works programs were mostly implemented through NGOs and CBOs. As o f August, the total value o f the programs was nearly $6 million, and the prograins have been implemented in all affected districts (see Table 4.5). Abeyratne and Weeratunge (2005) report that participation in the programs was most sought after by wage laborers and youth. According to Abeyratne and Weeratunge, these people considered public works as a temporary measure that would ease the immediate effects o f the crisis, expecting to return to their customary livelihoods as the crisis eased. Daily wage rates for participating in the program ranged from Rs. 300 to 400. It therefore appears that the principle that participants are paid slightly below the prevailing local market rate for unskilled agricultural labor has been violated (note that according to the Central Bank data, in December 2004 daily wages in agriculture were Rs. 322-377 for males and Rs. 227-266 for females). Indeed, Abeyratne and Weeratunge report that in certain sectors in some districts, labor shortages resulted from these programs (for example, in the paddy sector in Trincomalee district), raising daily wage rates by 20 to 30 percent. 60 Table 4.5: Public works orpanizedin tsunami-affected areas. 2005 Donor Agency Total Value Period of Location/ district Activity .. ." ............ _....... activity " .. ....." " ...."l."."l."",l ADF (Govt. of 25,000 Feb-Aug 2005 I Betticaloa, Debris cleaning & roads France) Trincomalee CARE Germany - 530,400 Apr-Dec 2005 Hambantota Debris, drainage, & EU irrigation USIAD 1,204,000 Jan-Sep 2005 Ampara, Debris, roads, irrigation, Hambantota,Matara env. Construction DFIDALO + FA 998,130 Apr-Sep 2005 Kalutara, Trinco, Infrastructure Galle, Batti & Jaffna construction UNOCHA Flash - 750.000 Jan-Sep 2005 5 Districts Infrastructure Appeal construction DCA - Germany 1 1 5.450 Feb-Jun2005 Ampara, Batticaloa Debris cleaning & roads JCCP Japan - 77,500 Jan-Apr 2005 Trincomalee Debris cleaning roads Spanish - Red 130,000 Jan-Aug 2005 Galle, Kalutara Debris, roads, drainage, Cross SP env. construction Govt. of Japan 800,000 Jan-Dec 2005 12 districts Debris cleaning roads, env. Improvement World Vision 100,000 Mar-Dec 2005 Galle & Jaffna Debris clearing & International & infrastructure USAID improvement OXFARM GB* 1,150,200 Feb-Dec 2005 Batticaloa, Rehabilitation of rural ..................................................................... Hambantota, Matara........ roads?...!!I"! ............... " " " TOTAL 5,880,680 ---*\-.--- Sozo.ce: ~ L O 'survey among donor agencies, September 2005. S Note: The data may be underestimated,becausenot all donor agencies respondedto the survey. 4.4 Reforming Safety Nets 154. Reforming safety nets requires the development of an overall strategy that will re-orient safety nets to contribute to both equity and growth. A four-prong strategy is proposed. First, cash transfers/social assistance should be better targeted to the chronic poor, with design features that reduce work disincentives (e.g., appropriate benefit levels, workfare). Second, safety nets should increasingly become "safety ~-opes"-nieans o f helping people to move out o f poverty and vulnerability-through improvements o f the human capital and earning possibilities o f the poor. Third, social assistance should also be linked to individual vuliierability-(orphans, at risk youth, abused women, and seasonal unemployment) by strengthening social welfare and care services. And finally, the response to aggregate vulnerability, such as natural disasters should be scaled up. (a) Reforming the Samurdhi System: Addressing Chronic Poverty 155. The Governmentof Sri Lanka has recognized the need to reform the welfare system, mainly by improving transparency in targeting the poor. Welfare Benefits Act No. 24 o f 2002 has paved the way for key reform in the sector. The Act's mandate is to provide the necessary legal framework for the payment of all welfare relief, to set out a transparent process by which the recipients o f such benefits can be identified, and to provide for the termination o f such benefit. The main thrust o f the Act is to (a) limit the program to the bottom 25-27 percent of the population (around 900,000 households), which will enable the increase o f the transfer amount; (b) achieve better accuracy in identifying the poor; and (c) make the transfers progressive, i.e., give larger amounts to poorer households. 61 Eligibility 156. The Act requires that eligibility of households for the program will be made more objective and transparent and be assessed periodically. I n this regard, the government has implemented a formula-based system in the North and East and is looking to implement it in the rest o f the country as well. Accordingly, households that have characteristics most closely associated with poverty-as determined by a transparent, proxy means test formula (PMTF>-will become eligible to benefits. The number o f households will depend on tlie available budget constraint. The selection o f eligible individuals will thus be made more transparent. Household eligibility is proposed to be checked every 1-3 years. However, higher administrative costs o f re-checking beneficiaries will need to be balanced against addressing vulnerability, as new beneficiaries may become eligible for assistance, and reduce leakage- some may move out o f poverty and should become ineligible for the benefit, making more frequent check-ups o f eligibility (for example, every 6 months) necessary. The Samurdhi Ministry is currently engaged in devising alternative income generation schemes - as opposed to handouts - for those who will transit out o f the cash transfers scheme as a result o f better targeting. 