Report No. 24300-CR Costa Rica Social Spending and the Poor (In Two Volumes) Volume i: Summary of Issues and Recommendations with Executive Summary October 31, 2002 Human Development Sector Management Unit Central America Country Management Unit Latin America and the Caribbean Region u Document of the World Bank CURRENCY EQUIVALENTS (Exchange Rate Effective June 13, 2002) Currency Unit = Costa Rican Colones US$ 1 = CRC 357.62 WEIGHTS AND MEASURES The Metric System is used throughout the report FISCAL YEAR January I to December 31 ABBREVIATIONS AND ACRONYMS BCCR: Banco Central de Costa Rica CCSS: Caja Costarricense del Seguro Social CEN-CINAI: Centros Integrales de Atenci6n Infantil CONPES: Consejo de Politica Econ6mica y Social DRG: Diagnostic Related Group EBAIS: Centros locales de atenci6n primaria de salud EAP: Economically Active Population ECD: Early Childhood Development Programs ENISO: Encuesta Nacional de Inversi6n Social FODESAF: Fondo de Asignaciones Familiares IMAS: Instituto Mixto de Ayuda Social IMM: International Monetary Fund INEC: Instituto Nacional de Estadisticas y Censos IVM: Programa de Invalidez, Vejez y Muerte de la CCSS GDP: Gross Domestic Product GNP: Gross National Product MIDEPLAN: Ministerio de Planificaci6n MOH: Ministerio de Salud MOE: Ministerio de Educaci6n MINVU: Ministerio de la Vivienda y Urbanismo NGOs: Non-Government Organizations OECD: Organizaci6n Econ6mica para la Cooperacion y el Desarrollo PHC: Primary Health Care PNUD: Programa de Naciones Unidas para el Desarrollo SIPO: Sistema de Selecci6n de Poblaci6n Objetivo WHO: World Health Organization Vice President LAC: David de Ferranti Country Director: Jane Armitage Lead Economist: Inderrnit Gill Task Manager: Helena Ribe COSTA RICA SOCIAL SPENDING AND THE POOR Volume I TABLE OF CONTENTS EXECUTIVE SUMMARY .......................................................................... 1. INTRODUCTION ..........................................................................1 2. POVERTY PROFILE: CHARACTERISTICS OF THE POOR AND VULNERABLE GROUPS ... 10 3. SOCIAL SPENDING: SECTORAL ISSUES AND OPTIONS ......................................................... 32 4. IMPROVING THE EFFECTIVENESS OF SOCIAL SPENDING .......................... .......................... 57 REFERENCES .......................................................................... 65 ANNEX 1 ......................................................................... 67 LIST OF TABLES Table 1: Key Economic Indicators in Costa Rica (1990 - 2000) .....................................................................3 Table 2: Public Spending in Social Sectors in Selected Countries, 1996 .......................................................3 Table 3: Social Indicators in Selected Latin American, Europe and Far Eastern Countries, 1998 ................4 Table 4: Forecasted Population by Age Group, 2000 - 2050 .................................................................... 6 Table 5: Typology of Latin American countries in terms of social spending, 1998 ........................................7 Table 6: Gini income elasticity for different values v .....................................................................8 Table 7: Illustrative impact on social welfare of four alternative programns, 1999 ..........................................9 Table 8: Poverty Incidence and Poverty Gaps, 1987 - 2000 ................................................................... 10 Table 9: Asset Accumulation by Families in Costa Rica, 1992 - 1997 ....................................................... 12 Table 10. Poverty Rates and Contribution to Poverty by Group, (1997 Household Survey) ....................... 13 Table 11: Characteristics of Poor and Non - Poor (Household Survey - 1997) ........................................... 14 Table 12: Labor Force Participation and Unemployment Rates of Heads and Non-Heads .......................... 15 Table 13: Investments in Assets by Poverty Level in Costa Rica, Urban Area, 1997 .................................. 16 Table 14: Investments in Assets by Poverty in Costa Rica, Rural Area, 1997 ............................................. 16 Table 15: Characteristics of Families in SIPO Data Base ................................................................... 18 Table 16: Characteristics of Families by Poverty Level ................................................................... 19 Table 17. Characteristics of Household Heads by Gender and Poverty Level, 2000-2001 ........................... 21 Table 18. Distribution of Population in SIPO by Nationality and Poverty Level ......................................... 22 Table 19: Characteristics of Children 0-5 Years of Age ................................................................... 23 Table 20: Characteristics of Youth 13-18 Years ................................................................... 24 Table 21: Characteristics of Youth 14-18 Years ................................................................... 26 Table 22. Characteristics of Youth 19-24 Years ................................................................... 26 Table 23. Characteristics of People 65 and More Years Old ................................................................... 28 Table 24: Summary of Critical GAPS in Social Services of the Poor and Vulnerable Groups in Costa Rica, 2000-01 .................................................................... 29 Table 25: Evolution of Key Education Indicators, 1990-99 ................................................................... 33 Table 27: Comparative Health Status Indicators ................................................................... 38 Table 29. Coverage of ECD Programs in Costa Rica, 2000-01 ................................................................... 50 Table 30: Budget Allocations by FODESAF Main Assistance Programs, 1990-2000 (%) ......................... 52 Table 31: Cumulative Distribution of Benefits of Key Social Programs by Income Quintile 1999 ............ 53 Table 32: Cumulative Distribution of the Bono de Vivienda by Income Quintile ....................................... 54 LIST OF FIGURES Page Figure 1. Secondary Completion and GNP 5 Figure 2. Secondary Completion and Public Expenditure in Education 5 Figure 3. Net Enrollment Ratios and Education Expenditure 34 Figure 4. Health Expenditure and Health Outcomes 39 ACKNOWLEDGMENTS This report consists of two volumes. Volume I presents the report's main policy recommendations. It includes a poverty profile and identifies the main vulnerable groups in the country. It also contains a summary of the social sector expenditure review, which identifies the main efficiency and targeting opportunities to improve the effectiveness of social spending in reducing poverty, and a review of institutional issues which are important for improving coordination and complementarities among programs. Volume II contains a detailed review of social spending in education, health and nutrition, pensions and social assistance with specific recommendations for improving the effectiveness of the public sector programs in these sectors. The report is a join effort by Government of Costa Rica and the World Bank and was co-task managed by Helena Ribe (Sector Leader Human Development) and Roxana Viquez (former President of IMAS in Costa Rica). Volume I was prepared by Tarsicio Castafieda, James Cercone and Luisa Fernandez (Consultants) and volume II by James Cercone, Fabio Duran, Rodrigo Bricenio, Stephan Brunner, Juan Diego Trejos and Suhas Parandekar. The peer reviewers were Lynne Sherburne-Benz (Sector Manager, Social Protection) and Margaret Grosh (Lead Specialist, HDNSP). Valuable comments and suggestions were provided by Andres Rodriguez, Luis Carlos Hemandez, Francisco Esquivel, Ana Mercedes Brealey, Jose Elizondo, and Nohemi Araya (from Costa Rica), and by Indermit Gill (Lead Economist Human Development Sector), Carlos Felipe Jaramillo (Lead Economist), Kathy Lindert (Senior Economist), Joel Reyes (Senior Institutional Development Specialist), Maria Luisa Escobar (Senior Economist), Quentin Wodon (Senior Economist) and Ana Lucia Armijos (Senior Economist). Marlene Sims (Language Program Assistant) provided assistance in the production of the report. The previous draft of this report was reviewed by members of the Social Cabinet of Costa Rica and their comments have been taken into account in this report. We also greatly acknowledge the collaboration of other government officials and consultants in Costa Rica. Nonetheless, the opinions presented here and any errors are the sole responsibility of the authors and should not be attributed to the individuals or institutions acknowledged above. COSTA RICA: SOCIAL SPENDING AND THE POOR VOLUME I EXECUTIVE SUMMARY 1. Costa Rica made substantial progress in reducing headcount poverty and improving social sector indicators over the past decade. Driven by economic growth averaging 4.5 percent of GDP and strong investment in the social sectors, headcount poverty declined from 27 percent in 1990 to 21 percent in 2000, although the pace of reduction in income poverty slowed down during the second half of the nineties. While an increase in economic growth will almost certainly reduce poverty rates, Costa Rica's recent experience indicates that relying on economic growth and increased public spending alone will not yield sufficient reductions in headcount poverty. 2. There is also rising concem that fiscal pressure will limit future possibilities of increasing social spending pointing to the need for more efficiency in the sector. The fiscal deficit averaged 3.8 percent of GDP over the last decade. Expenditures are highly inflexible (composed mostly of salaries, other recurrent expenditures and debt service payments), and revenues have not increased accordingly. Thus, governments have had to recur to foreign and domestic borrowing to cover deficits. The extemal debt, for instance, increased from about 6.2 of GDP in 1990 to about 20 percent of GDP in 2000. A byproduct of the fiscal pressures is the potential stagnation of social investment and a weakening of the institutional structures required to meet the challenges facing the social sectors. 3. While coverage of education and health remains nearly universal in the case of basic education and access to healthcare services, there are worrying signs that social sector indicators and the efficiency and effectiveness of social spending began to plateau or decline in some cases. Between 1990 and 2000, Costa Rica fell from 28th to 48th in the Index of Human Development. In education, there has been a small decline in secondary enrollment, and increases in repetition rates. Across the board, education levels are worse for those in the lower income quintiles. In health, performance continues to improve in terms of life expectancy infant mortality and health systems performance, but there is increasing dissatisfaction with the system as waiting lists grow daily and patients requiring more sophisticated treatment for chronic diseases are turned away. 4. Over the last ten years, public spending grew seventy percent in real terms and the social sectors increased their relative participation from 59 percent of public spending to 63 percent, with spending in education and pensions growing more rapidly than spending in health and social protection. Given the country's tight fiscal situation and the high level of public debt, there is little room for continued increases in social spending to address the service gaps and coverage needs. 5. Costa Rica spends more on social services overall (education, health and nutrition, social security and social assistance) than other countries in Latin America i and considerably more than the average for other countries at similar stages of development outside the region. The level of spending and the universal focus of many of these programs, however, mask the true effectiveness of these programs in terms of reaching the poor and vulnerable groups living in rural areas and city slums. Despite recent growth in the economy, many of the poor and vulnerable do not have access to social programs. Finally, the inflexible legal framework that guides social spending budget allocation distributes resources without considering the changing conditions and needs of the poor. 6. Further improvements will require reforming social spending and programs that aim to reach the poor. Without improvements in the efficiency and effectiveness of social programs additional resources are not likely to lead to improvements in access or effectiveness in reducing poverty. The report shows that despite increasing spending in nearly all sectors over the second half of the decade, progress in extending coverage and reducing poverty stagnated. New instruments and approaches will be required that focus on obtaining value for money in the social programs. 7. Scope of this report. This report addresses three key issues. First, who are the poor and vulnerable groups in Costa Rica? Second, what has been the impact of social spending on programs that aim to improve the welfare for the poor? Third, what improvements are required and how can government spending be used more effectively to reach the poor and to contribute to reducing headcount poverty. We address the third issue keeping in mind that the fiscal deficit in Costa Rica is high (close to 4% of GDP). 8. By answering these questions, the report will address three main objectives. The first is to present an in-depth, multi-dimensional analysis of poverty. The second is to examine the effectiveness of government policies and spending on the poor in the social sectors. The third is to use the empirical findings to identify options and priorities for improving social sector spending and reducing poverty in the future. Policy options are outlined not only in general, but also for the specific themes and sectors covered. PROFILE OF THE POOR 9. The report starts with a profile of the poor. Using data from household surveys, augmented with information from a database containing specialized variables characterizing poor households (SIPO), the first section provides a rapid overview of who the poor are today. This overview provides policymakers with key information on who and where the poor are and will lead to more informed decision-making when public policies have to be made regarding how social spending can effectively contribute to alleviate poverty. The rapid poverty assessment shows that the poor have the following general characteristics: 10. As in other Latin American countries, poverty in Costa Rica affects more rural than urban residents. Over 65 percent of the poor in Costa Rica reside in rural areas which have only 49% of the population. Poverty incidence is highest in families whose heads are working in agriculture, self-employment, micro enterprise and in domestic service. This group represents over 60 percent of the total poor. 11. Irrespective of urban and rural differences, even the poorest Costa Rican families have an important level of accumulation of physical assets as indicated by the high percentage of families owning their own house (75%), although the quality of ii housing is far from adequate. Regarding quality of housing, about 107,000 (65 percent) of the houses of the extreme poor and poor have floors made of substandard materials, 98,000 (59 percent) have leaking or inadequate roofs, and 101,000 (61 percent) have walls made of substandard materials. Also, about 25,000 (15 percent of extreme poor and poor families) do not have in-door water connections; their water sources are wells, rivers, and rain, among others. There is a Government subsidy program to help the poor improve housing, but only about 9 percent of the poor reported having received a housing voucher from the govermment. 12. Poor families are younger, have higher dependency rates and most have no or only one income eamer. Poor families are larger than non-poor families (4.7 family members vs. 3.7 for non-poor) and have a higher number of people younger than 12 years of age (1.7 vs. 0.9), who demand both more expenditures and time from their parents. 13. Women heads are a large proportion of the extreme poor population. Women head an estimated 48 percent of families in extreme poverty compared to only 33 percent in near-poor families. 14. One of the most striking facts of extreme poor families is the extremely high unemployment rate for all family members. Unemployment rates of extreme poor heads and spouses are 9-10 percent, while those of children are over 20 percent. The unemployment rates of youth 19-24 years old in extreme poverty are especially high (21.9 males and 27.3 for females). It has been estimated that about 8,100 extreme poor and poor youth are unemployed and low-skilled, and are good candidates for youth training programs. 15. The extreme poor and poor, when occupied, work primarily in temporary activities, while the near-poor work primarily in permanent activities. Thirty six percent of extreme poor heads, 9 percent of their children older than 12 years of age, and 32 percent of poor heads and 15 percent of their children work in temporary activities. By contrast, 57 percent of heads and 32 percent of children of the near-poor work in permanent activities. 16. The problem of poverty is more one of incomes and labor force participation than unemployment. The incidence of poverty is highest in families where only one or no family member is working. These two groups are the most frequent in poor families and account for almost 80 percent of all income poor people in the country. Furthermore, 16 percent of the income-poor families in Costa Rica have no income earner in the family (wage earner, pensioner or renter). The incidence of poverty is high among the unemployed (52.4 percent), although this group is a relatively small percentage (4.0 percent) of the poor population. 17. Education levels are much lower for the population in the lowest income quartile group. The percentage of people in the poorest 25 percent of the income distribution who completed primary education was only 62 percent in 1999, about the same rate of 1990, while the percentage of the richest 25 percent is much higher and has shown some improvement in recent years. In urban areas, only about 15 percent of the heads of the poorest income quartile have more than primary education, in contrast to 77 percent of the heads in the highest 50 percent group. In rural areas, most household heads in all income groups have only primary education. Regarding children's human capital, only iii about 62 out of 100 children in the lowest income quartile complete primary education, while 82 percent of children in the richest quartile finish their primary education. And, while only 15 percent of youngsters in the lowest quartile complete 11 years of education, 60 percent (still a low number) of children belonging to the top quartile of the income distribution do so. 18. For secondary education, the completion rates of the poorest are less than one fourth of those in the top 25 percent. Only about 15 percent of the poorest children complete secondary education, a level considered as a threshold to break intergenerational poverty cycles in countries of Latin America. The situation is far worse for children 14-18 years of age, where 47 percent are out of school. Of those not attending school, about half of them work while the rest stays inactive. It has been estimated that about 33,600 youth 13-18 years of age are out of the school system and should be induced to return to school or to receive special training to upgrade their skills for the modemizing Costa Rican economy 19. The poverty profile shows that despite a universal health system the roughly 30 percent of poor are having trouble accessing the CCSS programs for the uninsured and poor. This reflects problems with knowledge and access. Health conditions also vary considerably between the non-poor and the poor. The results show that more than half of the poor and extreme poor had a self-declared health condition at the time of the survey. More importantly, a majority of the remaining people in extreme poverty indicated having a permanent physical or mental limitation. Over one third of the elderly in extreme poverty had a permanent limitation, and another 10 percent a temporary limitation. While the country has made the epidemiological transition, the health system has yet to fully adjust to the changed nature of health care needs. These results underscore the need to shift the health system towards management of chronic problems, disabilities and mental illness. 20. Despite the relative strength of the healthcare system and the progress in extending primary education, children 0-5 years of age are a highly vulnerable population group in Costa Rica. About 84,000 of them (95 percent of the poor group) are not attending CEN-CINAI or Hogares Comunitarios centers. Most children stay at home under the care of mothers, older siblings, or relatives. Over 80 percent of mothers of these children have only primary education and may be unaware or unable to provide adequate stimulation and the motor and social skills their young children need to start breaking the poverty cycle. 21. The elderly over 65 years of age in extreme poverty are also a highly vulnerable group because of their ill health and low incomes. About 46 percent of extreme poor people older than 65 years of age reported permanent or temporary physical or mental limitation. About 14,000 elderly in extreme poverty and poor (about 50 percent of group) are not covered by contributory or assistance pensions. However, recent government figures for 2000 and 2001 indicate that the number of non-contributory pensions awarded to the poor has increased considerably and that the number of uncovered elderly may be about 7,000, by the end of 2001. iv THE WAY FORWARD 22. The past decade has demonstrated that significant improvements are possible in the socioeconomic condition of the poor when economic growth is combined with cost- effective social programs that are well designed to reach the poor and to contribute to welfare improvements. But progress in poverty reduction and social sector performance has been uneven, and there are important gaps between rich and poor. Continued improvements will have to rely on better use of expenditures, as the scope for further increases in social spending are limited by fiscal and financial constraints. If the success of the past is to be replicated, the government of Costa Rica should undertake structural changes to improve overall effectiveness of programs in the areas of education, health, and pensions that offer opportunities for improving the condition of the poor and of social assistance programs to ensure that they reach the most vulnerable groups of the population. The report also highlights that significant improvements could be obtained without major additional spending and that one of the key elements to achieving greater effectiveness of social spending is an improved institutional framework that will coordinate the implementation of social programs, and greater flexibility in programs to respond to the needs of the poor. Each of these issues is addressed throughout Volume I of the report and in greater detail in Volume II. IMPROVING THE EFFECTIVENESS OF SOCIAL SECTOR SPENDING Education 23. Costa Rica has made major strides in the education sector, including nearly universal primary coverage and low illiteracy, and has gone further than most countries in introducing modem technology, mainly at the primary level. There have been some improvements in secondary education enrollment, increasing from 40 to nearly 50 percent by 1999 and further increasing to 54 percent by 2001 '. Increasing spending per student over the decade has contributed to improving coverage rates. Despite the increasing spending and increasing coverage rates, there are signs that performance decreased over the decade. 24. Although secondary coverage improved, little progress was made reducing the gap between rich and poor. From 1995 to 2000, the enrollment rates for the poorest 20 percent of the population did not improve and the poor remained roughly half as likely to enter secondary school as the richest 20 percent. 25. There are particularly alarming results in the areas of repetition, dropout rates, and low secondary completion rates. Key figures are: * Overall more than 10 percent of all students grades 1 to 6 repeat at least one grade, with higher rates for repetition in 1st and 7th grades (16 %). Decreasing The total secondary enrollment rises to 62 percent if other secondary programs such as IPEC are considered in the matriculate. v repetition has a great impact since cost reduction and resource allocation can be used for improving quality and coverage equity. * Each year nearly 23,000 students drop out at the primary school level and the 19,000 dropouts at high school level. * Only one third of the Costa Rican youth 20 years old have completed high school education. The differences are even more acute between regions and income quartiles. Only 15% of the youth 20 years old from the lowest income quartile have completed 12 years of education, while the proportion is close to 60%, or 4 times higher for those in the highest income quartile. Costa Rica's weak performance regarding secondary completion rates is highlighted by cross- country comparisons of OECD and Latin American Countries. Table 4.7 in the text shows that the secondary completion rates for Costa Rica are nearly 50 percent below Latin American leaders such as Chile and Colombia and only one third as high as Japan and Hungary. 26. The recommended measures for primary education include, among others: (i) strengthening the teaching and leaming of reading, writing and math skills in first through sixth grades; (ii) making more effective use of leaming assessment findings; (iii) improving in-service teacher training; and (iv) strengthening community and parental participation in school-related matters. 27. If improvements in value for money are to be obtained, structural changes will have to be made in the following directions: (i) update the existing curriculum; (ii) improve teacher training; (iii) increase secondary education supply through a menu of options, including traditional schools, tele-secundarias, distance learning, open-access education, virtual schools and vocational training; (iv) promote the collaboration with the private sector to strengthen the relevance of secondary education; and (v) help parents and students, especially in rural areas, to finance part of opportunity costs of studying through scholarships, conditional cash transfers, and other supplemental financing schemes. 28. Progress among the poor has been limited over the past five years as compared with the top income groups. Consequently, in addition to efforts to improve coverage and quality described above, the Ministry of Education needs to improve the targeting and equity of programs, such as school feeding, transportation and scholarships. 29. The net cost implications of the recommended primary and secondary education measures are relatively modest when compared with existing resources for education. In total, it is estimated that the cost increases of suggested measures would sum to between 6 and 7 percent of current expenditures. These increases could be financed by improving internal efficiency, through reduction in repetition rates, for example; and by reducing poorly targeted programs, which distribute resources to populations that are less needy. Crude estimates show that repetition in primary school costs about $21.2 million per year or about 10 percent of the education budget for primary education, while the costs for secondary education repetition are equivalent to $14.8 million per year. vi Health 30. Over the past decade, Costa Rica has made substantial progress in improving the health of the population. Costa Rica consistently ranks in the top 40 countries in the world in terms of life expectancy (76.7 years), infant (10.3 per 1,000 in 2001) and maternal mortality and health systems performance (36 according to WHO). As opposed to education where Costa Rica ranked considerably behind its peers regarding secondary enrollment in terms of value for money, in the health sector Costa Rica has actually decreased spending while extending access and improving productivity. Despite these improvements, the health sector is starting to feel increasing strain as Costa Rica's epidemiological and demographic profile changes and the opportunities to use the private sector to generate efficiency, quality and competition have started to expand. Costa Rica should respond to both these changes. 31. Throughout the 90's, the Government objective was to contain spending and to focus on allocating resources more efficiently to improve quality of service, equity and efficiency of service delivery. In the face of a rapidly rising GDP, and the re-estimation which increased GDP, total public health spending decreased slightly over the decade to just under 5.5 GDP in 1999. The Caja Costarricense de Seguro Social (CCSS) offers healthcare to all and has obtained coverage of nearly 90 percent of the population. Nonetheless, nearly 30 percent of the poor are uninsured and have problems accessing basic services. For those that do get access, waiting lists are a problem in the public as average waiting times over 12 months for several services. The pressure of a free health system is increasingly evident. 32. Costa Rica needs to continue the reform process started in the early 1990s to reduce waste and improve services by: (i) consolidating the Ministry of Health role as steward and policy maker and strengthening health education and public health programs; (ii) improving administrative and budgeting procedures in the CCSS and consolidating the culture of performance agreements with hospitals and other health providers to increase coverage of primary health care and improve efficiency of the system; (iii) promoting changes in the system to increase management and financial autonomy of health providers in the context of the 1999 Law on Deconcentration; (iv) reducing current disparities in expenditures levels by region which do not favor poor regions; (v) introducing changes in the pharmaceutical procurement system to produce savings in storage and procurement; (vi) strengthening the CCSS' collections system and promoting changes in the financial information systems; and finally, (vii) developing altemative delivery arrangements through public-private partnerships that complementing public services by the CCSS and gradually introduce managed competition in the healthcare market. 33. The cost implications of the continued reform process have been estimated at about US$70 million over five years, but these costs are about half the savings that can be obtained from the reforms over the same period. The Costa Rica Health Sector Strengthening and Modernization Project will produce substantial savings for the sector, primarily through reductions in hospitalization costs resulting from ambulatory surgery ($12 million per year), pharmaceutical management and reduced overhead from increasing decentralization ($12 million per year). Improvements in pharmaceutical supply chain management alone could save over $30 million per year by reducing unnecessary inventory costs. Even under conservative assumptions regarding the effectiveness and speed of change, direct benefits would exceed $24 million per year. vii Social Protection 34. In pensions, Costa Rica has made some significant reforms in the past decade. The majority of the reforms addressed the most pressing financial problems posed by the diversity of benefits among the many different pension schemes. The approval of the Ley de Protecci6n al Trabajador was an important step towards establishing a capitalization based pension system, as a second pillar (the first being the public system), under the supervision of the recently created Pension Superintendence. While this latter reform could be criticized for not going deep enough in reforming the existing pension system, it has the advantage that, compared to reforms in other countries, it has a limited fiscal impact. Part of the fiscal impact of the reform will be the institutional strengthening expenses, including public information campaigns, strengthening the Superintendence of Pensions, and establishing the centralized contributions collection system to be run by CCSS. Another fiscal impact could arise from the implementation of the new rules for the means-tested old age pensions and its fiscal cost. 35. There are important issues remaining for the future. These include: (i) reducing evasion and sub-declaration of salaries; (ii) adjusting the calculation of benefits to more accurately reflect eamings capacity, through providing incentives for contributing a greater share of actual eamings in the earlier years of the eamings cycle; (iii) extending coverage and improving targeting of the non-contributory pensions; and (iv) consolidating reforms that will guarantee the long-term sustainability of the pension schemes for Government workers. If reforms are not undertaken, as the population ages sharply over the next 30 years-due to falling fertility rates and rising life expectancy- the increasing financial imbalance could lead to pension fund insolvency. 