Report No. 40084-UY Uruguay Income Transfer Policies in Uruguay Closing the Gaps to Increase Welfare October 18, 2007 Argentina, Chile, Uruguay Country Management Unit Human Development Department Latin America and the Caribbean Region Document of the World Bank TABLE OF CONTENTS Acknowledgements............................................................................................................ vi Executive Summary.......................................................................................................... vii 1. Introduction..................................................................................................................1 2. Why Income Transfers in Uruguay?............................................................................4 2.1. Social Risks, social protection and income transfers.......................................... 4 2.2. Income transfers and the fragmented welfare state............................................. 6 2.3. Protection versus Incentives: implications and trade-offs in the labor market... 7 2.4. Income Transfers in Uruguay: A blurred line between Social Insurance and Social Assistance ............................................................................................................ 9 2.5. Income transfers as indirect policy tools: CCTs, "Contraprestaciones" and others 11 2.6. Income transfer programs in Uruguay: the rationale and tools......................... 13 3. The current income transfers policies........................................................................17 3.1. The origins ........................................................................................................ 17 3.2. The "Contributory" programs........................................................................... 18 3.3. "Non-contributory" programs........................................................................... 20 3.4. The institutional framework of income transfer programs ............................... 21 3.5. Coverage ........................................................................................................... 23 3.5.1 Coverage of active workers in contributory programs ............................. 23 3.5.2 How many beneficiaries?.......................................................................... 25 3.5.3 Coverage gaps and overlaps...................................................................... 32 3.6. Benefit amounts and welfare impacts............................................................... 37 3.7. Fiscal costs and Sustainability .......................................................................... 42 3.7.1 Social Public Expenditures ....................................................................... 42 Financing income transfers....................................................................................... 46 3.7.3 Sustainability.................................................................................................... 47 4. The new "plan de equidad social". Potential Impacts and costs................................49 4.1. The impact of Citizen Income........................................................................... 50 4.2. Changing the structure of the FA benefits........................................................ 52 4.3. Extending the FA coverage and non-contributory Pensions............................. 54 4.4. Comparing alternative targeting criteria........................................................... 56 5. Conclusions and Looking forward: The main challenges for income transfer policies...................................................................................................................63 5.1. Coverage gaps................................................................................................... 64 5.2. Fiscal and Social Sustainability ........................................................................ 66 5.3. System Integration ............................................................................................ 67 6. References..................................................................................................................70 iii Boxes Box 1. Welfare Reform in OECD Countries................................................................... 8 Box 2. The Rationale for Conditioning Transfers in Uruguay...................................... 13 Box 3. The 1996 Pension Reform ................................................................................. 20 Box 4. Estimating Social Public Expenditures in Uruguay........................................... 45 Tables Table 1. Social Risks and Income Transfer Programs, by Age Group............................ 14 Table 2. Overlaps of Income Transfer Programs, by Income Quintile and Number of Programs.......................................................................................... 36 Table 3. Average Benefit Amounts of Income Transfer Programs, late 2006................. 37 Table 4. Income Transfers Incidence on Poverty Measures in 2006............................... 42 Table 5. Targeting Criteria: Children under 18 Years Old by Participation in the FA Program and Poverty.............................................................................. 58 Table 6. Targeting Criterion: Individuals Older than 65 years by Perception of Pensions and Poverty..................................................................................... 58 Table 7. Simulation Scenarios: Impacts on Coverage, Social Indicators and Costs................................................................................................................... 60 Table 8. Additional Simulation: Impact of Further Increases in Family Allowances Benefits on Poverty and Spending................................................. 61 Table 9. An Integrated Income Transfers System to Provide Adequate Coverage to All.................................................................................................. 66 Figures Figure 1.Poverty, Extreme Poverty and Gini Coefficient in Uruguay, 1991- 2006...................................................................................................................... 1 Figure 2.Contributors to BPS Pension System, 1988-2006 ............................................. 24 Figure 3.Occupied Labor Force Formality Rate, 2001-2006........................................... 24 Figure 4.Labor Force Formality for Latin American Countries, mid 2000s.................... 25 Figure 5.BPS Pension System Coverage, 1985-2006 ...................................................... 26 Figure 6.Pension Systems Coverage, 2001-2006, by Type of Benefit............................. 26 Figure 7.Pension System Beneficiaries, by Income Quintile (2006) ............................... 27 Figure 8.Unemployment Insurance Beneficiaries............................................................ 27 Figure 9.Unemployed Insurance Coverage...................................................................... 28 Figure 10. Unemployed Insurance Coverage, by Quintile of Per Capita Income (2006).................................................................................................... 29 Figure 11. Family Allowances Beneficiaries (as of December of each year), 1991-2006 .......................................................................................................... 29 Figure 12. Family Allowances System Coverage, by Program Type, 2001- 2006.................................................................................................................... 30 iv Figure 13. Family Allowances Beneficiaries Distribution by Quintile of Income per Capita and Type of Benefit............................................................. 30 Figure 14. Citizen Income Applications and Coverage by Poverty Status, 2006.................................................................................................................... 31 Figure 15. Citizen Income Beneficiaries Distribution by Quintile of Income per Capita........................................................................................................... 32 Figure 16. Gaps and Overlaps of Income Transfer Programs 2001-2006........................ 33 Figure 17. Gaps of Income Transfer Programs 2001-2006, by Income Quintile .............................................................................................................. 34 Figure 18. Overlaps of Income Transfer Programs 2001-2006, by Income Quintile .............................................................................................................. 34 Figure 19. Households with Citizen Income Benefits, by Type of Overlap with Other Programs, 2006................................................................................ 35 Figure 20. Comparison with LAC Programs: Distribution of Beneficiaries by Income Quintile for Non-contributory Family Allowances and Citizen Income................................................................................................... 36 Figure 21. Income Transfers Value, 1991-2006 (as a percentage of poverty line) .................................................................................................................... 38 Figure 22. Distribution of Income Per Capita­with and without non- contributory transfers­2006............................................................................... 39 Figure 23. Concentration Curves for Non-Contributory Transfers in 2006..................... 40 Figure 24. Concentration Curves for Contributory Transfers in 2006............................. 41 Figure 25. Public Social Expenditures, by Country, average 1990-2000 ........................ 43 Figure 26. Public Social Spending, by Category, 1985-2005 .......................................... 43 Figure 27. Income Transfer Expenditures, by Program, 1990-2006................................ 45 Figure 28. Financing of BPS Income Transfer Programs, 1991-2006............................. 47 Figure 29. Projected Financial Flows: BPS Pension System 2010-2050......................... 48 Figure 30. Simulation Results: Actual Data for 2006 and Simulated Results without IC Transfer............................................................................................ 51 Figure 31. Simulation Results: Scenarios without IC and with New Decreasing FA Benefit per Child....................................................................... 53 Figure 32. Simulation Results: Scenarios with FA Benefit per Child and Additional Benefit by Years of Schooling......................................................... 54 Figure 33. Simulation results: Scenarios with Revised FA Benefit Structure and Extending Coverage to All Qualifying Children ........................................ 55 Figure 34. Simulation Results: Scenarios with Extended Family Allowances Coverage and Extended FA and Non-Contributory Pensions Coverage ............................................................................................................ 56 Figure 35. Simulation Results: Scenarios with Extended Family Allowances Coverage and Extended FA and Non-Contributory Pensions Coverage ............................................................................................................ 59 v Acknowledgments This report was prepared by the Social Protection Unit, Human Development Department for Latin America and the Caribbean of the World Bank. The work was led by Rafael Rofman (LCSHS) and the team included Juan Martin Moreno (LCSHS), Evelyn Vezza (LCSHS), and Gaston Blanco (LCSHS), with the permanent support of Sarah Ruth Bailey (who edited this document originals in English and Spanish) and Febe Mackey in the Buenos Aires office and Sylvia Albela, Maria Ines Ferres, and Valeria Bolla in Montevideo. While working in a parallel project, William Reuben (LCSSO) provided solid advice and comments. Several consultants, including Andrea Vigorito, Rodrigo Arim, Marisa Bucheli, Veronica Amarante, and Esther Hennchen collaborated preparing background reports and participating in many discussions. Jesko Hentschel, from his position as Human Development Sector Leader for the Southern Cone countries, provided a critical contribution during the whole process, generating ideas, discussing drafts, and supporting the team, professionally and personally. Officials and technical staff of several Government agencies in Uruguay provided invaluable support during the preparation of this report, discussing ideas, providing access to data, and participating in many meetings and electronic exchanges. Among those, Minister Marina Arismendi (MIDES), Vice-ministers Jorge Bruni (MTSS) and Ana Olivera (MIDES), Ernesto Murro (President, BPS), Alicia Melgar (Director, INE), and their respective staff were particularly helpful. The team worked under the guidance and supervision of Helena Ribe (Sector Manager, LCSHS), Jesko S. Hentschel (Country Sector Leader, LCSHD), David Yuravlivker (WB Representative for Uruguay), and Axel van Trotsenburg (Country Director, LCC7). The peer reviewers were Martin Rama (EASPR), Ana Revenga (EASHD) and Emil Tesliuc (HDNSP), who offered many useful comments. vi Executive Summary 1. As Uruguay emerges from the serious social crisis of the late 1990s and early 2000s, the attention of policy makers and analysts has progressively shifted from providing emergency responses to redesigning the country's income transfer policies to make them more consistent, effective and sustainable. The purpose of this report is to contribute to the policy dialogue currently underway in Uruguay regarding the design and operation of these income transfer policies. While the report summarizes the Bank team's findings and analyses, much of the work for this study has already been shared with Uruguay's authorities, through many instances of formal and informal dialogue. The team has supported and participated in public and closed discussions about the performance of current programs and policy options for the future, and these are reflected in the report. 2. This report considers the rationale, coverage, impact, and costs of today's four main income transfer programs, including the pension system, unemployment insurance, family allowances, and citizen income. Analyzing the coverage and impacts of these programs with microdata from the first semester of 2006 national household survey, the report identifies significant coverage gaps and program overlaps, discusses the underlying rationale that guides their design, and simulates the impacts that several alternative strategies could have on poverty, equity, and financial sustainability of these programs. 3. The main goal of this report is to present a detailed discussion of policy options and implications in the area of income transfers, aiming at facilitating a better-informed policy making process in Uruguay. As authorities in Uruguay are preparing a comprehensive policy proposal to introduce a new program to alleviate poverty and improve the living standards of many Uruguayans in a sustainable way, this report focuses on this debate. However, the report also acknowledges that the overall reform agenda includes other transfer programs, such as unemployment insurance and pensions, presenting the links and interactions between the two areas. 4. Social and equity policies in Uruguay go beyond income transfers programs. On the one hand, there are social programs such as PANES, and the future Social Equity Plan that includes non-monetary components that are equally as relevant. As well as these, other policies, such as the universal health insurance or the recently implemented tributary reform also have an important impact on income distribution in Uruguay. Although this report concentrates on transfer programs, this does not mean that the relevance of those other actions mentioned is ignored. 5. The report discusses the performance of Uruguay's income transfer programs against four basic criteria: coverage, impact, sustainability and (across all of them) consistency. Effective income transfer programs are expected to offer protection to most individuals and families exposed to social risks and unable to protect themselves, providing transfers that suffice to alleviate significantly the impact of social shocks within a fiscal and social sustainable framework. Equally important, these programs must be consistent, by ensuring that their design, targeting, and institutional structures do not present unintended overlaps that would result in inefficiencies, or gaps, vii leaving some vulnerable groups unprotected. The programs should also be consistent with labor market development, aiming at promoting beneficiaries' self-sufficiency by providing incentives and supporting access to formal jobs. 6. Analytical approaches to income transfer programs include two alternative views. Traditionally, analysts have studied Social Insurance schemes (or "contributory") and Social Assistance schemes (or "non contributory") separately. Programs are classified in one of these categories considering differences in aspects such as financing schemes, target populations, and policy goals. More recently, some analysts have preferred to consider them as part of a single system, considering that they have a common core rationale, under which Government mandate or provide transfers to individuals or households to compensate for the impact of social risks. 7. In Uruguay, differences between Social Insurance and Social Assistance are not so clear, since social insurance, contributory programs are only partially financed by contributions. Many current beneficiaries of these programs have not contributed consistently in the past and some current contributors will not receive benefits in the future. Since most social insurance and social assistance financing come from the same source (general taxes), and the contributory aspect of traditional social insurance programs is rather weak, this report adopts a comprehensive approach, considering all income transfer programs as part of a single system. 8. Considering income transfer programs as part of a single system does not mean that they should be one single scheme with one set of goals, financial arrangements, target populations, or management arrangements. Since different groups in the population are affected by different risks, it is reasonable to tailor income transfer programs to their needs. However, it is important to maintain an integral approach to these policies, to ensure that no undesired coverage gaps or overlaps occur, or that programs that should complement each other end up competing to protect the same population. - The current situation 9. Uruguay's income transfer programs have been traditionally organized around the labor market, in the form of social insurance schemes that provide income substitution or complement for formal workers. Pension schemes were first introduced during the XIX century and slowly expanded to include most workers. Other transfer programs, such as family allowances (introduced in the early 1940s) and unemployment insurance, expanded and complemented the social protection system for formal workers and their families. 10. These programs have provided ample coverage to most formal workers and their families, at a relatively high cost. Nearly 85 percent of the population aged 65 and more receive a pension in Uruguay. This high coverage, together with the older demographic profile of the population, results in costs that have been difficult to finance. Pension expenditures in Uruguay reached, in the mid 1990s, a maximum of around 14 percent of GDP, seriously limiting the fiscal space for other social interventions. 11. As fiscal pressures and declining labor market formality made evident that the traditional approach was becoming increasingly insufficient to protect all viii Uruguayans from different social risks, authorities began to slowly shift the income transfers strategy towards a more targeted approach, beginning in the mid 1990s. The first actions in this direction were the introduction of a targeting criterion for contributory family allowances in 1995, and the reduction of public exposure to fiscal risks produced by the pension system, an effect of the 1996 pension reform, at the same time that coverage of the elderly stagnated and began to slowly decline. Since the economic crisis had a clear impact on social indicators and, at the same time, limited fiscal resources, the trend accelerated and new, non-contributory schemes with significant coverage and budget were introduced to protect those excluded from the older schemes. Non-contributory old age and disability pensions have provided a basic income to a small number of beneficiaries for decades, but larger programs targeted to low income younger workers and families were first created less than ten years ago. First, in late 1999 households with no formal workers were given limited access to the family allowances program, a measure that was then expanded in 2004. More importantly, in early 2005 an emergency social plan ("PANES") was introduced, aiming at ameliorating the social impacts of the economic crisis that affected Uruguay in the early 2000s. The main (but not only) component of this plan was an income transfer program called "Citizen Income" (Ingreso Ciudadano - IC), which was rapidly implemented and provided supplementary income to nearly 9 percent of Uruguay's households, including more than half of those living in extreme poverty. 