157. All homeless households are to be included in the program automatically,without applying the PMTF (which would be difficult to calculate in any case). This would not lead to any significant change in the coverage and other estimates-homeless households are only about 1.3 percent o f all applicant households in pilot areas, and the average household size is much lower for homeless households, which translates to relatively low amounts per household using the payment scheme described previously.68 158. Ensuring exit clauses-helping people to move off eligibility rolls. The program design should also include methods to help households (who can do so) to move out o f the program. In some countries, exit is forced by limiting the duration o f benefit (e.g., US) and with certain conditions for re-entering welfare rolls) or through improving the households prospects for earning income and helping their children escape poverty (through conditional cash transfer programs, e.g., Mexico and Brazil). Adequacy and work incentives 159. The benefit will consist of two components: a fixed component that is an equal amount for all eligible households, and a variable component that has a fixed amount per member or appropriately defined subsets o f members. The variable amount has been defined as per vulnerable member o f the household, where the vulnerable are children aged between 0 and 15, elderly aged 66 and above, and those who are disabled or permanently ill.The rationale for this scheme is that such individuals cannot, or slioiild not earn income for the household, and are also the ones most affected by shocks to household incoine. 160. The reform aims to increase the amount of support within existing budget through better targeting. Siniiilatioiis show that with tlie current transfer budget held constant (Rs. 9 bn.), a program targeted to around 27.5 percent o f tlie population (which corresponds to the cut-off being considered by tlie Welfare Benefit Board), would cover a little over 900,000 households. Each household on average would receive around Rs. 830 per month, compared with the current average grant o f Rs. 320 per household. It would also be important to avoid erosion o f the benefits by inflation, so period adjustments o f benefits in line with inflation rate should be considered. 68 Twenty four percent of homeless householdscompriseof a single member, comparedto one percent o f those living in homes. Five percent o f homeless households comprise o f a 6 or more members, comparedto 37 percentof those living in homes (World Bank 2005). 62 161. The increase in benefit level may create disincentives to work and should be carefully monitored. Ifthe average householdtransfer is significantly raised, this may create disincentivesto take a job and may create a poverty trap. The option mentioned above, raising the benefit to Rs. 830 per month, may create siicli disincentives (such a transfer would amount to slightly more than a third of monthly wages of unskilled workers-daily wage earners in the tea, rubber or coconut sectors earn a minimum wages of Rs. 3,025, assuming they work 25 days per month). The impact of the increase in benefits will therefore have to be carefully monitored. Similarly, the eligibility for benefit needs to be monitored on an on-going basis (see above). Workfare programs(discussed in chapter 3) can help reduce disincentive by making benefit receipt for able bodied poor contingent on participation in small scale works programs. 162. Benefits will be de-linked from savings and insurance. The practice of deducting at source will be discontinued. Likewise, the Samurdhi social insurance scheme will cease to be a mandatory scheme and will operate more like a voluntary, contributory scheme. However, Samurdhibeneficiaries, as well as other poor who require loans will continue to be able to obtain consumption loans from the system so long as they have participated in the group savings scheme and establishedtheir credit worthiness. In this way, Samurdhi will become a strictly re-distributional transfer to the poor based on clear eligibility criteria. Such poor can participate in the ancillary programs o f credit and insurance ifthey so wished, on a voluntary basis rather than a mandatory basis. However, some form of subsidy to the poorest for the very may well be required(see discussion in chapter 3). Targeting 163. The reform introduces proxy means test to determine eligibility. Two critical changes are to be effected. First, tlie introduction of a proxy means test formula (PMTF) as the selection criteria. This method sliould provide a more objective selection criteria than direct means testing under the circiinistances where subjective reports o f personal income cannot be easily verified (Box 4.3). Second, the selection of eligible individuals will be done by the Welfare Benefits Board, with decision-making delegated to the District Secretary, Divisional Secretary, or the Grama Niladkari (GN), Le., though decentralized administration. The PMTF has already been tested in a pilot targeting survey covering 48,000 households in 114 GN divisions, backed up by a field survey of a sub sample of GN divisions (World Bank 20051). 