36. In social assistance, Costa Rica has a well-established social assistance network designed to protect vulnerable groups and to deliver a wide gamut of social programs. There are however critical gaps in social assistance of key vulnerable groups, as many of the benefits do not reach the poorest individuals. Among the critical gaps in social assistance are the inadequate coverage of programs for poor children under five years of age, who may lack adequate education and stimulation, and assistance for the elderly poor not covered by the formal pension system. Institutional strengthening and modemization would help to avoid duplications, reduce costs, and offer better services and they could be complemented by delivery systems involving communities and NGOs to complement public sector agencies. 37. The importance of increasing the effectiveness of social assistance programs is underscored by the fact that Costa Rica spends between 1.5-1.8 percent of GDP in social assistance (a much higher percentage than that of other similar countries). Most of the social assistance spending is financed through FODESAF, a social fund created by earmarked taxation. Currently about 77 percent of FODESAF's funds are allocated by law to a few programs, some of them not of the highest priority for the current needs of the poor, and a myriad of other programs. FODESAF finances over 30 programs ranging from early child education (CEN-CINAI, Hogares Comunitarios, other), to school welfare programs, to housing subsidies and subsidies to purchase land and engage in micro-enterprise and or agricultural activities. IMAS has also a number of proprietary programs, sometimes directed to the same population groups and activities. These programs are often fragmented and uncoordinated. There have been recent improvements in program coordination and a few programs have been eliminated, but viii efforts are needed to change the inflexible system, often resulting from outdated laws and institutional arrangements. 38. Targeting of most safety net programs has been weak. According to household surveys statistics for 1999, only about 24 percent of subsidies involved in the public provision of early child education have gone to the poorest 25 percent of the population. Only about 25 percent of housing subsidies have gone to the poorest 25 percent of the population. Overall, the share of social assistance programs resources received by the poorest 25 percent of the population just approximates their share in the population, implying little or no re-distributive impact of spending. Targeting of assistance programs has improved recently with the introduction of the SIPO system to select beneficiaries (see below) for a number of programs, including the school voucher and scholarship programs, non-contributory pensions, school stipends to poor children and other direct income support programs operated by IMAS. 39. One social assistance program of high priority has been the De la Mano program created in 2000 to increase coverage of early childhood education to poor children 0-5 years of age. This "umbrella" program seeks to promote coverage of early childhood development care in the CEN-CINAI (Centros de Atenci6n Infantil Program (operated by the Ministry of Health), the Hogares Comunitarios program (operated by IMAS), and some others centers operated by the Ministry of Education and other government agencies. Overall during 2000-1 coverage increased from about 27,600 (about 5.5 percent) to about 44,600 (8.6 percent) poor 0-5 years olds. However, it has been estimated that 84,000 children remain without coverage and are under the care of poorly educated mothers, grandparents or older siblings. 40. The returns to investing in human capital since early ages are well documented in the literature. Early childhood education provides a number of benefits including, early stimulation and development of motor and social skill for children and access to quality child care for mothers who want to participate in the labor market or get training. The main focus of ECD programs in Costa Rica should be, however, in early stimulation and development and not food assistance and health care. Malnutrition in Costa Rica is quite low even in poor families and health care is currently being provided by the local health centers under the EBAIS program of the CCSS. The concentration on child stimulation and development will provide costs reductions and will permit a more rapid expansion of coverage of poor children. 41. Coverage of elderly poor with non-contributory pensions is still not universal despite significant resources spent for that purpose in Costa Rica. A large number of pensions are given to non-poor pensioners as the allocation criteria has been until recently influenced by political favoritism and other considerations. It has been estimated that an additional 7,000 elderly poor remained uncovered by the end of 2001. 42. The cost implications of expanding ECD programs for poor children and non- contributory pensions for the elderly poor are also relatively modest and appear to be affordable within the current social assistance budget. It has been estimated that increasing ECD coverage to all children in a mix of center-based care and conimunity and parental care, would cost about US$26 million per year. The current budget of the CEN-CINAI program is about US$19 million per year, and it is estimated that an additional US$ 11 million would be required to cover all poor children. These are rough estimates that assume that non-poor children are not attended in the public programs. ix Costs could be reduced firther if the CEN-CINAI programs are changed to reduce food subsidies and health care provided by the CCSS. In relation to covering the estimated 7,000 elderly poor still not covered, the annual cost could amount to about US$3.1 million which can be financed by shifting resources from non-poor pensioners that currently receive a public pension. 43. The SIPO program was created to identify beneficiaries of major social programs operated by IMAS, thereby improving the targeting of many social assistance programs. In 2000, the SIPO database of potential beneficiaries contained about 808,500 people, representing about 21 percent of the Costa Rican population. There are major challenges for SIPO. First, there is still a lack of a normative act mandating the use of the system by all institutions and social assistance programs. In the absence of such a normative, institutions will likely prefer to use their own targeting methods, often subjective and amenable to manipulation. Second, specific measures should be taken to update the system. Third, the applicability of SIPO for identifying the "new" poor may be limited. The survey form may be too long and costly to apply. The point score system based in weights of household characteristics describes better "structural poverty" than the "transient" poor. This, however, could be amended if, during crisis, a shorter version of the survey form is applied, giving larger weights to unemployment, under-employment and income variables. Not all programs, however, can or should be targeted with SIPO; some programs such as nutrition programs for child and mother would need other instruments. 44. In summary, spending priorities in social assistance need an urgent change. Priority actions are in three fronts: (i) introduce legislative changes to FODESAF's law, so that the government can change spending priorities according to needs and direct more resources to priority vulnerable groups, and be able to respond to crisis times; (ii) consolidate efforts to improve coordination of programs and eliminate duplications; and (iii) improve targeting, coverage and design social assistance programs, such as the CEN-CINAI, and (iv) complete coverage of the poor by non-contributory assistance pension programs. Improving Institutional Coordination and Program Evaluation 45. The impact of social spending in Costa Rica could be improved, particularly its impact on the poor, with the creation of a policy-making mechanism charged with setting social policy, exploiting synergies, and reducing programs overlaps. The government should consider possibilities for improving coordination of social policies by exploring the creation of a Council for Social Policy coordination under the direction of the President, with a technical secretariat to define priorities and regularly evaluate the impact and targeting of social programs. Such an entity would define programs and their goals, beneficiary populations, and the institutional arrangements to carry out key strategies. Efforts in this direction have been undertaken in recent administrations, but they have been hampered by institutional rigidities and laws which have undermined coordination efforts and the reform or elimination of ineffective programs or institutions. In this regard, allowing FODESAF greater flexibility to channel resources to vulnerable groups and to respond to changing conditions, particularly during crisis, will greatly contribute to improving the effectiveness of social spending. 46. Monitoring and evaluation systems should be strengthened. The lack of an effective M&E system for the social sectors is an impediment to ensuring value for x money in social spending. This is most evident in programs such as the CEN-CINAI where, had a good evaluation program existed, resources could have been saved by finding the most cost effective alternative for child care. Different modalities of child care, some including only parental education, others including center-based stimulation and early child education, among others, could have been evaluated to determine their impacts and relative costs. Monitoring and evaluation is key to evaluate progress, evaluate impact of programs and make more efficient decisions on programs to improve social indicators of the poor. The practice of establishing base-lines and measuring the impact of specific programs should be implemented routinely, in order to draw lessons from experience which can be used to modify programs and improve their impact. In addition, instruments such as Living Standards and Measurement Surveys and improved household surveys provide valuable information on key target groups and on the outcomes of the main social programs, complementing the annual household surveys currently carried out by the Statistical Institute. At the same time, evaluation should include ex-ante and ex-post evaluation of social sector programs using cost- effectiveness and cost-benefit analysis to assess programs. 47. The collective impact of the actions presented in this report would contribute to progressive improvements in the impact of social spending and to ensuring that poverty reduction is not tied exclusively to rapid economic growth. Clearly, improvements should focus on how to improve the impact of existing resources and ensuring that any additional resources that may be allocated to the social sectors are targeted to the poor. xi xii 1. INTRODUCTION Scope and Organization of this Report 1.1 Poverty rates declined over the decade by 30 percent in an environment of continuous economic growth, averaging 4.5 percent, and increases in social sector spending. However, economic growth has stagnated to less than 2 percent in 2000 and 2001, fiscal pressures threaten continuing spending increases and poverty reduction has declined. While an increase in economic growth will almost certainly reduce poverty rates and improve social indicators, Costa Rica needs to strengthen existing programs and to develop instruments to reduce poverty and improve stagnating social indicators of the poor. 1.2 Social spending in Costa Rica, at about 18 percent of GDP, appears to be already high for its economic level and fiscal means. The fiscal deficit averaged 3.8 percent over the last decade while the public debt has reached levels of over 40 percent of GDP in 2000, signaling the need for major improvements in the efficiency of public spending and targeting of social spending. 1.3 In order to respond to this situation, this report addresses the challenge of poverty reduction through three key dimensions. First, who are the poor? Second, what has the impact of social spending been on programs that aim to improve welfare for the poor? Finally, what improvements are required and how can government spending be used more effectively to reach the poor and to contribute to reducing headcount poverty. 1.4 The objectives of this report (Volume I) are threefold. First, present a poverty profile. Second, review efficiency and targeting issues of social spending on education, health and nutrition, pensions, so as to identify actions that can be taken to improve social services and have more impact on indicators of the poor. Third, identify possible options and alternatives Costa Rican governments have for addressing identified issues, including improving the institutional policy making and executing framework, and the financing and delivery of social services. 1.5 This report consists of five chapters. After the introduction, chapter 2 contains the poverty profile using household survey information from earlier studies, and the System for Targeting Social Spending (the SIPO system in Spanish) database for 1999-2000. This latter information source is uniquely rich in that it contains information on over 207,000 poor families which have been identified as potential beneficiaries of social assistance programs. Chapter 3 contains a summary of key issues in education, health, social security and social assistance programs which are affecting the efficiency, equity and effectiveness of these programs in Costa Rica. This review is based on the results of the social spending review presented in Volurne II. Chapter 4 discusses the main institutional issues which appear to be affecting the effectiveness of social spending in improving social indicators of the poor. Finally, Chapter 5 summarizes the main recommendations of the report. 1.6 Before we present the poverty profile and social spending chapters, this introductory chapter contains a review of the economic and social context by examining recent trends in main economic variables, social indicators, and social spending. Social indicators and social spending figures for Costa Rica are compared with those of selected Latin American and non-LAC countries to assess the relative position of Costa Rica in the 1 region and in other regions of the world. Also, in this chapter, we make a brief review of main demographic trends of Costa Rica, which will help identify future needs for social services. Economic Context 1.7 Costa Rica's economy has grown at an average of 4.5 percent over the past decade, with a low point in 1996. In recent years, however, the economy has shown a much slower growth (2.2 percent in 2000 and an estimated 0.3 percent in 2001) (Table 1). Growth was strong in 1998-99 (8.3 percent on average), in large part the result of the construction and a major increase in production by two new Intel plants.2 hIflation was about 14.4 percent per year on average during the last decade. Average minimum real wages increased by about 1.7 percent per year in 1990-99. The most dynamic sector of the economy has been manufacturing, especially after 1996 with the establishment of Intel and other industries, now accounting for 23 percent of total production. The growth of tourism has also been impressive leading the service sector to account for over 19 percent of total production in recent years. By contrast, agriculture has been loosing ground as a share of total GDP from about 16 percent of GDP in 1990 to about 14 percent in 1999.3 1.8 Open unemployment was 6 percent on average per year during the last decade and employment grew at a rate of about 3 percent per annum absorbing the rapidly growing labor force bom after the 1960s and 1970s. The most dynamic sectors for employment creation have been manufacturing, trade, services and construction. Agriculture has shed employment in absolute and relative terms. While in 1976 agriculture accounted for over 35 percent of total employment, it accounted for only 20 percent in 1999 (GOCR, 2000a). 1.9 The external current account deficit improved to 3.1 percent of GDP in 1999 from 5.3 percent in 1998. The historical trade deficit has become a surplus due largely to Intel exports (about $2.5 billion in 1999), while profit remittances have increased substantially. The deficit in the current account was more than covered by net private and official capital inflows. In 1999, inflows of direct foreign investment were $624 million, equivalent to almost 4 percent of GDP. 1.10 The persistence of fiscal deficits (including Central Bank losses) is one of the main outstanding problems of the Costa Rican economy. The fiscal deficit averaged 3.8 percent of GDP over the last decade. Expenditures are highly inflexible (composed mostly of salaries, other recurrent expenditures and debt service payments), and revenues have not increased accordingly. Thus, governments have had to recur to foreign and domestic borrowing to cover deficits. The external debt, for instance, increased from about 6.2 of GDP in 1990 to about 20 percent of GDP in 2000. The fiscal situation could worsen with the potentially large increments in earmarked expenditures for education due to the recent upward revision in GDP figures (the Constitution establishes that expenditures in the education sector should be at least equivalent to 6 percent of GDP). A byproduct of the fiscal pressures is the potential stagnation of social investment and a weakening of the institutional structures required to meet the challenges facing the social sectors. 2 A wide gap has emerged in Costa Rica between GDP and GNP figures because large foreign investment flows have led to profit remittances equivalent to 8 percent of GDP. Thus, while GNP rose by 8.7 percent in 1998, it showed no growth in 1999 and 2000 as profit remittances picked up. These figures are from World Bank (2000). 2 Table 1: Key Economic Indicators In Costa Rica (1990 - 2000) :1ndicat6oi 1990 1996 1997'l 1998 1999; :2000 GDP growth (%) 3.5 0.9 5.6 8.4 8.2 2.2 Inflation (%, CPI end-period) 27.2 13.9 11.2 12.4 10.1 11.0 Open Unemployment (%) 4.6 6.2 5.7 5.6 6.0 6.0 Overall public sector balance (% GDP) -2.0 -4.1 -2.6 -2.0 -3.6 -3.7 Current account balance (% GDP) -7.4 -3.8 -4.9 -5.3 -3.1 -3.1 Direct foreign investment (% of GDP) --- 0.04 0.03 0.04 0.04 0.03 Domestic public debt (% GDP) 18.3 24.5 23.1 21.8 25.1 23.2 Extemal public debt (% GDP) 6.2 24.4 21.8 21.0 20.3 19.9 Sources: BCCR, IMF, Costa Rica, Informe Estado de la Naci6n, Informe 4. 1.11 Social spending in Costa Rica is much higher than in other LAC countries. Total social spending was 18.2 percent of GDP, as compared to 14.5 percent for selected LAC countries, in 1996, the latest year for which comparable information has been assembled (Table 2). Adding social welfare expenditures for Costa Rica, social spending is as high as 20 percent of GDP, that is, over five GDP percentage points higher than in the selected LAC countries in the Table.4 Thus, within the acute fiscal constraint Costa Rica now faces, continued improvements in social indicators and poverty reductions will have to rely on better use of expenditures, as the scope for further increases in social spending are very limited. Table 2: Public Spending in Social Sectors in Selected Countries, 1996 (Percentage of GDP) Ed ' --Mation- i6jeiltli I.S , Housinig .Tot'al - ' - ' . ,,: , ecurityad .and Weifare Argentina 3.5 1.8 8.0 0.8 14.1 Bolivia 7.4 2.2 4.5 0.2 14.3 Brazil 3.0 1.7 11.0 0.6 16.2 Chile 3.2 2.5 7.1 1.2 14.0 Colombia 5.4 2.0 4.9 0.7 13.0 CostalRica, ;' - >:5.P , i .8'. -. 6.3 - b.4i-= 18.2. Dominican Republic 2.0 1.7 0.7 2.1 6.5 Indonesia 1.3 0.4 1.1 3.0 5.7 Korea 3.7 0.2 2.2 0.5 6.5 Malaysia 5.0 1.4 1.5 1.4 9.3 Mexico 3.8 0.5 3.0 0.5 7.8 Panama 4.6 5.6 5.7 1.2 17.1 Philippines 3.4 0.5 0.5 0.3 4.7 Turkey 3.5 0.7 1.4 0.4 5.1 Uruguay 2.1 1.9 19.4 0.5 23.9 Latl,An ena ;* l . : J27r.2 . 7.1 'i , i. ;1`.54 Aveiage" > ] m,! ;:i, ,p -, .' - ,,, 8,Far!E.astern,AverageY 3 ' 13 , 1.3. 66 O.v'er,all[,ve,rage*..':i '3.8 >*...x'.2t.0i . .H.5 1 ;'9.- VOi'$'cX.8 ' * These averages include only the countries in the table of each group. Source: World Bank, 2000b, based on Government Finance Statistics Yearbook (1998) of the IMF. 4The figures for Costa Rica decrease substantially if the new corrected figures for GDP are used. However, aggregate social spending in Costa Rica continues to be 1-3 percentage points higher than for the rest of LAC countries. 3 Social Indicators 1.12 Costa Rica has achieved great progress in social indicators. Infant mortality at 10.3 per 1,000 live births and child malnutrition are among the lowest in Latin America (Table 3)5. Access to potable water is nearly universal. Illiteracy rates for youth, at 2 percent for both males and females, are among the lowest in Latin America. Coverage of preventive and curative health care is high as 90 percent of the population has access to healthcare through the CCSS. Table 3: Social Indicators in Selected Latin American, Europe and Far Eastern Countries, 1998 Country r &Iiifai' 'C-hild', -QAdulffIlliteracy : Net Mortalivty Mal- Rte, : ,., e Enrollment (pe1,0 nutrition', (Q-5jlmd,)~r ,, l, (perlO,O, 'nuttion (,15'ano,ver), '(il5-24 yrs: old), ! econdary,. 'live,births) . 1992-98 .Male 'rile .Male,.; Female .1,7 Argentina 19 2 3 3 2 1 77 Bolivia 60 8 9 22 2 7 40. Brazil 33 6 16 16 10 6 66 Chile 10 1 4 5 2 1 85 Colombia 23 8 9.0 9.0 4 3 76 Costa Rica l04 5 5.0 5.0 2 2 612 Dominican Rep. 40 6 17 17 10 9 79 Indonesia 43 34 9 20 2 4 56 Korea 9 n.a. 1 4 0 0 100 Malaysia 8 20 9 18 3 3 64 Mexico 30 n.a. 7 11 3 4 66 Panama 21 6 8 9 3 4 71 Philippines 32 30 5 5 2 1 78 Turkey 38 10 7 25 2 7 58 Uruguay 16 4 3 2 1 1 84 LAC & Carib.3 31 8 11 13 7 6 66 Europe & CA3 22 8 2 5 1 2 81 EU3 5 n.a. n.a n.a n.a n.a 96 '. Weight for age measure (% of children under five years of age). Data are for the most recent year available. 2 Taken from the 2000 Census. 3. It includes weighted averages of the ratios for all the countries in the group. CA: Central Asia 4 This figure is for 2001 Source: World Bank (2000c). The World Development Indicators, 2000. 1.13 In education, Costa Rica has achieved universal enrollment in basic education and 60 percent enrollment in secondary education. These results lag behind averages for LAC where net enrollment rates in secondary education are 66 percent and, and completion rates of secondary education are nearly 33 percent. In 1995, the latest year for which information is available for a number of countries, only about 30 percent of Costa Rican males and 35 percent of females 22-24 years of age, had completed their secondary education, as compared to 36 percent of males and 40 percent of females on average for selected Latin American countries. Moreover, Costa Ricans 22-24 years of age had significantly lower secondary education completion rates than people of the same age group in a country with similar per-capita GNP and public expenditures on education, as those of Costa Rica (see in Figures 1-4 the distance between the dots corresponding to Costa Rica and the continuous lines). The completion rates of males and females in Costa Rica are over 22 percent lower than those of a similar country in per-capita GNP, and 5The figure for 2001 was 10.8 per 1,000 live births. 4 about 34 percent for females and 50 percent for males lower than those of a similar country in public expenditures on education.' Figure 1. Secondary Completion and GNP - --- -- -- -'-'------- - --- 5 _ _ . . _ _ 00 a-.-*---~---.-. _.0 !$ i" -5 - - - - - - - - -f-3- - -- 25~~~~ -,,,lw -__ -,,, -1 - - - - .~ - - - - - -*CI 01 0 - - - . * I c - -- -- --- - ,..:02 60 II0 *1.0.0 0 IS .NO~S.-------- 35 - - - - - -- -- -- 0 Z0 1 6 ) 0 10000 12000 0 20 t3 O 1,000 1200 O5P09C,0.(U05I00.WI') Source: Table 3 Figure 2. Secondary Completion and Public Expenditure in Education 6Us : i- ______ _ _ _ _ _ ________ NO . ___ _ _ _ -______ - 435 - 15 -__' -= _ '- - -- 20 _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ * L.p.035 6 2 3 0 2 5 4 5 a 7 a I X 3 *0 5 7 I E4..P~.b0o EV.di5.%GD255 159 Source: Tables 3 and 4. 1.14 The rapid increase in the working age population will put great pressures on the economy to insert the newcomers into the labor market, without great rises in unemployment, and/or falls in relative wages. The rate of employment creation will have to remain at levels of 2-3 percent per year-a rate similar to that of the 1 990s-during the coming decades for the economy to be able to accommodate the new labor force entrants with the increments in labor force participation that normally occur with development. 6 These figures are calculated from regressions as the percentage difference between the estimated and observed completion rates for Costa Rica. 5 1.15 Taking advantage of the new opportunities created by the increase in labor force and decline in dependency rates will, thus, critically depend on the fate of the economy. If economic performance continues to be satisfactory, workers and the government can produce savings to finance the increased expenditures of the aging population. The number of direct contributors to health and pension regimes will increase producing a reduction in beneficiary-per-worker ratios and more revenues for those regimes, for a long period of time. Some of these demographic effects have already started to show up in the health and pension regimes of the CCSS, in the 1990s. The number of family member per active contributor has declined from 1.88 in 1990 to 1.46, in 1999. This provides an opportunity for assessing possibilities of lowering already high payroll contribution rates- and, thus, the tax on labor--so as to promote additional employment, and to continue reforming the pension system to make it fully funded and able to meet the challenges of the future, as suggested below. Table 4: Forecasted Population by Age Group, 2000 - 2050 (Figures in Thousands) Age Group (Years) Year ofForecast Period (%) .* - - - e= :.* 2000. 2010 2020 2030 2040 . 2050, 0 - 4 408 440 438 447 447 448 (%) (10.0) (9.0) (8.0) (7.0) (7.0) (7.0) 5 - 19 1,227 1,279 1,326 1,322 1,333 1,336 (%) (31.0) (27.0) (24.0) (22.0) (21.0) (20.0) 20 - 59 2,014 2,625 3,026 3,252 3,408 3,422 (%) (51.0) (55.0) (55.0) (54.0) (53.0) (50.0) 60 and more 291 436 697 1,028 1,292 1,588 (%) (7.0) (9.0) (13.0) (17.0) (20.0) (23.0) Total 3,943 4,781 5,488 6,050 6,484 6,793 (%) (100.0) (100.0) (100.0) (100.0) (100.0) (100.0) Dependency ratio (%)° 96.0 82.0 81.0 86.0 90.0 99.0 Source: Centro Centroamericano de Poblaci6n, Universidad de Costa Rica. 2001. '.Calculated as the number of people 0-19 years of age plus the population over 60 years of age divided by the population 20-59 years of age. 1.16 The demographic impact on poverty reduction is likely to be mixed. On the one hand, the poor, as well as the non-poor, are likely to have lower dependency rates and lower family consumption and expenditure pressures. On the other hand, the poor may have great difficulties in successfully competing in the now tighter labor market, given their highly deficient levels of education and training, in the face of a rapidly changing and modernizing economy. To help the poor compete and enter into this economy, major and sustained efforts will need to be made to improve their human potential and skills, as indicated in most of this report. Do the Poor benefit from Government transfers and programs? 1.17 Interventions designed to improve the education level of the poor, their employment opportunities, and their access to health, basic infrastructure services and better housing are key to offset the gap in standards of living between the them and the rest of the country. Since the mid 1990s, there has been a substantial increase in real terms in funding for the social sectors in Costa Rica as in other countries. In this section, after briefly documenting the increase in public social spending, we assess with the available 6 data whether the poor benefit more or less than others from a few Govermnent transfers and programs for which data on targeting are available. 1.18 National and local governments have a fundamental role to play in reducing poverty and improving social indicators. At the national level, stable and sound macroeconomic and labor policies are one of the most important contributions that governments can make for the reduction of poverty. A second important contribution comes from the impact of public spending on the social sectors. According to CEPAL (2001), social public spending per capita in Latin America increased on average by 50 percent between 1990-91 and 1998-99. The magnitude of the increase differed between countries, but the increase was generalized. In Colombia, Guatemala, Paraguay, Peru and the Dominican Republic, spending more than doubled. Increases in social spending tended to be smaller in countries with higher initial levels of spending (Argentina, Costa Rica, Brazil, and Panama), with increases averaging around 30 percent over the decade. In Costa Rica as in other countries, the increase in spending was made feasible by economic growth (a higher level of per capita GNP leads to higher spending per capita holding the share of spending in GNP constant), and a reorientation of fiscal priorities towards the social sectors. 1.19 Today, in comparison to many other countries, the level of public social spending in Costa Rica is relatively high. Table 5 proposes a typology of the countries in terms of their fiscal pressure and their fiscal priorities. Countries such as Argentina, Brazil, and Uruguay have high levels of social spending as a share of GNP because they combine high fiscal pressure with an emphasis on the social sectors for expenditures. By contrast, countries such as El Salvador, the Dominican Republic, and Peru have low levels of social spending as a share of GNP because they combine low fiscal pressure with low levels of priority for the social sectors. Table 5: Typology of Latin American countries in terms of social spending, 1998 Fiscal pressure: Expenditures as Fiscal priority: Share of total public spending going to social sectors share of GNP Less than 40% Between 40% and 60% More than 60% More than 30% Nicaragua (12.7) Costa Rica (16.8) Argentina (20.5) Colombia(15.0) Brazil (21.0) Panama (19.4) Uruguay (22.8) Between 20% and 30% Honduras (7.4) Bolivia (16.1) Chile (16.0) Venezuela (8.6) Less than 20% El Salvador (4.3) Guatemala (6.2) Dominican Republic (6.6) Mexico (9.1) Peru (6.8) Paraguay (7.4) Source: CEPAL (2001). The numbers in parenthesis correspond to the share of social spending in GNP 1.20 The extent to which the poor benefit from public social spending in Costa Rica depends on the type of spending considered. While spending for primary education tends to be pro-poor, spending at the upper secondary and tertiary levels level tends to be pro- rich. At the same time, several categories of spending are targeted to poor areas and households. 1.21 One simple way to summarize the distribution of public spending is to provide estimates of the Gini Income Elasticity (GIE) of the various transfers and programs (see Wodon and Yitzhaki, 2002, for a description of the GIE). A GIE of one means that program benefits or transfers are distributed in the same way as per capita.income, so that 7 an increase in benefits or transfers would not affect the inequality in per capita income. A GIE larger (smaller) than zero means that per capita income and program benefits are positively correlated, hence the households in the upper (lower) part of the distribution of income benefit more from the increase in spending. A GEE of zero means that all households benefit in the same way on a per capita basis from an increase in spending for the program, hence the increase in spending is progressive since it is distributed more equally than per capita income. A GEE between zero and one means that the distribution of spending for the program is positively correlated with per capita income, so that the households who are better off receive a higher share of the benefits on a per capita basis, but the increase in spending is still progressive because it is not as unequally distributed as per capita income. 