12. The current income transfer programs in Uruguay are managed by a complex set of institutions. Several ministries, government agencies, independent institutions and commercial firms participate in the design, operation, and supervision of the different income transfer programs. There are several asymmetries among these institutions, regarding their level of financial and political independence, their internal organization and their capability to deliver what is expected from them. Therefore, some conflicts are unavoidable, even within a climate of collaboration, as has been the case in recent years. This report recognizes the advances made in inter-institutional collaboration, including the work of the Inter-institutional Social Security Commission (which has recently begun a National Dialogue process) or the activity of the National Social Policy Coordinating Council. 13. The family allowances reform of 2004 and the introduction of the social emergency program PANES provided important tools to reduce the coverage gaps of income transfer programs in Uruguay. While until the early 2000s nearly fifty percent of households of the poorest quintile had no access to these policies, the proportion began to decline when the Government approved a family allowances system reform in 2004, and accelerated when the IC program was implemented. By early 2006, the coverage gap in the first quintile was below 25 percent. This improvement also affected households in the second quintile of the income distribution, but had no effect on the richest households, indicating that the new programs were well targeted. ix Gaps of Income Transfer Programs 2001-2006, By Income Quintile Percentage of households with no access to income transfer programs ("gaps") 60 2001 '02 '03 '04 '05 2006 50 40 e ga ntecr 30 pe 20 10 0 I II III IV V Source: Bank's staff calculations based on ECH and ECHA. 14. The efficacy shown by the new programs (in particular, IC) to reach many of the poorest households in Uruguay suggests that IC has been one of the most successful income transfer programs in terms of targeting effectiveness, if compared with the experience in other countries. More than 75 percent of beneficiary households belong to the poorest quintile of the population. This percentage is much higher than what was achieved in other programs in Latin America and around the world, especially when considering that IC was designed and implemented in a very short time. 15. On the other hand, IC was implemented on top of previously existing programs, resulting in an important level of overlap between income transfer schemes. Eighty percent of IC beneficiaries are also recipients of other income transfer benefits, in most cases (74 percent) family allowances. This overlap is legal, but might indicate that there is an efficiency problem in terms of design and management of different programs with similar goals and targeted populations. Households with Citizen Income Benefits, By Type of Overlap with Other Programs 2006 (in percent) CI + others; OnlyCI; 5,9% 20,1% CI + FA; 73,9% Source: Bank's staff calculations based on ECHA. x 16. While family allowances and citizen income are clearly the best targeted programs, their impact on poverty is small, due to their benefit amounts. If non- contributory family allowances were to be immediately discontinued, the percentage of poor Uruguayans would grow by 0.8 percentage points (assuming no behavioral responses), and in the case of IC the impact would be 1.2 percentage points. On the other hand, the national pension system has had a much larger impact, as it explains a difference of almost 12 percentage points on poverty incidence, even if the program is not directly targeted to the poor. The impacts on equity are similar, and only when considering extreme poverty incidence and gaps does the IC program become relevant. Since its inception, the Government indicated that PANES was a temporary program, aimed at overcoming the social emergency. It was set to last two years, until mid 2007. As this deadline approaches, it has become clear that a new, more integrated approach to income transfer policies is desirable in Uruguay, to close coverage gaps, limit overlaps, maximize social impacts and ensure fiscal sustainability in the short and medium term. 17. Having understood these problems, the Government in Uruguay has focused its efforts on the design of a new income transfer program, as part of a wider social policy called Plan de Equidad Social (PES). The new program would improve the performance of the current programs while ensuring its fiscal sustainability. While the details of this new program are being defined and announcements are expected for the second semester of 2007, a few characteristics have been discussed informally and in media reports. 18. The new program could operate as a reformed family allowances scheme, reducing the overlaps problem. The program would not only focus on the extreme poor, but would also aim at reducing overall inequity. To achieve this, family allowances benefits would significantly increase for most beneficiaries to maximize their impact on welfare and make them more attractive to potential beneficiaries. Benefits would continue to be assigned to children, but reduced as the number of children in the household increase, to recognize the role of economies of scale within the families. Also, benefits would grow as children accumulate years of schooling, with the goal of providing incentives to complete secondary education. 19. In addition, the program would aim to expand coverage to all children that qualify to receive benefits. Nearly 125,000 children that should be receiving the benefit were not registered as of early 2007. While enrolling all these children may prove difficult, there is some clear space for improvement. Finally, a possible component of the reform could be an expansion of coverage of the non-contributory pension scheme, to include those who already qualify for it and, possible, anticipate access from the current 70 years of age to 65 years old. 20. Simulating the impacts of these proposals, the report estimates that, if all are successfully applied, poverty would decline by three percentage points, while extreme poverty could be reduced by more than 50 percent. The report presents results of simulations for the proposed reforms, showing the impact that they could have on extreme poverty, poverty, income distribution, coverage gap for the poorest quintile and total spending. Results are presented in a cumulative way, as each component of the proposed reform is added in steps to the scenarios. The expected impact of PES on income distribution would still be small. The coverage gap for the first quintile would xi decline by nearly 50 percent, thanks to the expansion of family allowances benefits to all qualifying children. 21. To fully implement the new PES, additional fiscal resources would be necessary. Total expenditures in PES, implemented as simulated for this report, would amount to between twice and thrice the amount currently spent on IC benefits. Additional expenditures would be between 0.2 and 0.8 percent of GDP. As seen in the figure, this would represent a relatively small proportion of the current expenditure on income transfers, at approximately 2-8 percent. While this fiscal effort should be manageable over time, the already limited space to increase spending in these programs could make it unfeasible or force authorities to limit its coverage in the short term, reducing its effectiveness. Simulated impact of PES Proportional Changes in Extreme Poverty, Poverty, Poverty Gap, Gini coefficient, Total Fiscal Expenditures on Income Transfers, and Poverty and Coverage Gap for the First Quintile Actual Extreme poverty 1.5 PES 1 Annual cost Poverty 0.5 0 Gini Poverty gap Coverage gap Source: Bank's staff calculations based on ECHA. - The way forward 22. After implementing a wide coverage program in response to the social emergency through PANES, the government is now moving towards a more strategic approach. PANES was a "pure endowment" program, designed and implemented in the emergency, very efficient to reach the poorest and reduce extreme poverty incidence, but with no strategy to tailor benefits according to households' needs or conditions. The Government is now adopting a life cycle approach, designing PES to focus on children and elderly needs. This is a welcomed move, as it will result in a more xii effective policy. However, additional reforms will be necessary to increase these policies' impact, since the proposed PES design has two limitations: a coverage gap for working age poor adults, and benefit amounts that are insufficient to eliminate poverty in many households. 23. The biggest obstacle to advance further in the income transfers system reform is fiscal: if, in addition to what was included in the simulations, PES were to include a component for working age adults and increase benefits, how could the Government finance these additional expenditures? There are clearly three options (collect additional taxes/contributions; redistribute resources currently spent on income transfer schemes; or redistribute resources currently spent on other areas). These alternatives have important political, economic, and social costs and restrictions, and their impacts should be carefully assessed before formulating any policy proposal. 24. Another aspect that would need to be considered in order to maximize effectiveness of these policies is the incentives structure generated by them. Income transfers reforms may affect the labor market and the financial structure of some of the programs. On the one hand, some individuals' preference to join the labor force may be affected by these programs, if the benefit amount is high enough to discourage them to look for a job, or they may become informal workers if the program does not allow to work while receiving benefits. On the other hand, different authors have indicated that when non-contributory programs offer benefits that are similar in level and quality to those of contributory programs, then some workers may prefer to become informal. While there is no empirical evidence showing that any of these effects are relevant in Uruguay, policy makers might want to remember that they are a possibility, and design the programs to limit their impact. 25. Finally, institutional arrangements could become an obstacle to successful implementation of PES and other reforms in the mid term. If the political consensus that facilitates collaboration among the different institutions is not sustained over time, a serious risk could emerge, affecting the overall effectiveness of the programs. Political authority to define strategies is spread among several of the institutions involved in the system, and some of them have their own internal political processes, which could result in internal and inter-institutional conflicts. 26. Once PES is designed and implemented, the next task for Uruguay's policy makers will be to consider what to do about the "contributory" income transfers, aiming at increasing their effectiveness (in particular, with regards to the unemployment insurance), creating fiscal space, and sustaining coverage. Problems such as coverage and effectiveness of the unemployment insurance, the expected decline of pension coverage in the medium term, or the overall fiscal sustainability of the income transfer policies could be the focus of future policy debates in Uruguay. Reform proposals to (i) expand access to unemployment insurance to most unemployed workers; (ii) increase transparency in the financing scheme of these programs, disentangling their contributory and non-contributory components; (iii) free some resources to reinforce the antipoverty components of the income transfer policies could be develop and implemented; and (iv) find sustainable strategies to revert the expected decline in pension coverage over the next few decades. This is the main challenge for the next years. xiii 27. The next few months will be critical for the future of Uruguay's income transfer policies. As the Government advances in its work to define the components and financing of PES, it will be essential to maximize the new program's impact while recognizing the existence of these remaining challenges and start addressing them. Income transfer policies have had an important role in Uruguay to provide poverty relief and improve income distribution during most of the twentieth century, and the main challenge for policy makers and Uruguay's society as a whole is to ensure that they continue to do so in an effective and sustainable way into the new century. xiv 1. INTRODUCTION 1. Income transfer policies were first introduced in Uruguay more than one hundred years ago, as the earliest forms of pension funds were created. Since then, the income transfer programs have evolved and expanded to protect most formal workers and their families. This arrangement was very effective for decades, limiting poverty and improving equity, but the combination of growing fiscal pressures and diminishing formality in the labor force resulted in growing problems to provide adequate social protection to all Uruguayans. 2. After decades of being one of the few countries in the region with high income equity and very low poverty incidence, Uruguay suffered a sustained deterioration of these indicators during the 1990s, and especially as a result of the economic crisis of the early 2000s. The Gini coefficient began to grow in 1993 and, nine years later, had gone from 0.41 to more than 0.45, as shown in Figure 1. Poverty and extreme poverty were stable during the 1990s, but also suffered a sharp increase, beginning in 2000, as both indexes grew by more than 100 percent in four years. In this context, the income transfer programs became central in the social policy strategy in Uruguay. They provided a tool to ameliorate the crisis impacts on the most vulnerable individuals and families, and to provide a structural social safety net for those who, even after the worst of the crisis passed, continue to live in difficult conditions in a post-crisis environment. Figure 1. Poverty, Extreme Poverty and Gini Coefficient in Uruguay, 1991-2006 35 0,47 Extreme Poverty Poverty Gini 30 0,46 25 0,45 roop 20 0,44Gi ni % 15 0,43 10 0,42 5 0,41 0 0,40 1991 92 93 94 95 96 97 98 99 2000 01 02 03 04 05 2006 Source: INE (2006) 3. Income transfer policies in Uruguay include traditional social insurance programs, unemployment and family allowances schemes and some recently created poverty alleviation schemes. This collection of programs is not fully articulated, as in many cases it is the result of accumulated reforms and expansions, leaving some space for coverage overlaps and gaps, both in terms of policy goals and population groups. 1 4. While some analysts prefer to consider social insurance and social assistance separately, this report studies them as part of a single system. The line between social insurance and social assistance programs in Uruguay is blurred, as some programs traditionally considered part of the former cannot be considered contributory in strict terms. Nearly 50 percent of pension expenditures and all unemployment insurance payments are financed by general taxes and treasury transfers. In addition, many contributors to social security will have no access to benefits, due to the strict qualification requirements. Thus, while this report recognizes differences in financing, targeted population, and program goals among different programs, it considers them as part of a system integrated by programs set by the State to transfer income to individuals and households to alleviate the impact of social shocks. 5. The national pension system represents the main component of Uruguay's social protection policies in terms of coverage and expenditures. The system was originally established more than 100 years ago, as a constellation of independent occupational contributory schemes, and slowly expanded to provide near universal coverage. It also included other benefits, such as family allowances and health and unemployment insurance. Until the mid 1990s, these contributory programs represented the bulk of the income transfer policies, both in terms of coverage, expenditures, and political relevance. While suffering from inefficiencies and other problems, these programs represented the core component of Uruguay's social policy strategy, having a strong impact on poverty and equity. Income transfer policies also provide tools to help restore social inclusion. Several of the effects of social exclusion (children in the streets, geographical segmentation of poverty, etc) need a more tailored and multidisciplinary approach. Nevertheless, the establishment of a safety net that helps to prevent vulnerable people from falling into poverty may act as an instrument to reduce exclusion. Income transfer programs can function as facilitators of more targeted and specific approaches, necessary to bring social, educational and cultural standards closer to what most Uruguayans expect. 6. As in other Latin American countries, sustained unemployment and, more importantly, prevalence of labor informality resulted in the progressive exclusion of a significant proportion of the population. The combination of fiscal pressures and declining coverage drove Uruguay's authorities towards a trend to target income transfer programs to the poorest and most vulnerable sector of the population. Eliminating family allowances for households earning more than 10 minimum wages (in 1995), and the pension reform of 1996 (which reduced the public sector's liabilities in the pension system) were two clear steps in this direction. More recently, this trend was deepened, as two new programs were introduced. In 1999, the government enacted a law granting family allowances to poor households, regardless of their labor market insertion, and further expanded this program in late 2004. 7. After taking office in early 2005, the new Government created a social emergency program, named PANES. The main component of PANES was an income transfer program, called Citizen Income ("Ingreso Ciudadano" ­ IC), and several other social promotion components were included in the plan. Being an emergency program, PANES was scheduled to operate for only two years. PANES will eventually reach an end at some point during 2007. However, it is highly unlikely that the income transfer 2 policies in Uruguay will then revert to the prevailing model before 2005. The social needs that motivated the intervention are still present, albeit less dramatically, and deep institutional and political changes produced by the introduction of PANES will be unlikely reverted. In late 2006, the Government announced that a new program (Plan de Equidad Social ­ PES) would be implemented. The precise contents of this "post- PANES" policy are yet to be known. 8. While income transfer policies in Uruguay have had a valuable impact, and recent reforms (including what is know about the future PES) seem to be moving in the right direction, there are some important challenges, in terms of coverage, fiscal and social sustainability, and system integration, that the current and future authorities will need to address. The main challenges in these three areas are discussed in the report, identifying risks and tools to manage them. One core message of the report is that income transfer programs need to be designed, financed, and managed in an integrated way, to increase effectiveness and sustainability, and to avoid problems originated in insufficient coordination among institutions. 