164. The targeting formula is to be adapted to conflict areas. The pilot results did not uncover any evidence to suggest that tlie PMTF should not be applied in the conflict areas (World Bank 2005). However, a number o f issues were raised by workshop participants on the need to change the way certain variables were defined and interpreted for the North and East, which would in turn imply a change in the instructions for the application form for the region. These included, for example, instructions about the definition of cultivable land, which is a variable in the PMTF. In the North and East, this definition will have to be reflect a possible impact of conflict-namely that land, cultivable in the past, cannot be cultivated next season because it may now have mines. 63 Box 4.3: Developing a proxy means test formula In principle, conducting a direct "means test" that correctly measuresthe earnings of a household is the best way to determine eligibility when the poor are the target group, as is the case with Samurdhi. In practice, however, such straightforward means tests suffer from several problems, the most important being that verifying incomes o f households is very difficult in developing countries where reliable income records do not exist. Thus the idea o f using a Proxy Means Test Formula (PMTF) that avoids the problems involved in relying on reported income is appealing. This involves using information on household or individual characteristics correlated with welfare levels in a formal algorithm to proxy household income or welfare. These instruments are selected based on their ability to predict welfare as measured by, for example, consumption expenditure of households. The obvious advantage o f proxy means testing is that good predictors o f welfare-like demographic data, characteristics of dwelling units and ownership of durable assets (as long as these cannot be changed or moved, respectively)-are likely to be easier to collect and verify than are direct measures like consumption or income. This has led to PMTFs being widely used around the world for targeting safety net programs, and the Steering Committee set up by the Government o f Sri Lanka to guide the technical aspects of the reform process has also decided to adopt a PMTF to identify the target group for Samurdhi benefits. Source: Sri Lanka Poverty and Social Impact Analysis, World Bank (20051). 165. While not considered under this reform, community participation methods can be used to validate eligibility lists created by PMT formula targeting process. Such a validation would be particularly important, as Coady et al. (2004) fi nd that the method o f targeting per se is virtually irrelevant for it effectiveness. For example, means tested programs scored among the best and among the worst among the 100 programs evaluated by the authors. They identify community participation as one o f the rare approaches which does make a difference. Community participation is not new to Sri Lanka, and was practiced quite successfully in selecting beneficiaries for the Janasaviya Program, the Samurdhi's predecessor (World Bank 1995; see Box 4.4). It is worth stressingthat effective community participation is also one the promising methods to include in the program the hard core, hard-to-reach poor. The government's welfare reforin as articulated in Sri Lanka New Development Strategy (Ministry o f Finance and Planning 2005) also mentions the adoption o f community screening methods as part o f its proposed welfare reform. Thus, combining community participation with PMT may yield a stronger targeting approach. 