1.22 Overall, when comparing programs or categories of transfers, the program or transfer that is most redistributive is the program with the lowest GIE. While other factors than the GIE should be taken into account when deciding which programs or transfers to cut or expand, the GIE does provide a good basis for ranking the redistributive impact of alternative policies. 1.23 Table 6 provides estimates of the Gini income elasticity of selected social sector programs. The estimates for the parameter v=2 correspond to the impact of the program on the standard Gini index of inequality. Estimates for higher values of v put more weight on the bottom part of the population when assessing the redistributive impact of transfers but for the sake of this analysis a parameter of v=2 is considered. The results show that school vouchers are better targeted than school feeding programs (larger negative GIE). Overall, programs from CEN-CINAI are better targeted than school vouchers or school feeding, and within the CEN-CINAI programs, the milk program appears to be the best targeted. The targeting performance of non-contributory pensions is similar to that of the school programs, but lower than that of the CEN-CINAI programs. As far as the housing vouchers are concerned, while they were not well targeted in the early 1990s, targeting performance has improved since then, even though it remains below that of other programs. In most cases, the ranking of the programs does not change when the weight placed on the poorer member of society (the parameter v) increases. But in one case--child care--the redistributive assessment of the programs is lower with more weight on the poorer members of society. Table 6: Gini income elasticity for different values v. v=2 v=4 v=6 v=8 School voucher -0.97 -1.19 -1.27 -1.30 School feeding -0.65 -0.78 -0.83 -0.85 CEN-CINAI Childcare -0.52 -0.47 -0.41 -0.35 Milk -1.26 -1.80 -2.12 -2.29 Food rations -0.80 -0.98 -1.04 -1.05 Total -1.16 -1.62 -1.88 -2.03 Pensions -0.85 -1.14 -1.32 -1.41 Housing Bonds 86-90 -0.05 0.06 0.15 0.22 90-94 -0.09 0.00 0.04 0.06 94-98 -0.31 -0.23 -0.18 -0.15 98-99 -0.67 -0.78 -0.79 -0.78 Source: Authors' estimation using 1999 household survey results. 1.24 Beyond targeting performance, the impact of programs on beneficiaries should also be taken into account when evaluating alternative options. While there are no good 8 estimates of these impacts (such as increased school attendance, or better nutrition and their impact on productivity at adult age), emphasis can be placed on the type of considerations that should be taken into account for a more comprehensive assessment of the various programs. More precisely, in order to take into account long-term impacts, one may use a social welfare function concept, and assess the impact of programs on social welfare, that is on both the income of households and the inequality in the distribution of income, taking into account estimates of the income value of the long-term impacts of programs. Details on the methodology are provided in Wodon et al. (2002), but the basic idea is to argue that the impact of a program on social welfare is proportional to two factors: a) the program's income multiplier effect; and b) its distributional characteristics as measured by one minus the product of the GEE and the Gini for per capita income. This type of framework is used in table 7 in order to compare three of the programs mentioned earlier. 1.25 It has been suggested that school vouchers have positive behavioral effects (because they tend to increase school attendance, and thereby education endowments, with positive long-term impacts on future wages), while food subsidies may not have such effects, or at least may not have them to the same extent. In the case of housing as well, general subsidies may have only limited long-term behavioral effects. When programs have long term positive effects, we may take this into account through the concept of multiplier effects over time, so that for example a transfer of one colon today may lead to benefits of two or more colons for the beneficiary household. Let us assume for the sake of the illustration that this multiplier effect is equal to 3 in the case of school vouchers. Assuming no multiplier for the milk and housing programs, and using as a social welfare function the mean per capita income times one minus the Gini index of inequality, one gets the ranking presented in table 7, with school vouchers now clearly being the best program, followed by milk subsidies and housing transfers. Table 7: Illustrative impact on social welfare of four alternative programs, 1999 Growth impact.per Gini Income Gini index of Overall impact on dollar spent Elasticity inequality social welfare (A) (B) (C) A*(l-B*C) School vouchers 3 -0.97 0.5 4.455 Milk 1 -1.26 0.5 1.63 Housing 1 -0.67 0.5 1.335 Source: Authors' estimation using 1999 household survey results. 1.26 The broader message of table 7 is that the long-term positive effects of social programs may matter as much if not, more for improving social welfare than their immediate distributional characteristics. In order to take this message into account in the design of social programs, it is however necessary to have estimates of the long term impacts of programs, which are often lacking in Costa Rica, as in other countries, due to the weak monitoring and evaluation framework for social programs. 9 2. POVERTY PROFILE: CHARACTERISTICS OF THE POOR AND VULNERABLE GROUPS 2.1 This chapter contains a description of recent trends in income poverty and asset accumulation of the Costa Rican population over the last decade. The analysis includes a profile of the poor based on household surveys and the SIPO database (see Annex 1 for description of the SIPO database). The profile of the poor can be helpful to identify ways in which social spending can improve social indicators of the poor and vulnerable groups in Costa Rica. Trends in Income Poverty and Asset Accumulation by Families in Costa Rica 2.2 Costa Rica has made significant progress in reducing poverty in the last ten years. From 1987 to 2000, the incidence of poverty, or headcount ratio, declined by over 30 percent (Table 8). Even more impressive has been the reduction in extreme poverty from about 10 percent in 1987 to about 6.4 percent in 1998.7 Impressive results are shown in the "poverty gap" that measures how poor are the poor compared to the poverty line. While the average incomes of the poor were about 11 percent lower than the poverty line in 1987, average incomes were only about 8 percent below the poverty line in 1999. Less impressive declines in poverty rates are shown in poverty rates that use a different method of calculation, the basic needs index (NBI), from 47 percent in 1989 to 42 percent in 1997 (Trejos, 2000).8 Money incomes have increased more rapidly than access to other basic services included in the NBI Index. Table 8: Poverty Incidence and Poverty Gaps, 1987 - 2000 Years- [eadcount PoPve i eap. roity;a't$1 -Pov'ert- Gap,' o Poerty at"S2 oPoverty (At~~~~~~~~m rt.4, Po2rty (At Poer y L -(Al Poverty Per-Capita (At Sl Per-' er-Capita p'er. ap (At $2 -:.- -Lde)' :'- Line;-.: pei.y . Capliai - D Pie-Capita) 1987 29,0 11,3 - - - - 1988 28,4 11,3 - - - - 1989 28,3 11,0 - - - - 1990 27,1 10,7 - - - 1991 31,9 13,0 - - - 1992 29,4 11,4 - - - 1993 23,2 8,7 - - - - 1994 20,0 7,4 - - - l995 20,4 7,5 - - - - 1996 21,5 8,4 9.6 3.2 26.3 10.1 1997 20,7 7,4 - - - - 1998 19,7 6,8 - - - 1999 20.6 8.0 - - - 2000 21.1 - - - - Source: Ouinto Informe del Estado de la Naci6n cited in GOCR (2000) and Sauma (2001) for poverty rates and World Bank (2000a) for international poverty rates at $1 and $2 dollars per-capita per day. ' The number of people below the $1 dollar per-capita per day (which may be considered extreme poverty) has been estimated at 9.6 percent, while the number below the $2 dollar per-capita per day (poverty), at 26.3 percent in 1996 (World Bank, 2000a). The Basic Needs Index (NBI) method is an alternative form of designating a household to be poor. A household is poor if its housing is made of substandard materials, if it lacks access to safe water or sanitation, if the housing is overcrowded (two or more members per room), if there is a high dependency rate, or if a child aged 7-11 is not attending school. If any of these needs are not met the household is considered to be poor. If two or more needs are present, the household is determined to be in extreme poverty. 10 2.3 The pace of the reduction in income poverty has, however, slowed considerably after 1994 despite strong economic growth (over 4.5 percent in 1990-2000). Some observers have suggested this is an indication that the recent economic growth, largely associated with export industries (Intel Corporation), has had weak links with the rest of the economy, especially the rural economy. Yet, another possible reason may be that many poor Costa Ricans may be unable to take advantage of employment opportunities and higher incomes offered by the growing economy due to their poor training and low education.9 2.4 Consistent with the sharp decline in poverty over the last ten years, between 1992 and 1997, Costa Ricans improved asset accumulation, by increasing housing ownership, improving housing quality and purchasing durables. Table 9 displays the data for 1992 and 1997, the most recent year for which comparable information on housing characteristics is available. Three out of four Costa Rican families owned their own home in 1997, as compared to a little more than one in two in 1992. Nearly four in five families had basic services and possessed high priced comfort durables, such as color TV, refrigerators and washing machines, in 1997. 2.5 In contrast to rapid increases in physical assets, advances in human capital assets of parents and children have been modest in the last few years (Table 9). The number of family heads who have completed at least primary education has remained more or less constant, at about 37-39 percent, from 1992 to 1997. Net enrollment rates of children 0-5 years old in Early Child Development Programs (ECDP) at 6-8 percent are very low for countries of income levels similar to those of Costa Rica. Attending these programs is critical for the children to improve their health, nutrition and early stimulation status. In addition, despite the historically high enrollment rates in primary education, the number of students who by age 14 complete six or more years of education was only 76 percent in 1999, about the same rate of 1990. The number of children, who by age 20 have completed 11 years of education --secondary education-- or more, is also low at 33 percent in 1999 (27 percent in 1990). 2.6 Education levels are much worse for those in the lower income quartiles. The percentage of people in the poorest 25 percent of the income distribution who completed primary education was only 62 percent in 1999, about the same rate of 1990, while the percentage of the richest 25 percent is much higher and has shown some improvement in recent years. For secondary education, the completion rates of the poorest are less than one fourth of those in the top 25 percent. Some improvements in the completion rates of the poor have occurred in recent years. In rural areas, the completion rates of secondary education are much worse than in urban areas, although they show some improvement from 1990 to 1999. These generally low levels of education attainment of Costa Rica's children and youngsters undermine their ability to take advantage of the high growing technology and other modem sectors of the economy.'0 9 Some of the measurement problems of comparing GNP figures, which are calculated from National Accounts, and poverty rate changes, which are calculated from Household Surveys, are discussed by Deaton (2001) in the context of similar discrepancies between growth and poverty in India. '° Net enrollment rates for primary and secondary education for Latin America and many other countries are in World Bank (2000a). Table 9: Asset Accumulation by Families in Costa Rica, 1992 - 1997 :- . .'.Asseis - e :-- -: -.1992 .:1997 %.ChanFe Phvsical Assets - Own a house' 51.7 75.2 45.5 - House with basic services 74.6 84.6 13.4 - House of good/regular quality 74.8 84.7 13.2 - House has comfort durables: TV 48.6 77.5 59.5 Refrigerator 60.2 82.6 37.2 Washing machine 50.7 76.3 50.5 Human Capital of Head of Household With some secondary & more 39.4 36.9 -6.4 Human canital of children ECDP for 0-5 Years Olds2 6.0 Complete primary3 77.4 76.4 -1.3 First quartile (25% poorest) 62.3 62.1 -0.3 Top quartile (25% richest) 90.6 92.4 1.9 Rural areas (all) 69.1 69.9 1.2 Complete secondary 26.6 32.7 22.9 First quartile (25% poorest) 11.6 15.0 29.3 Top quartile (25% richest) 49.6 59.3 19.5 Rural areas (all) 16.2 20.2 24.7 Source: IMAS, 2000c. '. The figures on ownership for 1992 are from the ENISO Survey and correspond to ownership of the "site", while those for 1997 coffespond to ownership of the house. The poor are more likely to own the house but not the site, especially in slum areas. However, these cases are not more than 2.5%, according to figures from SIPO. 2, Net enrollment figures include children in CEN-CINAI centers, Hogares Comunitarios, and other public childcare programs for 1999. 3. Figures for 1990 and 1999 from Trejos, J. (2000). A Profile of The Poor 2.7 This section first presents a poverty profile, based on household surveys and published studies and then provides greater detail from the population included in the SIPO database 2.8 As in other Latin American countries, poverty in Costa Rica affects more rural than urban residents. Over 65 percent of the poor in Costa Rica reside in rural areas. This is because the proportion of people living in rural areas is still large in Costa Rica and because there is a higher incidence of poverty in rural than in urban areas (Table 10). 2.9 The incidence of poverty (that is, the percentage of poor in a given income category) is highest in families where none or only one family member is working. " I These two groups are the most frequent in poor families and account for almost 80 percent of all income poor people in the country. Furthermore, 16 percent of the income poor families in Costa Rica have no income earner in the family (wage earner, pensioner or renter). The incidence of poverty is high among the unemployed (52.4 percent), although this group is a relatively small percentage (4.0 percent) of the poor population. "A good description of characteristics of the poor by activity and occupational category is in Trejos (2001). 12 2.10 Poverty incidence is highest in families whose heads are working in agriculture, self-employment, micro enterprise and in domestic service. This group represents over 60 percent of the total income poor. Agriculture employment, however, at the aggregate economy level, has been rapidly declining from about 35 percent in 1976 to 20 percent in 1999, while employment in commerce and other services has been growing rapidly (GOCR, 2000a). The rapid growth rate in non-agricultural sectors has created opportunities to move out of agriculture into more skill demanding jobs of the modem sectors of the economy. Table 10. Poverty Rates and Contribution to Poverty by Group, (1997 Household Survey) -~~ -'- ~~~P6(~~rtyae'' -oniributlon 'to .~~~~ t- ¢ ; - r 1 ' , t",*,;s ' 'PovertyL _~~ , * s * r- ,, * ' ' .. ' (IncIdeiiei'" Srivey-1997 Headcount 20.7 - Region -- 100.0 Urban 16.3 34.6 Rural 24.1 65.4 Persons Working, in Family -- 100.0 None 45.6 22.9 1 24.1 56.2 2 11.1 15.1 3 9.5 4.2 4 and more 8.5 1.6 Income Earners, In Famiy' -- 100.0 None 73.1 16.3 1 25.8 60.7 2 11.1 16.7 3 8.7 4.5 4 and more 7.5 1.7 Head's Activity -- 100.0 Unemployed2 52.4 4.0 Inactive (not in labor force) 31.2 30.7 Occupied3 17.3 65.3 Head's Industry -- 100.0 Agriculture 31.8 29.8 Non-agriculture 12.5 35.5 Micro enterprise 17.1 19.8 Domestic service 34.9 2.7 Self-employed 19.8 13.0 Source: Trejos, 2001. 1. Includes employed, pensioners and renters. 3 This is not the open unemployment rate, but the share of unemployed in total number of heads. 3. This is the share of occupied people in the total number of heads. 2.11 Poor families are younger, have higher dependency rates and most have none or only one income earner. Poor families are larger than non-poor families (4.7 family members vs. 3.7 for non-poor) and have a higher number of people younger than 12 years of age (1.7 vs. 0.9), who demand both more expenditures and time from their parents (Table 11). Poor families also have a smaller number of people who are active in the labor market (1.2 vs. 1.6), and have a smaller number of people working per family than non- poor families (1.1 vs. 1.5). 13 Table 11: Characteristics of Poor and Non - Poor (Household Survey - 1997) 1 * i .. ; ~-'.'atl6nai- Non-ioor Poor. - Demographics Persons per Family 4.1 3.7 4.7 Persons in Working Age 3.0 2.8 3.0 Persons Active 1.6 1.6 1.2 Persons < 12 Years of Age 1.1 0.9 1.7 Working People per Family 1.5 1.5 1.1 Dependants per Worker 1.7 1.5 3.4 Characteristics Heads Percent Women Heads (%) 20.7 19.6 27.1 Head's Age (Years) 45.9 46.0 48.6 Head's Education (Years) 6.9 8.3 4.7 Source: Trejos (2001) and authors' calculations. '. The characteristics of this group are those of the non-poor according to NBI and income criteria (Trejos, 2001). This is a smaller and higher income group than the non- poor according to income alone. Thus, the indicators in the table may be better than those of the income non- poor, and the difference in poor/non-poor may be somewhat exaggerated. 2.12 All labor force participation rates (for heads and non-heads) for the poor are much lower than those of non-poor families (Table 12). Participation rates of male heads are 7 percent and those of non-head males 15 percent lower than those of the non-poor counterparts. Participation rates for females are even much lower than those of the non- poor counterparts. For poor female heads, participation rates are 30 percent lower than for the non-poor, and for non-heads, participation rates are more than 50 percent lower than for the non-poor. There may be many factors explaining the lower labor force participation of the poor, including the high fertility of poor women, their low education, and the possible lack of childcare services, among other factors. 2.13 Unemployment rates are high for the poor. Poor heads and non-heads face unemployment rates which are more than three and a half times higher than those faced by their non-poor counterparts. While unemployment rates for non-poor male and female heads are reasonably low --less than 3 percent-- they are significantly higher for the poor (5 percent for males and 10 percent for females). The situation is far worse for poor non- heads and is similar for males and females. Almost one in four poor non-heads in the labor market were unemployed as compared to one in fifteen in non-poor families, in 1997. 14 Table 12: Labor Force Participation and Unemployment Rates of Heads and Non-Heads, (Household Survey - 1997) '*'' . 'i iStlon-[i| Nou-Pooir t '~ or Labor Force Participation Heads 79.6 80.3 69.3 -Male 86.9 86.3 80.4 -Female 51.7 55.4 39.5 Non Heads 39.8 41.6 26.4 -Male 59.7 56.8 48.2 -Female 29.4 34.7 16.3 Unemployment rate (total)2 5.7 3.7 13.2 Heads 2.0 1.1 5.8 -Male 1.6 0.8 5.0 -Female 4.4 3.1 9.9 Non Heads 9.5 6.5 23.0 -Male 10.0 7.7 22.3 -Female 9.0 5.6 23.9 Source: Trejos (2001) and author's calculations. '. The characteristics of this group are those of to the non-poor according to NBI (Basic Needs Index) and income criteria calculated by Trejos (2001). This is a smaller and higher income group than the non-poor according to income alone. Thus, the indicators in the table may be better than those of the income non- poor, and the difference in poor/non-poor may be somewhat exaggerated. 2 2 This is the open unemployment rate. 2.14 Costa Rican families show some accumulation of assets in 1997, even for the poorest of the poor.12 There are, however, great disparities by poverty level for both, urban and rural areas (Tables 13 and 14). While only about 27 percent of the poorest urban families (approximate first quartile of the income distribution), and 55 percent of the poorest rural families, own their own home, about 79 percent of urban and 85 percent of rural families in the richest 50 percent group do so. Access to basic services is deficient, especially in rural areas for the poorest quartile of families. Also, the indicator for housing quality is highly deficient in urban and rural areas for the poorest 10 percent of the population. 2.15 Fewer of the poor complete primary and secondary education. In urban areas, only about 15 percent of the heads of the poorest income quartile have more than primary education, in contrast to 77 percent of the heads in the highest 50 percent group. In rural areas, most household heads in all income groups have only primary education. Regarding children's human capital, only about 62 of children in the lowest income quartile complete primary education, while about 82 percent of children belonging to the richest quartile do so. And, while only 15 percent of youngsters in the lowest quartile complete 11 years of education, 60 percent (still a low number) of children belonging to the top quartile of the income distribution do so. 12 Poverty level is defined according to a point score system developed by IMAS to select people for the different programs of IMAS. The point score system gives a poverty score to each family in the sample (poor and rich), according to the socioeconomic characteristics of the families, using principal component analysis. In practice, each characteristic is calculated a weight, and these weights are added up to arrive at the total point score for the family. The lower the score, the poorer the family is. The technical aspects of the methodology are presented in Elizondo et al. (2000). 15 Table 13: Investments in Assets by Poverty Level in Costa Rica, Urban Area, 1997 Assets 0-.4% .2.4%/e 50% :-,1.6.2% 25.0%: 51:0% -''Total I Podrest .. Richest.! :. Number of Families 1,176 7,059 14,706 47,647 74,411 149,410 294,115 Physical Assets Own a house 27.5 40.4 50.3 62.5 74.5 79.1 73.2 House has basic servicesi 14.3 60.8 83.3 94.4 98.6 99.5 96.0 House good/ regular condition 36.3 35.8 62.1 79.1 92.1 99.0 90.5 House has comfort items: - TV color 18.4 48.9 59.2 75.5 87.0 96.2 87.3 - Refrigerator 22.4 32.5 59.8 82.8 91.3 97.8 90.0 - Washing machine 19.9 36.7 57.6 75.5 84.3 93.6 84.9 Human Capital of Head Complete primary & less 100 92.8 83.2 75.2 66.0 22.9 47.0 Secondary (some+ complete) 0 7.2 16.8 23.5 30.8 38.0 32.1 Higher education 0 0 0 1.1 3.2 39.1 20.9 Human Capital Children' Ql (poorest quartile) Q2 Q4 Total Primary and more 62.1 79.9 92.4 81.6 Secondary and more 15.0 18.4 59.3 43.7 Source: IMAS, 2000b. '. Water, electricity, sanitation. 2, Figures for national level, 1999. Calculations by quartile made by Trejos (2001) based on household surveys. The third quartile not included. Table 14: Investments In Assets by Poverty in Costa Rica, Rural Area, 1997 Aissets :'-" .: -: 1.8%. 4.3% -8.7% 29.3% 28% -48.0%= ...'Total Poorest' .est Number of Families 6,777 16,190 32,757 35,016 105,425 180,353 376,519 Physical Assets -Ownahouse 55.2 51.5 63.7 71.7 72.7 85.9 77.0 -House has basic services' 9.5 29.5 48.7 57.7 75.2 90.7 73.2 -House is good/ regular 14.4 35.6 52.4 68.6 82.4 96.1 81.8 -House has comfort: - TV color 9.9 19.2 32.3 47.6 63.5 87.3 67.8 - Refrigerator 9.7 20.1 37 60.5 75.3 92.6 75.3 - Washing machine 3.0 20.1 29.6 53.9 65.0 85.9 67.8 Human Capital, Head Primary (some and complete) 100 98.3 96.1 94.7 91.4 62.7 78.9 Secondary (some and complete) 0.0 1.7 3.7 5.3 8.4 26.0 15.8 Higher education 0.0 0.0 0.0 0.0 0.0 11.1 5.3 Human Capital Children' Primary and more 69.9 Secondary and more 20.2 '. Water, electricity, sanitation. 2 Refers to 1999. Figures from Volume 2 of this report with data calculated by Trejos (2000). Source: IMAS (2000b). 2.16 In summary, the income poor have large families to support and low labor market participation rates for both heads and non-heads. In addition, poor households face much higher unemployment rates than the non-poor. While unemployment may be discouraging labor force participation, especially of poor non-heads, other factors may be also present, including the low education, the high fertility of the poor, the lack of child care, among 16 other factors.'3 Even the poorest Costa Rican families have an important level of accumulation of physical assets as indicated by housing ownership, quality of housing, and purchase of durables. The situation is, however, not the same for human capital accumulation of the poorest-only about 15 percent of the poorest children complete secondary education, a level considered as a threshold to break intergenerational poverty cycles in countries of Latin America. In 1999 the majority of poor (25 percent of the poorest) children 15-17 years of age (45.7 percent, male) were working instead of attending school and this figure is nearly twice as high for adolescents in rural areas as compared to urban teens. 2.17 The poor are lagging behind in several aspects: * There are major gaps in education coverage. The problems begin with the youngest children, 0-5 years of age, where social spending is directed to mniddle and upper income children. * There are major differences in education attainment of youth ages 13-18 between the top and bottom quintiles of the income distribution as a result of high repetition in primary and secondary education and dropouts. In this age group, only about 15 percent of youth complete secondary education in the bottom quintile compared to 60 percent for the top quintile. * The poor participate much less in labor market activities than do the non- poor, especially those who are not household heads, and when they do participate they have substantially higher unemployment (3 to 4 times higher) than those of the non-poor. This is most likely the result of their low education and skills. * Poverty appears to be related to the disintegration of the family. Single women head nearly fifty percent of extreme poor families. This fraction is much larger than poor and near-poor families, at 31 percent and 33 percent, respectively. Over 90 percent of women heads of all poverty groups have no spouse, as compared to only about 10-20 percent poverty levels for family with male-headed households. * There are a still large number of elderly poor who have not been covered by non-contributory pensions despite the fact the Costa Rica spends sufficient resources to reach the elderly population with non-contributory pensions. Expanded Poverty Profile Using the SIPO Database 2.18 This section presents a more detailed analysis of the family and individual characteristics of the poor and vulnerable groups and of their main social service needs based on the SIPO database. 14 Families are classified in three groups: (i) the extreme poverty group, includes families whose per-capita income is below the cost of the basic 13 Some people have even suggested that the existing great number of social programs for poor families may be having an adverse effect on labor force participation of women. Unfortunately there are no studies to support or deny this hypothesis. 14 Some methodological issues conceming SIPO's representation of the population as compared to household surveys and the 2000 population census are presented in Annex 1. 17 food basket; (ii) the poor group includes families whose per-capita income is below the cost of a basic consumption basket (poverty line) and above the cost of the food basket; -and (iii) the near-poor group includes all families with per-capita incomes above the poverty line. About 46 percent of families are classified as extremely poor, another 34 percent as poor, and another 20 percent as near-poor (Table 15). The near-poor are those with income over the poverty line but who reside in the areas surveyed by SIPO. This latter group is not a representative sample of the non-poor population. 2.19 While the majority of people in SIPO live in urban areas, the majority of the extremely poor reside in rural areas. As indicated in the methodological considerations Annex 1, however, the urban-rural definition in SIPO is different from that used in household surveys and it is, thus, not possible to compare these results with those of household surveys. 2.20 Women heads are a large proportion of the extreme poor population. Women head about 48 percent of all families in extreme poverty, compared to only 33 percent in near- poor families (Table 15). The family size of the extreme poor families is much larger than that of poor and near-poor families. Table 15: CharacteristIcs of Families in SIPO Data Base .Eiiiedie Poor.' - :- - oo . . r Near- poor - .Total. - No. Families 95,047 70,385 42,164 207,596 (46%) (34%) (20%) (100%) No. Persons 395,919 284,541 128,125 808,585 (49%) (35%/o) (16%) (100%) Persons / family 4.2 4.0 3.0 3.9 By Area 100.0 100.0 100.0 100.0 Urban 47.0 60.0 59.0 53.0 Rural 53.0 40.0 41.0 47.0 By Gender Total 100.0 100.0 100.0 100.0 Male 46.2 49.2 51.1 48.1 Female 53.8 50.8 48.9 51.9 By Gender Heads 100.0 100.0 100.0 100.0 Male 52.0 69.0 67.0 61.0 Female 48.0 31.0 33.0 39.0 Source: Author's calculations using SIPO data base 2.21 The poorest families tend to be younger than the near-poor families. Over 65 percent of children of extreme poor families are younger than 12 years of age, as compared to about 44 percent of children of near-poor families (Table 16). And while only about 8 percent of children in extreme poor families are older than 19 years of age, about 31 percent of children in near-poor families are in that age group. 5 Children of near-poor families stay longer in school, and stay longer with their families before they migrate or have a new household. As shown below, older children at home in near-poor families participate more in the labor market and help them move out of income poverty. 15 Household surveys provide information only of children living in the household at the time of the survey not of the total number of children a family has had. 18 Table 16: Characteristics of Families by Poverty Level Characteristics- - Extreme Poor: ' oor ; - - Ner-poor Hesads Spouse .Chfldren Headi Spokiie Chlldreni Heads Spouse Children. Age Groups 0-5 Years - - 23.6 - - 19.6 - - 16.5 6-12 - - 42.1 - -- 38.5 - - 27.5 13-18 2.7 1.5 26.6 0.6 2.5 28.1 0.7 4.2 25.2 19-24 8.3 6.9 5.3 6.5 10.1 9.4 10.0 13.3 17.9 25-30 11.7 17.4 1.0 12.6 17.5 2.1 12.5 13.7 6.6 31-40 30.5 39.0 0.8 31.8 35.3 1.4 22.1 24.3 4.3 40-60 29.6 25.6 0.6 34.0 26.7 0.9 36.6 35.3 1.9 61-plus 17.2 9.5 0.1 14.6 7.9 0.1 18.1 9.2 0.1 Total 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 Education' 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 Some primary 50.2 46.5 43.6 39.7 35.7 32.1 35.4 31.5 22.0 Complete primary 34.7 41.1 32.3 38.4 42.4 35.4 35.6 38.9 36.4 Some secondary 11.4 9.5 20.9 15.3 15.7 26.6 15.6 17.5 28.1 Complete second. 2.5 2.3 1.9 4.6 4.6 3.3 7.2 6.9 6.7 Higher 1.0 0.6 1.2 2.1 1.5 2.5 6.0 5.1 6.6 Activity' Occupied2 53.3 7.1 12.8 77.5 12.8 26.0 78.2 28.8 48.2 Unemployed3 9.3 10.2 23.9 2.1 3.1 9.8 1.4 1.3 4.5 Inactive 41.2 92.1 83.2 20.8 86.8 71.2 20.7 70.8 49.5 - House wife 21.4 88.6 3.7 8.1 83.8 2.8 6.4 66.6 2.3 - Student 1.1 0.2 57.7 0.1 0.3 53.3 0.1 0.5 37.4 Occupied' Permanent 17.4 1.7 3.4 45.3 5.8 11.2 56.6 19.9 31.9 Temporary4 35.9 5.4 9.4 32.2 7.0 14.8 21.6 8.9 16.3 Occupational CategoryI 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 Unpaid worker 0.6 2.2 16.9 0.2 1.2 3.7 0.2 0.7 1.5 Public employee 1.6 2.3 0.6 6.3 5.3 1.6 12.4 17.2 6.3 Private employee .27.4 26.5 27.2 50.2 41.9 49.1 56.4 51.2 65.0 Self employed 69.8 68.8 54.9 42.5 51.0 44.9 30.2 30.4 26.7 Employer 0.6 0.2 0.4 0.8 0.6 0.5 0.8 0.6 0.6 Source. Author's calculations with SIPO database. 1. Children are 12 years of age and older. Children refer to code 3 (Hijos) in survey form. 2 This is the share of occupied people in population older than 12 years of age. 3. This is the open unemployment rate. 4. Temporary includes seasonal and occasional. 