9. Uruguay's Government is currently analyzing several reform strategies for the social sectors. These include a health reform, the introduction of a new poverty alleviation program, possible changes in pensions and unemployment insurance, and also a tax reform, that impacts the income transfer policies in several aspects. This report focuses on the income transfer policies and, among those, it discusses the possible impacts of the new Plan de Equidad Social. Other reforms are equally important, and the Bank has been contributing to their analysis and implementation through different instruments. In particular, the Government has recently initiated a National Dialogue to discuss possible changes in pensions and unemployment insurance, aiming at ensuring adequate coverage in a fiscally and socially sustainable way. 10. The purpose of this report is to contribute to the debate around the design of the income transfer policies for the medium and long term. Policy makers and experts are preparing a policy proposal to redesign the main poverty alleviation programs. The report offers a contribution to this process, considering these policies from several different perspectives. First, it discusses the underlying rationale behind these policies. Second, it provides a concise description of the currently existing programs and their impacts. Third, it assesses the potential impact of the new PES. Finally, it analyzes the relevance and implicit policy options of what are considered the main challenges for the programs in the medium and long term, this report represents a contribution to this process. 11. The report is divided in five sections. Section two presents the conceptual discussion; section three describes the current policies, including an assessment of coverage, impacts, and fiscal effects. The fourth section presents a simulation to assess the potential impacts of the new Plan de Equidad, and the fifth section discusses conclusions and the central challenges for the future. 3 2. WHY INCOME TRANSFERS IN URUGUAY? 12. Income transfer programs have been one of the main tools that governments used to implement their welfare policies in the last 150 years. While there are many differences in approach, nearly all countries around the world have some type of income transfer program. Differences across countries are wide, regarding coverage, design, impacts, costs, or institutional setting. Some programs are targeted to specific groups (by age, poverty status etc.), others are universal; in some cases the transfers are conditioned to certain behavioral responses by beneficiaries, or require them to participate in some type of public works program. Benefit levels and fiscal sustainability range from extreme generosity (and fiscal difficulties), to some cases where benefits are so low that they have no poverty impact at all. 13. Regardless of differences in design and targeting, most programs respond to a common rationale: modern States must provide reasonable protection against social risks to their citizens. As in many other sectors, programs are the result of a combination of conceptual principles, fiscal, social, and institutional restrictions, and political negotiations. This combination of determining factors sometimes results in system fragmentation, when some social sectors have access to comprehensive and effective protection, others have limited access and some have no protection at all. This fragmentation is sometimes mirrored by a similar problem with the institutional structure responsible for running the programs, resulting in increased inequities, and sustainability risks for the policies. 14. This section presents a discussion on the rationale of income transfer programs. The section does not aim at developing a full conceptual framework around the policy goals and instruments of income transfer policies. Instead, it provides an overview of the main concepts that support the existence of income transfer policies, substantiates the analytical framework adopted in the document. It considers the motivations that policy makers have to design and implement social protection and income transfer programs, discussing the risks and problems generated by the systems' fragmentation, identifying possible impacts that these programs might have on labor markets, and assessing the role of conditions on this type of programs. Finally, the section discusses the role that income transfer programs should have in a consistent social risk management framework in Uruguay. 2.1. SOCIAL RISKS, SOCIAL PROTECTION AND INCOME TRANSFERS 15. Individuals are exposed to the risk of suffering contingencies everywhere and at all ages. These contingencies may have a very low probability of happening, may have a small impact on welfare, may have no impact beyond the individual or be adequately covered by market institutions. In these cases, risks are considered "facts of life", and individuals and households are usually left alone to prevent them or cope with their consequences. In other cases, contingencies may have important social implications, because the impact of a potential loss would affect many individuals or families, there may be no reasonable mechanisms to prevent or cope with them individually, or the indirect social impacts may be too large. In those cases, social mechanisms to reduce the 4 risks or alleviate their consequences become necessary, and State intervention occurs through the implementation of social protection systems. 16. The level of intervention by State institutions to manage social risks depends on three dimensions: the frequency of the potential loss (or probability of occurrence), the magnitude of the potential loss, and the social impact of the loss. Adapting the discussion of World Bank (2007) and others about the "comprehensive insurance framework" approach, it is possible to say that the higher the frequency and the magnitude of the potential loss, the higher will be the need for action to prevent or compensate. On the other hand, the higher the social impact of this loss, the higher the need for proactive intervention by the State. These interventions can range from mere regulation of some markets to ensure reliable availability of private mechanism to households to protect themselves, to direct provision of transfers or services. These interventions comprise what is usually known as social protection policies. 17. Definitions of Social Protection abound, but they are not always fully satisfactory. They generally refer to the concept of social risks discussed above and, in general, include public policies and actions related to social insurance and social assistance. However, most definitions are broad, including references to social insurance and assistance, as well as health, education, and other social policies. In the 2001 "Social Protection Strategy" document, prepared by the Social Protection Department of the World Bank, it was defined as "the set of public interventions aimed at supporting the poorer and more vulnerable members of society, as well as helping individuals, families and communities to improve their risk administration" (Holzmann and Jorgensen, 2001). The Asian Development Bank has defined it as "the set of policies and programs designed to combat poverty and vulnerability through the promotion of efficient labor markets, diminishing population's exposure to risk and augmenting its capacity for self- protect from risks and interruption or loss of income". More recently, Bertranou (2005) defined it as "the set of interventions from public and private institutions that intend to alleviate households and people from the burden implied in a series of risk or needs". 18. The lack of a convincing conceptual definition for social protection led some authors to define it by enumerating the programs included. For example, in a book about Uruguay's Social Protection system prepared for the ILO, Ferreira-Coimbra and Forteza (2004), discussed the social protection system considering the collection of social security programs, education, health, and housing policies, and proposed a categorization of components in transfers and services, depending on whether the programs distribute money or provide access to basic services. This last definition is also wide (as, in fact, it includes almost all social policies), but provides a valuable distinction between transfers and services. 19. Following this distinction between transfers and services, income transfer programs can be considered part of the social protection policies, as they provide resources to individuals and households to ameliorate the effects of social risks. Not all social risks can be managed through income transfer programs, as some may be not caused or result in financial problems for households. However, many risks, including some of the most serious ones, have a direct impact on households' ability to obtain necessary goods and services, and adequate income transfer programs may solve these problems. 5 2.2. INCOME TRANSFERS AND THE FRAGMENTED WELFARE STATE 20. Income transfer policies have a long history in Latin America. The earliest programs were introduced at the onset of the Twentieth Century. Uruguay, together with a few other countries, was among the forerunners, developing social insurance and transfer schemes to protect workers against social risks such as old age poverty. While these programs developed rapidly and covered a growing proportion of the population, they did not reach all Uruguayans. This problem has been described by De Ferranti et al (2003) and Fiszbein (2006) as the "truncated welfare state" syndrome. 21. The basic message behind the "truncated welfare state" concept is that most countries in the region have focused on provided adequate protection to a limited part of the population, while failing to include those most vulnerable. During most of last century, social protection programs, including old age pensions, health insurance, unemployment insurance, and family allowances, have been targeted to workers in the formal labor market and their families. However, they generally neglected to provide similar protection to those who, for different reasons, were not part of the formal labor force. Only in recent years the attention of policy makers began to focus on this problem, and different solutions have been proposed and attempted. Among those, there were two general approaches: some countries tried to include informal workers in the formal social protection systems, while others aimed at implemented independent programs aimed at this population. 22. The first approach to overcome this truncation, expanding formal coverage, has had limited success. For example, efforts to include self-employed workers in contributory pension schemes compulsorily have had very limited success. Data from household surveys show that, among twelve countries in the region, only one had a coverage rate over 20 percent for the independent workers in the early 2000s. Furthermore, four had coverage levels between 10 and 20 percent and the remaining seven were able to include in their pension systems less than 10 percent of their self- employed workers (Rofman and Lucchetti, 2006). Different strategies have been attempted, making participation compulsory, offering incentives, or even subsidizing participation, but results have been weak in most cases. 23. The second approach, creating new programs aimed at informal workers and their families has been more successful. Coverage of recently implemented income transfer programs aimed at this group have been important in most countries in the region, including Mexico, Argentina, Ecuador, Colombia, Uruguay, Chile, and several Central American countries. However, in most cases these new income transfer schemes had no direct linkage with the more traditional social insurance programs, resulting in even more truncation of the welfare state. In these cases, some workers receive wide access to pensions, health insurance, unemployment insurance, and other social programs (such as housing, education, and even tourism in some cases). At the same time, a second group of workers--including in many cases the poor and unemployed--received a more limited set of programs, including income transfers (sometimes linked to health, education, or public works conditions), basic health, food and nutrition, etcetera. 24. The consequences of this fragmentation in the social protection policies are serious in terms of effectiveness and sustainability. On the one hand, in most countries 6 the truncation resulted in an important gap in coverage, as some informal workers that are not poor enough to become the subject of the highly targeted programs receive no protection at all. In addition, fiscal sustainability of these policies as a whole becomes a serious challenge, since social insurance programs continue to grow and spend an increasing proportion of public budgets, crowding out the fiscal space for other necessary income transfer programs that are forced to compete for resources with them. A third problem is that, in many cases, the new programs are negatively considered by the media, politicians, and the population in general, as they are seen as charity schemes, or questioned due to the potential negative incentives that they might generate among beneficiaries, thus stigmatizing the beneficiaries and weakening their political and social support. Finally, fragmentation in design is usually replicated in the institutional structure, as agencies in charge of managing the programs are separated and focused exclusively on their own target population, creating further inefficiencies in the system as a whole. 2.3. PROTECTION VERSUS INCENTIVES: IMPLICATIONS AND TRADE-OFFS IN THE LABOR MARKET 25. The impact of income transfer programs cannot be considered exclusively by measuring income changes, as other indirect effects may be equally important. As individuals and households receive income transfers, they perceive an "income effect", as the total household income increases, resulting in an improved standard of living. The magnitude of this effect is obviously linked with the benefit size, since the larger the benefit the larger the improvement in welfare will be. However, a second, indirect effect must be considered, as income transfers act as incentives to change behaviors through several channels. 26. Individuals and households respond to economic incentives if and when benefit amounts are large enough. Two types of incentives have been often discussed in the literature. First, if benefit access is linked to some type of conditionality, then beneficiaries may change behavioral patterns in order to obtain these benefits. This is the rationale behind the conditional cash transfers programs, discussed below. A second incentive effect has to do with the labor market. If benefits were high enough, it would be reasonable to expect that some individuals will prefer to apply and receive a benefit to have a job and receive a salary. 27. The literature on the negative labor market incentives involved in welfare programs is extensive, but debates are far from being settled. The theory is clear: at any given level, some individuals will always prefer to receive benefits instead of earning a salary at the labor market. This is a well-known moral hazard problem, discussed since the late XVIII century, when Thomas Malthus attacked the poor laws in England. This effect has been debated at length in some developed countries, where analysts and policy- makers are worried about the potential negative impacts of income transfer programs in labor supply. Much of the debate on welfare reform in the US and western European countries since the early 1980s have been based on this argument. Even as recently as in 2006, an OECD report found that in Norway, Switzerland and Poland, some workers 7 might prefer to receive sickness and disability benefits than to reenter the labor force, due to the particularly generous design of their systems (OECD, 2006). Box 1. Welfare Reform in OECD Countries Western European countries were the first to introduce formal welfare programs as part of their public policies. During most of the twentieth century, welfare policies expanded, including more beneficiaries and expanding benefits. Other OECD countries, such as the US, followed their example, albeit with a more restrictive approach. These programs, considered a core component of the social contract in most of these countries, began to be questioned in the last couple of decades as fiscal pressures to finance them mounted and concerns about their impact on labor markets expanded. Social safety nets in the US have experienced a varied number of changes since its inception in the thirties with the New Deal. The pioneer Aid to Families with Dependent Children (AFDC, 1935) was not aimed at relieving poverty but to support households headed by single women. Social Security was introduced a year later, and other food, income transfers and health insurance schemes were adopted by the federal government in the following twenty years. During the 1970s, the number of beneficiaries of welfare programs (in particular those directed to poor households) expanded rapidly, and incentives to join or return to the labor market were introduced. As incentives became requirement and enforcement was strengthened in the 1980s, long-time welfare dependence became rare. Finally, the 1996 reform replaced the historical ADFC with a new program (Temporary Assistance for Needy Families-TANF) that focused on finding jobs for welfare beneficiaries, in a model that is known as "welfare to work". Reforms in Europe in recent years have been in the same direction. Countries with extended social safety nets and a tradition of generosity, such as the Netherlands or the UK also moved towards a "work first" approach. In both countries the traditional welfare system, that provided income relief to the poor, unemployed, and disabled, became a job placement system, offering training, job search support, and strong incentives for beneficiaries to return to the labor market. The full impact of these new policies is still unknown, as policy reforms are recent. However, it seems clear that these programs have been able to reduce welfare dependence for some families, in a context of sustained labor demand. Many analysts and policy makers around the world are carefully watching these experiences, trying to assess their medium term impacts and their replicability in less developed economies and weaker labor markets. 28. The actual impact of income transfer systems on labor force markets (and, indirectly, on GDP) it is not completely clear and seems to depend on a number of design details. Lindert (2003), for example, has argued that (i) negative incentives for labor supply are sometimes not caused by the transfer programs, but by their targeting strategies, and (ii) reduced labor supply has little (if any) effect on GDP, as it is compensated by higher productivity among those who continue to work. There are also other important aspects to consider, such as the amount transferred to beneficiaries, compared to the normative and effective minimum wages that may have an important effect on the relevance of these incentives. In any case, it seems clear that program design 8 should consider the risk of introducing adverse incentives too strong to be controlled or compensated, resulting in an overall negative impact of the policies. 29. Concerns about negative incentives impact refer not only to the question of labor supply, but also labor informality. Two types of incentives have captured the attention of policy makers and analysts in the context of the prevailing informality in Latin American labor markets. First, as in the case of labor supply, some workers may decide to apply for benefits that are targeted to the unemployed or inactive, while hiding the fact that they have a job in the informal sector. In addition, firms may prefer to hire their employees informally, to avoid paying payroll taxes and other legal requirements, specially if these taxes and requirements seem excessive and enforcement is weak. 30. The debate about the impact of these incentives in Latin America has been extended. While empirical analyses that actually measure these impacts are scarce, the topic has been present as policy makers discussed and implemented non-contributory income transfer programs throughout the region in recent years. The debate, of course, is not about whether these incentives exist or not, but about whether their impact is significant and, if so, it is high enough to compensate other social or economic positive effects that these programs may have. 31. Actual cases seem to indicate that bad incentives are more often the consequence of poor design or implementation, than caused by the mere existence of social protection programs. For example, Colombia's "regimen subsidiado" is a health insurance scheme that is provided only to poor, informal workers, who lose their right to this benefit once they enter into formal employment, even if they leave this job shortly after. This rule makes formal employment less desirable for those who value access to the health insurance, thus promoting informal employment. In general, social protection programs targeted to the unemployed or inactive poor exclusively tend to have this negative effect, and should be revised. 