166. Strengthening administration of the Samurdhi program. Cash transfers are to be distributed through individual accounts created in post offices (instead o f Samurdhi banking societies as is the current practice), preferably in the name o f the female member o f the household (particularly in areas where alcoholism is a problem). Timely distribution o f benefit will also be important, given that some o f the benefits noted above are delayed, which cause undue hardship to eligible poor. Potential recipients are required to provide all information relating to the income and asset status o f themselves and their families. To increase transparency, the names o f eligible persons would be published in the media a time frame o f I O days given for receipt o f objections, which would then be investigated. Any person or public officials found to be declaring false information would be subject to trial and if found guilty subject to a fine o f 5-10,000 and imprisonment o f up to one year. Divisional Secretaries are responsible for ensuring the update o f names including omissions and deletions to the list o f eligible persons (Centre for Poverty Analysis 2004) While fines and threats may reduce false claims, the verification o f eligibility and monitoring o f claims will still need to be undertaken, particularly if governance remains an issue. This could be strengthened via random audits, and fair complaints procedures. Community groups might also be used to help target and monitor programs and ensure benefits are delivered to those identified by PMT/community-based targeting methods. 64 Box 4.4: Community participationin targetingwelfare benefits-SriLanka's experiencewith Janasaviya Many non-poor receivedthe income transfer in the first round o f the Janasaviya Program. For example, the original income ceiling o f Rs. 700 o f Round 1 was found not to apply in practice, as a random survey of nearly 300 beneficiary households found that nearly 30 percent had incomes above Rs. 1,500 per month before entering the program. But having first used income-based indicators to select beneficiaries, the Janasaviya Program subsequently replaced these with a set of detailed, region-specific criteria identified with the assistance o f the support team members and the Grama Niladhari under the supervision of the Divisional Secretary. These were framed on observable variables such as the ownership o f land and consumer durables, and yielded valuable division-specific information on divisional economies, and the kind of income-generating opportunities that were available. Officials then resorted to community-based screening methods to improve targeting and village-level committees used the new criteria to identify eligible families, and lists o f those eligible in descending order of poverty were publicized. Anonymous complaints were then gathered and evaluated by the Divisional Secretary, and were thereafter inquired into by an investigative team comprising of government officials, NGOs and others among the general public. These inquiries were usually made at a public gathering or "Janahamuwa", where the selected family and others were questioned in order to verify that they were indeed eligible for Janasavi benefits. Fine- tuning of the screeningprocess improved targeting efficiency considerably. Source: Gunatilaka (1997) (b) Options for Helping the Poor Escape Poverty 167. Recent advances have moved social safety net programs from purely helping the poor to cope with poverty to helping them to escape from poverty, while Samurdhi reforms discussed above help transform the program to one which is purely redistributive. Examples o f such innovations include making income transfers conditional on the access o f the poor to health and education services, particularly for children, and boosting the income earning opportunities o f the poor via access to micro- finance. Many programs, for example in Latin America and in Turkey, that started out on a pure cash transfer basis have been now transformed into programs that simultaneously enhance human capital by adding measures to help the poor escape poverty to cash transfer assistance. 1 68. Hirntnn Cripitnl Developnzent. While the above describe Samurdhi reform does not address this issue, cash transfers could be used as incentives to reach the small share of children who are not enrolled in school and reduce dropouts, and also to help the poorest participate in nutrition based programs given malnutrition remains an important challenge for Sri Lanka. 169. Improving school participation of poor children and reducing early drop-outs from schooling. Our analysis identified financial constraints as an important factor behind lack o f enrollment and early drop-outs from schooling (see chapter 2). A conditional cash program-the program that makes income transfers conditional on continued school enrollment-may help to prevent such dropouts. Such a program would be targeted to five and six year olds, to prevent late entry to school, and to eleven to fourteen year olds (those attending lower secondary grades), to prevent them from dropping out, and would be available in selected poor, rural and urban, areas. Evaluations show that these programs can raise school enrollnient and attendance rates (and improve child health and nutrition, if that i s the program's objective)-see Box 4.5. Such programs need to be well targeted, so that there is transparency aboiit \\ho receives tlic transfers. The level o f benefits and the number o f beneficiaries is also easily aci.jiidecl and e~tcndedin times of crisis. Ihese programs may also need to be coupled with 'flexible' timing in the provision of schooling/tio~i-fonnaleducation to meet the needs of working children, and 65 with progi`aiiih that providc hearing and \. isual testdaids to disabled children to increase their enrollment