2.22 Education attainment of all family members in extreme poor and poor families is very low. Over 85 percent of heads and spouses of extreme poor families and about 78 percent of heads and spouses of poor families have only primary education (some and complete), as compared to about 71 percent for heads and spouses in near-poor families in SIPO data base. The situation of children over 12 years of age is not much better. Only about 24 percent of extreme poor and 32 percent of poor children have more than primary education, as compared with 41 percent for near-poor children in SIPO. A strong positive relationship appears to exist between mothers' and fathers' education and that of their 19 children, as has been found in econometric studies for many LAC countries (Castafieda- Aldaz, Moran, 2001).16 2.23 All family members in extreme poor families have low labor force participation rates, including family heads, spouses and children older than 12 years of age. The participation rate of extreme poor heads is only 58.8 percent as compared to 79.2 percent for poor heads, and near-poor heads 79.3 percent. The differences in participation rates between extreme poor, poor and near-poor are even higher for spouses and children. Extremely poor spouses participate very little (7.9 percent), with poor (13.2 percent) and near-poor spouses (29.2 percent) participating at considerably higher rates. This may be because poor women have more children younger than 12 years of age than near-poor women, and because of their lower education levels. 2.24 Employment rates (employed in relation to people 12 years and more) of all family members are significantly lower for the extreme poor (53 percent) than for the poor and near-poor (77 and 78 percent respectively). The employment rate of near-poor spouses and children are about four times higher than those of the extreme poor spouses and children, as near-poor families have a substantially larger proportion of children older than 19 years of age (in the family) than do the extreme poor and poor.17 2.25 The extreme poor and poor, when occupied, work primarily in temporary activities, while the near-poor work primarily in permanent activities. Thirty six percent of extreme poor heads, 9 percent of their children older than 12 years of age, and 32 percent of poor heads and 15 percent of their children work in temporary activities. By contrast, 57 percent of heads and 32 percent of children of the near-poor work in permanent activities. 2.26 All family members (heads, spouses and children) tend to work in the same occupation category, but the occupation categories vary widely by poverty group. All members of extreme poor families work primarily in self-employment, while poor families work in self-employment and in the private sector. By contrast, near-poor families work primarily are more diversified, working in, the private sector, self-employment and the public sector.18 2.27 Unemployment rates are very high for all family members of extreme poor families. Unemployment rates are particularly high for spouses and children (non-heads), as was reported in previous section using the 1997 household survey. Poverty appears to be related to the disintegration of the family, as single women head nearly fifty percent of extreme poor families. This fraction is much larger than in poor and near-poor families, at 31 percent and 33 percent, respectively. (Table 17). 2.28 Families headed by women, especially in extreme poor families, have lower labor force participation rates. While the labor force participation of male heads is over 75-80 percent for all three groups, the participation rates of women heads reach only 36 percent. 16 We do not attempt to test this hypothesis with the SIPO data base because the near-poor in SIPO data base is not a representative sample of the non-poor population, and thus we will have, most likely, severe sample selection bias. " This explains the apparent paradox that the children of the near-poor work more than the children of the poor, when the opposite result may be expected. The main difference is thus the age structure of children of poor and near-poor families, which may be a key factor in explaining income poverty. The children of the near-poor are older, more educated and contribute to family income. 18 Unfortunately, the SIPO data base does not distinguish by sector of activity (agriculture, industry, services, etc.) or by size of industry, as it is done in household surveys. 20 They also have higher unemployment rates than those of male heads in extreme poor families. The lower participation and higher unemployment rates of female heads are not compensated by higher participation of other family members. 2.29 Families headed by women have fewer members active and or working in the labor market than do poor and near-poor families. They are also smaller than those headed by men for all poverty groups, the largest difference being for the extreme poverty group (4.9 members vs. 3.3 for the near-poor). When female heads and children work, they do so primarily in temporary jobs and in self-employment. These jobs provide not only less security, but also less benefits as health insurance, and others. Health insurance rates for female heads are much lower in extreme poor (less than 70 percent) than for near-poor (84 percent) and male headed households are also much more likely to be insured. Table 17. Characteristics of Household Heads by Gender and Poverty Level, 2000-2001 tCharacteriftic :- E:;-." Extreme Poor, :. :;:kPoor :. Near-poor; -. - Total:.- ,~ ~ ~ - ; ~ :;.-;-Males F.emales: izMafgle.Fmales' Males...Femaies. -~Males Femde Number 49,115 45,932 48,617 21,768 28,170 13,994 125,902 81,694 Heads (%) 52.0 48.0 69.0 31.0 67.0 33.0 61.0 39.0 With spouse (%)' 89.0 8.0 92.0 12.0 80.0 12.0 88.0 10.0 Education (Years) 4.3 4.8 5.2 5.3 5.8 6.1 5.0 5.2 Age (Years) 44.9 40.2 42.4 43.2 43.3 44.8 43.6 41.8 Persons per family (No.) 4.9 3.3 4.4 3.3 3.3 2.5 4.4 3.2 Children 0-12 (No.) 2.0 1.5 1.5 1.1 0.7 0.5 1.5 1.2 Persons in working age 2.7 1.7 2.7 2.0 2.4 1.8 2.6 1.8 Activity Occupied (%)2 74.0 32.0 87.0 57.0 86.0 62.0 82.0 44.0 - Permanent (%) 27.0 7.0 52.0 30.0 61.0 47.0 45.0 20.0 -Temporary (%)3 47.0 25.0 35.0 27.0 25.0 15.0 37.0 24.0 Unemployed(%)4 8.6 10.8 2.1 2.1 1.5 1.0 4.4 5.6 Inactive (%) 19.0 64.0 11.0 42.0 13.0 37.0 14.0 53.0 Workers per family (No.) 0.9 0.4 1.2 1.0 1.4 1.2 1.2 0.7 Persons active per family 1.0 0.5 1.2 1.0 1.5 1.2 1.2 0.8 Dependants per worker5 2.3 1.9 1.5 1.2 0.7 0.7 1.6 1.3 Dependants per eamer 2.1 1.7 1.4 1.1 0.7 0.6 1.5 1.3 Single with children (%)6 4.0 77.0 4.0 75.0 4.0 65.0 4.0 74.0 Insured (%) 68.7 70.2 75.7 76.0 79.1 84.8 73.7 74.3 Source: Author's calculations using SIPO database. These are people married or in consensual union. 2 This is the share of occupied people in the total of heads. 3. Includes seasonal and occasional in SIPO occupational categories. 4. This is the open unemployment rate. 5. Dependants are those less than 12 years of age and the older than 60 years of age. 6 Heads who are single, divorced, separated, widows and have children or grandsons. 2.30 Although housing ownership is relatively high, the quality of housing is far from adequate. Very few (9 percent) of the extreme poor and poor families in SIPO reported having received a housing subsidy (Bono). Regarding quality of housing, about 107,000 (65 percent) of the houses of the extreme poor and poor have floors made of substandard materials, 98,000 (59 percent) had leaking roofs made of substandard materials, and 101,000 (61 percent) had walls made of boards and other similar materials. Also, about 21 25,000 (15 percent of poor families) have no indoor water connections. Their water sources are wells, rivers, and rain, among others. 2.31 Migrants. About 94.8 percent of people in the SIPO population are Costa Ricans, about 4.9 percent are Nicaraguans and the rest (0.3 percent) are from other Central American and other countries (Table 18). It may be expected that the number of migrants may be underestimated if people are fearful of reporting their true nationality, or if they refuse to be surveyed because they believe they are not eligible for the social programs of IMAS or other social welfare institutions, or other reasons. Table 18. Distribution of Population In SIPO by Nationality and Poverty Level ) 2 .!Eeme.-P-ooT'.;'3>. IItSPAd Near-Poor: -Total r Costa Ricans 376,515 269,414 120,528 766,457 Men heads 51.2 69.1 66.4 60.4 Women heads 48.8 30.9 33.6 39.6 Education (years) 3.6 4.4 5.4 4.2 Unemployment rate (%) 12.5 4.2 2.4 6.6 Nicaraguans 18,322 14,206 7,069 39,597 Men heads 56.6 68.6 71.7 63.8 Women heads 43.4 31.4 28.3 36.2 Education (years) 3.3 4.4 5.2 4.0 Unemployment rate (o)' 10.6 3.4 1.7 5.6 Other Nationalities 1,067 915 520 2,502 Men heads 58.5 68.0 70.0 64.4 Women heads 41.5 32.0 30.0 35.6 Education (years)2 4.8 6.3 7.1 5.9 Unemployment rate (%)2 8.1 4.9 2.2 5.0 Total 395,904 284,535 128,117 808,556 (100.0) Source: Author's calculations using SIPO database. '. This is the open unemployment rate. 2, These are simple averages of the other nationalities. 2.32 In general, the characteristics of Costa Ricans in the SIPO database appear to be somewhat worse than those of other nationalities. For instance, the proportion of women heads is higher for Costa Rican families than for other nationalities, especially for the extreme poverty group. Also, the unemployment rates of all active persons older than 12 years of age are higher for Costa Ricans than for other nationalities, especially for the extreme poor group. Finally, while the average number of years of education is similar for Costa Ricans and Nicaraguans in all three poverty groups, they are much lower than those of nationals of other countries, in all three groups. 2.33 These numbers are however only indicative, as they may hide several biases including the possibility that those accepting to be interviewed for SIPO may be the relatively better off migrants with a legalized status. It is interesting to note that the proportion of Nicaraguans in SIPO is 4.9 percent, similar to that reported in the Census (5.6 percent of total Costa Rican population), contradicting the expectation that Nicaraguans could be over-represented in the poor in Costa Rica. . 9. Further investigation of these issues may be done using more detailed information of the 2000 population 22 Vulnerable Groups in Costa Rica 1. Poor Children 0-5 Years of Age 2.34 According to figures from SIPO, only about 3 percent of extremely poor children and 7 percent of poor children attend Hogares Comunitarios or CEN-CINAI Centers, the two major programs of this kind for low income families, as compared to 15 percent of the near-poor children.20 (Table 19). This group of children faces several risks, including lack of early physical and intellectual stimulation and poor motor and social skills.21 According to information from SIPO's database, over 80 percent of extremely poor children 0-5 years of age have mothers with only primary education (incomplete plus complete), as compared to 54 percent for the near-poor and about one-third of extremely poor children have mothers who have no health insurance, as compared to only 13 percent for the near-poor. 2.35 Most mothers of extremely poor and poor children have low labor force participation rates, in part to care for their small children and probably because of their low education. In addition, when extremely poor women participate in the labor market, their open unemployment rate of more than 11 percent is much larger than that of the other groups. Almost 80 percent of mothers of extremely poor and 75 percent of mothers of poor children are inactive, as compared to about 50 percent of near-poor mothers. Table 19: Characteristics of Children 0-5 Years of Age Clihiaicterlstici i. Extreme Poor. -.Poor.:- Near-p ooi . 'Total, Number of Children 0-5 57,010 31,262 9,725 97,997 Number of children per mother 1.4 1.3 1.2 1.3 Mother's Education' 100.0 100.0 100.0 100.0 Incomplete primary 38.7 26.9 16.5 32.6 Complete primary 43.0 44.5 38.6 43.0 Incomplete secondary 14.5 20.8 24.3 17.5 Complete secondary 2.8 5.5 11.2 4.5 Higher education 1.1 2.4 9.4 2.3 Mother's Activity' Occupied3 18.3 25.0 49.5 23.5 Unemployed4 11.4 2.4 0.9 6.1 Inactive (mainly housewife) 79.5 74.4 50.1 75.0 Mother's Health Insurance' 100.0 100.0 100.0 100.0 Insured 67.3 77.3 86.7 72.4 Not Insured 32.7 22.7 13.3 27.6 Attendance ECD Centers 100.0 100.0 100.0 100.0 Hogares Comunitarios 1.7 5.2 13.4 4.0 CEN-CINAI Centres 1.4 1.5 1.4 1.5 Other Centers 1.3 1.8 2.4 1.5 Not Attending 95.5 91.5 82.8 93.0 Source: Author's calcuEations from SIPO's database. Mother's characteristics are those of heads or spouses with children 0-5 years of age. Not included are the characteristics of mothers of those 0-5 years old children who appear as grandsons/daughters of family head. In SIPO, there were about 1,800 grandsons/daughters 0- 5 years of age not covered by ECD programs. 2. This is not a representative sample of the non-poor. '. This is the share of occupied people in the total of mothers with children less than 5 years old.'. This is the open unemployment rate, 20 Coverage of this group is overestimated because not all children of this group are included. 21 Research on this topic in developed and developing countries has documented significant benefits of ECD programs to mothers-who may be able to work and or study-older siblings, who can devote more time to studying instead of caring for younger children, children themselves when they grow older, who achieve greater levels of schooling and earnings, suffer less unemployment, and to society for lower levels of crime and expenses of the justice system (Karoly et al., 1998). 23 2. Youths, 13-18 Years Old 2.36 Of all youth 13-18 years old in SIPO's database, over 52 percent are extremely poor and about 36 percent are poor (Table 20). About 80 percent of their mothers have only primary education. Extreme poor and poor mothers have lower participation rates in the labor market, than near-poor mothers (24 percent for mothers of extreme poor vs. 44 percent for mothers of near-poor). Table 20: Characteristics of Youth 13-18 Years VCfajrktWi1ltk- EWtOrWiiei. - 1Nsor m . ' Near-poo'r Totai Number of young 13-18 64,182 44,650 14,768 123,600 No. Children per mother 1.7 1.6 1.4 1.6 Mother's Education' 100.0 100.0 100.0 100.0 Incomplete primary 43.6 33.3 28.4 37.9 Complete primary 40.8 42.7 40.5 41.4 Incomplete secondary 12.1 16.7 16.5 14.2 'Complete secondary 2.7 5.4 7.5 4.3 Higher education 0.9 2.1 7.1 2.0 Mother's Activity' 100.0 100.0 100.0 100.0 Occupied2 24.3 30.4 43.9 28.8 Unemployed3 6.7 1.8 0.7 3.7 Inactive (mainly housewife) 74.0 69.1 55.8 70.1 q 0of*Education (Youngsters), Males Females Males Females- NMaies Females 'Total' Primary 76.3 73.3 70.6 64.8 64.0 55.7 70.6 Secondary 19.8 23.8 26.7 33.5 33.4 42.1 26.5 Higher 0.1 0.2 0.1 0.2 0.2 0.3 0.2 Activity (Youngsters) Permanent job 2.8 0.7 5.2 1.8 10.2 3.4 3.0 Temporaryjob 8.8 2.1 12.4 3.1 13.3 3.4 6.8 Unemployed3 20.4 32.7 9.4 16.1 5.6 9.2 15.2 House wife 0.2 5.2 0.2 4.1 0.1 4.3 2.3 Student 67.9 68.6 67.6 73.1 64.4 73.8 69.0 Other inactive 17.4 22.0 12.8 16.9 10.6 14.6 17.0 Source: Author's calculations from SIPO's database. These may not all be the biological mothers of these children. These are, however, women household heads or spouses in families where those children live, and are, thus, responsible for them. 2 This is the share of occupied people in the total of mothers with children 13-18 years old. 3. This is the open unemployment rate. 24 Text-BoA V ,;hhiaho t Poweiry in 'Q o 's ; ig Results from 'a recent U1CIEF, and,UCR.study showt,hat 30 ,ift of all individual6supdeirl8 years of age are iblow' the poverty line If total, childhood poverty. accounits fobt a 500perqcnnt of all,p iy zy -i~ ifC:ta ~Rica. ,Acc,ording to this sudithe, y,t eyst bhildren *om. -i yirs, are`more 'li,e1y to.be ^poor' tepoveity te,torchi,9-ll years t yruA1 'children is even oFe seere 2n ral areas,32 j"ent of all children 4and ~ adolesents live,n poverty, cormpned to 27,percent in utban areas. This irip,lies tt two o.f qver ypoor, children-ieside in'rtaI areas. Based on these statistics, it is clear that any'podve`ty reduction sategy .hastoaddresstheissueof rural;.childhood poverty. . 1 r Other,relevaint egiultisffron the ut 'show that:'' e o ,-, e,'70,peidentrof the,ufidei,;l8'popula4tin uf pqvery'ls composed of children under 11 U. .,Neary half of the .childlfol ptoverty is 'concentr d in ithe cdertrl rlegion.- The nsk of beip`g poor .a a lcildd is drt1y reted to the ¢onditior of the' hou,seHld. Children livingip a n'ile:headed household " 50 percent less-likely to be ,poorthiania - i '.c.hil4Wen fema1e,hea,deid shou seh,olds (29 perc ' t:o mal~ and 43 ierce,t, for fbmnale,,headed ,hotehbold ii-r' a,,E, '.>o^,;-' 4- 'Nlkar1yI perc'a of the families whose, head of household is an. agriculuialbwo4treare t below 'the`pdo*etty threshold, twice the national average and rcpresentiil'a"fill20 ptcent 4. of all'poorfarnilies. .-" '~vlr~ thar~half- of the childhood population. (56/o) whose head of hoiA,ehold- ~is' an, , -ncia,ibihial'worker are -belqw the poverty threshold. Poverty amriong th,eadolescd.nt *5op ;a' '. ', p too18 years of'age, is worsenialg, decreasing from 21 percent i- 1,4to 24 W {s, , < +, ' < < ,',,ii i ; §e ; i' ,'06 V ; , ' A , 2.37 The education attaimnent of 13-18 year olds in the SIPO database is very low. Only 21 percent of youngsters in extreme poor families have completed at least one year of the secondary education cycle, as compared with 39 percent for the near-poor. Most of them are still attending the primary education cycle in an age when most of them should be in secondary education. About one-third of youngsters in extreme poverty are out of school and are either working (mostly in temporal jobs) or are inactive. Females have achieved generally higher education levels and stay longer in school, but most of them stay inactive not studying nor working, especially in extreme poor and poor families. When they are in the labor market, their open unemployment rates are very high (over 20 percent) especially for females in extreme poverty. 2.38 Over 22 percent of extreme poor--youth girls 13-18 years old--stay at home as inactive, as compared to about 15 percent in near-poor families, in SIPO. The proportion of youths out of school is even higher for youths 15-18 years old. About 47 percent of these youths are out of school for all three poverty groups as a whole, with some minor differences in the proportions of the different groups (Table 21). Of the youths out of school, the majority of males and females in extreme poverty stay inactive (23 percent for males, 28 percent for females). In contrast, about 35 percent of near-poor males work as compared to 18 percent of extreme poor males. Unemployment rates faced by these youths are very high (over 20 percent), especially for females in extreme poverty. 25 2.39 Only about 37 percent of all youths 14-18 years old in SIPO have completed at least one year of secondary education or higher. The situation is even worse for youth males in extreme poverty of whom only 29 percent have completed at least one year of secondary education or higher. Table 21: Characteristics of Youth 14-18 Years .;,- 't . Extreme_ Poor, - P,o7r ., ,- Near.poor - Total -"-f . C~ ' .~-:. N-':i- 1b1ales' Females N1ale Females Niales Femailes, -Toai Number ofpeople 15-18 20,481 21,613 15,352 14,863 5,955 5,775 84,039 FPeads 0.9 10.8 0.9 1.6 2.8 2.3 3.8 Level of education of youth Primary 66.9 63.8 60.8 54.0 55.9 48.9 60.7 Secondary 29.0 33.2 36.2 44.0 41.1 48.5 36.3 Higher 0.2 0.2 0.2 0.2 0.3 0.3 0.2 Activity of youth Permanent 4.3 1.0 8.1 2.9 15.3 5.5 4.8 Temporary 13.5 3.5 18.8 4.6 19.3 5.0 10.1 Unemployed' 20.8 31.0 9.2 13.7 5.7 7.5 14.7 House wife 0.3 15.4 0.2 13.0 0.2 20.2 7.8 Student 54.2 50.7 53.7 57.3 50.7 53.3 53.4 Other inactive 22.9 27.5 16.5 20.9 12.4 15.1 21.3 Source: Author's calculations from SIPO's database. '. This is the open unemployment rate. 3. Youths 19-24 Years Old 2.40 There are important differences by gender in this age group in at least four aspects (See table 22). First, females are over-represented in the youth in extreme poverty. A large number of females in extreme poverty are household heads (41 percent), compared to only 18 percent of males that are in extreme poverty. Second, the proportion of youth male heads increases as families move out of poverty, while the proportion of women heads decreases (from 41 percent for extreme poor to 17.5 percent for the near-poor). Table 22. Characteristics of Youth 19-24 Years vCharacterlstic-i .'-,:,'. -- '. -,.' ' Et"reme Poor; -, p -rr :-. -. :Nea-woor Total- ' - , *S,. 'Mw- ' 1stales. -Females llNales -Females Miaks ales: e s Total t . . .| . L - ,| - . I - . - - I Number of youth 19-24 9,796 14,902 12,909 12,194 10,294 8,160 68,255 Heads 18.4 40.8 22.3 13.7 27.0 17.5 24.4 Level of Education Primary 63.9 62.6 61.8 56.6 57.5 46.9 58.9 Secondary 25.8 29.4 29.8 35.4 34.4 42.1 32.3 Higher 3.5 3.0 4.4 5.0 5.5 9.3 4.9 Activitv Permanent 13.6 3.4 35.8 15.7 58.5 36.2 25.4 Temporary 35.1 10.0 36.1 11.7 25.7 10.1 21.3 Unemployed' 21.9 27.3 7.3 7.6 3.6 2.5 10.0 House wife 0.8 47.3 0.3 41.8 0.2 32.0 21.8 Student 20.9 16.7 15.3 17.5 8.9 14.5 15.8 Other 15.9 17.5 6.8 11.1 3.5 6.0 10.6 Source: Author's calculations from SIPO's database. T.lhis is the open unemployment rate. 26 2.41 Third, the extreme poor and poor youths 19-24 years old have very low educational attainment. Only about 29 percent of males and 32 percent of females in extreme poverty have completed at least one year of secondary education or higher, as compared to the near-poor (40 percent of males and 51 percent of females). A surprising result is that the proportion of students is higher for youths in extreme poverty than for the near-poor, for both males and females. This may indicate that the extreme poor may still be completing secondary education at this age, or are engaged in other non-formal education. The very high unemployment rates for the extreme poor may be related to their low education levels. Over 8,100 extreme poor and poor youth 19-24 years of age are unemployed and have low education. 2.42 Fourth, labor force participation rates of the extreme poor are much lower than those of the near-poor for both males and females (62 percent and 18 percent vs. 87 and 47 for males and females, respectively). This may be related to the very high unemployment rates for both males and females in extreme poverty (21.9 and 27.3 percent for the extreme poor), as many of these youths may be discouraged from participating in the labor market. However, as noted earlier, a high proportion of extreme poor and poor youths are still studying (probably repeating some years of secondary education) and a high proportion are inactive, as compared to near-poor families. When they work, most of the extreme poor males and females work in temporary jobs, while most of the near-poor work in permanent jobs. 4. The Elderly: Sixty Five Years and Older 2.43 Extremely poor elderly people show rates of high malnutrition and poverty risk because of their inability to work or the lack ofjob opportunities. Costa Rica has sought to address these risks by the introduction of a non-contributory pension regime to cover the poor not covered by the formal contributory regime. In the SIPO database there are about 35,500 people older than 65 years of age, of whom about 48 percent are extremely poor, 32 percent are poor and 20 percent are near-poor (Table 23). In total, an estimated 17 percent of the people in the SIPO database are over 65 years of age, compared to around 8 percent for the general population. The following are some of the main characteristics of this population group. * First, as expected, a large fraction of this group is women and they are over-represented in the extreme poor group where a large fraction of them are family heads. * Second, around 40 percent of elderly report a poor, self-assessed health status. Over one third of the elderly in extreme poverty had a permanent limitation, many report mental health problems and another 10 percent some temporary limitation. Insurance coverage is high, however, with only about 8 percent of the extremely poor in this group report having no health insurance. 1.1 * Third, only about 50 percent of the extreme poor and poor groups receive a pension (contributory or non-contributory). The percentages covered by non-contributory pension are higher for the extreme poor group (43 percent) and poor (24 percent) than 27 for the near-poor group (12 percent). However, about 14,000 extreme poor and poor elderly people were still not covered by any pension regime in 200022. Table 23. Characteristics of People 65 and More Years Old Characterlstics .: ' Extreme Pooor Por- Non-Poor Totil Number of people 17,002 11,302 7,258 35,562 Men 44.3 49.4 49.5 47.0 Women 55.7 50.6 50.5 53.0 Female Heads 45.9 34.7 37.4 40.8 Male Heads 54.1 65.3 62.6 59.2 Health Condition Permanent limitation 35.9 30.3 25.2 31.9 Temporary limitation 10.4 10.6 9.1 10.2 Good health condition 53.7 59.2 65.6 57.9 Health Insurance Insured 92.2 94.0 94.5 93.2 Not insured 7.8 6.0 5.5 6.8 Activitv 100.0 100.0 100.0 100.0 Permanent Job 2.0 3.6 6.9 3.5 Temporary Job 7.0 9.0 7.2 7.6 Unemployed' 14.5 6.2 3.6 9.0 House wife 24.1 25.2 22.5 24.1 Pensioner 50.3 50.9 52.8 51.0 - Contributory regime 6.7 25.7 40.0 19.5 - Non contributory regime 42.9 24.3 11.5 30.6 -Other 0.7 0.9 1.3 0.9 Other activity 15.0 10.4 10.1 12.6 Source: Author's calculations from SIPO's database. '.This is open unemployment calculated as the number of unemployed divided by the active population in group. 2.44 The main service gaps of the poor and vulnerable are as follows (Table 24). First, about 84,000 poor children who were not covered by public Early Child Development (ECD) programs, such as CEN-CINAI centers, Hogares Comunitarios or other programs. Most of these children are highly vulnerable because they stayed at home under the care of poorly educated mothers, older siblings, or relatives who may lack the skills to stimulate little children and help them develop the motor and social skills they need. Second, about 33,00 children 13-18 years of age who were out of the school system and were, thus, unable to complete higher levels of education to take advantage of the modermizing economy, leaving them highly vulnerable to high unemployment and low productivity and incomes. Third, about 7,000 elderly poor who were not covered by any assistance or contributory pension leaving them vulnerable to the effects of very low incomes. In addition a great number of poor families still lack adequate health care and suffer from highly inadequate housing conditions and lack of basic services. 22 Coverage may have improved recently with the increased targeting of existing pensions and the increase in the number of pensions awarded by the government in 2000-2001, which may not be reflected in the database that includes families surveyed from 1999-until early 2001. 28 Table 24: Summary of Critical GAPS in Social Services of the Poor and Vulnerable Groups In Costa Rica, 2000-01 (Figures from SIPO, Number of People) PopiNfon-Groups aiddSeirvii ., GAP, :.'Progra.. A. Education and training Children 0-5 years of age 84,500 Not covered by ECD programs of CEN-CINAI, Hogares Comunitarios, other. Major efforts are needed to target above programs to increase coverage and increase NGO and private sector ________________ _ .cooperationf. Children 13-18 years old out of school 33,600 Programs to induce children to return to school-and improve education quality. Scholarships conditional transfers, other. Unemployed young 19-24 years old 8,100 Youth training programs-financed by public funds and provided by private enterprises to ensure relevance of training. B .Health insurance gap -Men and women heads 28,700 CCSS with state funds. CCSS decentralization of care and imp rove information. -Elderly (65 years of age and older) 2,000 CCSS with state funds. CCSS decentralization of care and improve infonmation. C. Pension gaps Elderly (65 years of age and older) 7,000 Non-contributory pensions of the CCSS financed below poverty line by FODESAF. D. Housing and basic servicesgap Housing quality (number of houses) Bad/a floor 107,000 Housing vouchers for housing or housing Bad roof: 98,000 improvements financed by FODESAF. Previous Bad wall: 101,000 vouchers have not been targeted to poorest Only about 25 percent of subsidies have gone to poorest 25 percent of population in the 1990-99 period. Basic services (potable water) 25,000 (15% of Subsidized connections to basic services (number of houses) houses) have no in- provided by water companies-national, door water municipal, other. connections. /a: Bad refers to the Rresence of substandard materials in floors. roofs and walls adversely affecting the living conditions of ro_r families. Souree: Author's calculations based on SIPO data base. 2.45 Summary poverty profMle using SIPO. In summary, the main characteristics of the income poor and vulnerable groups (many of which are similar to those of comparable variables from household surveys, as seen in Annex 1), are as follows: * Children 0-5 years of age are a highly vulnerable population group in Costa Rica. About 84,000 of them (95 percent of the poor group) are extremely poor or poor and are not attending CEN-CINAI or Hogares Comunitarios centers. Most children stay at home under the care of mothers, older siblings, or relatives. Over 80 percent of mothers of these children have only primary education and may be unaware or unable to provide adequate stimulation and the motor and social skills their young children need to start breaking the poverty cycle. (See chapter 3 for strategies and cost estimates to increase coverage on ECD education for these children). * Extreme poor and poor families show very low education attainment of all family members. About 80 percent of parents have only attained primary education (some and complete). A large number of children 13-18 years of 29 age are out of the school system, although the proportion out school is similar for the extreme poor, the poor and near-poor in SIPO (about 30 percent). The situation is far worse for children 14-18 years of age, of whom about 47 percent are out of school. Of those not attending school, about half of them work while the rest stays inactive. A larger proportion of extreme poor females remain inactive in comparison with males. It has been estimated that about 33,600 youth 13-18 years of age are out of the school system. These students should be induced to return to school or to receive special training to upgrade their skills for the modernizing Costa Rican economy. Extreme poor and poor families show lower participation rates in the labor market of all family members than the near-poor. Extreme poor females participate the least in the labor market, as compared with the near-poor. The lower participation of children over 12 years of age for the extreme poor, as compared to the poor and near-poor, is somewhat surprising because it is commonly believed that poor children would leave school early to work and help support their families. Apparently, this is not the case in Costa Rica. In fact, as indicated by the profiles of the youth 14-18 and 19-25 years old, extremely poor youth participate much less than the poor and near-poor, and show a higher proportion who are studying. Extreme poor and poor families are larger, have more children under 12 years of age, and have fewer children over 19 years of age than near-poor families. Thus, dependency rates are higher for extreme poor and poor families than for the near-poor because of the higher number of dependants (children) and lower number of workers. One of the most striking facts of extreme poor families is the extremely high unemployment rate for all family members. Unemployment rates of extreme poor heads and spouses are 9-10 percent, while those of children are over 20 percent. The unemployment rates of youth 19-24 years old in extreme poverty are especially high (21.9 males and 27.3 for females). It has been estimated that about 8,100 extreme poor and poor youth are unemployed and low-skilled, and are good candidates for youth training programs. Women head 48 percent of the families in extreme poverty, compared to 33 percent for near-poor families. Famnilies headed by women are, however, smaller than those headed by men. Women heads participate more in the labor market than other women (spouses) but participate much less than do male heads. Most female heads work in temporary jobs, rather than in permanent jobs that offer job security and health insurance, among other benefits. The elderly over 65 years of age in extreme poverty are also a highly vulnerable group because of their ill health and low incomes. About 46 percent of extreme poor people older than 65 years of age reported perrnanent or temporary physical or mental limitation. About 14,000 elderly in extreme poverty and poor (about 50 percent of group) are not covered by 30 contributory or assistance pensions. However, recent govermment figures for 2000 and 2001 indicate that the number of non-contributory pensions awarded to the poor has increased considerably and that the number of uncovered elderly may be about 7,000, by the end of 2001. Most, if not all, of these people can be covered by re-allocation of current pensions from non-deserving to the deserving poor, as was done in 2000-2001. Most of the people in SIPO database (94.8 percent) are Costa Ricans, about 4.9 percent are Nicaraguans, and the rest are from other countries. The small proportion of other nationals may reflect under reporting, if they are fearful of being interviewed for SIPO, if they believe that SIPO is only for Costa Ricans or other reasons. In general, within each group, the characteristics of Costa Ricans appear to be somewhat worse than those of other nationalities in the proportion of women who are household heads and unemployment rates. Although ownership of housing is relatively high, the quality of housing is far from adequate. Regarding quality of housing, about 107,000 (65 percent) of the houses of the extreme poor and poor have substandard floors, 98,000 (59 percent) substandard roofs, and 101,000 (61 percent) walls. Also, about 25,000 (15 percent of extreme poor and poor families) do not have in-door water connections. Their water sources are wells, rivers, and rain, among others. Only about 9 percent of the poor reported having received a housing voucher (for housing or housing improvement) from the government. 