2.4. INCOME TRANSFERS IN URUGUAY: A BLURRED LINE BETWEEN SOCIAL INSURANCE AND SOCIAL ASSISTANCE 32. Policies and programs that promote, require, or provide income transfers to individuals or households have been traditionally categorized in two groups: Social Insurance and Social Assistance. The main differences between these programs are the financing arrangements (that is, who is paying for the transfers), the targeting principles (that is, who receives benefits), and the programs' objectives (that is, what are the goals of the program). 33. The conceptual criteria used to define these categories are clear. In social insurance schemes, participants ("contributors") are required to finance the schemes, usually through payroll taxes; only those who contribute have the right to receive benefits, and the systems aim at replacing previous salaries when an unexpected (or expected but unavoidable) loss of income occurs. On the other hand, social assistance programs are generally financed by general revenue, selection of beneficiaries is made considering their needs instead of their past contributions and benefits aim at alleviating poverty. Because programs in the first group require that participants contribute to 9 finance them, they are usually called "contributory", while those where no contributions are required are called "non contributory" 34. These differences, clear in theory, are less evident in practice. In fact, none of them is fully complied by the traditional "contributory" programs in Uruguay: a. The first condition (contributory financing) has become less strict as demographic aging and labor market trends affected these programs. In the early 1980s the unemployment insurance and family allowances programs lost their funding, and became dependent on tax transfers. Also, the pension programs began to generate growing deficits, that had to be covered with earmarked taxes or transfers from the treasury, and contributions have financed 50 percent or less of pension benefits in the last ten years. b. The second condition (the link between contributions and access to benefits) is weak in the case of the pension system. Many current retirees have received their benefits under particularly generous circumstances, since the information systems were inadequate and sworn statements were accepted as proof of past contributions. On the other hand, different analysts have indicated that, given current trends in the labor market and the vesting requirements to retire, many of those who are currently contributing will not retire under the pension system in the future. In other words, many of those who are retired did not contribute to the system, and many of those who are contributing will not retire. c. Finally, the third condition (the objectives of the programs) has also weakened over time. The pension program was designed as an income replacement scheme, where retirees would receive a benefit proportional to their past earnings, but legal reforms and implementation policies have resulted in a flattened benefit scale. As a result, nearly 20 percent of retirees and almost 70 percent of survivors received benefits below US$100 per month by the end of 2006. 35. Given the blurred limit between social insurance and social assistance, where "contributory" programs neither require contributions nor are financed by them, this report considers income transfer programs as part of a single system. This system includes policies aiming at providing a flow of income to individuals or households that, for some reason, are not able to obtain income from the labor market. These programs are not identical, as they grant different benefits to different populations, using different financial and institutional arrangements, but their commonalities are enough to justify their inclusion in a single report. The distinction between contributory and non-contributory programs is maintained in this report, to distinguish between those programs that aim explicitly to alleviate poverty and do not pretend to operate as an insurance scheme, from those that, at least in theory, operate within the framework of an insurance program. 10 2.5. INCOME TRANSFERS AS INDIRECT POLICY TOOLS: CCTS, "CONTRAPRESTACIONES" AND OTHERS 36. Access to income transfer programs is sometimes conditioned, responding to three possible goals. First, conditions are sometimes used to provide incentives to increase demand of some social services, usually those linked with accumulation of human capital (health and education). A second reason is to implement a self-selection method, to target benefits in an automatic way. Finally, a third possible reason is linked to a political economy problem: trying to increase the legitimacy of income transfer programs, policy makers introduce conditions to indicate that these programs are not giveaway schemes that promote welfare dependence. 37. The first motivation, to promote higher investment in human capital, has been common in Latin America in recent years. Most income transfer programs implemented in the region in the last few decades were organized as "Conditioned Cash Transfer" schemes. From the best know national programs in Mexico ("Oportunidades"); Colombia ("Familias en Acción"); Chile ("Chile Solidario"); Ecuador ("Bono de Desarrollo Humano"), and Brazil ("Bolsa Familia"), to less discussed schemes in other countries in the region (Honduras, Dominican Republic, Jamaica) and around the world (including European, Asian, and African countries), the core model of these programs is to provide an income transfer to poor households while promoting investment in human capital. 38. While conditional cash transfer programs provide basic income relief to the poor, in most cases their main goal is to provide incentives to stimulate demand of basic social services (generally health and education, and sometimes other programs). In fact, the direct impact on poverty is generally small. The International Poverty Centre estimated that Mexico's Oportunidades reduces poverty headcount by one percentage point, and Brazil's Bolsa Familia does it by two percentage points (Zepeda, 2006). Thus, these programs focus more on breaking the vicious circle of poverty in the long term than to alleviate it in the short run. Under that rationale, these programs may have a significant impact if several underlying assumptions are verified: first, there must be a serious problem of structural poverty, originated on insufficient accumulation of human capital in the long term. Second, the supply of these services should be adequate, both in terms of availability and quality; and third, there must be a problem of demand, originated in lack of interest, opportunity costs, or other factors that could be overcome with additional income. 39. Conditional cash transfer programs have been relatively small, in terms of fiscal implications. For example, the largest program in the region, Ecuador's BDH, spends approximately 0.55 percent of GDP, while the programs in the two largest countries, Mexico and Brazil, spent approximately 0.35 percent of GDP per year. Targeting strategies included, in most cases, a proxy means test, and evaluations have shown important impacts in terms of educational attainment and health status1. 1In a detailed discussion of CCT impact assessments, Rawlings and Rubio (2005) showed that the CCT programs in Mexico, Nicaragua and Colombia had clear positive impacts on education, health, and consumption. 11 40. The second approach to conditioning income transfers uses the conditions as a self targeting mechanism. The logic here is that if benefits are made unattractive for those who might have other options, then only those who are really in need of income support would apply for them. In this context, the condition would act as a deterrent to avoid granting benefits to individuals or households that should not be the target of the program. Many workfare programs, where beneficiaries of income transfer programs are required to work on public projects to qualify for benefits are designed with this logic, as it is expected that those who have access to a better paying, higher quality job, will prefer to keep it. 41. In some cases, income transfer programs include conditions as a combination of the two mentioned goals. This is the case of programs that grant benefits to unemployed workers, while requiring them to participate in training activities. By doing so, policy makers expect to increase beneficiaries' skills and, at the same time, reduce access for individuals who have an informal job or are not interested in re-entering the labor force. This approach has been increasingly adopted by developed countries, and known as the "welfare-to-work" model, and also in some of the countries in the region. 42. Finally, a less technical but equally relevant reason to introduce conditions into income transfer programs has to do with the political economy of these policies. When political concerns about possible moral hazard and welfare dependency problems arise as relevant issues in a country, policy makers sometimes propose the inclusion of conditions in income transfer programs with the goal of making them more politically acceptable. 12 Box 2. The Rationale for Conditioning Transfers in Uruguay Uruguay has had conditioned cash transfers for many years, as family allowances benefits have always been subject to school enrolment. In the original law that introduced them in 1944, school assistance was not a requisite for children younger than 14 years old, but was required to extend coverage beyond that age. This original requirement evolved into a general requirement to receive health controls or attend school (BPS, 2006), for all children younger than 18 years old. While the motivation for this requirement is not explicit in the legislation, there is informal evidence (including household survey data) indicating that it is rarely enforced. The conditions currently included in income transfers in Uruguay do not seem to respond to any explicit rationale framework. The first reason discussed in this report (increase social services demand) does not seem applicable in the case of health services or primary schooling, since coverage in both cases is nearly universal (meaning that there are no significant demand problems). On the other hand, conditions could be more effective to increase school attendance at the secondary level, where dropout rates are too high, if benefit amounts and enforcement policies are appropriate. Nearly fifty percent of children from households in the poorest quintile fail to complete their secondary education, and new incentives to continue studying would have an effect. Under this approach, income transfer programs are often considered part of a long-term strategy to build human capital among particularly disadvantaged groups, thus requiring a very effective targeting strategy. The second possible reason to introduce conditions (self-targeting) does not seem to fully apply in the case of Uruguay. This strategy renders results when the condition(s) required to access benefits is linked to services or actions that higher income individuals would reject, such as participating in community services programs, enrolling in training or activities that due to their schedule would prevent them from maintaining a formal job, etcetera. While workfare and training programs were created as part of PANES, participation was voluntary. This is not the case of conditions established for family allowances or IC in Uruguay. The third reason, linked with programs' legitimacy and political economy considerations, is somewhat more relevant than the others. In a clear example of this approach, President Vazquez stated in early 2005, referring to the then new PANES program, that "... [PANES] is not about welfare dependence, since beneficiaries will have to [...] participate in community works and send their children to schools and take them to regular health controls..."2 2.6. INCOME TRANSFER PROGRAMS IN URUGUAY: THE RATIONALE AND TOOLS 43. Different individuals are exposed to different social risks depending on their age. In order to identify the most relevant social risks affecting Uruguay's households, it is interesting to consider the life cycle approach developed by Arriagada and Hall (2000). This approach considers the risks affecting different age groups, to then indicate what 2President Tabaré Vásquez in `La Republica' (5/25/2005) as cited by Hennchen, (2007) 13 tools or programs are being provided or could be implemented to reduce them or ameliorate their impacts. Social protection offers risk prevention tools to reduce the probability of suffering a loss due to a social risk, mitigation tools, to reduce the potential loss if the risk cannot be avoided, and coping, to relieve the loss impact on population's welfare. Through an adequate mix of these tools, SP systems should be able to offer adequate protection to the population. 44. The main income transfer programs in Uruguay analyzed in this report, aim at both preventing, mitigating and coping with social risks. By identifying specific risks affecting different age groups, it is possible to assess the effectiveness of the existing transfers programs and the potential impacts of possible reforms. Thus, the population is classified in six age groups, from pre-school children to the elderly. Table 1 shows, for each age group, the main social risks that affect them and how could they be addressed with an income transfer program. In addition, the table shows the level of poverty incidence, an indicator of the magnitude of other social risks affecting the age group, the most relevant income transfer programs aimed at the group, the coverage level of these programs, and an indicator of the effectiveness of targeting of these programs for each age group. Table 1. Social Risks and Income Transfer Programs, by Age Group Poverty Other Age Current Income transfer Coverage Main Social Risks Incidence relevant group programs index (2006) risks Poverty, access to pre-school 0-5 49.3% 22%(1) Family allowances, IC 70.2% education, basic health Poverty, access and quality of 6-12 48.5% 0.5%(2) Family allowances, IC 69.3% education Poverty, access, dropouts and 13-17 39.1% 24.7%(3) Family allowances, IC 56.8% quality of education Unemployment Insurance, IC, Poverty, Unemployment, 18-59 23.8% 35.2%(4) Disability and survivors' 8.4% Disability, Death benefits Poverty, Health, Lack of family 60-69 12.1% 38%(5) Contributory pensions, IC 64.6% support network Poverty, Health, Lack of family Contributory and non 70+ 6.8% 45%(6) 92.9% support network contributory pensions, IC Notes: (1) Indicates the percentage of children aged 4-5 who do not attend pre-school (2) Indicates the percentage of children aged 6-12 who do not attend primary school (3) Indicates the percentage of children aged 13-17 who do not attend secondary school (4) Indicates the percentage of working adults aged 18-60 with informal jobs (5) Indicates the percentage of elderly living alone or with other older family members (6) Indicates the percentage of elderly living alone or with other older family members Source: Bank staff 45. Poverty as a risk and Citizen Income as an income transfer program are included for every age group. Other risks and programs are age-specific, resulting in some level of program overlapping, as discussed earlier in the report. In addition, some risks and programs refer to individuals, while others are focused on households (thus 14 affecting all members, regardless of their age), creating some additional "noise" in the life cycle analysis. 46. Besides poverty, basic health and access to pre-school education are considered the main risks of the youngest group (0-5 years old). As it often happens in other countries, young children have the highest poverty incidence rate, at almost 50 percent. Other indicators show better performance, as is the case of pre-school education enrolment, since 78 percent of children aged 4 and 5 years old are included in the system. The main income transfer programs protecting this population in Uruguay are family allowances, both contributory and non-contributory, and citizen income. More than 70 percent of children in this age group are covered by these programs, and coverage seems to be concentrated among poorer children. 47. Children aged six to twelve years old are exposed to poverty levels very similar to those of younger kids. However, they have a better access to State- provided social services, as evidenced by the nearly universal coverage of primary education. Family allowances and IC provide coverage to 69 percent of these children, with a slightly better focus on the poorest groups. On the other hand, teenagers (ages 13 to 17) are also exposed to poverty risks, as well as access to and quality of secondary education (where the main risk is dropping out before completion). Poverty incidence is lower in this group than at younger ages, at 39 percent, but nearly 21 percent of teenage children appear to have dropped out of school. This group has access to the same income transfer programs than younger children (family allowances and IC) but their coverage is somewhat lower, at 57 percent. The lower coverage might be caused by the school attendance conditionality included in both programs, which intends to act as an incentive to increase education demand but might be operating as a mechanism to exclude poor teenagers and their families from the programs. 48. The main social risks affecting the adult population are linked to the labor market. Unemployment, disability and death are events that interrupt income flows for workers and their families, exposing them to poverty. Poverty incidence in this group is much lower than among children, at less than 24 percent. Sixty five percent of the working population has a formal job, which provides them with income and protection, through the contributory social security system, against some of these risks. However, this protection is not complete, given that unemployment insurance has time-limited benefits. Also, the remaining 35 percent of the working population does not have access to social security programs, making them more vulnerable to the social risks. Citizen income provides an additional layer of protection for all members of this age group, as beneficiaries qualify by taking their household income level into consideration, regardless of their employment status. More than 8 percent of this population receives at least one income transfer, since it is the most common one. Finally, the high correlation between poverty and coverage in this group is due to the prevalence of IC, a highly targeted program in terms of poverty. 49. By age 60, many Uruguayans qualify for retirement in the pension system. The main risks affecting the elderly in Uruguay are, as in other countries, structural poverty (as workers retire from the labor market and their consumption becomes dependent of savings or transfers), health problems, and lack of family or social support networks. Poverty incidence in the 60-69 years age group is much lower than average, at 15 12.1 percent. This group has access to two different income transfer programs: contributory pensions, for those qualifying to receive one, and IC. The combined coverage of these programs is high, since almost two thirds of the population receives a benefit. Moreover, it is important to consider that some of the population that continue to work in this age group have made enough contributions to the pension system to qualify for a benefit, but chose to postpone retirement. 50. Finally, the oldest group considered (integrated for those aged 70 and more) is exposed to risks similar to those of the 60-69 years old, but it is affected by a significantly lower poverty incidence, at below 7 percent. This population is also protected by the contributory pension system and IC, and they also have access to the non-contributory pension program. Coverage in this group is high, as almost 93 percent of the population receives a transfer. 51. Of course, income transfers are not the only possible response to the social risks identified in table 1. Other policies have a role in preventing and alleviating them, such as health insurance and education access programs, public health, nutrition, labor markets and general macroeconomic policies have a role and can significantly influence the magnitude and incidence of social shocks over the population's welfare. 