and prctcnt dropouts. Box 4.5: Conditional cash transfers as a way to improveschoolattendance, health and nutrition of the poor Conditional cash transfers provide money to poor households conditional on various desirable behaviors, such as sending children to school and visiting health centers for growth monitoring, immunizations, and nutrition interventions. These programs have been implemented in increasing number of countries, particularly in Latin America and Caribbean (Brazil, Honduras Jamaica, Mexico, Nicaragua), but also elsewhere(Turkey, for example). Mexican "Progresa" in 1999 reached 2.3 million families, and in Brazil (Bolsa Escola and PETI) and Jamaica (PATH) conditional cash transfers are widely used to promote health and education of children (World Bank 2004b). Evaluationsshow that conditional cash transfers are an effective means for promoting human capital accumulation among the poor households, as well as for poverty reduction. There is clear evidence of success in increasing enrollment rates, improving preventive health care and raising householdconsumption (Rawlings and Rubio 2004), as well as enabling households to make previously unattainable investments in income generating activities (Gertler, Martinez and Rubio 2005). For example, the Mexican program (Progresa) increased primary school attendance by more than 2 percent and secondary enrolment by more than 8 percent, while it led to an increase in health visits by some 20 percent. The Brazilian Bolsa Escola reduced school drop out from 5.6 percent to 0.4 percent(Rawlings and Rubio 2004). 170. Combating child malnutrition. Currently, the government applies a four-prong approach to fight child malnutrition: (a) measures to address specific nutrition problems, (b) health interventions, (c) direct food consumption based measures to ensure adequate nutrition intake among households and individuals, and (d) poverty reduction programs (World Bank 2005a). However, there is little evidence about the effectiveness o f these public nutritional interventions. Given the lack o f solid and robust research, these nieasiires may also fail to address some important aspects o f malnutrition. Therefore, more work is needed to understand and tackle the causes o f malnutrition and design interventions that will help address this problem. Assuming income will be a major constraint in program participation of the poorest children, to ensure their participation (and potentially that o f their mothers), the Samurdhi program could be made conditional on participation o f poor children in the appropriately designed nutrition programs. 171. Increasing Income of the Poor. There is growing evidence that linking the poor with programs that help them generate income, e.g., training and micro-finance, help these groups escape poverty. We have already discussed training of poorer informal sector workers in chapter 2, and focus here on the provision o f micro-finance programs to the poor. Samurdlii, too, has developed such programmes. But forced participation has been damaging to overall objectives. Hence, delinking the microfinance programs from the income transfer program, giving the poor a choice about whether to participate, and ensuring the autonomy o f the other programs may yield more positive outcomes. In addition, linking the poor to comnii~nitybased self-help groups may also be useful in helping them escape poverty. 172. Microfinance for the poor: strengthen the micro finance industry and pilot innovative financial services to the very poor. Given the wide outreach o f microfinance institutions, enhancing their services to the very poor may provide new ways to the poor to escape poverty. To achieve this goal, efforts must be made in two directions (see Annex 4.1 for the description and analysis o f key problems of the microfinance industry in Sri Lanka). First, improve conditions o f the industry by strengthening the regulatory framework to ensure the security o f deposits and probity o f microfinance organizations. An independent regulatory body with representatives o f major stakeholders on its governing board needs to be established with sufficient safeguards to prevent its capture either by the government or by the 66 inicrofinance providers themselves. There is also a need for greater transparency about the extent of subsidy received by the niicrofinance providers. And the Samurdhi Bank system needs strengthening o f its governance to gain the autonomy from the political establishment. Second, efforts should be redoubled to rigorously assess the impact o f microfinance on poverty, especially o f the very poor clientele. Moreover, while promising examples exist about how to weave in financial service to standard safety net programs, not enough knowledge exists about the methods and conditions that bring success in microfinance (see an example o f an "opportunity ladder" provided by financial services in Box 4.6). Piloting various methods o f providing financial services cum training and provision o f selected business services to the very poor segments o f population, and rigorously