31 3. SOCIAL SPENDING: SECTORAL ISSUES AND OPTIONS23 3.1 Social spending is a key instrument for improving social indicators of the poor and vulnerable groups. Social spending in Costa Rica is now among the highest of any middle income country, as a share of public spending and as a share of GDP. Over the past decade, efforts to promote universal programs in health, education, social assistance and pensions have obtained impressive results in terms of coverage and outcomes and now rival many OECD countries. These efforts have come at a significant cost and over the past five years results appear to have slowed, particularly in education. This chapter identifies ways in which social spending can be made more effective for improving social indicators of the poor and vulnerable groups identified in previous chapter. 3.2 Over the last ten years, public spending grew seventy percent in real terms and the social sectors increased their relative participation from 59 percent of public spending to 63 percent, with spending in education and pensions growing more rapidly than spending in health and social assistance. Given the country's fiscal situation for the next several years, there is little room for continued increases in social spending to address the service gaps and coverage needs. Major efforts will be required to improve efficiency, effectiveness and targeting of the current social spending to achieve improvements in social indicators of the poor and vulnerable groups in the next few years. 3.3 While the decade of the 90's focused on developing programs that aimed at extending the coverage of education and health and increasing spending in all aspects of social spending, the 21st Century will have to focus on improving value for money-or getting better results with the same amount of public resources. These efforts will have to consider governance issues as related to the social sectors in general, institutional issues related to the design and management of programs, and finally targeting of social spending. Efforts focused individually on any of these issues are unlikely to maximize value for money. A multi-dimensional approach will clearly be needed. 3.4 Recent reform efforts have made considerable progress in improving access and efficiency in the social sectors and in maintaining, or increasing, expenditure levels. The relative performance among sectors and the results on different income groups is varied and demonstrates the importance of maintaining clear and targeted policies. The previous section profiling the poor using household surveys and the SIPO database show that important gaps remain between the rich and poor and improvements in social spending will be required to better target programs to the needs of the poorest citizens. The following is a summary review of major reforms, the results accomplished so far, and the major issues remaining for the future. Volume 2 presents the sector analyses and recommendations in greater detail. 23 This Chapter is a summary of Volume 11. 32 Education 3.5 Costa Rica has made major strides in the education sector, including nearly universal primary coverage and low illiteracy, and has gone further than most countries in introducing modem technology, mainly at the primary level. In fact, in the early part of the 90's important advances were made in reducing inequality for the poorest quintiles, however, in many cases the results have stagnated or grown more slowly than improvements for the better-off. Overall, progress has not been maintained in recent years, and there are strong indications of slippage, as reflected in high repetition and dropout rates, especially in the first grade of primary (16.0 and 6.6 percent) and in the seventh grade (16.4 percent and 17.1 percent) in 1999, the slow progression from sixth to seventh grade, and low learning achievement outcomes as measured by standardized annual testing. Table 25 displays the key indicators regarding coverage for primary and secondary education. Table 25: Evolution of Key Education Indicators, 1990-99 indicator s . flnisnlshprlmrdu- ..tiona % fini sh secondary . : .- . . -1990 . 1995. . 990.19 19 995 1990 .--11995. 1999: Total Country 77.4 79.0 76.4 37.3 44.2 44.1 26.6 30.4 32.7 Income levei4 Cuartil 1 62.3 73.4 62.1 22.3 34.1 23.5 11.6 13.7 15.0 Cuartil 2 78.7 76.3 79.9 27.6 31.6 37.2 10.7 16.1 18.4 Cuartil 3 89.0 80.3 81.4 42.9 44.1 41.8 27.2 31.6 28.3 Cuartil 4 90.6 93.1 92.4 66.7 74.8 68.9 49.6 42.2 59.3 Region Metropolitan Area 88.3 85.7 78.2 59.5 54.8 59.4 42.4 41.8 48.6 Urban Central Area 89.8 85.7 87.8 55.8 63.9 69.6 37.1 43.8 48.5 Rural Central Region 81.6 78.2 80.9 24.8 31.4 36.0 21.3 27.6 27.8 Chorotega Region 68.8 79.7 70.3 28.1 43.1 33.2 20.3 30.0 27.6 Central Pacific 71.8 74.1 71.8 30.3 35.9 26.2 15.3 - 20.1 14.7 Brunca Region 64.9 72.3 64.5 22.7 36.8 26.5 17.7 12.1 19.8 Atlantic Region 59.6 65.2 65.0 17.6 33.7 28.2 10.4 15.0 7.0 North Region 58.0 74.6 66.8 9.5 32.9 37.4 . 14.7 13.7 10.5 Zone Urban 86.9 84.6 81.6 55.0 56.1 57.9 37.4 39.9 43.7 Rural 69.1 73.6 69.9 19.6 31.1 29.0 16.2 20.4 20.2 Sex Male 76.8 75.0 75.4 38.2 42.1 38.3 22.9 25.5 28.0 Female 78.2 83.6 76.5 36.3 46.2 49.4 30.0 35.6 36.1 1/ Percentage of the population of 14 years with 6 or more education years approved 2/ Percentage of the population of 18 years with 9 or more educaton years approved 3/ Percentage of the population of 20 years with 11 or more education years approved 4/ According to per capita domestic income. Cuartil 1 includes the 25% of the families with the lowest per capita incomes Source: JD Trejos, based on the Household Survey of the Statistics National Institute 33 3.6 Spending per student over the decade has increased substantially for all education levels but the outcomes are mixed. While the distribution of resources is progressive for primary education, at the secondary and university level a disproportionate share of resources is allocated to the wealthy. Distributional analysis indicates that 70 percent of the resources destined to primary education go to meet the needs of the poorest 50 percent of the population; at the university level, however, 70 percent of the resources are allocated to the wealthiest 30 percent of the population. 3.7 The distributional problems are exacerbated by problems with value for money, as shown in the following graph, which compares education expenditures with the net enrollment rates for primary and secondary education. The figures compare the relative performance of Costa Rica with the average for all medium income countries (See Figure 3). The metric is divided into four segments: those countries with low spending and poor outcome; low spending and high outcomes; high spending and low outcomes; and high spending and high outcomes. Using this measure, it is clear how Costa Rica obtains limited value for money in secondary education. The figures show that in primary education, Costa Rica ranks above the average for middle-income spending but also spends more than the average. In secondary education, the outcomes are less favorable. Spending continues to outpace the middle-income average but outcomes, in terms of net enrollment rates, are considerably below the average. Significant changes are required to shift to a high outcomes, average spending environment, which would reflect value for money. Figure 3. Net Enrollment Ratios and Education Expenditure i i CN AL MY KR ~'PH MX I 9 ~ ~ j T h Y E C o H U M D T NOU a. NAt ^RO PO. MD 'iLTLV 0r UY R' EG CR 0 R EG C X * _ ,s BW M ~se a,KR - 81 jMC |; CL c EP ALG E .i * - ~~ ~~~~~~ ~~~~~E TRc 0MX MD 84 CT ' / SA Z 0r BW CR 'NA LB liltl11111111 1111MC I 0%V VE BR 2.0 2.5 3.3 3.7 4.1 4.7 5.1 5.4 6.1 7.3 2.0 3.1 3.7 4.2 4.8 5.2 5.5 7.3 Education Expenditure as a % of GNP Education Expenditure as a % of GNP 3.8 The differential between rich and poor has not been significantly reduced over the decade. While the educational achievement of the poor improved, the gap between the richest 25 percent and the poorest 25 percent did not close. At the primary level, the percentage of children finishing primary school remained constant at 62 percent for the poorest quartile, while the richest 25 percent increased to over 92 percent. Secondary level attainment improved slightly' increasing over the decade from 11 to 15 percent for the poorest quartile but the richest quartile increased from 49 percent to 59 percent. Finally, rural students are only half as likely to finish secondary school. 34 3.9 At the same time, Costa Rica's economy and future sources of growth are changing dramatically with the advent of high-tech foreign investment (Intel and associated industries). This presents an opportunity and a challenge for Costa Rica. It needs not only to rapidly expand access, but at the same time reorient its education system to take advantage of the opportunities brought about by this sudden infusion of high-tech investment-and this relates not only to secondary, but to higher education and the education of adults which are working in jobs that are changing, or will change, with new technology. With a whole new set of service industries at its doorstep, Costa Rica needs to further develop its human capital in order to maximize opportunities. 3.10 Improving value for money in education will require a multi-planked strategy. The education policies that the Ministry of Education should consider in order to improve the efficiency, quality and equity of education are: (i) increasing the coverage and improving the quality of early childhood development programs, especially of poor children; (ii) strengthening the quality of basic education (grades 1 to 6); (iii) strengthening the quality and increasing the coverage of secondary education (grades 7-11/12), especially upper secondary for rural people; and (iv) improving equity through better targeting of subsidies such as the school-based feeding program and scholarships. 1. Increasing coverage of pre-primary education 3.11 Early childhood education has proven worldwide to be a cost-effective intervention to improve children's performance in basic education in particular and in society at large. The current pre-primary coverage for poor children under 5.5 years in Costa Rica at 8.6 percent (2001) is very low for the country's development and social spending levels. In addition to the CEN-CINAI and the Hogares Comunitarios programs described below as part of the social protection network, the Ministry of Education is responsible for pre- primary education covering 5.5 to 6.5 year olds. The program covers about 70,000 children (82 percent coverage based on figures from Ministry of Education). The Ministry of Education should aim to: (i) increase pre-primary coverage primarily in rural areas; (ii) improve quality by introducing a pedagogical component; and (iii) increase parental involvement. 2. Improving quality in basic education 3.12 The Ministry of Education urgently needs to establish the reduction in repetition and drop-out rates as a priority in basic education (grades 1-6), especially in first and sixth grades, increasing learning achievements and introducing informatics. The data indicate that completion rates among primary school graduates increased from 77 percent in 1990 to only 79 percent in 1999. But this improvement hides the acute differences between. income levels. While 92 percent of all children in the highest income groups completed primary school, only 62 percent in the lowest income groups finished primary school. To improve completion rates a pilot Program of Excellence was implemented in 100 urban and rural schools showing alarmingly high repetition and dropout rates, and low leaming scores in 2000. The Program of Excellence includes: (i) strengthening the teaching and leaming of reading, writing and math skills in first and sixth grades; (ii) making a more effective use of leaming assessment findings; (iii) improving in-service teacher training; (iv) establishing informnatics laboratories in 45 of these schools; and (v) strengthening community and parental participation in school-related matters. The Ministry of Education needs to expand the interventions in the Program of Excellence once they have been 35 evaluated and shown good results. Further, improved targeting to groups with particularly low outcomes would make more effective use of limited funds. 3. Increasing coverage and improving the quality of secondary education 3.13 Costa Rica's progress in Table 26: Comparative Secondary achieving universal primary Countries Secondary education coverage is not matched in Rates lower (grades 7-9) and upper Costa Rica 28.3 5,770 secondary (10- 1/12). Net France 85.0 21,897 secondary (10-1 1/12).NetHungary 91.9 10,479 enrollment rates for secondary Japan 94.9 24,041 education (13-19 years old) are only Spain 73.0 16,730 about 61 percent (below the mean for United States 78.2 30,600 Argentina 41.5 11,324 LAC) and far lower than OECD or Chilel/ 47.4 8,370 Asian economies.24 Moreover, table Colombia 43.2 5,709 26 shows that Costa Rica ranks Panama 39.1 5,016 considerably behind other LAC Urupuay 39.4 8,280 countries in terms of secondary 1/ datos 1998 completion rates-Chile (47% and Fuente: Education at a Glance 2001, OECD and Colombia (43%)--and achieves only Panorama Social de America Latina 2000-2001, CEPAL one-third of the performance of leading countries like Hungary (92%) and Japan (94%). This reflects, to a large extent, high dropout and repetition rates, especially in grade 7 and extremely low transition rates from basic to post-basic education (from grade 9 to 10). Enrollment is especially problematic for the poor. For adolescents between 16 and 18 years, nearly 50 percent are no longer in school and are not working. 3.14 If improvements in value for money are to be obtained, structural changes will have to be made in the following directions: (i) revise existing curriculum; (ii) improve teacher training; (iii) increase secondary education supply through a menu of options, including traditional schools, tele-secundarias, distance learning, open-access education, virtual schools and vocational training; (iii) promoting the collaboration of the private sector to strengthen the relevance of secondary education; and (iv) helping parents and students, especially of rural areas, to finance part of opportunity costs of studying through scholarships, conditional transfers, and other supplemental financing schemes. 4. Improving targeting and equity 3.15 The government needs to improve the targeting and equity of education, in particular of programs directed at poor students which are run by the Ministry of Education, such as the school feeding, transportation and scholarships. The school lunch program is not based on targeting and there has been no evaluation of its impact." About one-third of users of the transport subsidy do not need it, and other students are not covered. Other programs, such as the scholarship and school bond programs, appear to be better targeted to the poor in recent years when the SIPO system started to be used to select beneficiaries (GOCR, 2000a). To better improve targeting, most, if not all, education 24 The figure for Costa Rica is from the 2000 population Census. 25 The government has recently requested the UNDP Office in Costa Rica to undertake a thorough review of all welfare programs run by the Ministry of Education. This review will include revision of administrative matters, targeting and costs, among others (UNDP, 2000). 36 welfare programs (including the school lunch program), should consider using SIPO for selecting their beneficiaries. 5. Cost implications of the reforms and priorities suggested in Education 3.16 It is important to note that the interventions mentioned above are already included in the plans of the Education Ministry; for instance, the 2001 figures are part of the initial budget proposals of the Ministry of Education. In total, it is estimated that the cost increases of the reforms proposed would sum to between 6 and 7 percent of current expenditures. These increases could be financed by two main sources: (a) improving internal efficiency, through reduction in repetition rates, for example; and (b) through the constitutional imperative to increase the-Ministry of Education's budget to 6 percent of GDP. The Ministry of Education budget projections include a roughly 11 percent increase in dollar terms for the budget from the year 2000 to 2001, and the same rate of increase for 2001 to 2002. Careful attention should to be paid to balance the increasing investments with potential benefits. Improvements in efficiency and quality are likely to free-up resources currently tied up to finance inefficiency, such as high repetition rates, and poorly targeted programs, which distribute resources to populations that are less needy. As an example, crude estimates indicate that repetition in primary school costs about $21.2 million per year or about 10 percent of the education budget for primary education, while the costs for secondary education repetition are equivalent to $14.8 million per year. With a reasonable target of achieving a 50 percent reduction in repetition, total savings would amount to nearly $20 million, or ten percent of the annual MOE's budget. 3.17 Reducing repetition would allow the government to redirect spending to the new priorities and further improvements in coverage. According to priorities listed, more spending needs to be allocated to pre-primary education to increase coverage and improve quality-providing relatively more resources to school materials rather than salaries, and improving secondary education. Costa Rica's education spending is heavily biased towards salaries (93 percent), as compared with Chile (67 percent), Brazil (82 percent) and OECD countries (80 percent). The structural inefficiency in this area may have to be resolved by increasing use of the private sector. Efficiency gains could be obtained by establishing contracts with private schools, or managers that would be able to reduce the overall percentage of funds allocated to salaries and improve quality. 3.18 But improving the "supply" side of education by spending more in pre-primary, primary and secondary education may not be enough to reduce repetition and drop-out rates and increase the education level of the population. Much more will need to be done to improve the "demand" for education, especially by the poor in rural areas. The demand for education may be low because parental education is low, and working opportunities are increasing. Thus, parents and youngsters need to be convinced that education is a good investment, leading them to spend more in education (opportunity costs plus direct costs) relative to alternative investments, and this has to be done by offering higher quality education and providing information on the returns to schooling.26 Also, the government may have to provide direct incentives for poor students and their families, so that they are able to forego their opportunity costs and provide for direct costs of education. This may include conditional cash transfers-similar to the education scholarship program 26 The lack of information on the returns to schooling may be important for parents and children with low education. Regression analysis indicates that the probability of repetition and drop-outs is much lower for children of more educated parents (Hernandez, 2000b). 37 introduced by the current government, perhaps increasing the amount and coverage of such a program. 3.19 In summary, the main challenges for the education sector are: a) increasing coverage of pre-primary education to reach 100 percent of all school aged children; b) improving quality of primary education-and reducing repetition and drop-out: and c) improving quality and coverage of secondary education-reducing repetition and drop- outs, which are costly for society and the Costa Rican economy. Inproving quality and coverage of secondary education will require investments in teacher training, technology and modem methods to reach dispersed populations, as well as measures to improve the demand for education, including direct subsidies or scholarships to families so that they can forego the opportunity and direct costs of schooling, especially in rural areas. Net enrollment rates in secondary education are significantly lower in Costa Rica than in countries with similar per-capita GNP and expenditures in education. Completion of secondary education appears to be a key threshold to break the poverty cycle in Costa Rica and other Latin American countries (Castafieda-Aldaz-Moran, 2001). Health 3.20 By the year 2000, Costa Rica has obtained substantial progress in improving the health of the population. Costa Rica consistently ranks in the top 40 countries in the world in terms of life expectancy (76.7 years), infant (10.3 per 1,000 in 2001) and maternal mortality and health systems performance (36 by WHO). Table 27 shows a comparison among countries of key indicators. Table 27: Comparative Health Status Indicators 1nC ;3 . -Jl 'zar, 'TtI* Male -"4s Costa Rica 2001 10.3 14.8 76.7 74.2 79.3 Mexico 1997 31 38 72.6 69.6 75.6 Chile 1997 11 13 75.4 72.4 78.4 Argenbina 1997 22 24 73.3 69.7 76.9 Brazil 1997 34 44 67.2 63.5 71.3 Jamaica 1997 12 14 74.7 72.5 76.9 United States of 1997 7 10 76.8 73.5 80.2 America United Kingdom 1997 6 7 77.5 75.0 80.0 Source: Costa Rica: Indicadores Basicos 1999, MS The World Bank: World Development Report 1999/2000 Pan American Health Organization. Health StaUstics from the Americas, 1998 edition. 3.21 Although Costa Rica's health indicators are among the best in the region, and coverage of health services is nearly universal, the health sector faces important challenges to contain cost, improve efficiency of resource use and preserve financial solvency in the future. The health sector is starting to feel increasing strain as Costa Rica's epidemiological and demographic profile changes and the opportunities to use the private sector to generate efficiency, quality and competition start to expand. Over the past decade, the Government's objective has been to maintain spending, currently at about 5.5 percent of GDP, while allocating resources more efficiently to improve quality of service, equity and efficiency of service delivery. Costa Rica has high quality outcomes with 38 regard to the expenditures. The following figures compare the expenditure with the outcomes obtained, using life expectancy and infant mortality as proxies for health status. Figure 4. Health Expenditure and Health Outcomes GOl 80.0 e A00G A? C' ~cDM GA 70.0 9 EE 70.04 @2~ ~ ' au.0 PO ZA NA« 0 V So.oB V A GY KI Ul ~ ~ ~ GYEG PG Bo a %oGHal Ent 60.0 T O N ~ AZA ",PVCR KR CL H" ~SKCzt CT 0.0 2.5 5.0 7.5 0.0 1.5 3.0 4.5 6.0 7.5 9.0 10.5 12.0 Healt Expenditure as a % of GNP Health Expenditure as a % of GNP 3.22 As opposed to education where Costa Rica ranked considerably behind its peers regarding secondary enrollment in terms of value for money, in the health sector, Costa Rica has actually decreased health spending relative to GDP, lowering spending from over 6 percent of GDP in 1996 to 5.3 percent in 1999. In comparison with other middle-income countries, Costa Rica consistently spends more than the average but obtains considerably better results than the other middle-income countries. The pioblems are related to how to improve these results with fewer resources through improved access and quality of care, how to increase equity in the distribution of resources and how to create competition in the system to reduce the crowding out of the private sector to free up resources for use in other social programs. To address many of these issues, Costa Rica initiated a reform process in the mid-1990s. While the reform process has yielded many early returns, there is an unfinished agenda regarding the consolidation of these reforms and the adjustment of regulatory, purchaser and provider organizations to adapt to their new role in the reformed system. The following paragraphs highlight the key reforms promulgated over the past decade and outline the challenges ahead. 3.23 Under the reform process started in the mid- 1 990s, the Ministry of Health completed the shift from a regulatory, financier and provider of services to become a regulatory and policy formulating agency (Ministerio rector). This process is still underway and will require the strengthening of the MOH and a restructuring of its functions and interrelations with other sector organizations. During the last decade all remaining health care provision and financing became the sole responsibility of the social security agency (CCSS). Changes in the marketplace, such as the increasing importance of the private sector, underscore the need to improve the regulatory function of the Ministry. In addition, CCSS began a decentralization program to increase accountability, improve performance in service delivery, encourage non-government provision of services, and 39 increase the cost efficiency of pharmaceutical procurement. The reforms are well- advanced but need to be consolidated and deepened. 3.24 The reform process is bringing about important institutional and health care delivery changes, especially through the creation of health areas, the reorganization of the primary health care model with the introduction of the EBAIS, and the adoption of new management schemes. The primary health care system was reorganized into more than 90 health areas, each one with responsibility for the health of a population averaging 45, 000 people. Within the health areas, smaller teams of physicians, nurses and technicians work closely with the population to apply PHC care. Through these changes the system initiated a transformation from a supply-based model to a population-based model. 3.25 As a complement to the changes in the health care delivery network, over 120 management contracts for health areas and major hospitals (Compromisos de Gesti6n) have been successfully signed. These contracts encourage an important transformation among hospitals and the medical community regarding management and resource use. Management contracts are beginning to generate innovation in hospital management; budgets are starting to be prepared on the basis of health outcomes, and inputs used to generate results. Over the past several years, average length of stay has declined from 6.5 to 5.4 days and hospital productivity has increased substantially. In addition, reductions in hospital infections and process reengineering have led to annual savings in excess of $10 million per year. More progress can be made in these areas. 3.26 Waiting lists continue to remain a problem in the public sector. With average waiting times over 18 months for several services, the pressure of a free health system is increasingly evident. Improvements in monitoring of processes and evaluation of results are becoming an integral part of the health care delivery system. While this represents welcomed progress, much more needs to be done to consolidate and extend these reforms. The CCSS would need to consolidate a budget allocation system based entirely on performance outcomes, and design incentive packages for health workers that can encourage productivity and efficiency. 3.27 The decentralization process, initiated with the introduction of management contracts is greatly enhancing the institutional autonomy of service providers. The recently approved decentralization law (ley de desconcentraci6n) takes the process further by introducing stronger incentives for institutional autonomy. By the beginning of 2001, 12 hospitals functioned on an autonomous basis and a formal plan has been established to gradually expand the decentralization to the remaining hospitals. Throughout this process, there will be great challenges related to the technical, managerial and institutional aspects required to manage such a change and to adapt its own organizational structure in order to support the decentralization process. Building upon what has already been accomplished and tailoring the instruments to the changes introduced by the new legislation, it is nrecessary to improve management in the hospitals, to strengthen the management contracts so that they become more outcome based, and to develop and to approve a consistent evaluation framework of those contracts. The main areas that will require continued efforts to improve value for money include: 40 1. Improving sector organization: Strengthening the stewardship role of the MOH and institutional changes in the CCSS 3.28 Despite continued efforts to restructure the Ministry of Health, the role of regulation or stewardship is weakly defined. One of the key roles of the Ministry of Health as a policy making and regulatory body is to establish the rules that lead the rest of the system to achieve optimal performance. After having devolved all provider, and most financing, functions to the CCSS, the MOH has yet to establish a clear role in the new system. In this context, the MOHI has an important role in the monitoring and follow up of the sector programs, as well as carrying out economic evaluation studies of health programs. These types of studies would encourage better expenditures usage and would increase efficacy, efficiency and effectiveness in the sector's institutions. 3.29 The MOH's policymnaking capacity is weak due to the lack of definition regarding the scope of responsibility, the need for strengthening hum'an resources in the areas of health economics, public health and management. Traditional areas that fall under the purview of the MOH include accreditation of providers, licensing and certification of personnel, quality control of pharmaceuticals and food products, epidemiological surveillance, technology assessment, and the development of health policy. Most of these areas are undeveloped in the MOH, with the exception of a strong program of epidemiological surveillance, and adequate staffing does not exist. In addition, the MOH will have to develop capacity to carry out ex ante and ex post evaluation of social programns using modem tools such as cost-effectiveness and cost-benefit analysis. The ongoing changes in the epidemiological profile will require renewed support to health education programs to initiate changes in lifestyle and behavior among the Costa Rican population. Strong programs to reduce risk factors, such as alcohol consumption, smoking and unhealthy diets, complemented with improved control of chronic diseases, including hypertension and diabetes, would have an important impact on the future incidence of the higher-cost chronic diseases that characterize the new epidemiological profile. In addition, efforts by MOH to strengthen health education for communicable diseases among the open population need to be consolidated and extended to the CCSS. 3.30 One of the major accomplishments of the health reform program has been to require CCSS to manage risks differently, through the separation between the health and pension regimes. Also, within the health insurance branch the separation between financing and delivery of health services (primary services and hospitals) within the CCSS has been a key element to generate an intemnal market among CCSS hospitals to drive efficiency and quality improvements. CCSS's institutional structure, however, is not yet fully consistent with this formal separation. For instance, the Financing Division of the CCSS (health regime) should be in charge of all financing aspects, such as the collection of contributions, invoicing and evasion control, however, the Administrative Division retains control over key functions. Another division should be in charge of the purchasing of services from providers (health areas and hospitals). At present, most processes are managed by distinct divisions thereby increasing delays and confusion and minimizing the Possibility of improving accountability. This lack of internal organizational clarity is likely to worsen as a result of the Deconcentration Law, which will require some decentralization of CCSS functions. Clearly, a centralized organizational structure will not be able to efficiently perform its functions under the new rules. Organizational restructuring is an imperative to increase efficiency and clarify responsibilities at the different levels. If these changes were not implemented efficiently, the expected benefits 4 1 from the reforms already accomplished and those that would arise from increased autonomy at the service provision level could be seriously compromised. 