52. Overall, poverty declines with age, while income transfer programs coverage presents a "U" shape. This difference is caused in part by the sharp difference in average benefits across programs. While children receive approximately UR$230 per month from the family allowances program, the elderly receive between ten times that amount from the non-contributory pension scheme, or twenty times from the contributory pension scheme. 16 3. THE CURRENT INCOME TRANSFERS POLICIES 53. Income transfer policies in Uruguay involve several programs and institutions, ranging from traditional unfunded old age pension schemes to modern poverty alleviation income support programs. Some of these programs have a long history, while the most recent ones were introduced in the last couple of years. This section presents a short description of these programs' origins and evolution, a basic discussion of their main design components, their coverage, and financial situation. 3.1. THE ORIGINS 54. As in other Latin American countries, income transfer programs in Uruguay have evolved in a sort of bottom-up process (Mesa Lago, 1978). Military forces and some guilds of civil servants were the first groups to receive some type of social protection, in the form of a pension scheme, on the first half of the XIX century (Bertranou, 2005). From then on, other groups of workers started setting their own pension funds. The order of accession to these funds can be considered as a weighted index between pressure capacity and organization ability. (Bertranou, 2005). While these programs were mostly designed as contributory schemes, as early as 1919 a non- contributory pension system was established (Saldain and Lorenzelli, 2002). 55. The pension system consolidated into a national and nearly universal program in 1954 when the State intervened and converted social security into a top- down policy. Still, a few independent pension funds remained, surviving until today3. An important parametric reform was introduced in 1989, when the indexation rule was changed from prices to wages, and in 1996, a structural reform introduced a funded component in the system. 56. Other social protection programs were introduced in Uruguay after the pension scheme, expanding the scope of social insurance programs. In the early 1940s, family allowances were first granted for workers in manufacturing and commerce, and were later expanded to other formal sector workers. Similarly, earlier versions of unemployment insurance schemes were introduced for civil servants and formal private workers in the late 1910s and mid 1920s, and successive coverage expansions and reforms resulted in the current system, which covers most formal workers. Other programs, such as health insurance, illness and maternity benefits were also introduced as part of a progressive trend to expand and integrate the social protection system for formal workers4. 57. Only in recent years, have governments begun to look more carefully on policies that would provide better protection to those outside the formal labor market. In 1999, a first step to include households with no formal workers in the family 3At present, there are five pension funds running in parallel to BPS. Two of them are managed by the State and cover the military and police, while the other three are independent, covering banking employees, independent professionals, and public notaries. 4There is extensive literature on the history and evolution of social protection in Uruguay. For a detailed discussion of these trends see, for example, Ferreira-Coimbra and Forteza (2004), Saldain and Lorenzelli, 2002), Bertranou (2005), Bucheli (2004), Bucheli and Amarante (2006), Vigorito and Arim (2006) 17 allowances program was taken. This was further expanded in 2004, and citizen income, a new income transfer program aimed at the extreme poor (regardless of their employment status) was introduced in 2005 as part of the social emergency program PANES. 58. Nowadays, the main income transfer programs in Uruguay include pensions, family allowances, unemployment insurance and the means-tested citizen income program. Some of these programs were designed as contributory schemes, while others were designed as non-contributory. In practice, differences between these two groups are less clear, and respond more to tradition than to actual differences in financing or benefit targeting. Next, a short description of each program is presented, indicating their basic design characteristics, such as contribution requirements, requirements to qualify for benefits, and benefit level. 3.2. THE "CONTRIBUTORY" PROGRAMS 59. The pension system is the main social insurance program in Uruguay, while family allowances and unemployment insurance are much smaller. Most civil servants and private sector workers, including those self-employed, are required to participate in the national pension scheme. Only those enrolled in one of the five smaller occupational schemes are not required to join the national system. Contributory family allowances are paid to formal salaried workers, providing a monthly stipend per child, maternity benefits provide income to working mothers while on maternity leave, and unemployment insurance provides income replacement to formal workers who lose their jobs. 60. The national pension system, managed by the "Banco de Previsión Social" (BPS) is a multipillar scheme. All employers' contributions and a part of employees' contributions are used to finance a PAYG scheme that provides retirement benefits to individuals older than 60 years of age with at least 35 years of contributions. Benefits are proportional to past salaries (although with a redistributive formula), and grow with age. Part of the personal contributions of workers earning more than a threshold are directed to individual accounts, managed by commercial firms in a scheme very similar to those existing in other Latin American countries5. In those cases, retirement benefits will be the combination of a reduced PAYG benefit from the public scheme and a flow of payments in the form of an annuity, financed with accumulated funds. Also, older individuals (aged 70 or more) with at least 15 years of contributions can obtain a reduced benefit. Finally, disability and survivors' benefits are paid by the PAYG or the funded scheme, depending on the worker's enrolment. 61. There are five independent occupational pension schemes that are run separately from the BPS national program. Police officers and the military have their own schemes, run directly by government agencies and financed through the national budget. The other three funds cover financial sector employees (by far, the largest fund in this category), independent professionals and notaries. They are all run as defined benefit 5Approximately 37 percent of all contributors to the pension scheme were also contributing to the funded scheme by the end of 2006. 18 pay as you go schemes, with retirement requirements similar or slightly less strict than those of the BPS. 62. "Contributory" family allowances can be claimed for children younger than 18 years old by salaried workers in the private sector, civil servants, retirees and unemployment insurance beneficiaries. Benefits are paid per child, and there are two levels, depending on workers' income. Children of workers earning less than six BCP ("Base de Prestaciones Contributivas", an index used to adjust all social security variables, valued at UR$1,636 in early 2007) receive an allowance of 16 percent of that value, while the children of workers earning between six and ten BCP receive 8 percent of that value. Children of workers earning more than 10 BCP receive no benefit, with a few exceptions6. Benefits are financed and paid for by the BPS in the case of private sector employees, retirees and unemployment insurance beneficiaries (although no earmarked contributions are defined to finance them), or directly by the State, in the case of civil servants. 63. Unemployment insurance provides benefits for up to 6 months. Private sector workers with at least 12 months of formal employment are covered. Benefits are equivalent to 50 percent of previous wages, with a maximum of 8 BPC and a minimum of 0.5 BCP. The system is financed by BPS, with no explicit contributions. The program also covers formal workers who are temporarily suspended, a policy that has generated controversies in the past7. Other smaller income transfer programs include two programs that replace wages for a limited period for workers on maternity leave or on extended sick leave. 6For workers with more than 3 children there is a scale of requisites that increases the earning ceiling in 1 BCP per child (i.e. to claim for 4 benefits the earnings ceiling is 11 BCPs, for 5 it is 12 BCPs, and so on up to 20 BCPs. 7For a discussion about the role and performance of the Unemployment Insurance in Uruguay, see Bucheli and Amarante (2006) 19 Box 3. The 1996 Pension Reform Following the lead of other countries in the region, Uruguay enacted a pension reform in 1996. As in neighboring Argentina, the reform in Uruguay created a true multipillar scheme, as all workers would continue to participate in a government run PAYG scheme, and some of them could direct part of their contributions to one of several competing pension funds managed by a commercial firm. The reform in Uruguay preserved a more important role for the State than in other countries, partly because the system was already more extended and mature (which would result in a higher transition cost) and partly because of political restrictions. Not only all workers continued to contribute part of their payroll taxes to the PAYG scheme, but the largest managing firm in the funded scheme is a joint venture of the Banco Republica (the main state-owned commercial bank), the Banco de Previsión Social (the national social security agency) and the Banco de Seguros del Estado (the main state-owned insurance company). With some exceptions (young, high-wage earning workers), workers had the opportunity to choose to make part of their contribution to the new capitalized system. Responding to a strong publicity campaign and the offer of important advantages in terms of benefits, most workers chose to join the funded scheme, thus creating a true multipillar system. Together with the introduction of a funded pillar, the reform also modified several parameters in the system. The most relevant changes included a reduction in the expected replacement rate, an increase in the required vesting period to 35 years (one of the highest in the world), and an increase in the minimum retirement age, all measures designed to respond to the fiscal pressures that the system was generating at the time of the reform. The reform was effective in controlling the growing trend in expenditures. As discussed later in section 3.7, the growing trend of social security expenditures stopped in 1996. However, this was achieved at the cost of a declining trend in coverage, as the number of beneficiaries of the pension system has declined since 1996 both in absolute and relative terms. Medium and long term projections of the system indicate that this decline could continue in the future, and the topic has become central in the policy debate in recent times. 3.3. "NON-CONTRIBUTORY" PROGRAMS 64. Opposed to the nominally contributory programs discussed in the previous section, there are a few explicitly non-contributory schemes that provide income transfers to individuals and households in Uruguay. These programs are directly financed from general revenue funds, are targeted to populations that, for different reasons, are exposed to a high social risk (particularly, with regards to their poverty status), and do not require any work history or past contributions to qualify. 65. This group of income transfers includes three programs: the traditional old- age pension, the family allowances scheme introduced in 1999, and Ingreso Ciudadano, a component of the social emergency PANES plan. The three programs have a targeting mechanism to select beneficiaries, considering either the current income of households (in the case of family allowances and old age pensions) or a proxy-means test to assess poverty status (in the case of IC). 20 66. Non-contributory pensions are granted to individuals aged 70 or older who do not qualify for a pension, have insufficient means (including those of close relatives) and have lived in Uruguay for at least 15 years. Benefits are flat, at slightly below UR$3,000 per month, and are indexed to the BPC. The program is managed and financed by BPS, as part of the national pension system. 67. Family allowances coverage was extended to children of poor, informally employed families in 1999. In a departure from previous practice, Law 17139 granted family allowances to children living in low income households (less than 3 minimum salaries, later replaced by the BPC), if their father had been an unemployment insurance beneficiary but exhausted the six-month limit or they are solely supported by their mother. Benefits were set at the same level of the contributory family allowances. Enrollment in this new program was slow until a reform introduced in late 2004 extended coverage to all children of low income households, regardless of the employment status of their parents. 68. Citizen income is the income transfer component of PANES, the social emergency program that the new administration implemented as it took office in early 2005. The program aims at providing income relief and other services to the extreme poor for a period of two years. IC distributes a monthly benefit equivalent to approximately UR$1,400 per household. The program was originally targeted at the population in the first quintile of those below the poverty line, or nearly 8 percent of the total population, and the selection of beneficiaries was made through a proxy means test, which included home visits from social workers and volunteers to collect household data and apply the proxy model to select beneficiaries. 3.4. THE INSTITUTIONAL FRAMEWORK OF INCOME TRANSFER PROGRAMS 69. The design, financing, and management of income transfer programs in Uruguay involve a complex web of public and independent institutions. The Social Security Administration ("Banco de Previsión Social"-BPS), Ministry of Social Development ("Ministerio de Desarrollo Social"-MIDES), Ministry of Labor and Social Security ("Ministerio de Trabajo y Seguridad Social"-MTSS), Ministry of Economy and Finance ("Ministerio de Economía y Finanzas" - MEF), Central Bank of Uruguay ("Banco Central del Uruguay"- BCU), Planning and Budget Bureau ("Oficina de Planeamiento y Presupuesto"-OPP), Pension Fund Managing Companies ("Administradoras de Fondos de Ahorro Previsional"-AFAP) and five agencies in charge of the independent pension schemes are involved in the program's management. 70. Responsibilities of MTSS and MIDES include the design of policies and programs, and implementation of some of them. All policy debates and proposals regarding contributory schemes (such as the pension system, unemployment insurance or family allowances) are centralized in MTSS. MIDES is responsible for the design and implementation of the social emergency program PANES, including citizen income transfers and the execution of social inclusion programs, monitoring and evaluation, and coordination of social policies in general. 71. OPP and MEF play a role in the fiscal aspects, as they manage the national budget and transfer resources to finance the programs. They are also usually 21 involved in policy discussions. OPP also coordinates a social security policy working group, with representatives of all public institutions to advance policy reform proposals. 72. On the implementation side, BPS is by far the largest agency in Uruguay, as it manages most aspects of all income transfer programs, as well as other areas, such as the health insurance scheme for private workers. BPS operates as an autonomous public agency, directed by a Board with representatives from the Government and social sectors (workers, employers, and retirees). The AFAPs are commercial companies in charge of managing assets and accounts in the second pillar of the national pension system, which is supervised by the BCU; and the managers of three occupational pension funds (Bank employees, notaries, and professionals) are independent agencies run by boards selected by their members. 73. The large number of institutions related with the income transfer programs is, by itself, a potential risk for the system. Some agencies, such as the BPS, BCU, and of course the managing firms for the funded pension scheme have an important degree of autonomy. While this is good to ensure technical rigour and independence from political pressure, it also creates a potential risk, since coordination problems or differences in approach might result in conflicting actions. Interestingly, while the number of institutions has been high for decades, there never was one Ministry or agency in charge of overseeing the whole income transfers system. 74. The complexity of the institutional framework comes not only from the number of involved agencies and ministries, but also from the fact that there are some important asymmetries among them. While the ministries and OPP are part of the central administration and have a clear political dependency from the President, the Central Bank has an important degree of autonomy, and BPS is governed by a board with political representatives (from the government and opposition parties) and elected delegates from workers, employers, and beneficiaries. The AFAPs, on the other hand, are commercial firms. 75. The asymmetries are not only caused by differences in administrative autonomy, but also by differences in economic and human resources availability. While these differences are unavoidable and, in some cases, necessary, there are some extreme cases that may affect how the whole system operates. The Vice-ministry of Social Security, for example, is a small unit of less than 10 people, in charge of formulating and overseeing all social security policies in the country. On the other hand, BPS, the agency responsible for implementing most of them, has more than four thousand employees, and manages a budget of as much as 13 percent of Uruguay's GDP. 76. In 2005, the new Ministry of Social Development was chartered to coordinate social policies implemented by the Executive Branch. This legal provision, important as it signalled the Government's interest in ensuring proper coordination of all agencies, was not fully implemented, as other institutions maintained their authority and autonomy. Several positive actions have been taken since then. A Social Cabinet was created soon after the new Ministry, and the Minister of Social Development was named coordinator of this Cabinet. Also, the OPP organized a "Sectoral Commission for Social Security", to discuss sectorial policy proposals, in a reduced structure that includes some of the Social Cabinet participants. 22 3.5. COVERAGE 77. This section presents a revision of coverage levels for the main income transfer programs discussed before. Data for this analysis come from administrative records in BPS and MIDES, and from household surveys carried out by the National Statistical Institute since 2001. Data from both sources are not always consistent, due to differences in definitions, registration criteria, and statistical errors. However, both sources show similar trends and combined they provide a solid basis for the analysis. 78. Measuring coverage and impact of income transfer programs is not simple, as there are many alternative approaches and indicators. First, it is important to consider that in the case of contributory programs, coverage may be defined considering those who are contributing to the different programs, thus earning the right to receive a transfer in the future. On the other hand, the goal of income transfer programs is not to allow individuals to accrue rights to receive those transfers, but to actually make the transfers, regardless of their contributory or non-contributory financing scheme. Thus, to assess these programs' coverage it is necessary to consider those accumulating rights and those receiving transfers, in absolute terms, or as a percentage of the total or targeted population. This section presents data on coverage of active workers in contributory schemes, to then move to discuss coverage of beneficiaries, in both contributory and non- contributory schemes. 