evaluate them (including analyzing the effects o f poverty reduction by longitudinal surveys), is therefore needed. Box 4.6: Combining public works with financial services Under the Rural Maintenance Program o f CARE, Bangladesh, poor women have been recruited for a four-year period to maintain earthen village roads as part of a public works program. They have to enroll in a compulsory savings plan, setting aside 22 percent o f their earnings, and they have to participate in a comprehensive cycle of life-skills training (including basic numeracy, health, nutrition, and income generatingactivities). The program is active in approximately 90 percent of districts in rural Bangladesh,employing over 40,000 women with 10,000 o f them graduating every year. Seventy nine percent of these women continue with self-employment activities, and 66 women graduates have been electedto local governmentoffice to date. The program has successfully met the twin challenges of targeting the poorest and ensuring program graduation. Equipping theni with necessary skills and helping them to generate substantial savings preparedwomen to start a microenterprise. Source: Hasheini (2005). (c) Helping the Poor Cope with Vulnerability 173. Aside from addressing chronic poverty, safety net programs can help address other individual vulnerabilities and help individuals cope against large covariate shocks. Strengthening these programs may not be fiscally neutral and fiscal cost will require careful consideration prior to implementation. A(1dressing individua1 vuInerabiIities 174. Disability/illness of main breadwinner, or other household member, is likely one of the largest health shocks affecting families. As noted above, addressing disability requires a comprehensive approach and also much more facilities and resources. Such an approach is embodied in Sri Lanka's national strategy for disability, and it will have to be assessed against the overall budget envelope. The efforts in this area shoiild include: (i) improved information and awareness about the problems o f the disabled.69This requires providing more and better information about the incidence and type o f disability, public awareness campaigns which would inform the public o f the benefits of preventing disability, and help reducing the stigma about the disabled and including them in the society; (ii)improved prevention- developing measures to improve road safety, safety at the workplace, improving nutrition; (iii)facilitating participation o f disabled fully in the socio-economic life o f their community, including in education and health programs (through, for example, provision o f hearing aides or glasses, or improved emergency treatment and care o f mental illnesses, trauma and depression); and (iv) better coverage o f disabled in safety net programs (and ex ante in disability insurance programs, see chapter 3). Perhaps most 69 It was found, for example, that the lack o f disability awareness negatively impacted the inclusion o f peoplewith disabilities at all phases of the recent tsunami relief efforts. 67 importantly, ensuring the participation o f the disabled in employment/training programs to help them accessjobs will be critical for improving their well being and helpingthem escape poverty. 175. Social welfare and care services. Attending the needs o f other vulnerable groups, such as alcohol and drug abuse, orphans, and street children, requires strengthening social welfare and care services, which is a very useful complement to cash transfer programs. These services provide or facilitate the provision o f counseling for victims o f trauma, alcoholism, divorce, loss o f parents, and other social problems. While services can be outsourced to private providers/NGOs, the government agencies need to set standards o f care, license providers, and regulate the provision o f services. The scaling up o f these services will also require adequate financing, with the benefits o f provision reflects against the long term consequences o f inadequate care. Sri Lanka already implements programs in this area (including NGOs), but these programs would need to be adequately funded and strengthened to address these iinportant vtiInerabiIities, Addressing Aggregate Vulnerabilities 176. To address clisnster/(lrouglzt relief, the government could consider a package of interventions: cash/livelihood grants, workfare, and social welfare and care services, to be scaled up in case of disasters. One o f the painful lessons of tsunami has been the lack o f programs prepared in advance and readily available for relief and recovery response. A social protection component o f such an intervention would systematically prepare and design programs to protect the affected population (appropriately defined), ready to be rolled out in terms o f emergencies, and form an integral part o f the country's disaster prevention and coping strategy. Specifically, the safety net component would (i)meet immediate basic needs by ensuring food, shelter, and medical aid for immediate relief o f disaster victims; (ii) distribute cash grants