2. Improving the efficiency, equity and coverage of health care delivery 3.31 Despite the considerable effort to reorient health service provision away from costly inpatient care and more toward disease prevention and ambulatory care, there is still a need to develop, approve and implement, specific strategies to replace inpatient with outpatient care where appropriate. In 1999, fewer than 5 percent of all hospital discharges were resolved in an ambulatory setting. With minor investments in training, equipment and irifrastructure (remodeling), the CCSS could increase ambulatory interventions to nearly 20 percent of all discharges. Benefits would include cost savings of more than US$12 million per year, improved quality and greater patient satisfaction. In addition, new, cost-effective service delivery strategies-such as networks for home care and hospice care-are required to address the challenge of caring for a growing elderly population. 3.32 This is important to increase cost efficiency, patient satisfaction and adapt to the country's evolving demographic profile. The latter is rapidly changing which, in combination with population mobility, will require increased focus on elderly care. Caring for the elderly is starting to move away from the core family responsibilities of the past and will increasingly need to rely on social programs. At present, there is almost no network to provide home and hospice care for the elderly. Also, to improve efficiency and promote better quality services, users should be able to select providers, especially in specialized services. 3. Improving the efficiency and decentralization of pharmaceutical procurement 3.33 Given that pharmaceuticals represent 12 percent of CCSS health expenditures (nearly 0.07 percent of GDP), and inefficiencies in the present system are apparent, an altemative procurement system could yield large savings and efficiency gains. Introducing changes in procurement that aim to reduce inventories which average 18 months for key drugs would yield savings in excess of $9 million per year based on the elimination of unnecessary warehouse space and reduced financial outlays. Further, there is considerable scope to improve the pharmaceutical procurement system so that it is more consistent with the changes introduced as part of the health reform process. The adoption of management contracts, the Decentralization Law, which provides greater autonomy in purchases to service providers, and the increasing role of public/private partnerships, will require the introduction of new mechanisms to evaluate and control purchases and to devolve responsibility to providers. 3.34 The Government is moving towards the adoption of a decentralized approach to the purchase and distribution of pharmaceuticals, based on price/quality bidding processes originating in the health care delivery facilities themselves, however, considerable work needs to be done before the gains are realized. This will require not only the design of such a system, but also the reorganization of the personnel at the central level that perform these activities. The Government will surely obtain important savings and efficiency gains from this altemative procurement system. 42 4. Improving the allocation of resources and financing of health care delivery 3.35 Although the country does not Table 28: Territorial Distribution of PHC aim to reduce govermnent spending on Spending, by Health Zone health, there is consensus among policymakers about the need to allocate Area de Salud-lowest per capita resources more equitably and efficiently spending: Prez Zeled6n (in colones) 1,202 as a way of obtaining better healthcare Area de Salud-highest per capita indicators and limiting future escalation spending: Lim6n (in colones) 60,154 in health spending. Despite nearly Average (in colones) 15,791 universal coverage (above 85 percent of Standard deviation (in colones) 10,331 the population has access), there is room for improvement in assuring that good % of Areas with expenditure per for improement inassuring hat good capita > 2 s.d. above average 28.6 quality services reach disadvantaged populations, and that government % of Areas with expenditure per resources are targeted well. Table 28 capita < 2 s.d. below average 71.4 shows that there are important disparities in the per capita allocation of financial resources per region and health area for primary health care. While the national average is estimated at $48 in 2001, there are still three health regions that lag behind: North Central, South Central and Brunca. Notoriously, in the Brunca region because of the variation in the per capita allocation among its health areas, the average reaches only $27. Meanwhile, there are three health regions, Chorotega, Central Pacific and Huetar that allocate per capita resources for primary care well above average. 3.36 Efforts to improve value for money center on changing incentives and resource allocation mechanisms so that existing resource obtain the greatest outcomes possible. The introduction of a contractual logic based on annual performance agreements has not been fully supported by the types of changes in the resource allocation model that would ensure greater equity and better targeting of resources. In the first phase of financial reforms, subjective mechanisms to increase allocation to underserved areas were introduced. The next step will be to extend the preliminary design to a formula-based plan for resource allocation, taking into consideration population size, adjusted for risk and poverty. Updating and refining the Sistema de Identificaci6n de la Poblaci6n Objetivo (SIPO) developed by IMAS, offers an opportunity to improve targeting of public subsidies; however, implementation of the system will require the formulation and introduction of regulations, as well as continuing technical support and the CCSS Board approval. 3.37 At the same time, the mechanisms for reimbursement of hospital providers have not fully made the difficult transition from a historical budget to a performance-based system, where case-mix adjusted-production and quality outcomes will determine annual hospital revenues. The first phase of reform provided the tools to make the transition in the form of a comprehensive and nationwide system of Diagnostic Related Groups (DRGs), whereby the performance of each hospital is known on a monthly basis. The next phase of reforms would have to gradually adjust individual hospital financing in terms of outcomes and production of services, using the DRG system to adjust historical budgets. 3.38 CCSS also needs to continue working on its invoice system in order to make it more efficient and to control evasion. Evasion results from sub-declaration of salaries and from non-payment of contributions, especially in small (less than 10 employees) rural enterprises, where only about 25 percent are affiliated to the CCSS, and small (with less 43 than 10 employees) industrial enterprises, where only about 36 percent are affiliated to the CCSS. It has been estimated that the cost of evasion for sub-declaration and non- contributions amounts to over 26 percent of revenues of the CCSS for health care. The recently approved Labor Protection Law will place an increased burden on CCSS to improve its collection system, which will require strengthening its invoicing system and critical improvements in the financial information system. This is important for maintaining the financial sustainability of the health plan of the CCSS over the coming years when population ages and health expenditures will most likely increase more rapidly than in the past. 5. Encouraging private sector participation and improving the regulatory framework 3.39 Changes in health delivery have been introduced mainly at the primary health care level in rural areas, but an important agenda remains for urban areas, where private sector participation can play a major role. CCSS is trying different types of public-private partnerships for health delivery, including diagnostics, treatment and surgical interventions, which seem to be proving beneficial in efficiency and quality terms. By 1999, over 6 percent of the population was served by health cooperatives and other private organizations under contracts with the CCSS. Recent studies show that these models offer important opportunities to increase efficiency, improve quality and reduce the need for public investment and expenditures. Increasing the role of the private sector, in a thoroughly regulated environment, offers a unique opportunity to leverage public resources and to obtain improved outcomes within the same resource frontier. 3.40 There is an urgent need to develop a regulatory framework for private participation both under public private partnerships and in the open private market. According to recent estimates, Costa Ricans spent about 1.9 percent of GDP in medical consultations, dental care and medicines in 1998. This high private sector health spending is likely to continue or even increase as the country continues to attract foreign investment and the CCSS is not able to keep pace with increased demand for specialized services. In fact, Costa Rica may be able to benefit from allowing creation or expansion of specialized care and health insurance services (now the exclusive monopoly of the Instituto Nacional de Seguros, INS) in its own territory taking away services which may have been provided outside the country. 6. Cost Implications of Health reforms 3.41 While most of the major reforms have already been set in motion and future efforts will focus more on consolidating the changes and changing culture in the sector, additional investments will be required to invest in primary care and to consolidate changes in other levels of the system. Estimates provided by the government during the preparation of the second health modernization project, indicate that the second phase of reforms would have an estimated cost of roughly $70 million over 5 years. These costs would cover: (a) investments in primary care to consolidate the primary care model, extending access to 100 percent of the population ($10 million); (b1) investments in personnel, training and information systems to improve the stewardship function of the MOH ($5 million); (c) investments in upgrading hospital equipment and developing ambulatory solutions ($40 million); (d) implementation of management reforms and information systems to improve quality and efficiency in the supply chain, hospital management and purchasing function ($5 million); and (e) continued support to implement nationwide the improvements in the 44 financial management system and to introduce changes related to the Ley de Protecci6n al Trabajador ($10 million). Over five years, these costs would be equivalent to less than 2 percent of the estimated annual health budget for the CCSS. 3.42 While the cost burden is easily assumed within the annual increases in the CCSS budget, the implementation of the reforms highlighted in this section offer important insights into the potential savings that would be introduced as a result of the reforms. This would continue with the spirit of the first phase reforms that produced annual savings of more than $10 million which offset many of the capital investments that were made to expand coverage. 3.43 The savings would come from changes in the following areas and produce the following expected savings. Changes in the supply chain regarding pharmaceutical purchasing and distribution could save $9 million per year through reduced inventory costs, declines in the rental of warehouse space and reductions in the business process time. Increasing the share of ambulatory surgeries and strengthening diagnostic capacity would yield annual savings estimated at over $12 million. Improving the financial management system and supporting efforts to reduce evasion could potentially reduce expenditures and increase revenues. If evasion were reduced by just 5 percent, the annual increase in revenue would exceed $7 million (assuming the cost of evasion is $144 million per year). In summary, these three reforms would produce annual savings in excess of $28 million per year. Over five years, the savings would be nearly twice the estimated cost of the reforms and allow the CCSS to continue producing more with fewer resources. While these figures are estimates, they provide guidance on how changes translate into savings thereby allowing the CCSS to continue to strive for greater value for money. 3.44 In summary, the principal challenges of the health sector over the next few years, include: * Strengthen the role of the Ministry of Health as a regulatory agency with a strong capacity for policy development, establishment and enforcement of quality standards and broadened responsibility to promote health education and public health programs. * Efforts to expand coverage to rural areas and to metropolitan populations that have limited access to health services will contribute to increasing equity and to improving the access of the poor to basic services. The SIPO data shows that nearly 30 percent of the extremely poor do not have health insurance, despite universal access. Efforts should be made to introduce alternative delivery arrangements for the primary health care model to be implemented in the metropolitan areas, to ensure adequate funding, to target the poor with PHC programs, to develop incentives for providers in rural areas and to re-orient the regional supply of services in function of a national health plan based on the regional health needs of the population, rather than on the historical regional distribution of services. * Consolidation of a culture of performance agreements, and shift from a historical based budget to a performance based budget. Along these lines, the establishment of mechanisms and regulatory changes to allow for the introduction of performance-based incentive payments to personnel 45 constitutes a considerable challenge to improving work place efficiency and quality. * Continue promoting changes in the system to increase management and financial autonomy of health providers in the context of the 1999 Law on Decentralization, including changes in finance, management and organization. * Introduce changes in the pharmaceutical procurement system that aim to improve operating efficiency and adapt the supply chain to recent changes under the Law on Decentralization. * Target evasion and strengthen efforts to implement the Ley de Protecci6n al Trabajador, including strengthening the CCSS' invoicing system and promoting changes in the financial information systems. * Promote improvements in clinical management in hospitals by professionalizing the management of information systems, making hospital chiefs of staff accountable for clinical and financial results and generating a. culture of clinical management. * Improve coordination with universities regarding human resource development for the long-term needs of the population. Social Protection 3.45 Pensions. The pension sector also underwent significant reforms in the past decade. The approval of the Ley de Protecci6n al Trabajador was an important step towards establishing a capitalization based pension system, as a second pillar (the first being the public system), under the supervision of the recently created Pension Superintendence. Reforms between 1992 and 1995 reduced the number of special pension schemes and achieved a more unified benefit framework. At the same time, the reforms reduced the future burden on public expenditure and the state subsidy necessary to cover pension expenditures. 3.46 By the end of the 90's, an estimated 56 percent of the economically active population (EAP) was covered by the National Pension System. This represents an overall decline in coverage from 66 percent of the EAP at the start of the decade. However, the coverage of informal sector workers has increased to 27 percent. The aging of the population represents a significant challenge in public policy over the coming years. Coverage of the CCSS public pension pillar currently covers only 36 percent of the population over 60 years, while an additional 35 percent is covered under the non- contributive scheme (RNC). The introduction of the complementary schemes should increase these percentages over the coming years while expanding access to RNC will improve overall coverage of the pension system. 3.47 The first step in the reform process was the unification and standardization of the special pension regimes in the public sector. Before the reform process started, there were more than 19 special regimes, which have been reduced to two (teachers and judicial sector workers). Today, only 5 percent of all pension contributors belong to special regimes. In 46 addition, the requirements for pensions and benefits of these regimes were brought closer in line with those of the IVM (Disability, Old Age and Death) of the CCSS. In 1995, a law was approved to regulate a third, voluntary private pillar. 3.48 The current reform continues this gradual approach by focusing on reforms to the second pillar (the capitalization based system). The Government prepared the recently- approved pension law based on a wide consultation process (Foro de Concertaci6n). The reform addresses some of the outstanding issues through a mixed pension system. The main public pension pillar, administered by the CCSS, will consist of the existing IVM regime with minor parametric changes to be determined. The second pillar is a new mandatory, fully funded pillar, with individual accounts managed by pension fund managers. 3.49 The law is rather general, leaving specific aspects of the system to the regulations. The law gives more powers to the CCSS to increase collection, strengthens penalties for evasion, redirects some labor taxes toward pension savings (mainly the severance scheme), introduces a supplementary funded pension scheme, extends a means-tested old age pension scheme, and sets up a single supervisory system for the pension systems. For the introduction of a funded supplemental pension scheme it redirects labor taxes of about 4.5 percent of salaries progressively over 2 years toward pension savings. The law does not specify fee structures and requirements for switching between funds. 3.50 Although Costa Rica's pension reform could be criticized for not going deep enough in reforming the existing pension system, especially the first pillar, it has the advantage that, compared to other reforms that have been undertaken, it has a limited fiscal impact. The new funded second pillar does not receive any contributions from a reduction in size of the first pillar and, therefore, there is no financing gap for the payment of current benefits. Part of the fiscal impact of the reform will be the institutional strengthening expenses, including two public information campaigns, strengthening the Superintendence of Pensions, and establishing the centralized contributions collection system to be run by CCSS. Government finances will also be affected by the reallocation of the 3 percentage points from the severance payment scheme to the labor capitalization funds. The severance payment contribution had been just a book entry, but now the 3 percentage points have to be paid in cash. This impact will also be felt by enterprises that did not pre- fund the severance payment liabilities. Another fiscal impact could arise from the implementation of the new rules for the means-tested old age pensions and its final cost. If the revenue for these non-contributive pensions (lottery, tax on public enterprise profits) does not reach 3 billion colones a year (in real terms) the Treasury will have to cover the shortfall. 3.51 In summary, the main challenges in the pension sector include: * Reducing evasion and sub-declaration of salaries. * Adjusting the calculation of benefits to more accurately reflect earnings capacity, through providing incentives for contributing a greater share of actual earnings in the earlier years of the earnings cycle. * Improving access to minimum pensions and improving equity in the distribution of these benefits. 47 * Improving the level of benefi&s of the non-contributory pensions to cover levels guaranteed to ensure welfare of the elderly. * Consolidating reforms that will guarantee the long-term sustainability of the public pension schemes. As the population ages sharply over the next 30 years-due to falling fertility rates and rising life expectancy-the financial imbalance would lead to pension fund insolvency. 3.52 Social assistance. Costa Rica has a well established social assistance network designed to protect vulnerable groups and to target them with a host of social programs. There are, however, critical gaps in social assistance of key vulnerable groups and there is a need to improve targeting, efficiency and monitoring/evaluation of ongoing programs. There is also a need for continued efforts in institutional strengthening and modernization, to reduce costs and offer better services and the adoption of new delivery systems involving communities and NGOs to complement public sector agencies. Among the critical gaps in social assistance is the inadequate coverage of programs for poor children under five years of age, who lack adequate education and stimulation programs, and assistance for the elderly poor not covered by the formal pension system. 3.53 Conceptually, social assistance programs are important not only because they can help protect the poor and vulnerable groups during normal and crisis times, but also because they can be used to improve the utilization and effectiveness of social programs (such as education and health) by the poor to help them deals with shocks. Thus, social assistance programs can function as a safety net to cushion the poor and vulnerable groups from the adverse effects of a shock, or as a coping mechanism, to reduce the chances that people do not have the capacity to deal with the consequences of shocks, and or to reduce the chances that the adverse shocks occur. By doing this, people can adopt self-protecting strategies, which reduce the chances that such shocks in income or consumption occur, as when people get education and training that help them reduce the probabilities of unemployment.27 Also, when ill, poor people can get health protection from health insurance schemes or government sponsored health care, as in Costa Rica. 3.54 Costa Rican assistance programs have been, however, designed primarily to serve the first purpose (coping inechanism), and have placed less attention to the potential benefits that the poor and vulnerable people would get by protecting their children and their families of future shocks through better education and training. This can be seen in the large shares of subsidies allocated to transfer programs such as housing, food distribution, and others, and the fact that most transfers benefit middle income groups who do not need them for improving their education and health outcomes. In what follows of this section, we identify ways in which assistance programs can be made more pro-poor and better designed to help the poor insure and self-insure against futures risks. 3.55 Costa Rica's social assistance network consists of three pillars. The first pillar is the Fondo de Desarrollo Social y Asignaciones Familiares (FODESAF) which serves as a financing agency for most social assistance programs, nationwide. FODESAF receives budgetaryfunds from the Ministry of Finance based on earmarked tax revenues. FODESAF's resources amounted to about 1.8 percent of GDP in 1998-2000 (an average of 27 For a discussion of the concepts of market insurance and self-protection in the context of social policy see De Ferranti, et al. 2000. 48 1.7 percent of GDP in 1990-2000).28 The. second pillar is the Instituto Mixto de Ayuda Social (IMAS), which manages an array of social programs for poor and at-risk groups. IMAS enjoys great flexibility in program design and financial arrangements, being able to deal with social and economic emergencies and provide direct income support to families. The third pillar are line ministries such as health, which operates nutrition and child care centers (CEN-CINAI), education, which is responsible for school lunch, scholarship and other social welfare programs for students, and the housing ministry, which is responsible for the housing voucher program. In addition, there are a great number of private non- profit organizations and foundations that receive funds from FODESAF and/or IMAS to support social programs for the elderly, handicapped, abandoned children and other at-risk groups. 3.56 The following are some actions that the government could take to improve the social assistance system. 1. Increasing coverage and targeting of ECD programs for 0-5 year olds 3.57 The coverage of 0-5 year olds with early childhood development (ECD) programs in Costa Rica at about 8.6 percent (2001) is low for its level of development and high social spending About 84,000 poor children 0-5 years of age have been found in the SIPO database as being not covered by any of the existing Early Child Development Programs. There are three main public programs financed by FODESAF for this age group. The first is the CEN-CINAI program operated by the Ministry of Health which covers about 18,200 children 2-5 years old (2001) with integrated health care, nutrition, early stimulation and pre-school education services. The CEN-CINAI program also provides powdered milk and food rations to poor pregnant mothers and their families (about 14,000 mothers and 97,000 children). The second is the Hogares Comunitarios program operated by IMAS, which provides community-based child care services to about 6,600 poor children. The program follows child-care models common in other LAC countries that provide nutrition and care services to children living in poor neighborhoods, offered by trained community mothers; qualifying families receive monthly subsidies from IMAS to pay the community mothers, who are not on the government payroll. The third is the program of the Ministry of Education that covered about 11,000 children. Finally there are a few privately run publicly funded programs attending about 8,700 children in 2001. 3.58 The government of Costa Rica created in the year 2000 the De la Mano program aimed at increasing coverage and improving the targeting of existing ECD programs. De la Mano is also charged with designing and promoting new approaches to increasing parental awareness of the benefits of ECD education, improve parental education and make wide dissemination of ECD teaching materials for patents, school children and communities. The coverage of ECD programs for poor families increased significantly from about 27,600 (5.5 percent of 0-5 years old) children attended in the CEN-CINAI, Hogares Comunitarios, the Ministry of Education and other programs in 2000, to about 44,600 (8.6 percent) in 2001 (Table 29). The De la Mano program distributed education materials to parents, teachers and secondary school children, and financed radio programs and newspaper separates to educate parents and communities in child development practices. Using SIPO, a study was made to match regional needs with available supply of ECD education by communities, NGOs, religious and local organizations, and the private 28 Government figures reported in GOCR, 2000a. There are some differences in the ratio of expenditures to GDP with other studies because Costa Rica is currently reviewing national account figures for GDP. 49 sector. Based on this study, a strategy is being prepared to further increase coverage and the quality of ECD education, which can be a key input for the next Administration taking office in 2002. Table 29. Coverage of ECD Programs in Costa Rica, 2000-01 lCD Programs 2000 -; ,;, -2001-EnhdJune 200iEnd September Estimated Population 506,000 517,126 520,423 0-5 years olds Public Programs 19.054 28.786 35.878 CEN-CINAI 13.831 13.213 18.200 Ministry of Education 1.140 11.100 11.100 Hogares Comunitarios 4.083 4.473 6.578 Other-NGOs 8.548 8.736 8.737 Total attended 27.602 37.522 44.615 % Covered total 5.5% 7.3% 8.6% Source: Figures from IMAS, De la Mano Program. 3.59 Cost of increasing ECD coverage to poor children. The De la Mano program has proposed several alternatives for increasing coverage of the estimated 84,000 children without coverage, by the end of 2001. One viable alternative considers a four year expansion plan and two modalities of child care. One modality is the traditional CEN- CINAI model which includes center-based care including full day care provided by trained child care personnel. The other modality is the application of an "informal model" in which parents are trained in early childhood education and stimulation to attend their children at home. Parents receive training in groups by a central group of professionals organized by De la Mano. 3.60 The per child cost of the informal model is US$65 dollars per year, as compared to that of the CEN-CINAI model of US$ 590 dollars per year. Assuming a 50-50 percent split between the two care models for the four year expansion plan (adding 25 percent of children each year), total costs amount to US$26 million in the fourth year when full coverage of poor 0-5 years old children is attained. Current spending on the CEN-CINAI program amounted to about US$19 million in 2001, so that only an additional US$11 million would be needed to achieve full coverage in 2002 and following years. This requires, however, that CEN-CINAI stops benefiting non-poor children (about 75-80 percent of current beneficiaries) to concentrate on the poor. These costs can be reduced substantially if the CEN-CINAI model is rationalized to emphasize early stimulation and development of children decreasing or eliminating the child nutrition component and leaving the health component to the EBAIS of the CCSS. Currently, malnutrition is very low in Costa Rica and there is no need for major food and nutrition component of the program. Also, as indicated in other sections of this report, the CCSS is successfully implementing the local health primary health service component un the EBAIS system which could well assume all health and child growth monitoring components under the CEN-CINAI system, avoiding duplication and saving resources. 2. Increasing coverage of non-contributory pensions for the elderly poor (over 65 years of age) 3.61 Despite Costa Rica's high coverage of the social security system, a large number of elderly poor remain uncovered and in a state of poverty or extreme poverty. The CCSS, with funds provided by FODESAF, provides a large number of non-contributory pensions 50 directed at poor elderly people not covered by the contributory system, and other special groups, such as widows, handicapped persons and people with brain paralysis. The monthly pension amounted to 12,000 colones in 2001 (or US$37 dollars) representing about 20 percent of the minimum wage or about the cost of the food basket. 9 According to figures from SIPO, about 14,000 people (about 50 percent of elderly below the poverty line) were uncovered by any pension regime in 2000. 