3.5.1 Coverage of active workers in contributory programs 79. Coverage of contributory programs (in particular, pensions, unemployment insurance and contributory family allowances) can be considered by assessing how many workers are actually contributing to the system, thus accruing rights to receive these transfers. The contributory social insurance programs in Uruguay have had a high coverage for decades. Administrative data shows that the number of contributors to BPS has been quite high in the last two decades, with some volatility due to economic cycles and crises. Figure 2 shows how the total number of contributors has been high and growing, with a small slowdown in 1995 (when the "tequila crisis" hit the economy) and a more important decline in 1998-2002, as the last economic crisis developed. The recovery in the number of contributors registered since 2003 brought the figure to record levels by the end of 20068. 8Data from BPS registration records cannot be rigorously compared with labor force, as BPS counts the number of jobs instead of individuals, leading for double counting among workers with more than one job. In a preliminary estimation, BPS staff indicated that this effect could explain as much as 15 percent of the reported figures. 23 Figure 2. Contributors to BPS Pension System, 1988-2006 1200 1000 ) 800 000s ni( s 600 butori 400 ontrc 200 0 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 Source: BPS (2006, 2007) 80. The decline in the number of contributors to BPS that took place during the economic slowdown of the early 2000s was mirrored in other schemes, resulting in a drop in the proportion of the occupied labor force with access to jobs that provide social insurance coverage. As shown in figure 3, the proportion of workers with access to pension coverage declined by almost five percentage points in three years (from 64 percent in 2001 to 59.3 percent in 2004). However, the recovery since 2004 has been substantial and, by early 2006, the proportion of occupied workers contributing to social security was back to the pre-crisis levels. Figure 3. Occupied Labor Force Formality Rate, 2001-2006 70 0 8 60 8 5 3 64, 62, 3 64, 60, 61, 59, 50 e ag 40 entc 30 per 20 10 0 2001 2002 2003 2004 2005 2006 Source: Bank's staff calculations based on ECH and ECHA. 81. These high coverage figures put Uruguay at the top among countries in the region. This high coverage can be partly explained by the relatively high per capita GDP in Uruguay, but also by a long tradition of compliance with labor market regulations. As shown in figure 4, coverage of the pension systems, estimated at 64.8 percent for 2006, puts Uruguay together with Chile and Costa Rica at the highest levels in the region, well 24 over other countries, including much larger economies such as Argentina, Brazil or Mexico. Figure 4. Labor Force Formality for Latin American Countries, mid 2000s Percentage of occupied labor force with contributions to social security 100 90 80 7, 7, 8, 70 64 64 64 e 60 ag 3, entc 50 5 48 1, 8, 8, 39, per 40 9, 6, 36 35 31 30 4, 9, 26 27 4, 2, 19 19 20 8, 15 10 12 10 0 BO PY PE NI GU CO EC SA VE MX AR BR CL CR UY Source: Rofman and Luchetti (2006) and Bank's staff calculations based on ECHA 2006. 3.5.2 How many beneficiaries? 82. An alternative, and more common, approach to consider income transfer programs coverage is to compare the number of beneficiaries with the population that could potentially claim them. Measuring coverage with this criterion is also difficult, partly due to data sources problems (sometimes registries are incomplete or inconsistent, and surveys may also have errors), but also due to differences in the definition of who could or should request a benefit. This section considers the coverage of each income transfer program, showing data from registries, when available, or household surveys. 83. More than 85 percent of the population aged 65 years or more in Uruguay receives a pension benefit. This includes benefits from the contributory and non- contributory old age pension schemes, and the rate has been very stable since the mid 1990s (Rofman and Lucchetti, 2006). These rates reflect the high level of compliance with the pension system among active workers, as discussed above, and a generous system that, in the past, made access to benefits possible with a relatively short history of contributions. 84. Registry data show that BPS paid 710 thousand pension benefits in 2006, a figure that has been slowly declining in the last decade. This data does not include other funds, which are much smaller. More than fifty percent of them are retirement benefits in the contributory scheme, almost forty percent are disability or survivors' benefits and the remaining 9 percent are non-contributory pensions (including old age and disability). The total number of benefits has slowly declined since the historical record of 730,000 was reached in 1998. This reduction, shown in figure 5, was faster among retirees from the contributory scheme, which has been declining by almost one percentage point per year since 1998. 25 Figure 5. BPS Pension System Coverage, 1985-2006 Number of beneficiaries, by program 800 Old Age Survivors + Disable Non contributory 700 600 ds 500 na 400 ous ht300 200 100 0 1985 87 89 91 93 95 97 99 01 03 05 Source: BPS (2006, 2007) 85. There is a significant level of benefit overlapping within the pension system, as many retirees also receive a survivor's benefit. Considering the population aged 65 and more, nearly 70 percent receives retirement benefits, and 30 percent receive disability, survivor, or non-contributory pensions. However, 15 percent of this population receives more than one benefit, and 15 percent receive no benefits at all. Most of the duplication is explained by women collecting retirement and survivors' benefits. The proportion of overlapping benefits has been growing in recent years, it went from 14 percent of the elderly in 2001 to almost 16 percent in 2006 (figure 6). Figure 6. Pension Systems Coverage, 2001-2006, by Type of Benefit Percentage of total population aged 65 and more 100 only retirement both only survivors, disability or non contributory 80 17,7 17,8 18,0 17,2 18,0 16,7 ega 60 14,0 14,9 14,8 15,0 15,6 15,9 ntecr 40 pe 54,2 54,4 53,2 53,8 52,7 52,7 20 0 2001 2002 2003 2004 2005 2006 Source: Bank's staff calculations based on ECH and ECHA. 86. While overall coverage is quite high, there are some important differences by income level. Figure 7 shows the distribution of pension system beneficiaries by quintile of per capita income in 2006. If individuals from all income levels had similar access to benefits, then each quintile would include 20 percent of the beneficiaries. Instead, only 7 26 percent of beneficiaries are in the first quintile, while 54 percent are in the top two quintiles. In other words, most uncovered elderly are among the poorest population. Figure 7. Pension System Beneficiaries, by Income Quintile (2006) 30 25 5, 2 4, 26 27, 20 23 e ag entc 15 2, 16 per 10 8 5 6, 0 I II III IV V Source: Bank's staff calculations based on ECH and ECHA. 87. The unemployment insurance system has the lowest coverage among the income transfer programs. The number of beneficiaries since the early 1990s has widely varied, from 15,000 to 35,000, generally following the economic cycle, as the number of beneficiaries grew when the economy slowed down. These figures, however, are very low if compared to the total number of unemployed workers, as the number of beneficiaries was never more than 6.5 percent of the unemployed9. Figure 8. Unemployment Insurance Beneficiaries 40 35 ) 30 000s 25 ni( esi 20 aric 15 enefi B 10 5 0 1993 94 95 96 97 98 99 2000 01 02 03 04 05 2006 Source: BPS (2006, 2007). 9As Bucheli and Amarante (2006) show, the low coverage is caused by the restrictive conditions established to apply for benefits, and the limited period of coverage. Coverage among those who formally qualify for the benefit is high, but the problem is that many unemployed workers do not satisfy all requirements to apply for a benefit. 27 88. Unemployment insurance is not only low, but many of its beneficiaries are not actually unemployed. Workers temporarily suspended by their employers can also receive this benefit. Since 2001, the proportion of UI beneficiaries that are not actually unemployed oscillated between 50 and 75 percent (figure 9). As a result, the actual coverage of the insurance is even lower than the figures reported in the previous paragraph. Only 4.9 percent of the unemployed in 2002 received a benefit, and the rate has declined to almost 2 percent in 2006, as the total number of benefits declined and the proportion received by suspended workers increased. Figure 9. Unemployed Insurance Coverage Percentage of unemployed population and percentage of beneficiaries who are not unemployed 6 80 % by 74.7 70 of de ec 5 56.5 55.6 56.1 54.6 60 ben ervoc anru 51.3 ef 4 une nsi 11 50 89 m iciar edyolp 4. nte 3 4. 65 40 pl ies 3. oy m ed who m 30 neu oylp 2 75 2. 44 2. 16 20 em 2. are of un1 no % 10 t 0 0 2001 2002 2003 2004 2005 2006 UI coverage UI beneficiaries who are not unemployed Source: Bank staff, based on ECH and ECHA. 89. Urban, middle income, older, and male workers are more likely to be protected by the unemployment insurance scheme. Men make up a large majority among unemployment insurance beneficiaries. Also, coverage in cities is much higher than in rural areas, and beneficiaries' households tend to be headed by individuals older than 40 years (66.8 percent) and include children younger than 18 years (54 percent). The structure of UI beneficiaries by income is interesting, as it is closer to a neutral distribution than other programs. Figure 10 shows that most beneficiaries seem to be middle income workers, excluding the poor groups (who are probably not covered due to long term unemployment or informality) and the richer quintile, who are less exposed to the risk of losing their jobs. 28 Figure 10. Unemployed Insurance Coverage, by Quintile of Per Capita Income (2006) 30 25 8. 6. 26 7. 20 23 22 age entcr 15 8. pe 14 10 2. 12 5 0 I II III IV V Source: Bank staff, based on ECH and ECHA. 90. The third program, family allowances, covers more than 550,000 children. The system reached around 400,000 children of private sector workers in the early 1990s. As the system was reformed to reduce expenditures and targeted to low and middle income formal workers in 1995, the number of beneficiaries fell by nearly 20 percent, and stabilized at around 350,000 children in the second half of the 1990s, as shown in figure 11. The introduction of non-contributory benefits in 2000 compensated a fall in contributory benefits (caused by the decline in formal employment), and the expansion of this program coverage in 2004, together with a recovery of formal coverage in the private sector, explained an unprecedented record coverage of more than 550 thousand children by late 2006. Figure 11. Family Allowances Beneficiaries (as of December of each year), 1991-2006 600 ) 0s 00 500 in( Contributory Non contributory esi 400 ari ic efne 300 B se 200 nc waoll 100 A lyi 0 amF 1991 92 93 94 95 96 97 98 99 2000 01 02 03 04 05 2006 Source: BPS statistical data. 91. When coverage of the family allowances program is measured as the proportion of children younger than 18 receiving a benefit, the expansion of coverage observed during the current decade is very important. By 2001, 29 approximately 47 percent of children were included in the system. Five years later, this figure had grown to 67 percent, mostly due to increases in non-contributory benefits (14 of the 20 percentage points). Also, figure 12 shows that the recovery in private employment also contributed to the overall increase in coverage. Figure 12. Family Allowances System Coverage, by Program Type, 2001-2006 Percentage of children under 18 years old receiving a benefit 70 60 18.4 20.6 8.4 10.4 50 8.9 6.6 aget 40 16.9 15.7 17.7 13.8 12.3 enc 10.9 30 per 20 29.2 30.7 29.2 32.1 32.7 33.8 10 0 2001 2002 2003 2004 2005 2006 Contributory (private sector) Contributory (civil servants) Non contrib. Source: Bank staff, based on BPS data, and civil servants estimates by Arim & Vigorito (2006). 92. Since the family allowances program is targeted by design, it is reasonable to expect a significant concentration of beneficiaries in the lower quintiles of the income distribution. Indeed, 70 percent of beneficiaries of contributory allowances are in the lowest two quintiles of the distribution, while only 11.5 percent are in the top two quintiles. The concentration is higher for the non-contributory benefits, as 75 percent of beneficiaries are in the lowest two quintiles. Figure 13. Family Allowances Beneficiaries Distribution by Quintile of Income per Capita and Type of Benefit 60 0. Contributory Non Contributory 50 57 40 age 6. entcr 30 37 4. pe 29 20 1. 4. 24 19 10 9 9. 3 7 8. 9. 7 9 0. 3. 0 I II III IV V Source: Arim & Vigorito (2006) based on ECH and ECHA 30 93. Finally, the newest income transfer program, Ingreso Ciudadano (IC) has reached almost 80 thousand households in its two years of history. The program was introduced to provide a basic income transfer to the poorest households. As indicated in the earlier documents discussing it, the program aimed at covering households living in extreme poverty, estimated to be around 4 percent of the population, or 40,000 households, in 2004. As the targeting instrument was designed and the program announced, more than 200,000 households applied to receive benefits. Among those, approximately 80,000 where accepted, duplicating the initial estimate. 94. Since the program was explicitly targeted at the poorest households in the country, the program's coverage should be considered by assessing the percentage of households living in poverty or extreme poverty that receive benefits. Registration data indicate that nearly 9 percent of all households in Uruguay received an IC benefit in 2006. According to the household survey, almost 86 percent of households living in extreme indigence applied for the benefit, but it was granted to only 56 percent (due to the application of the proxy means test). Regarding moderately poor households, 54 percent applied and 18 percent received it. Non-poor household participation was very small, since only 5.8 percent applied, but was granted to less than 1 percent (figure 14). Figure 14. Citizen Income Applications and Coverage by Poverty Status, 2006 As percentage 100 80 9. 85 e 7. 60 agt 55 encr 7. 40 53 pe 1. 20 18 8 7 5, 0. 0 extreme poor moderate poor non poor Source: Bank's staff calculations based on ECHA. 95. Thus, citizen income is, by far, the best targeted program among those considered in this report. Figure 15 shows that 75 percent of beneficiaries belong to the lowest quintile of the income distribution, and another 20 percent are in the second lowest quintile. On the other hand, less than one per cent of beneficiaries appear to be in the top forty percent of the income distribution. 31 Figure 15. Citizen Income Beneficiaries Distribution by Quintile of Income per Capita As percentage 80 70 4. 75 60 50 age ntec 40 per 30 20 5. 10 19 3 7 1 4. 0. 0. 0 I II III IV V Source: Bank's staff calculations based on ECHA. 3.5.3 Coverage gaps and overlaps 96. The previous section presented historical trends and current data on coverage for each individual income transfer program, and this section considers how they have interacted, in terms of coverage. Since these programs were designed independently, with different objectives, target population, and enrolment procedures, there is some space for gaps and overlaps in coverage. This section reviews the combined coverage of all programs, indicating the magnitude of coverage gaps (that is, the population not covered by any transfers program) and overlaps (those who are covered by more than one program) among them. In a sense, an assessment of coverage gaps produces an estimation of the level of collective efficacy of the income transfer programs, as it indicates how many potential beneficiaries are reached by them, while assessing the overlaps indicates how efficient these programs are. 97. Slightly over forty percent of households in Uruguay received no benefits from any income transfer program in 2001, a proportion that was reduced by 7.5 percentage points by 2006. This proportion was very stable until late 2004, when the non- contributory family allowances were introduced. This change produced a quick drop of five percentage points in this coverage gap. Part of the change may have been explained by the introduction of PANES and its income transfer program, although this process was slow until late 2005. However, the impact of the citizen income program became clear in 2006, when the coverage gap declined another 2.5 percentage points (figure 16). 32 Figure 16. Gaps and Overlaps of Income Transfer Programs 2001-2006 Percentage of households with no access to income transfer programs ("gaps") and households receiving transfers from two or more programs ("overlaps") ­ as percentage 50 2001 02 03 04 40 05 2006 30 entagec per20 2006 10 05 2001 02 03 04 0 Gaps Overlaps Source: Bank's staff calculations based on ECH and ECHA. 98. Overlaps between programs were also very stable in the early 2000s, and only grew when the new family allowances reform and PANES were implemented. The stability on coverage gaps during the early 2000s can also be seen in the incidence of program overlaps. The proportion of households receiving benefits from more than one program was, in 2001-2004, very stable around 4 percent. Most of these households combined pension system benefits and family allowances, with few cases with unemployment benefits and family allowances. With the extension of the non- contributory family allowances scheme in 2004 and the introduction of Citizen Income in 2005, many families who were already receiving other benefits were included, increasing the percentage of households with overlapping benefits to 6.4 percent in 2005 and 11 percent in 2006. 99. The levels and trends observed in coverage gaps and overlaps had important differences by income level. Of course, not all coverage gaps represent shortcomings for the income transfer policies, since in some cases there is no need for social protection. Figure 17 shows that the gap tended to slightly grow for the poorest quintiles in 2001- 2003 (probably reflecting the decline in employment and labor force formality, which reduced access to family allowances benefits), while the richest quintiles improved their situation. The economic recovery after 2003 and the expansion of non-contributory programs had a strong impact among the poorest: the proportion of households with no transfers declined from fifty percent to less than half that figure in three years. This improvement also affected less vulnerable households, as the size of coverage gaps in the second and third quintiles declined by 32 and 19 percent, but had no impact on the higher end of the income distribution, as these households have no access to the two mentioned programs. 33 Figure 17. Gaps of Income Transfer Programs 2001-2006, by Income Quintile Percentage of households with no access to income transfer programs ("gaps") 60 2001 '02 '03 '04 '05 2006 50 40 age entcr 30 pe 20 10 0 I II III IV V Source: Bank's staff calculations based on ECH and ECHA. 100. Differences in coverage overlaps by income are also important. Until 2004, only 7.5 percent of first quintile households received benefits from more than one program. Figure 18 indicates that two years later, this figure had jumped to 28.5 percent. This sharp increase was caused by the introduction of IC, which could be received together with other transfers. In fact, 80 percent of households that receive IC benefits also received family allowances, unemployment insurance or pension benefits. The effect was also noticeable in higher income quintiles, but less dramatically, as most citizen income benefits were targeted to low income households. Figure 18. Overlaps of Income Transfer Programs 2001-2006, by Income Quintile Percentage of households receiving transfers from two or more programs ("overlaps") 30 2001 '02 '03 '04 '05 2006 25 20 age entcr 15 pe 10 5 0 I II III IV V Source: Bank's staff calculations based on ECH and ECHA. 101. The balance of gaps and overlaps shows that policies introduced in the last few years in Uruguay have been very effective in including some of the poorest households in the income transfers policies. However, they also show that this 34 improvement was achieved with a rather "inefficient" approach, as overlapping grew rapidly. As seen in figure 19, four out of five households receiving citizen income benefits by early 2006 were also receiving other transfers, in most cases family allowances. Figure 19. Households with Citizen Income Benefits, by Type of Overlap with Other Programs, 2006 As percentage CI + others, Only CI, 5.9% 20.1% CI + FA, 73.9% Source: Bank's staff calculations based on ECHA. 102. Programs' overlaps are not bad by themselves. If different income transfers programs, designed to protect individuals or households from specific, independent risks, happen to provide transfers to the same household, this would not necessarily be a problem. However, the analysis in the report shows that most of the overlaps observed in Uruguay are not caused by accumulation of shocks in the same household, but by overlaps in design, as the programs aim at protecting the same population from the same risks. This is clearly the case of family allowances and citizen income, since both provide additional income to reduce poverty incidence to poor households with children. 