to provide for the minimum subsistence needs of affected households (and perhaps to restart livelihoods and rebuild housing); (iii)facilitate income earning opportunities for poor during the post-crisis through access to micro-credit and grants to start up businesses; (iv) offer workfare programs for those able bodied poor who are not able to find ajob or increase their income (see Box 4.7 for key design elements o f a successful workfare program); and (v) link affected individuals to social welfare and care services that could help provide, for example, needed counseling and family placement for individuals suffering the trauma o f the loss of family members (e.g., orphans and widows), and help the disabled (with rehabilitation, orthotics, and psycho-social support). 177. Providing relief for conflict-affected groups and areas. This effort could have several major thrusts. First, helping demobilized soldiers (and perhaps to other previous active participants in combat) to reintegrate into the civil society through programs that support their employment or creation o f self- employiiietit. including labor market counseling and job training. Such programs are common in countries which have undergone civil conflicts in other regions countries which have undergone civil conflicts, e.g., in Bosnia Herzegovina (see Box 4.8). Second, addressing the special needs o f children-ex-combatants and others, for example, to ensure that these children attend school and obtain needed health care. Third, measures also need to be taken to integrate disabled children and youths into society to ensure access to treatment including trauma counseling and other psycho-social assistance (see above section on disability). Finally, community rehabilitation-of schools, roads and essential infrastructure through public works or workfare-could also be provided (at the same time, offering temporary income support to the unemployed or demobilized soldiers). 68 Box 4.7: Key design elements of a successful workfare program In designing public works programs, several general principles should therefore be followed. First, forgone earnings should be minimized by attracting workers who have low alternative earnings opportunities. Second, displacementeffects should be avoided, among others, by the careful selection of areas of activities. Third, while the program should in principle be open to anyone, wages shouldbe set low enough so as to trigger a self-selection mechanism through which only those in need participate. Low wages also encourage participants to search for a regular job. And fourth, to maximize the their intended effect on income generation by the poor, projects that require heavy non-labor costs should be avoided, particularly during crises. In general, public works should be more labor intensivethan required by pure profit maximization. Argentina's Trabajar program is a very well know example of a successfulworkfare program, illustrates the above design elements. Trabajar allocates funds across provinces based on the distribution of the unemployed poor. Proposals to use the funds are made by municipalities and non-government organizations. These proposals are approved at the regional level, based on a system o f points related to poverty in the area and the merits of the proposed project. The government pays for the costs of unskilled labor and the sponsoring units pay for the equipment, materials, and the skilled labor. The wages for unskilled labor are set at two-thirds of the average wage for the poorest decile in the capital city. In principle there are no restrictions on the eligibility of beneficiaries to participate in the program, but in practice there is rationing. The financing of the Trabajar program as a matching grant scheme not only induces local governmentsto commit to the project, but it also induces local governmentsto make use o f more labor. The use o f labor intensive approaches is thus enhanced through incentives to local governments rather than through instructions to contractors and engineers. Source: Vodopivec (2004) and de Ferranti et al (2000). 69 Box 4.8: Emergency labor redeployment project Following the Dayton Peace Agreement of November 1995, an estimated 425,000 soldiers were demobilized from the armies in Bosnia-Herzegovina. Their reintegration into civilian life-supporting them to find employment or to create sustainable jobs and businesses-was facilitated by the two successive Emergency Demobilization and Reintegration Projects implemented by the World Bank. The projects provided labor market counseling services and promoted on the job training and self employment in small business activities and in agriculture. In one of the projects, 5,426 ex-soldiers participated, of which 1,520 took jobs under the on-the-job training component; 3,450 started farming (the project supported, for example, the purchase of cattle and farming machinery), and the rest undertook vocational training. The table below summarizes some demobilization programs targeting ex combatants or aiming at reducing military in other countries: Projecti SI