3.62 Some of the reasons for the gap in coverage and high leakage include, the lack of information and outreach of the poorest prospective beneficiaries, and the fact that social workers granting pensions at the CCSS have verified only about 60 percent of the applications received.30 However, recent figures from the government indicate that major increments in coverage and targeting have been achieved in 2000-2001 which have not been reflected in the original SIPO database. According to these figures, over 10,000 pensions that had been awarded to non-deserving applicants in the past have been re- directed to the poor in the SIPO and that new pensions have been given to uncovered poor. These recent figures indicate that only about 7,000 elderly poor may have been without coverage by the end of 2001. Unfortunately, SIPO has not been updated to include these new figures. 3.63 Cost of increasing coverage of non-contributory pensions. As in many other social assistance programs, Costa Rica appears to have enough funds dedicated to non- contributory pensions. Thus, the additional 7,000 elderly poor not covered by the end of 2001 can be covered with re-allocations of pensions currently in the hands of non- deserving people, continuing what was done in 2000 and 2001, and choosing new applicants with the SIPO system. As seen later in this section, non-contributory pensions have been badly targeted to the poor and many recipients do not qualify when the SIPO system is applied. 3. Improving coordination and efficiency of programs by strengthening the role of FODESAF 3.64 Unlike many other LAC countries, Costa Rica has relatively few institutions managing and operating social assistance programs. There is only one single institution, the FODESAF, charged with financing social assistance programs. FODESAF is, thus, in a unique position to influence institutions and programs to improve efficiency and coordination by introducing production-based financing mechanisms and promoting program coordination. Some of these actions have been introduced since 1999, when FODESAF instructed programs and institutions to provide targets to be achieved and to rate programs and institutions according to the attainment of these targets. These efforts will need to be, however, consolidated and formalized in the institutional and legal framework of FODESAF to be effective. 3.65 FODESAF has some structural problems. First, most FODESAF allocations (about 77 percent) are fixed allocations determined by old laws which also often earmark tax 29 The poverty line is calculated by the Institute of Statistics based on the cost of a basic food basket and other necessities. In June 2001, the cost of poverty line was set at c$26,084 urban and c$20,094 rural (US$81 and 62) per capita per month. The cost of the basic food basket (extreme poverty line) was set at c$11,965 urban and c$10,200 (US$37 and 32) per capita per month (exchange rate used c$320 per one dollar). 30 Information contained in the document "Evaluaci6n del Programa Regimen No Contributivo de Pensiones Por Monto Bisico", Draft FODESAF, 2000. 51 proceeds to certain programs irrespective of the changing needs of the poor. This makes it impossible for new governments to change budget allocations according to new priorities or needs of the poor. Old priorities remain in place with no possibility of changing them. For instance, as observed in Table 30, too much may have been spent on housing (and or contributory pensions) and too little may have been spent on child care programs (see allocations for Ministry of Health vs. those for BANHVI and CCSS). Also, and because of this situation, FODESAF has little flexibility to address the needs to the "new" poor in crisis times. Thus, FODESAF law needs to change to allow flexible allocations of funds to different programs according to needs and the changing needs of the poor. 3.66 Second, FODESAF has little effective control over the efficiency and impact of programs it finances, especially allocations mandated by law to programs and institutions. These institutions are not accountable to FODESAF. This limits the capacity of FODESAF for effective monitoring and evaluation of those programs, and reduces incentive based financing, and corrective measures, if needed, as a result of such monitoring and evaluation. Third and finally, FODESAF has limited ability to plan for resource spending since a large percentage of budgeted funds are not effectively disbursed by the Treasury during the course of the fiscal year. In fact, only about 63 percent of budgeted funds, on average for 1990-2000, have been disbursed. Disbursements as a share of budgeted funds have, however, increased to about 66 percent since 1999 (Table 30).31 Table 30: Budget Allocations by FODESAF Main Assistance Programs, 1990-2000 (%) items - . - 1990 -- 1995 . 1998 -1999 2000 CEN-CINAI, MOH 22.2 11.4 6.0 7.0 6.8 School Feeding, MOE 6.8 11.1 12.6 13.9 13.1 BANHVI-Housing Bond 29.1 26.9 31.7 37.2 32.3 IMAS, Assistance programs 1.7 11.4 9.8 4.1 10.5 Patronato Nacional de la Infancia 4.1 4.5 3.8 5.1 5.1 Assistance pensions, CCSS 19.4 18.6 22.9 21.1 20.7 Other 16.7 16.1 13.2 11.6 11.5 Total 100.0 100.0 100.0 100.0 100.0 % Spent vs. Budgeted 60.2 58.6 62.6 65.9 66.8 Source: Figures provided by FODESAF. 3.67 The goal of the government should be to continue streamlining institutions and consolidating programs financed by FODESAF. Some program consolidation has already taken place since 1999. Some scholarship programs which were administered by IMAS have been given to FONABE (Fondo Nacional de Becas), under the Ministry of Education. This has reduced administrative costs, has permitted increments in coverage of subsidies, and has contributed to better targeting of subsidies. 4. Improving targeting of assistance programs 3.68 As shown in Table 31 below, targeting of some important social assistance programs is highly deficient. This is the case, for instance, for the child care program (CEN-CINAI), where only 24 percent of beneficiary children belonged to families in the 31 The fall in the share of CEN_CINAI is mainly due to the reduction in health personnel charged to the program when the MOH transferred health centers and posts to the CCSS, as a result of the health sector reforms. 52 poorest income quintiie in 1999. Also, only about 45 percent of beneficiaries of non- contributory pension benefits belonged to the elderly poor in the poorest income quintile. For children in school, about 62 percent of subsidies of the school lunch program went to the poorest 50 percent of children. The school voucher program was, however, much better targeted to the poorest students, as 76 percent of benefits went to the poorest 50 percent of children, indicating that there is the technical capacity to improve targeting if the political will to do so exists. The improved targeting of the school voucher program has occurred because of the use of SIPO to select beneficiaries of this program. Table 31: Cumulative Distribution of Benefits of Key Social Programs by Income Quintile 1999 CEN-CINA] Non- + -Schodl' Schobl- . - . v -Contrlbutory - Qntilei 'vo'uch-er' 6eding- . Pensions * , . -- -. . ~. ' Cbildcare Milk -Food Rations: Total ; 1 41 34 24 61 37 56 45 2 76 62 58 85 69 81 69 3 92 82 79 95 89 93 84 4 98 94 96 98 93 98 96 5 100 100 100 100 100 100 100 Source: Plan de Solidaridad, based on Encuesta de Hogares de Propositos Multiples, 1999 3.69 The housing subsidies were poorly targeted but performance has improved. In fact, by design, the housing program, which accounted for almost one third of FODESAF resources in 2000, is not targeted to the poorest. People with less than four minimum wages can apply to the program and, until recently, cooperatives and other intermediary institutions could apply for benefits for their associates. Household surveys indicate that only about 25 percent of the subsidies had been received by the poorest 25 percent of the population during the 1990-99 period (Table 32). Recently, major efforts have been made to improve targeting by giving favored treatment to applicants with less than two minimum wages and by reviewing applications to ensure that they correspond to the poor and meet eligibility requirements. Table 32 shows some improvements in targeting over the last few years, as the share of subsidies received by the poorest 25 percent of the income distribution has increased to about one third of the subsidies, in 1998-1999. 3.70 It is doubtful, however, that the current form and spending size in housing subsidies is justified in Costa Rica. First, as seen above, Costa Ricans have high preferences for investing in their own house and making improvements with own resources. Housing ownership is among the highest in Latin America even by the poorest families. Second, most current housing needs are related to improving the quality of the house, improving floors, walls, roofs and connections to basic services rather than absolute lack of housing. This can be financed with much lower subsidies than the current US$3,700 dollars on average of the housing vouchers, and for many, with loans that can and should be recovered. Third, subsidies should not be provided to finance private goods (such as housing), but public goods or private goods which provide positive externalities for the rest of the population. This is the case of education, especially of early childhood education, in 32 However, the SIPO system appears not to have been used for school bond program in 2002, so that targeting has probably worsened in this year. 53 which Costa Rica presents significant gaps. Thus, of the about US$60 million dollars per year Costa Rica is spending in housing subsidies, over half to one third should be allocated to programs such as those in education to significantly increase coverage of ECD education for poor children. Table 32: Cumulative Distribution of the Bono de Vivienda by Income Quintile Quintile J990-1994.i:: -99494-1998 - 1998-1999 1 18.8 21.7 31.9 2 39.0 46.8 65.2 3 66.1 73.8 83.0 4 86.2 91.6 94.4 5 100.0 100.0 100.0 Source: GOCR (2000b). 3.71 In general, one of the main problems of targeting social assistance programs has been the use of many different targeting mechanisms and instruments (even for similar programs). This has led to inequities, favoritism, duplication of benefits and, in many cases, exclusion of people who need the benefits the most. The creation of the SIPO system in 1999 addresses some of these problems by proposing to be the single most important targeting instrument for most of the programs which are to be directed to the poor (means-tested), as seen below. Not all programs, however, can or should be targeted with SIPO. Some programs, such as nutrition programs for child and mother, would need other instruments. 3.72 The SIPO system is managed and operated by IMAS at central and regional offices which are in charge of maintaining the database and of keeping it up to date. By end 2001, the system contained a database of about 207,000 families who are potential beneficiaries of programs.33 A description of SIPO is in Annex 1 below. Families in SIPO can be classified according to poverty lines or according to a weighted poverty index (a proxy means test), which has been constructed on the basis of selected demographic and socioeconomic variables. The main characteristics of SIPO are as follows: * Uniform and objective. People in all parts of the country and for specific programs are measured with the same standard and criteria. * Easy to administer and apply. The system is simple and people at IMAS local level can administer and update it. * Cost effective. The cost of application and administration of the system (that is, to apply the formulary and enter data into computers) is low (8-10 dollars per family) relative to the amount of the benefits to be targeted and to cost of similar programs in other countries. It has been estimated that 33 People are included in the database in a three-step process. First, poor areas are identified using NBI-based poverty maps from census information and surveys, such as the poverty map developed by the Planning Office. Second, using statistical analysis of survey data for 1997, the variables most strongly associated with poverty were selected, along with the weights they exhibited in statistical analysis. Third, the selected variables were placed in a questionnaire that was administered to all households in selected poor areas, generating the database of potential beneficiaries (poor people who do not live in the selected area can apply for inclusion in the database). The variables were used to construct a poverty index. Program eligibility begins from the poorest up to a predetermined cut-off point on the index, established according to the program's budget and target population. 54 over 240 million dollars (most of FODESAF and IMAS budget) can be targeted using the SIPO system.34 3.73 The SIPO system can also be used, along with other criteria, to better target subsidies of programs such as the CEN-CINAI, the De la Mano program, the school lunch program, transportation subsidies, among others. Because the SIPO system is Web-based and is ready to share information with other government agencies, SIPO can become the platform of a single national registry of beneficiaries of assistance programs in Costa Rica. This will be critical for eliminating duplication of beneficiaries and to better coordinate and design social programs. With this, entire families (and even communities) can be traced along with the benefits they have received. Evaluation studies can be designed to assess the impact of the programs and to identity the best and most cost-effective interventions. 3.74 There are some critical issues that need to be solved if the SIPO system is going to be used by agencies and survive in the future. First, there is still not a normative act (Decree, Law) which specifies responsibilities for program use, updating of the system, and financing of the system. For instance, many agencies are applying the formularies to their prospective beneficiaries, in some cases, without quality control and supervision of IMAS affecting the quality of the information. Second, there is still a lack of clarity regarding updating procedures, determination of coverage and the duties and responsibilities of agencies using the system. For instance, there is still not a definition of when and how changes in family circumstances will be updated and how to avoid manipulation by people to be eligible for programs. Also, there is still not a definition of how other agencies update SIPO's registries when they provide the benefits and if other than the updating agency has access to records to cross check benefits and beneficiaries. Third, the SIPO system has been recently used to include some of the people affected by special circumstances, such as the "banana, coffee and fishing crisis" in some areas, who in normal times would not be included in the data base and whose point scores are much higher than those of the poor. This affects the credibility of the point score system which has been designed for the structural poor. IMAS is looking for ways to design a special module to include the "new" poor, but procedures are still not in place. 3.75 In summary, although Costa Rica has a functioning social assistance institutional network and spends a considerable amount of resources for the poor and most vulnerable groups, there are still a considerable number of poor to be attended. The impact of this network has been limited by a series of factors. First, there has been a lack of strategic planning which determines priorities in groups to be attended, duration of attention and interventions most effective. Second, the FODESAF Fund, which finances most assistance programs directed to the poor, has little flexibility in budget allocations to change expenditure priorities according to changing needs of the poor. Third, there has been a lack of institutional coordination, both in the design of programs and in the operation of such programs. A multiplicity of programs result in duplication of efforts and waste of resources. Fourth, there has been a lack of uniform criteria for selecting beneficiaries for such programs. Recent efforts have been made to address this problem with the introduction of the SIPO system in 1999, which has started to be used for targeting several programs, including the school bond, scholarships and non-contributory pensions with notable improvements in targeting. These targeting efforts need to be continued in the 34 Recently, the Comptrollers Office has issued directives to use SIPO for selecting beneficiaries of a great number of programs, including some of potable water and sanitation. 55 future with appropriate measures taken by IMAS for meeting the complex challenge of updating and maintaining the SIPO targeting system. Monitoring and Evaluation Systems 3.76 There is a lack of an effective M&E system for the social sectors and this is an impediment to ensuring value for money in social spending. Monitoring and evaluation are key to evaluate progress, evaluate impact of programs and make more efficient decisions on programs to improve social indicators of the poor. For instance, had a good evaluation program existed in the CEN-CINAI and other child care programs, resources could have been saved by finding the most cost effective alternative for child care. Different modalities of child care, some including orily parental education, others including center-based stimulation and early child education, among others) could have been evaluated to determine their impacts and relative costs. The practice of establishing base- lines and measuring the impact of specific programs should be implemented routinely, in order to draw lessons from experience which can be used to modify programs and improve their impact. In addition, instruments such as Living Standards and Measurement Surveys and improved household surveys provide valuable information on key target groups and on the outcomes of the main social programs, complementing the annual household surveys cuirently carried out by the Statistical Istitute. 3.77 In 2000, the Government began efforts to monitor progress in meeting targets set for the different social programs, and started to regularly publish the results to improve accountability of program administrators. It also contracted with the local UNDP Office the evaluation of major social programs in the education sector including the school lunch program, scholarship and voucher program, among others. These efforts need to be continued, along with the introduction of formal scientific evaluation of major social programs to evaluate their impact and cost effectiveness. 56 4. IMPROVING TIE EFFECTIVENESS OF SOCIAL SPENDING 4.1 This section summarizes the recommendations of this report and their associated costs for improving the social indicators of the poor in Costa Rica by focusing on improving the efficiency and targeting of social spending. The spending priorities include the increase in coverage of ECD education and early stimulation to poor children, improvements in education quality to reduce repetition and dropouts in primary and secondary education and increase coverage of non-contributory pensions to the elderly poor. Most, if not all of these spending needs, can and should be financed with re- allocations and redistributions of existing social spending, as well as efficiency improvements to produce value for money rather than with increments in social spending, as will be summarized below. 4.2 Increasing coverage of early child education programs for the poor. In 2000, Early Child Development (ECD) programs, such as CEN-CINAI centers, Hogares Comunitarios, or other programs failed to reach an estimated 84,000 poor children that require social assistance. Most of these children stayed at home under the care of poorly educated mothers, older siblings, or relatives who may lack the skills to adequately stimulate young children for developing their motor and social skills which are critical for success in school and later in life. Coverage of these children can be increased through several means including expansion of the number of children in the CEN-CINAI (where excess capacity exists) and Hogares Comunitarios centers and the introduction of non- formal models that rely on community and parental education in childcare stimulation and early education. The introduction of effective child care programs will not only benefit children, but also their mothers who can then be trained and can participate more in the labor market. It has been estimated that covering all uncovered children would cost about US$ 26 million dollars per year, using a combination of center-based care and non-formal care by parents. Since costs of CEN-CINAI are already high at about $19 million per year, only an additional $11 million will be needed to reach full coverage of poor children. This assumes that about 80 percent of current beneficiaries of CEN-CINAI who are not poor and are dropped from the program to make room for poor children). However, these costs can be reduced substantially if the existing childcare programs are rationalized to emphasize their education and stimulation components reducing or eliminating their current nutrition and health care components. Under nutrition is very low now in Costa Rica and the health component is being provided by the local health services (EBAIS) of the CCSS. There is also an urgent need to experiment with different modalities of child care and evaluate their cost effectiveness to determine the best altemative which is affordable for Costa Rica. 4.3 Improving completion rates and quality of primary and secondary education. Despite high spending on education, Costa Rica is lagging in education coverage and attainment relative to countries of similar income and education expenditures. To improve the quality of primary education-and reduce repetition and drop-out rates--the govemment needs to invest more in teacher training, technology in schools, school 57 materials. To improve access by the poorest, better targeting of the school lunch and other education welfare programs as well as other programs such as scholarships on conditional cash transfers should be considered in order to ensure benefits reach those who need the subsidies the most. In secondary education, net enrollment rates of the poor 13-18 years olds are 53 percent (as compared to 61 percent for the national average from the 2000 Census). To Improve quality and coverage of secondary education-and reducing repetition and drop-outs-the government will require supply-side investments in teacher training, technology and modem methods to reach dispersed populations in public schools. The government may need to consider measures to induce more demand for education by providing subsistence subsidies, stipends, scholarships, to poor students, so that they can forego the opportunity and direct costs of schooling, especially in rural areas. Some of these programs have already been introduced with success recently under the Administration of President Rodriguez 4.4 The increased cost of the proposed reforms is estimated to sum up to between 6 and 7 percent of current expenditures. These increases could be financed by, (a) improving intemal efficiency, through reduction in repetition rates, for example; (b) the constitutional imperative to increase the Ministry of Education's budget to 6 percent of GDP; and (c) improve the distribution of public funds between university, where spending is highly regressive, and primary and secondary education where spending is more pro-poor. Improvements in efficiency and quality are likely to free-up resources currently tied up to finance inefficiency, such as high repetition rates, and poorly targeted programs, which distribute resources to populations that are less needy. As an example, crude estimates indicate that repetition in primary school costs about $21.2 million per year or about 10 percent of the education budget for primary education, while the costs for secondary education repetition are equivalent to $14.8 million per year. 4.5 Improving the quality and health care coverage. Costa Rica's health care system has been increasing coverage of health services rapidly over the last few years, as a result of reforms begun in the mid 1990s. Overall coverage of preventive and curative health services has been estimated at about 85 percent of the population. Coverage among the poor, however, remains a problem. Major challenges remain for the health sector to improve value for money as it prepares for the aging population with a more costly chronic disease based epidemiological profile. In the health sector the government should consider the following actions. * First, strengthen the role of the Ministry of Health as a regulatory agency with a strong capacity for policy development, enforcement of quality standards and promotion of health education and public health programs. * Second, in the CCSS, continue expanding coverage to rural areas and to metropolitan populations that have limited access to health services by introducing alternative delivery arrangements for the primary health care, ensuring adequate funding. * Third, consolidation of a culture of performance agreements, and shift from a historical based budget to a performance based budget. 58 * Fourth, introduce changes in the pharmaceutical procurement system that aim to improve operating efficiency and adapt the supply chain to recent changes under the Law on Decentralization. * Fifth, target evasion and strengthen efforts to implement the Ley de Protecci6n al Trabajador, including strengthening the CCSS' invoicing system and promoting changes in the financial information systems. 4.6 Estimates provided by the government indicate that these reforms would have an estimated cost of roughly $70 million over 5 years. These costs would cover, investments in primary care to consolidate the primary care model, extending access to 100 percent of the population; investments in personnel, training and information systems to improve the stewardship function of the MOH; investments in upgrading hospital equipment and developing ambulatory solutions; implementation of management reforms and information systems to improve quality and efficiency in the supply chain, hospital management and purchasing function; and continued support to implement nationwide the improvements in the financial management system. Over five years, the costs would be equivalent to less than 2 percent of the estimated annual health budget for the CCSS. These reforms would produce annual savings in excess of $28 million per year. This would nearly double the estimated cost of the reforms and allow the CCSS to continue producing more with fewer resources. 4.7 Improving access to old age pensions. The pension sector underwent significant reforms in the past decade. The majority of the reforms addressed the most pressing financial problems in terms of diversity of benefits among the pension schemes. Further progress is needed to increase coverage, currently at 56 percent of the working age population (PEA), and to consolidate the financial standing of the system. The actions that can be taken include, first, consolidating reforms that will guarantee the long-term sustainability of the public pension schemes. As the population ages sharply over the next 30 years, the financial imbalance would lead to pension fund insolvency. Second, adjusting the calculation of benefits to more accurately reflect earnings capacity, through providing incentives for contributing a greater share of actual earnings in the earlier years of the earnings cycle. Third, reducing evasion and sub-declaration of salaries. And fourth, improving access to minimum pensions and improving equity in the distribution of these benefits. 4.8 Improving non-contributory pensions for the poor. In 2000, the SIPO database reported that about 14,000 extreme poor and poor elderly (about 50 percent of group) had no assistance pension. About 46 percent of extreme poor older than 65 years of age reported permanent or temporary physical or mental limitation. Recent figures from the government indicate, however, that the number of non-contributory pensions awarded increased significantly in 2000 and 2001, so that coverage of elderly poor has considerably increased and that only about 7,000 elderly poor remained to be covered by the end of 2001. It has been estimated that covering these remaining poor would cost about US$ 3.1 million per year. Most of this coverage could be achieved without additional spending if targeting is improved. 4.9 Improving relevance and targeting of social assistance programs. Social assistance priorities should shift towards programs that combine immediate income assistance to meet the needs of the poor and vulnerable groups, with investments in human capital for the future, and target subsidies to those who really need them. To achieve these 59 objectives, the government can consider reducing transfers to housing subsidies and other such transfers and allocate government funds to public goods or private goods with high externalities, such as education and early child education which are more likely to reach the poorest. Critical for this re-allocation of spending priorities is the need to change FODESAF's law, which currently assigns specific percentages of spending to programs that are of doubtful priority for the current needs of the poor. 4.10 In housing data shows that housing quality is the main gap for poor families, as reflected in the high numbers of poor families who have houses with floors made of substandard materials (107,000 or 65 percent), roofs (98,000 or 59 percent) and walls (101,000 or 61 percent). In order to improve effectiveness, the Government should reassess if the current policy of giving Bonds for new housing is the most appropriate way to meet the current needs of the poor population. Other options, such as providing loans for housing improvements could be considered. 4.11 One of the sources of ineffective targeting of public resources through social programs has been the lack of unified criteria for identifying beneficiaries for social programs. While no single system can capture the multiple characteristics of the poor and their precise needs, a system which characterizes the poor according to a wide number of characteristics associated to poverty is useful to reduce duplications of benefits, identify poor who are left out of the benefit system, and to establish targets for social indicators of the poor. Agencies implementing social programs can work with the same database of potential beneficiaries and record the benefits provided, thus, contributing to produce a single registry of beneficiaries with a de-facto "accounting" system for beneficiary families. 4.12 Although some improvements have recently been made to targeting social assistance programs to the poor, more needs to be done. This includes extending the application of the SIPO selection system for most programs that are means tested, such as CEN-CINAI, non-contributory pensions, and school lunches, among others. The SIPO system is based on state of the art Internet technology and, by the end of 2001, contained over 207,000 families or about 808,500 people, representing about 21 percent of the Costa Rican population. To be effective, however, the SIPO system should be mandated for all institutions managing targeted social programs35 and would need to be kept up to date and include all of the potential beneficiaries, a task which will be a great challenge for IMAS. The SIPO database is most useful for describing the main characteristics of the poor, and for the design of programs more attuned to local needs and conditions. 4.13 Three issues are critical for the maintenance of the SIPO system in the future. First is updating of the system, which is a common problem of such systems.36 IMAS has already produced updating procedures and has applied them regularly, but continuous 35 Several countries have systems similar to SIPO to help target social programs directed to the poor. These countries include Chile, that pioneered such systems with the introduction of the Ficha CAS system in the early 80s, Colombia that introduced the SISBEN system in the mid-1990s to target social spending particularly health insurance subsidies, Mexico that uses a means-test system for PROGRESA and Nicaragua for the Red de Protecci6n Social Program. 