103. Table 2 shows that, among the poorest quintile of the Uruguayan households, more than one fifth are not receiving any income transfer, while a quarter of them receive two benefits. The coverage gap in the first quintile has sharply declined since the early 2000s, showing the good performance of the new programs in that area. However, it is also clear that there is still space to improve, reaching those who are still excluded from the benefits. 35 Table 2. Overlaps of Income Transfer Programs, by Income Quintile and Number of Programs As percentage Income quintile # of programs covering each household I II III IV V Total 0 22.7 26.0 30.5 40.3 50.6 34.0 1 48.8 61.2 61.0 55.6 48.1 55.0 2 25.2 11.8 8.1 4.0 1.2 10.1 3 3.2 1.0 0.4 0.1 0.0 1.0 0.1 0.0 0.0 0.0 0.0 0.0 4 Total 100.0 100.0 100.0 100.0 100.0 100.0 Source: Bank's staff calculations based on ECH and ECHA. 104. For an international perspective, the new non-contributive schemes in Uruguay have performed very well in terms of targeting. Other programs in the region, such as Brazil's Bolsa Familia; Chile Solidario; Mexico's Oportunidades; and Argentina's Jefes have been less effective. Figure 20 shows that Citizen Income is the best targeted scheme among the group considered, closely followed by the Brazilian program. This success, in a program that was designed and implemented in a few months, indicates a high effectiveness of the targeting strategy. When compared with programs implemented in Europe and the US, Uruguay's citizen income is still among the most effective in targeting the poor. Figure 20. Comparison with LAC Programs: Distribution of Beneficiaries by Income Quintile for Non-contributory Family Allowances and Citizen Income Percentage of total beneficiaries in each quintile 80 70 60 50 40 30 20 10 0 1 2 3 4 5 BR Bolsa Familia (2004) CL SUF-Solidario (2003) MX Oportunindades (2002) AR Jefes de Hogares (2003) UY Citizen's Income (2005) UY NonContrib Family Allowances (2005) Source: Bank's staff calculations based on ECH and ECHA. 36 3.6. BENEFIT AMOUNTS AND WELFARE IMPACTS 105. Understanding the level and distribution of coverage in income transfer programs is an important step in assessing their impact. However, adequacy of these benefits is equally important, because if benefits are insufficient to affect the welfare of beneficiaries, the programs become ineffective. By considering both coverage and adequacy, it is possible to assess the impact of these policies on income distribution and welfare. 106. There are significant differences in the value of benefits provided by each program. Table 3 shows that contributory pension benefits are, on average, equivalent to two minimum wages, or 1.5 poverty lines. Since actual benefits are defined considering each individual's earnings history, age, and length of service, the variation around this average is significant. Minimum benefits are less than one fifth of the average, and maximum are over twice that value. Unemployment insurance benefits are somewhat lower on average, at 1.2 times the minimum wage, also with some variance, while survivors' and non-contributive pension benefits are slightly below the minimum wage. The two transfer programs partially or totally targeted to the poor, citizen income and family allowances, provide significantly lower benefits. CI benefits are equivalent to 49 percent of the minimum wage, while family allowances (per child) barely reach 8 percent. Table 3. Average Benefit Amounts of Income Transfer Programs, late 2006 % of minimum % of poverty Program in current U$ wage line Contributory Pensions $U 4,554 152% 147% Non Contributory Pensions $U 2,567 86% 83% Family Allowances $U 228 8% 7% Citizen Income $U 1,482 49% 48% Unemployment Insurance $U 3,653 122% 118% Source: Bank's staff calculations based on data from BPS, INE, and legislation. 107. The impact of these programs on poverty level is defined by the relation between their benefit value and the poverty line. Only retirement pensions and unemployment insurance benefit were equivalent to more than what an average Uruguayan needs to be considered non-poor. Average retirement pensions have been around two poverty lines since the early 1990s, with a decline between the late 1990s and early 2000s and a recovery since 2003. Unemployment insurance benefits followed the same pattern, at a lower level, and other pension system benefits (contributory and non- contributory) fluctuated around 50 and 90 percent of the poverty line. Family allowances, on the other hand, have been consistently low, as they only provided 7 percent of the money necessary to avoid poverty. 37 Figure 21. Income Transfers Value, 1991-2006 (as a percentage of poverty line) 200% 180% 160% 140% 120% 100% 80% 60% 40% 20% 0% 1989 1991 1993 1995 1997 1999 2001 2003 2005 2007 contributory pensions non contributory pensions unemployment insurance Citizen Income family allow ances Source: Arim & Vigorito (2006), and Bank staff 108. The impact of income transfers on the income distribution can be observed by considering the income profiles with and without them. Figure 22 shows the income per capita distribution, with and without these transfers. Clearly, when transfers are included, the curve shifts to the right, increasing income of almost all households in Uruguay, but specially those below the poverty line. This movement suggests that income transfers have some effect on poverty incidence (as some individuals are shifted to the right side of the poverty line) and a significant impact on poverty gaps, increasing the welfare of those who are still poor. Of course, a complete incidence analysis should also consider how the programs are financed to assess the potential impact of reduced expenditures on the tax structure, but such analysis would require a number of assumptions that are beyond the scope of this report. 38 Figure 22. Distribution of Income Per Capita­with and without non-contributory transfers­ 2006 200 .0 51 00 .0 yti ns de 100 .0 50 00 .0 0 0 5000 10000 15000 ipcfcvl with transfers without transfers Source: Bank's staff calculations based on ECHA 109. There are two possible criteria to assess the distributional impact of transfer programs. First, the programs can be assessed considering whether they have a progressive or a regressive impact on income distribution. If the distribution of benefits across income levels is more progressive than the income distribution before the transfer, a program can be defined as progressive, since it improves the overall distribution. If the effect is the opposite (that is, the distribution of the benefit is worse than that of other income sources, resulting in a regressive impact on overall distribution), then the program is considered to be regressive. The second criterion focuses on whether most transfers go to the poor (in which case the program is considered "pro-poor"), to the non-poor (in which case it is considered to be "pro-rich") or neutral. 110. Figure 23 shows that non contributory programs have a clear progressive and pro-poor profile. The benefit concentration curves for non-contributory family allowances, non contributive pensions and citizen income benefits are above the pre- transfers Lorenz curve, indicating that the distribution of these benefits is more progressive than that of other income sources, and consequently they have an equalizing effect on income distribution. Since the curves are also above the diagonal perfect equity line, it is possible to conclude that poorer households are receiving a more than proportional share of all transfers, thus making these programs "pro-poor". This is especially strong in the cases of citizen income and non contributory pensions, since more than 90 percent of transfers go to the poorer 40 percent of the population, but less evident for non-contributory family allowances and pensions, as the curve is closer to the diagonal. 39 Figure 23. Concentration Curves for Non-Contributory Transfers in 2006 100% 80% 60% 40% 20% 0% 0% 20% 40% 60% 80% 100% Perfect Equity line Lorenz w/o transfers Citizen Income Non Contributoy Pensions Non Contributory FA Source: Bank's staff calculations based on ECH and ECHA 111. On the other hand, the contributory programs have a milder progressive effect, but they are not particularly pro-poor. Figure 24 shows that the three programs reduce the Gini coefficient, but they are very close to the diagonal, indicating that benefits tend to be distributed proportionally along the income distribution and that they cannot be considered significantly pro-poor. Contributory family allowances are more concentrated in the middle of the distribution (hence the "s" shape of the curve), while the other two are more proportionally distributed. 40 Figure 24. Concentration Curves for Contributory Transfers in 2006 100% 80% 60% 40% 20% 0% 0% 20% 40% 60% 80% 100% Perfect Equity line Lorenz w/o transfers Retirement Pensions Unemployment Insurance Contributory FA Source: Bank's staff calculations based on ECHA 112. Income transfers impacts have an important impact on poverty headcount and poverty gap, as they reduce them by 44 and 70 percent. Estimations presented in table 4 show that if no income transfer programs were available, poverty incidence in Uruguay would be significantly higher than the 27.4 percent estimated for the first semester of 2006, at approximately 40.1 percent. The poverty gap would also be much higher, indicating that transfer programs not only reduce poverty, but they also improve the situation of those who remain below the poverty line. Due to its very large size (in coverage and benefit amounts), contributory pensions have the strongest impact both in incidence and the poverty gap. If this program were to be closed immediately, the poverty headcount index would grow by 38 percent, or nearly 11 percentage points, while the gap would grow by more than 50 percent. Other programs, such as the contributory and non- contributory family allowances schemes, have very little impact on incidence and poverty gap, in spite of their large coverage. This is caused by the limited benefit amount that, as discussed above, is around 7 percent of the poverty line. On the other hand, the effect of the citizen income program is interesting, as it has little effect on the headcount index (closing it would increase poverty by 1.6 percentage points), but it has a more relevant impact on the poverty gap, indicating that its benefits are improving the situation of the poor, even if they are not enough to bring them over the poverty line. 113. The impact of these programs on extreme poverty has been even more important. Overall, if no income transfer programs were available in Uruguay, the headcount ratio for extreme poverty would be 2.2 times higher, and the poverty gap would be almost four times higher. Once again, the most relevant program is the contributory pension system, but in this case, the citizen income program is also 41 important, as its absence would produce an increase of 50 percent in the extreme poverty incidence and 75 percent in the extreme poverty gap. Table 4. Income Transfers Incidence on Poverty Measures in 2006 Considering all income sources except... Considering non- poverty non- contributory unemploy- all income all contributory contributory citizen measure contributory family ment sources transfers pensions family income pensions allowances insurance allowances Poverty Headcount 27.4 40.1 37.9 28.1 27.9 27.6 27.5 27.8 impact (%) 46.4 38.5 2.6 2.0 0.7 0.6 1.6 Gap 9.6 17.0 14.5 10.2 10.0 10.0 9.8 10.5 impact (%) 76.7 50.8 5.7 3.7 3.8 1.3 8.6 Extreme Poverty Headcount 2.9 9.2 5.6 3.4 3.1 3.6 2.9 4.3 impact (%) 218.8 94.4 17.4 7.7 25.8 2.4 50.2 Gap 0.6 3.2 1.8 0.8 0.7 0.9 0.7 1.1 impact (%) 393.8 173.4 28.1 9.4 35.9 4.7 75.0 Source: Bank's staff calculations based on ECHA 114. Of course, these impacts are, in a sense, an upper bound estimation, since no behavioral effects are considered. If income transfers were suddenly terminated, it is possible that some of the current beneficiaries would search and find other income sources to replace them, either through the labor market or other channels. If that were the case, then the impact would be smaller than reported. However, there is no evidence to indicate that such effect would have an important impact, especially considering that labor demand for these individuals is still relatively weak. 3.7. FISCAL COSTS AND SUSTAINABILITY 115. Coverage, adequacy and income distribution impacts of income transfer programs are three critical dimensions to assess them, but fiscal sustainability is also critical to ensure that the programs can deliver what they promised over time. This section includes a discussion on the magnitude of fiscal resources involved in the income transfer programs, considering recent trends, expenditures structure, and sustainability. Information on individual programs and aggregate data are presented, assessing the effect of these expenditures on the fiscal situation and the space that current or future income transfer programs may have to ensure adequate financing. 3.7.1 Social Public Expenditures 116. Uruguay's social expenditures are relatively high, as compared with the regional levels. As seen in figure 25, Uruguay was one of the three countries (with Argentina and Cuba) that spent more than 20 percent of GDP on social sectors, significantly over the regional average, at 14 percent. Social spending showed a growing trend during the 1990s, reaching a maximum of 25.1 percent of GDP in 2001. As the crisis affected all public spending, there was a sharp decline in the following years, to 19.5 percent by 2005. 42 Figure 25. Public Social Expenditures, by Country, average 1990-2000 % of GDP 30 25 01. 20 25 e 07 39 ag 34. 20. entcr 15 70. 50. 20. 18 17 pe 10 15. 39. 16 23 13 13 85 11. 5 48 21 02 50 53 08 37 8. 8. 8. 7. 7. 49 5. 6. 6. 4. 0 GU DR PE EC NI PY BO MX VE CO CH LAC CR PA BR UY AR CU Source: Bank staff, based on ECLAC (2005). 117. The main determinant of the recent reduction in social expenditures has been the decline in real wages that affected most civil servants in Uruguay as the economic crisis developed in the early 2000s. Since pension benefits are indexed to wages, social security expenditures also followed the same pattern. Overall, social security and assistance explains more than 60 percent of all social spending, while education and health (including public health and insurance) explain between 15 and 17 percent each, as seen in Figure 26. Figure 26. Public Social Spending, by Category, 1985-2005 % of GDP 30 Social Security and Assistance Education Health Other 25 2.5 2.1 2.2 2.0 3 4 2 9 20 4. 4. 4. 3. 1.7 6 1.5 1.5 2 1 3 4 3. gea 5 9 8 8 8 8 3. 3. 3 4 1. 1. 1. 3. 3. nte15 5 1.9 1. 1. 3. 3. 0 9 8 8 8 3 rc 4 1. 1. 2. 1. 1. 1. 3. 0 3 pe 1.7 1.6 1.7 1 1. 3. 3. 1. 0 1. 10 5 2. 1. 15.3 15.3 15.5 15.4 13.5 13.4 13.9 14.0 13.8 13.8 5 11.3 12.2 13.1 11.5 11.8 9.1 8.9 10.4 0 1980 85 90 91 92 93 94 95 96 97 98 99 2000 01 02 03 04 2005 Note: Data befote and alter 1999 are not completely comparable, due to methodological changes in the source. 43 Source: Own, based on OPP (2004), Ferrera-Coimbra and Forteza, (2004); MIDES-OPP (2006), and BPS (2006). 118. The four main income transfer programs in Uruguay (pensions, family allowances, unemployment insurance and citizen income) explained 52 percent of all social public expenditures in 2005, or 10.2 percent of GDP10. This figure is somewhat smaller than the total social security and assistance spending estimations, mostly because some programs run by BPS and other social assistance programs are not income transfers, but provide access to services (such as food, tourism, or social promotion activities). Most of these expenditures correspond to the pension program, including old age, disability and survivors' benefits. Income transfer expenditures grew to 10.8 percent of GDP in 2006, partly due to a recovery in the pension program expenditures and partly due to the full implementation of the IC benefit. 119. Pension expenditures explain most of income transfer programs spending, and have been driving the overall trend since the early 1990s. Total pension expenditures (including those from BPS programs and those at the military and police pension funds) rapidly grew from below 10 percent of GDP to almost 12 in the early 1990s, and then again to almost 13 percent of GDP in 1999-2001 (figure 27). As the crisis eroded the real value of benefits, there was a new decline to the pre-1990s levels, and in 2005 expenditures were 9.5 percent of GDP. As real benefits began to recover, expenditures also grew back to 9.9 percent in 2006. The introduction of the non- contributory family allowances in 2000 produced a significant change in the income transfer policies strategy, since this was the first sizeable program that was not linked to the labor market, although it barely affected the level and structure of spending. Non- contributory benefits slowly grew, and by 2006, the expanded non-contributory family allowances program and the citizen income program together explained 0.5 percent of GDP, or five percent of all income transfers. 10Administrative expenses at BPS and other programs amounted to an additional 1 percent of GDP. 44 Box 4. Estimating Social Public Expenditures in Uruguay Uruguay has a relatively short tradition regarding publication of social expenditures data. In 2004, the Office of Planning and Budget published a report consolidating information on social expenditures in the public sector for the period 1999-2003 (OPP, 2004). In the same year, ILO published a report on social protection that included social expenditures data since the 1980s (Ferreira-Coimbra and Forteza, 2004), and a team of independent consultants financed by IADB prepared another report (Flood et al, 2004). Unfortunately, none of these efforts was continued in a systematic way. MIDES and OPP prepared a joint report in 2006, reviewing the methodology applied by OPP in 2004, and including estimations for social public expenditures for the period 2002-2005. This new methodology introduced some changes in the scope of public expenditures, excluding social expenditures of state owned enterprises, with the exception of the Public Water and Sanitation Enterprise (OSSE) and including the BPS health insurance expenditures as part of the public sector. The 2006 report did not include local level expenditures, due to the lack of reliable data, and reviewed the series published by OPP for 1999-2001 to make them consistent with the new definitions. These reports provide important information about social expenditures, but they also show the limitations that affect this area of knowledge. Historical series start only in 1999, and some aspects of the most recently published data are not fully comparable with the previous report. Also, the lack of municipal level data, and the absence of a permanent program to publish periodic updates limit the impact of these effort. In this context, the need to set up a permanent program to collect and disseminate social expenditure data could be one of the responsibilities that MIDES takes on for the medium term. Figure 27. Income Transfer Expenditures, by Program, 1990-2006 % of GDP 15 Pension Unemployment Insurance Family Allowances Citizen Income 14 0.3 0.3 0.4 0.3 13 0.5 0.4 0.5 0.6 0.3 0.3 0.3 0.3 geatne 12 0.4 0.3 0.3 0.3 0.2 0.2 0.3 0.3 rc 0.3 pe11 0.3 0.3 0.2 8 8 6 5 0.3 0.3 9 9 9 12. 12. 10 9 12. 12. 0.2 0.2 5 4 0.3 0.4 11. 11. 11. 11. 0.3 0.2 0.1 5 11. 11. 9 0.2 0.4 10. 9 10. 0.2 7 5 6 9. 4 9. 9. 8 9. 8. 8 1990 91 92 93 94 95 96 97 98 99 2000 01 02 03 04 05 2006* Notes: (*) Data for 2006 is estimated, based on budget data. The "y-axis" is out of scale, to facilitate the presentation of the smaller programs' expenditures Source: Bank staff, based on ATPS-OPP (2006), Ferreira-Coimbra and Forteza (2004), and BPS (2006). 