36 Updating information is not only a technical problem but also a "political" one. Once families realize that the information they provide is used for determining eligibility for some programs, they may request updating with the aim of manipulating the information. So, there is a need to clearly establish which information can be updated during a reasonable period of time-say births and deaths, changes of residence, etc-and which information can not be changed for a period of time-education, housing conditions, etc. Updating too frequently is also very costly. 60 efforts need to be made in the future.3 Second, despite the usefulness of SIPO for many social programs, many institutions will likely prefer to use their own targeting methods, often subjective and amenable to manipulation, in the absence of a normative that establishes compulsory uses and procedures for use of the system. Third, the applicability of SIPO for identifying the "new" poor may be limited. The survey form may be too long and costly to apply. The point score system based in weights of household characteristics describe better "structural poverty" than the "transient" poor. This, however, can be amended if, in crisis times, a shorter version of the survey form is applied and larger weights are given to unemployment, under-employment and income variables. 4.14 Improving monitoring and evaluation of social programs. Monitoring and evaluation of social programs have been traditionally weak or non existing in Costa Rica, as mentioned earlier in the case of ECD programs Monitoring and evaluation are key to evaluate progress, evaluate impact of programs and make more efficient decisions on programs to improve social indicators of the poor. In addition, the practice of establishing base-lines and measuring the impact of specific programs should be implemented more routinely, in order to draw lessons from experience which can be used to modify programs and improve their impact. Costa Rica is in a good position to implement a Living Standard Measurement Survey (LSMS) every 3-5 years to gather information on consumption and the status of key risk groups and on the outcomes of the main social programs, complementing the current annual household surveys carried out by the Statistical Institute (INEC). To strengthen the capacity of the LSMS to monitor poverty groups and programs, the following information should be included on a regular basis: * Consumption of household and household members; * Nutrition indicators of sub-sample of households; * Coverage of programs for children in the birth to five years old age cohort; * Private costs of pre-school education; * Incidence and coverage of social protection programs, including income support programs, education- based transfers, community based strategies, labor market programs, feeding programs and low-income housing programs; * Demand and supply issues with respect to pre-primary and secondary education; and * Reasons why children are absent from school. Improving Institutional Coordination 4.15 One of the impediments to more effective use of social spending to improve social indicators of the poor in Costa Rica has been the absence of a unified policy making body capable of setting social policy strategy which details programs, targets to be reached by programs, beneficiary populations, and the institutional arrangements to carry out that 37 In the future there will not be a need for massive surveys of poor areas as was done in the past. Most poor families are already in the system, so that regular updates of key variables is the only thing required. This can be done by IMAS on a regular basis when people access some of the social services it provides, or on demand when people want to be included in the database to apply for social programs. This updating strategy will reduce the cost of SIPO significantly in the future. 61 strategy. Efforts in this direction have been undertaken in recent administrations, but these efforts have been hampered by institutional rigidities and laws which have undermined coordination efforts and the reform or elimination of ineffective programs or institutions. The following is a brief discussion of some key institutional issues and possible ways to address them. 4.16 First, there is not a single unified body or agency which establishes social spending priorities, coordination rules and procedures for the different agencies, and monitors progress in achieving targets in social indicators. No functional linkages among institutions or programs have been established which describe their roles and responsibilities and their complementarities and synergies. This is producing duplication and atomization of efforts by institutions, and overlapping functions and responsibilities. Also, while in many countries private and NGO agencies operate many social programs, in Costa Rica most social programs are operated by public agencies under historically based budget allocations rather than on performance-based indicators. 4.17 Second, there is much rigidity in laws-such as FODESAF's-in expending priorities, so that social institutions cannot respond to the changing needs of the poor or to results achieved. Over 77 percent of FODESAF funds are allocated by law to different programs irrespective of the changing needs of the poor population or the effectiveness of these programs. 4.18 Third, there has not been a clear definition of which spending or programs are to be targeted to poor or other special groups, and which spending is to be universal. This has led to generally un-targeted programs, which benefit middle and upper income groups, and, until recently, little developed targeting instruments. While Costa Ricans have been understandably reluctant to target most social spending for fears that it weakens social cohesion and alters the achievement of common social values and culture, it is also undeniable that universal programs are costly. The long standing practice of spreading expenditure very thinly to attend large numbers of people-such as in the school lunch program-is not helping the poorest students who need them most, nor the better off students who are getting less benefits (calories and proteins) than what they could get with the help of parents.38 This practice leads to ineffective interventions as the size of the transfers or service may be too little to have any significant impact. 4.19 The institutional arrangements that could mitigate the problems highlighted in this report include the establishment of a unified policy setting body, with responsibilities for setting social spending priorities, defining institutional coordination arrangements, defining targets to be achieved by each institution, and monitoring progress, and assessing the effectiveness of programs and of the operational arrangements and institutions. Different Administrations have tried alternative arrangements to solve the institutional issues by creating a "social authority", often headed by one of the Vice-Presidents, but these efforts have lacked continuity and have changed with changes in the administrations. This section offers a series of suggestions on how institutional arrangements could be modified to improve policy making and coordination of the different social sector agencies in Costa Rica. 38 Targeting the school lunch program does not mean that better off students do not get school lunch. In fact, all students could get a good quality meal, the only difference being that poor students would get it free or heavily subsidized while better off students can pay all or part of the cost of the meals. 62 4.20 In terms of policy making and establishing spending priorities, one of the altematives that could be considered is the creation of a Council for Economic and Social Policy (CONPES) composed of ministries from Finance, Education, Health, Labor, the President of IMAS, among others. This body would be presided by the President of the Republic and would review and approve social policy strategy and action plans pursued by the government. This Council would need to have the authority for allocating social budgets and, particularly, for defining expenditure priorities of FODESAF funds, according to the strategy and action plans. This would give the Council the authority and budget power to enforce coordination and to hold the line ministries accountable for results. The strategy, actions plans and financing needs could be prepared by a Technical Secretarial of that Council. 4.21 The Technical Secretariat of CONPES could be composed of high level govemment officials from the Ministry of Finance, FODESAF, and the Ministry of the Presidency.39 The Secretariat would have two main responsibilities. First, prepare key policy papers and action plans-for example, at the beginning of each administration-- specifying targets, how targets will be achieved and the funds allocated to those purposes. Second, prepare bi-annual reports-to be reviewed and approved by CONPES-on the progress on achieving specified targets, the impact of programs achieved so far, and specify adjustment measures and costs, if needed. These reports could be based on reports made by operating agencies, including government, local agencies and non-government organizations, and evaluation reports commissioned with academic organizations, communities and others. 4.22 The operating agencies for social programs include ministries, decentralized or deconcentrated government agencies, local governments, communities and a host of organizations that could be involved in the social strategy and in government plans. The work of these agencies, the targets to be achieved, the beneficiary populations, and the financial resources would need to be detailed in a matrix of functions and responsibilities along with the outcome and intermediate indicators to be achieved. One issue that deserves attention is the public-private mix in service provision of social services. Currently, most social services including education, child care services and health care are provided by public agencies. In some other countries such as Chile, Colombia, among others, the private sector and or local governments and organizations provide education, child care and health care services to the poor, with government financing through vouchers given to beneficiaries or government transfers. This practice has often reduced costs and has opened the door to potential cost sharing arrangements with beneficiaries and operating agencies reducing costs for the government and improving efficiency of programs. 4.23 Allowing greater FODESAF budget flexibility could also these resources to be better targeted and to be used for economic contingencies affecting the poor and "new" poor, resulting from economic downturns (not only to achieve long term improvements in social indicators of the poor). FODESAF could then accumulate funds during times of economic growth, and spend them in times of economic down turns (in employment programs, unemployment assistance, etc.). FODESAF could then finance and manage a portfolio of carefully designed social programs to address social problems of the "structural" poor, as well as finance social spending needs during economic crisis. This 39 A similar group named "Budgeting Authority" serves to set expenditure priorities of the whole public sector in Costa Rica. 63 will require greater budget flexibility for the FODESAF Fund, so that it can be used to respond to changing economic conditions. Within FODESAF, programs such as non- contributory pensions, ECD programs, scholarships, may have some minimum expenditure ceilings, while other expenditures may change every year, according to identified needs and priorities. 4.24 ;, Budgets would need.to be increasingly allocated to institutions on the basis of performance rather than on the system of historical budget allocations, as started to be done by the administration of President Rodriguez for social programs, such as scholarships, Schools Bonds, pensions and others. Payments on the basis of people attended by institutions and other service providers would induce them to provide more and better services to the poor (IADB, 1996). When there are many providers of services, including public and private NGOs or other local institutions, vouchers are a good alternative in that beneficiaries can choose the provider that offers the best service. Private agencies and NGOs can be very effective in complementing public services in childcare, health care, education (for instance, secondary education and youth training) and many other social services. 64 REFERENCES Castafieda, Tarsicio, Aldaz, Enrique and Moran, Ricardo. 2001. The Intergenerational Transmission of Poverty: Some Causes anid Policy Implications. IDB, unpublished, Washington D. C. Centro Centroamericano de Poblaci6n, Universidad de Costa Rica. 20001. Cuadro 2-2. Costa Rica: Proyecci6n de la Poblaci6n por Sexo y Grupos Quinquenales de Edad. 1995-2050. http://ccp.ucr.ac.cr/observa/estima/cuadro2-2.htm Cercone, James, Duran, Fabio y Bricefio Rodrigo. 2001. El Desempefno del Sector Social en Costa Rica Durante la Decada de los Noventa: Los Nuevos Retos, Instituto Latinoamericano de Politicas Puiblicas, Banco Mundial, SANIGEST. Deaton, Angus, 2001. Counting the World's Poor: Problems and Possible Solutions. Research Observer, The World Bank. Washington, D. C. de Ferranti, David. and Perry, G. 2000. Investing in Our Future. The World Bank, Washington, D.C. Duryea, Suzanne., and Szekely Miguel. 1998. Labor Markets in Latin America. A Supply -Side Story. Inter American Development Bank. Elizondo, Rafael, Villalobos, William and Poltronieri, Jorge 2000. Fundamentos Estadisticos y Matematicos para el Calculo del Puntaje en la Metodologia SIPO. San Jose, Costa Rica. FODESAF. 2000. Evaluaci6n del Programa Regimen No Contributivo de Pensiones Por Monto Bisico, Draft, unpublished. Government of Costa Rica (GOCR). 2000a. Evaluaci6n del Plan de Solidaridad. Draft unpublished. . 2000b. El Sistema de Subsidios a la Vivienda. Unpublished Document. 1 1999. Plan de Solidaridad 1999. Hemandez, Luis Carlos 2000a. Situaci6n de la Educacion en Costa Rica y Gasto Publico. Unpublished. . 2000b. Un Modelo para Determinar las Causas de la Deserci6n Escolar en Costa Rica. San Jose, Costa Rica. Unpublished. Inter-American Development Bank (LADB). Economic and Social Progress in Latin America, 1996 Report. Making Social Services Work. Special Edition John Hopkins University Press. Washington, D. C. 65 Instituto Nacional de Estadistica y Censos (INEC). 2001. Census Information. http://www. inec.go.cr Instituto Mixto de Ayuda Social (IMAS). 2000a. Avances y Perspectivas del SIPO/SAP. Proyecto de Modernizaci6n del Programa de Equidad en Educaci6n. . 2000b. Descripci6n o Perfil de Tres Poblaciones: Nacional, S.I.P.O. y Programa Superemonos del Afio 2000, Con Base en el Modelo de Puntaje SIPO-2000, por Zona Urbana y Rural. PNUD. Costa Rica. . 2000c. Actualizaci6n del Modelo Estadistico para el Calculo del Puntaje: Construcci6n del Modelo SIPO-2000. Versi6n 2. Preliminar. Costa Rica. . 2000d. Memoria del Seminario "Evaluaci6n del Plan de Solidaridad". Proyecto Agenda Nacional de Superaci6n de la Pobreza. Doc.2000 #3.1. . 1999. Manual para la Aplicaci6n de la Ficha de Informaci6n Social (FIS). Asesoria en Planeamiento y Desarrollo Institucional. Sistema de Informaci6n de la Poblaci6n Objetivo (SIPO). Karoly, Lynn A., Peter Greenwood et al. 1998. Investing in Our Children: What We Know and Don't Know About the Costs and Benefits of Early Childhood Interventions. Rand Corporation. Sauma, Pablo, 2001. Aportes Para Una Estrategia de Reducci6n de la Pobreza en Costa Rica. Economia y Sociedad, enero-abril del 2001. San Jose, Costa Rica. Trejos, Juan Diego. 2001. Elementos de una Estrategia Nacional para la Superaci6n de la Pobreza en Costa Rica. Proyecto Agenda Nacional de Superaci6n de la Pobreza (MTSS/PNUD/COS/97/G51). Doc.2001 #4.1. . 2000. La Evoluci6n de la Pobreza en Costa Rica durante el Decenio de los Noventa. San Jose, Costa Rica. Instituto de Investigaciones de la Universidad de Costa Rica. Unpublished. UNDP. 2000. Resumen de Avances de Proyecto y Sugerencias de Disposiciones Administrativas. Proyecto de Modernizaci6n del Programa de Equidad en Educaci6n. San Jose, Costa Rica. World Bank. 2000a. World Development Report 2000/2001: Attacking Poverty. New York, United States of America: Oxford University Press, Inc.. 2000b. Brazil: Selected Issues in Social Protection. Draft, Report No. 20054- BR. 2000c. The World Development Indicators, 2000. 66 ANNEX 1 SIPO, HOUSEHOLD SURVEYS AND THE 2000 CENSUS: SOME METHODOLOGICAL CONSIDERATIONS 1. The SIPO database has been assembled by IMAS in 1999-2001 on the basis of a census of poor areas-urban and rural--- previously identified by NBI-based poverty maps made by the Statistical Institute (INEC). By mid-2001 the database contained 207,000 families or 808,500 persons. The SIPO data base contains information on socio- demographic and labor force characteristics including age, education, kinship relations among household members, civil status, activity, work type, money income, among others, housing characteristics, and coverage of many social programs provided by government agencies. These include education, health, social security-contributory and non- contributory pensions-housing programs, and many of the social welfare programs. This information is key to identify gaps in service provision by different poverty groups, and to quantify the approximate extent of those service gaps. Thus, with SIPO it is possible to combine information on family demographic and labor force characteristics, and the effective access to most social services provided by the government. This is not possible with the annual household surveys made by the Institute of Statistics (INEC). 2. It is not known, however, how representative of the poor and extreme poor the SIPO information is, and if the database includes most of the poor. It can be presumed that some biases in SIPO may be present. First, people may declare a worse situation than is real, if they anticipate that by appearing in a poorer situation, they could get some of the benefits available from IMAS or other institution. Second, some of the very poor residing in far away rural dispersed areas may be under-represented in SIPO because they are harder to reach and include in the data base. In fact, gross estimates by IMAS (based on information provided by local offices) indicate that by end 2001, between 5-10 percent of the poor population is not included in SIPO. 3. While most of the definitions of demographic and labor force variables are the same in SIPO as in the household surveys, there are marked differences in some key variables, such as rural-urban distinction and the unit of analysis in the two sources of data. Regarding the rural-urban distinction, NEC's household surveys classify the area denominatedperiphery urban as a rural area, while SIPO classifies it as an urban area. This will, thus, lead to a greater proportion of the population classified as urban in SIPO than in household surveys, ceteris paribus. 4. Regarding the unit of analysis, while household surveys have as a unit of analysis the household, SIPO distinguishes households andfamilies. In most of the analysis with SIPO database we use families rather than households because families are the object of many of the social programs. Households differ from families in that the former are a group of people or families that share a house and cooing facilities, while families are joined by blood or kinship ties. A household can be composed of one or more families. The difference in unit of analysis has implications for calculations of several demographic and socio-economic variables. For instance, housing ownership appears higher in 67 household surveys than in SIPO because all families in a household share the same characteristics of the household head, while in SIPO the ownership of the house is recorded to the family who owns it. So, a household with, say, three families will appear owning a house if any one of them own the house. If the unit of analysis were the family, only one out of the three own a house. 5. In the absence of a scientific analysis of statistical representation of SIPO, we propose to assess if the poor in SIPO resemble the poor in household surveys by comparing demographic and labor force participation variables. For these comparisons we use poverty profiles made by Trejos (2001) using household survey information for 1997, with those made using SIPO. Also, we compare several variables such as age, education, possession of durables and availability of basic services of the Costa Rican population, as reported in the 2000 population census, with those reported in SIPO. The purpose is to observe if the population in SIPO has generally worse indicators than the general population, as expected. These comparisons are in Tables 26 and 28, below. 6. Based on this preliminary analysis, we found that the poor in SIPO resemble the poor in the 1997 household survey, and are much poorer than the general population in the 2000 Census. Some of the differences that exist in some variables between SIPO and the household survey, such as family size, dependency ratios, among others, can be attributed to the fact that figures from household surveys refer to households while those of SIPO refer to families. 7. Thus, on this basis we decided to use the SIPO database to undertake a more in- depth poverty profile (Chapter 3) than is possible with household surveys, and to assess the social service coverage gaps-or unmet basic needs-of the extreme poor and poor populations. Even though we have used SIPO to calculate numbers of families uncovered by each one of the social services considered, we caution that these are crude estimates and serve only for raising awareness about the possible magnitudes of people that are not covered by essential public services in Costa Rica. Also, some of the gaps started to be covered in 1999-2000 and are most probably not reflected in the SIPO database which was built in those years. 1. Comparisons of Poverty Profile Using Household Survey for 1997 and the Poor in SIPO Data Base 8. In comparing the characteristics of the poor in the 1997 household survey with those of the poor in SIPO database, the following conclusions can be drawn. 9. First, there are large differences in the proportion of the poor residing in rural areas. In SIPO, the majority of the poor reside in urban areas while the contrary occurs in the household survey. This may be explained by the difference in urban-rural definitions in the two information sources, mentioned earlier, but this may be something that IMAS should investigate more deeply, since it can also indicate that some rural areas may not be adequately covered in the SIPO data base and, thus, in social programs. It is well known that systems such as SIPO, may be biased towards urban areas because it is easier (cheaper) to cover them, and the cadastral information is usually updated in urban areas. 10. Second, the distribution of the number of persons working in the family is somewhat worse for the poor according to SIPO than according to the household survey. While about 23 percent of households in the survey had no-one working in the family, that 68 proportion was about 29 percent in SIPO. And, households where only one person was working represented 56 percent according to the survey, while they represented 59 percent in the poor in SIPO.40 11. Third, the number of persons per family, and most of the other demographic variables (except the number of people younger than 12 years of age), are lower for the poor in the household survey than in SIPO. This may be explained by the fact that the survey considers households, while in SIPO we derive the statistics forfamilies. Households have larger numbers of people because they include one or more families. At the aggregate level of all people in SIPO, while the number of people perfamily is 4. 1, the number of people per household is 4.9. 12. Fourth, the distribution of activity variables of family heads are very similar for the poor in the two sources of information. The proportion of poor heads who are unemployed appears to be higher in the survey (5.8 percent) than in SIPO (3.8 percent), but this can be explained by the fact unemployment was, in general, lower in 2000 than in 1997. 13. Fifth, there is a large difference in the percentage of women heads in SIPO (41 percent), as compared with those in the household survey (27 percent). The percentage of poor women heads in SIPO is similar to the percentage of women heads in extreme poverty reported in the 1999 household survey (39 percent).41 We believe that the 1997 figure may be too low and may reflect the fact that they refer to household heads and not family heads -there are much more families than households in Costa Rica. 14. Sixth, while poor family heads are younger in SIPO than in the survey, the education of the heads, expressed in years of education, is strikingly similar in the two different information sources. 15. Seventh, while the labor force participation rates of both males and females heads are similar in the household survey and in SIPO, there are marked differences in the two information sources in the labor force participation rates of non-heads for both sexes. These latter rates are over ten points lower in SIPO than in the household survey. Also, unemployment rates for heads are similar in the two information sources for both males and females, but unemployment rates differ markedly for non-heads. Although unemployment rates are still very high (over 12 percent in 2000), unemployment rates for non-heads are much lower in SIPO than in the household survey. This may indicate either that unemployment has sharply declined between 1997 and 2000 for non-heads, or that there may be large sampling errors in the survey informnation, as the number of persons used for calculations in the different cells is reduced with greater desegregations of the survey data. 16. In summary, we conclude that while there are some differences in the characteristics of the poor according to the 1997 household survey and SIPO, most of these differences may be explained by the fact that the survey uses as a unit of analysis households while in SIPO we use families. These differences, however, need to be further investigated by IMAS, in carefully designed statistical analysis, to address the question of how representative is SIPO, and to make sure that SIPO is not leaving out of the data base poor populations residing in far away and difficult to reach areas. 40 We were not able to make statistical tests of difference in means of this and other variables because we did not have all the necessary information in the published data using the household survey. 41 See GOCR (2000a). 69 Table Al: Contribution to Poverty by Group, According to Household Survey for 1997 and SIPO for 2000 Household Survey SIPO Region 100.0 100.0 Urban 34.6 52.3 Rural 65.4 47.7 Persons Working, In Family 100.0 100.0 None 22.9 28.7 1 56.2 58.9 2 15.1 9.9 3 4.2 1.9 4 and more 1.6 0.6 Income Earners, in Family' 100.0 100.0 None 16.3 8.7 1 60.7 73.3 2 16.7 14.6 3 4.5 2.7 4and more 1.7 0.7 Head's Activity 100.0 100.0 Unemployed2 4.0 3.8 Inactive (not in labor force) 30.7 32.5 Occupied3 65.3 63.7 Head's Industry 100.0 - Agriculture 29.8 - Non-agriculture 35.5 - Micro enterprise 19.8 - Domestic service 2.7 - Self-employed 13.0 - . Includes employed, pensioners and renters. 2 This is not the open unemployment rate, but the share of unemployed in total number of heads. 3. This is the share of occupied people in the total number of heads. Source: Trejos, 2001. Based on Household Survey for 1997, and author's tabulations from SIPO database. Table A2: Characteristics of Poor and Non - Poor, According to Household Survey for 1997 and SIPO Database for 2000 Total Non Poor'-= Poor - : -Household Survey SUPO Demographics Persons per Family 4.1 3.7 4.7 4.0 Persons in Working Age 3.0 2.8 3.0 2.3 Persons Active 1.6 1.6 1.2 1.0 Persons < 12 Years of Age 1.1 0.9 1.7 1.6 Working People per Family 1.5 1.5 1.1 0.9 Dependants per Worker 1.7 1.5 3.4 2.1 Characteristics Heads Percent Women Heads (%/6) 20.7 19.6 27.1 41.0 Head's Age (Years) 45.9 46.0 48.6 42.6 Head's Education (Years) 6.9 8.3 4.7 4.8 '.The characteristics of this group are those of the non-poor according to NBI and income criteria (Trejos, 2001). Source: Trejos (2001) and authors' calculations, and tabulations from SIPO Data Base. 70 Table A3: Labor Force Participation and Unemployment Rates of Heads and Non-Heads, According to Household Survey for 1997 and SIPO Database for 2000 Description - National Non-Poor' : Poor - . -. - .- 'Household Suney - SIPO- Labor Force Participation Heads 79.6 80.3 69.3 67.5 -Male 86.9 86.3 80.4 84.7 -Female 51.7 55.4 39.5 42.8 Non Heads 39.8 41.6 26.4 17.7 -Male 59.7 56.8 48.2 33.6 -Female 29.4 34.7 16.3 8.8 Unemployment rate (total)2 5.7 3.7 13.2 8.0 Heads 2.0 1.1 5.8 5.7 -Male 1.6 0.8 5.0 5.2 -Female 4.4 3.1 9.9 7.0 Non Heads 9.5 6.5 23.0 13.8 -Male 10.0 7.7 22.3 14.5 -Female 9.0 5.6 23.9 12.4 The characteristics of this group are those of to the non-poor according to NBI and income criteria (Trejos (2001). Source: Trejos (2001) and author's calculations with SIPO Data Base. 2. SIPO and the 2000 Population Census 17. As seen in Table 29, the main results of this comparison are as follows. First, the percentage of income poor in SIPO database is about 18 of the total population. Extreme poor are about 10 percent and poor about 8 percent. These percentages are somewhat different from those reported in the 1999 household survey of about 6.6 percent for extreme poverty, and about 15 percent for poverty (GOCR, 2000a). Second, the percentage of poor people residing in rural areas is much larger that the percentage of the general population, which was to be expected. Third, the poor in SIPO have a much larger proportion of children 0-14 years of age (48 percent) than does the general population (32 percent), and a much lower proportion of working age population. As a result, the dependency rate of the poor in SIPO is 1.10 as compared with that of the general population (0.60).42 18. Fourth, net enrollment rates are much lower in SIPO for all education levels than in the Census for the general population.43 The main differences between the poor in SIPO and the general population appear in the pre-primary education age group, 5-6 years of age (54.5 percent vs. 64.6 percent), secondary education age group, 13-19 years of age (53.6 percent vs. 61.3 percent), and higher education. Similar results are obtained for educational attainment of the population over 5 years of age. While about 53 percent of the general population had obtained only primary education level, about 71 percent of the poor in SIPO had done so. Only less than 15 percent of the poor in SIPO had attained secondary education level, as compared to about 25 percent in the general population. 42 Dependency rate is defined as the population 0-15 years of age plus the older than 65 years of age divided by the working age population 15-64 years of age. 4 For enrollment in SIPO, we use the variable "Asistencia a Centros de Atenci6n de Enseflanza". Variable No. 48 by age group. We include only attendance to regular education to be comparable with the published census information-Table 14 (INEC, 2001). 71 Fifth, the percentage of families with durables is over 25 percentage points lower for the poor in SIPO than for the general population, for all major ticket items such as color TVs, refrigerators and washing machines. Finally, access to basic services of electricity and running water are much lower for the poor in SIPO than for the general population, although the percentages covered with those services are fairly high, at over 80 percent of families. Table A4: Comparisons of the Poor in SIPO with the General Population According to the 2000 Census - National Census 2000, . SIPO - Poor'2000 Total Population 3'810.179 680,439 Urban (%) 59.0 52.3 Rural (%) 41.0 47.7 Total poor (%) -- 18.0 Extreme poor (%) 10.0 Age Group 0-14 years 32.0 48.4 15-34 years 34.9 27.8 35-64 years 27.6 19.7 65 and more 5.6 4.2 Enrolled in Education Pre-primary. 5-6 Years Old 64.6 54.5 Primary. 7-12 Years Old 95.7 94.8 Secondary. 13-19 Years Old 61.3 53.6 Higher. 20-29 Years Old 22.8 6.4 30 and more 4.6 0.7 Education level (Age 5 & more) Primary 52.8 70.8 Secondary 25.1 14.8 Higher 11.7 0.8 Possession Durables Color TV 84.9 57.1 Refrigerator 84.3 58.2 Washing machine 80.8 53.9 Access to basic services Electricity 96.8 86.1 Sanitary 89.5 -- Water from aqueduct 89.3 83.3 Source: INEC, 2001 and author's calculations based on SIPO. 19. In summary, all indicators point to the conclusion that the poor in SIPO are really much poorer than the general population of Costa Rica in the 2000 Census, as expected. 72