45 Financing income transfers 120. Income transfers in Uruguay are financed by a combination of contributions, earmarked taxes, pure public subsidies and inter-program subsidies. The financial scheme for income transfer programs in Uruguay is rather complex. Some programs are directly financed by the treasury, through the national budget, others have an important contributory component and others are implicitly subsidized, with no clear financial source. In one extreme, Citizen Income is fully financed by the national budget, as part of the Ministry of Social Development expenses (which, in turn, transfers the funds to BPS to process payments). On the other, the pension programs are defined as contributory schemes, but both BPS and the police and military schemes run significant deficits. In the early 1990s BPS was able to finance almost 80 percent of its spending on income transfer programs with contributions made by employers and workers, but this proportion began to decline as the population of pensioners grew faster than the number of contributors. When the pension reform of 1996 introduced a funded scheme, BPS self- financing was already at 53 percent, and the reform accelerated the process. By 2003, only 40 percent of BPS revenues came from contributions. The remaining funds were transfers from general revenues, either in the form of earmarked taxes or direct transfers from the treasury, needed to balance BPS finances. The declining trend in "genuine" funding stopped in 2003, and by 2006 the percentage of BPS' income transfer programs funding coming from contribution had recovered to levels similar to those of the late 1990s, around 50 percent. 121. General revenue transfers to BPS for the financing of income transfer programs have two components. First, the legal framework establishes that all or part of certain tax collections is to be automatically transferred to BP. These taxes include VAT, lottery taxes, and a special tax on manufactured imports ("COFIS"). Transfers of these earmarked taxes represented between 15 and 18 percent of all BPS funds during most of the 1990s and early 2000s, a figure that was not enough to close the financial gap between revenues and expenditures. In consequence, the Government was forced to make additional transfers to close the deficits. These transfers reached 36 percent of all revenues in 2002, an all time high, when it amounted to UR$12.3 billion (4.7 percent of GDP). The recovery in contribution and tax collection observed after 2003 produced a rapid decline in this figure, and by 2006 it was UR$8.5 billion, or less than 2 percent of GDP. 46 Figure 28. Financing of BPS Income Transfer Programs, 1991-2006 % of total revenue 100 80 60 gea ntecr pe 40 20 0 1991 1993 1995 1997 1999 2001 2003 2005 Contributions Earmarked Taxes Treasury Transfers Other revenues Notes: (*) Data for 2006 is estimated, based on budgeted funds. Source: Bank staff, based on BPS (2006 and 2007). 3.7.3 Sustainability 122. Given the current spending structure of all income transfer programs, a discussion about sustainability in the medium and long term must be, unavoidably, about the medium and long-term perspectives of the pension system. As shown in figure 27, in 2006 the pension system spent nearly 10 percent of GDP, while the other three programs together barely reached 1 percent of GDP. 123. This report does not focus on the issue of sustainability by itself. However, it is important to acknowledge the relevance of this problem, since no policy reform will be possible if the fiscal implications for the medium term are unsustainable. The 1996 pension reform introduced several limitations to the growth in expenditures over time. In fact, pension spending, as a percentage of GDP, has been stable during most of the last decade, with some growth in 1999-2002 (due to a decline in GDP), which was compensated after that when GDP recovered faster than pension expenditures. 124. Existing pension expenditure projections predict a declining trend in the next few decades. Projecting financial flows for pension system in the long term is not an easy task, as it requires a number of assumptions on the evolution of key variables, including labor market trends, inflation, macroeconomic situation, and real wage growth. Staff at the technical advisors office in BPS prepared a projection until 2050, showing the expected evolution of the main variables in the system. In this projection, BPS estimated that contribution collection will be very stable, with a slight decline caused by the maturation of the 1996 reform. Earmarked taxes are assumed to evolve similarly, but 47 expenditures would decline to around 5 percent of GDP, mostly due to a decline in coverage of survivors' benefits and the maturation of the new system, that provides lower benefits from the PAYG component to retirees. Thus, treasury transfers, which amounted to 25 percent of all BPS expenditures by 2004, would slowly decline and disappear by 2030. Figure 29. Projected Financial Flows: BPS Pension System 2010-2050 % of GDP 8 Earmarked taxes Contributions Expenditures 7 6 e 5 ag ntecr 4 pe P D 3 G 2 1 0 2010 2020 2030 2040 2050 Source: Bank staff, based on BPS (2005) and ECLAC. 125. The decline in expenditures projected for the next twenty years is partially caused by an expected decline in coverage among the elderly. The strengthening of vesting requirements introduced by the 1996 reform, together with a decline in formality brought the authors of the projection to estimate that the proportion of the elderly with access to pension benefits will slowly decline in the next couple of decades. If policy reforms are adopted in the near future to reduce this effect11, then coverage and expenditures would be higher. Overall, these projections seem to indicate that the system does not face serious sustainability problems, as long as no major reforms are introduced in the medium term. 11For example, several sectors are proposing to reduce the required number of years with contributions to the system from thirty five to thirty. 48 4. THE NEW "PLAN DE EQUIDAD SOCIAL". POTENTIAL IMPACTS AND COSTS 126. As the emergency social plan PANES approaches its closing date in late 2007, its main income transfer component, the citizen income program, will be phased out and a new policy will be implemented. In that context, Uruguay faces a unique challenge and opportunity to redesign its income transfer programs aimed at reducing poverty and inequity, in order to increase their impact and sustainability. The Government has announced that a new program, called the Social Equity Plan (Plan de Equidad Social- PES) will be implemented in early 2008, and details on its design have been under discussion and expected to be defined in mid 2007. While it has been clearly stated that PES will not be a simple continuation of PANES and its component programs, it is reasonable to consider it as the long-term structural social program designed to reduce poverty and inequity. 127. The new program might be not a simple substitute for PANES to ensure continuity of benefits to those in need of support, but could represent a shift of approach from the emergency towards a more strategic approach. PANES was designed and implemented in the context of a serious social emergency, when unemployment, poverty and extreme poverty reached historical records. Since the worst of the crisis has passed, the Government is now focusing on the design of a long-term strategy to support and protect those most vulnerable and poor. 128. The new approach recognizes an underlying rationale on the life cycle framework discussed in section 2. PES will focus mostly on children, since they are the poorest and most vulnerable, aiming at improving their living standards and, as a consequence, breaking the vicious circle of poverty. To enhance the chance of having a lasting impact, the proposed program would aim at providing a strong incentive to complete secondary education among children of poor families. In addition, the program might spend some resources to support the poor elderly that is not protected by the pension system. While poverty incidence among the elderly is relatively low, this action is justified by the fact that those who fall below poverty in this age group rarely have the opportunity to revert the situation through the labor market, as opposed to the situation of other, younger individuals. 129. The main challenge for the new policy will be around three core issues: coverage, sustainability, and system integration. The previous sections showed that the existing programs have some important gaps and overlaps in coverage, that the poverty and equity impacts are not as large as they could be, especially among children, and that the high fiscal cost of existing programs leaves limited space to introduce new schemes. In that context, the challenge is clear: how to design a new program that provides adequate coverage to the poor and vulnerable, avoiding inefficient overlapping with other income transfer programs and negative impacts on the labor market, and ensuring smooth implementation and fiscal sustainability. 130. This section aims at providing some possible answers to this challenge. It seems clear that any proposal should consider the role of the income transfer programs currently targeted at the poor. Thus, the focus is on non-contributory transfers (family 49 allowances, non-contributory pensions, and citizen income), which will most likely be part of the new PES. The other income transfer policies (including contributory pensions and unemployment insurance), will necessarily have an important role in the overall strategy, but they are not part of the current debate. 131. This chapter discusses possible impacts of the new program, considering its effects under several scenarios. The modeled options (which, of course, do not represent exactly what the Government will eventually propose as the basic design of PES12) are presented describing their main components, to then simulate the impact that they would have. Six outcome indicators are presented to assess these impacts, including poverty incidence and gap; extreme poverty incidence; coverage gap among the poorest quintile of the population; income inequality (Gini coefficient); and the fiscal implications, to show the cost and benefit trade-off involved in each case. The reforms are presented in a cumulative way, to show the incremental effect that each measure could have. 132. There are four types of reforms discussed in this section. Starting from the current situation, section 4.1 presents a first scenario considering the potential impact that closing down the IC program would have, if no other action were taken. This scenario is considered neither likely nor desirable, but it is presented to set up the following ones. Then, section 4.2 includes a set of two scenarios measuring the effects of changes in the amount paid to family allowances beneficiaries. The third set of scenarios considers the impact of expanding coverage of family allowances and non-contributory pensions to all individuals who currently qualify but, for some reason, are not receiving the benefits and, in the case of non-contributory pensions, to individuals aged 65 and more. Finally, section 4.3 analyzes the impact that a change in the targeting criteria could have. 4.1. THE IMPACT OF CITIZEN INCOME 133. Citizen income has had an important effect on extreme poverty. Established as a temporary program, PANES will reach an end sometime during 2007. This scenario statically simulates the impact that closing down IC would have, assuming that no other policy reform is introduced. While this is not a likely policy scenario (as the Government has already announced that the new Plan de Equidad Social will be introduced), it provides a baseline to discuss the impact of other possible reforms. Six indicators are considered to assess the impact of this and other policy options: extreme poverty headcount ratio, total poverty headcount ratio, poverty gap, coverage gap (for the first income quintile), income distribution (Gini coefficient), and annual cost, expressed in current Uruguayan pesos and percentage of GDP13. Official figures from INE (2006) for 12 However, they are close to what several media reports have indicted would be the design of the new program. On December 22, 2006, "Brecha" informed that the new PES would include a new family allowances program, with larger benefits that would decline as the number of children in a household increase. Also, on March 28, 2007 "El Observador" informed that the new PES would include, as income transfer components, an extended family allowances program to cover 490,000 children and 7,122 new non contributory old age pensions, including beneficiaries aged 65 and more. 13Poverty, extreme poverty and poverty gap (also known as FGT-1) were calculated for the total urban population in towns of 5000 or more inhabitants (Montevideo and Interior), according to the poverty line methodology of 2002. Gini coefficient was estimated using per capita real income (adjusted by CPI), 50 the first semester of 2006, together with annual expenditure reported for 2006 have been taken as a baseline for comparisons with the simulated results. For a clearer visualization of the simulated changes, results are presented in a table and a hexagon graph showing all indicators, normalized to one at the base scenario. 134. This scenario shows that citizen income is targeted at the extreme poor. According to INE (2006), 2.9 percent of Uruguayans were extremely poor in early 2006, and 27.3 percent were poor. Almost 23 percent of the poorest quintile of the population received no income transfer. The Gini coefficient was 0.456 and total expenditures (including pension programs at BPS, family allowances, citizen income and unemployment insurance) in 2006 were almost UR$44.9 billion, or 10.4 percent of GDP. These normalized figures are represented by the solid line on the pentagon figures in the right panel of figure 30. Without the income transferred by this program, extreme poverty would be 4.3 percent, some 1.4 percentage points, or 50 percent higher than the baseline case, as shown in the north axis in the pentagon figure. The coverage gap of the poorest quintile would grow by almost three and a half percentage points, to 27.2 percent. The other indicators would remain relatively unaltered, as poverty would increase by only 0.5 percentage points (or 1.7 percent); inequality would be slightly higher and aggregate annual cost from transfers would be reduced by UR$1,270 million, or 0.3 percent of GDP (approximately US$52 million). Figure 30. Simulation Results: Actual Data for 2006 and Simulated Results without IC Transfer Scenarios Baseline Baseline No IC No IC Extreme poverty Extreme 1.5 2.87% 4.26% poverty 1 Poverty 27.37% 27.81% Annual cost Poverty 0.5 Poverty gap 9.64% 10.44% 0 Coverage 22.87% 27.17% gap Gini Poverty gap Gini 0.4519 0.4564 U$ million 44,898 43,625 Annual Cost (as of 2006) Coverage gap GDP% 10.4% 10.1% Source: Bank's staff calculations based on ECHA without implicit rental values, for the first semester of 2006, using INE's methodology. The coverage gap, refers to the proportion of first quintile households that do not receive pensions, unemployment, family allowances or citizen income benefits. 51 4.2. CHANGING THE STRUCTURE OF THE FA BENEFITS 135. The second scenario measure is an increase in family allowances benefits, for those already receiving them. While coverage of family allowances has significantly extended in the last few years, the amount paid to beneficiaries is small, limiting its impact on households' welfare. Among the possible reforms that could be introduced with the new Plan de Equidad Social, it has been suggested that the amount paid by the FA program should be modified, at least for those below the 3 BPC income thresholds. The proposals involve an overall increase in the benefits, but also the introduction of two sliding scales, that would consider economies of scale within families and would promote permanence in the formal education system for older children. 136. Two separate scenarios are presented to assess the marginal effect of each proposal. The change would affect 257,000 children under 18 years old and living in households with income below 3 BPC, that are currently receiving benefits (while another 300,000 children would continue to receive the current values). The new benefit structure would grant UR$1,000 per month for the first (and older) child, UR$750 for the second, UR$500 for the third, and UR$250 for the fourth onwards. Since simulation scenarios are accumulative (for the simplicity of exposition), the previous scenario is considered as the basis for the current simulations. Thus, this change in the amount of the FA benefits is applied on the basis of a scenario without the transfer from IC program. 137. Increasing the amount of the FA benefits more than offsets the potential rise in extreme poverty resulting from the removal of IC, as it would bring the extreme poverty index to 2.23 percent, or 77 percent of the baseline level. Poverty would also decline, but by less than one percentage point. Coverage in the first quintile would not change, since the considered policy would only affect benefits of those already within the system14. Income inequality would also decline, and would be slightly better than in the baseline scenario, as the Gini coefficient would decline by almost 1.7 percentage points with respect to the current situation. Finally, this scenario is more expensive than the current situation. The cost of introducing this change in the family allowances benefit would be around UR$2,160 million (net of current expenditures on family allowances for the same households), 70 percent higher that the amount transferred as citizen income benefits during 2006. Thus, this reform implies an increase in expenditures of 0.21 percent of GDP in relation to the amount spent during 2006 on income transfer programs. 14The coverage gap presents the proportion of households that receive no transfers, in the first quintile of the income distribution. Given that this distribution is recalculated after each simulation, the population in the quintile changes and the coverage gap shows slight variations. 52 Figure 31. Simulation Results: Scenarios without IC and with New Decreasing FA Benefit per Child. Scenarios No IC FA No IC Extreme poverty dec.p/child FA dec.p/child 1.5 Extreme 4.26% 2.23% poverty 1 Annual cost Poverty Poverty 27.81% 26.94% 0.5 Poverty 0 10.44% 9.13% gap Coverage 27.17% 27.97% Gini Poverty gap gap Gini 0.4564 0.4487 Annual U$ million 43,625 45,870 Coverage gap Cost GDP% 10.1% 10.6% Source: Bank's staff calculations based on ECHA 138. Including a rising scale per completed year of education to promote higher attainment among secondary level students would have a minimum effect in terms of poverty. This scheme could be an addition to the program, with the objective of promoting school attendance in teenagers.15 Assuming that this would be implemented by increasing the benefits by 1 percent per year of schooling, a child in the first year of primary school would see his or her benefit increased by 1 percent, while those in the last year of secondary school would receive an additional 12 percent. This measure impact would be minor on poverty and equity, and have a limited cost, as it is possible to see in figure 32. The marginal effect on extreme poverty would be 0.07 percentage points; poverty would decline by 0.03 points, and the Gini coefficient would decrease by 0.003 points. The fiscal costs of the programs, meanwhile, would go up by UR$106 million a year. Because these differences are so small, they can barely be seen in the hexagon graph in Figure 32. However, it is important to note that the goal of this change would not be to have a direct impact on poverty or inequality, but would aim at producing a behavioral change among teenagers and their families, increasing their demand for education and the proportion that completes secondary school. Thus, an adequate assessment of its impact should consider the evolution of enrolment rates and not the static impact on poverty or equity indexes. 15It is possible to adopt different method for this objective. The possibility of granting a fixed additional sum for children who remain in school after a certain age, a differential allowance per school level, etc, is among these options and has been simulated in this section. The simulated option here represents only a possible approximation of the secondary schooling problem. 53 Figure 32. Simulation Results: Scenarios with FA Benefit per Child and Additional Benefit by Years of Schooling. Scenarios FA dec.p/child FA FA inc Extreme poverty FA inc p/school dec.p/child p/school 1.5 Extreme 2.23% 2.16% poverty 1 Annual cost Poverty 0.5 Poverty 26.94% 26.91% 0 Poverty 9.13% 9.07% gap Coverage 27.97% 28.01% Gini Poverty gap gap Gini 0.4487 0.4484 Annual U$ million 45,870 45,976 Cost Coverage gap GDP% 10.6% 10.7% Source: Bank's staff calculations based on ECHA 4.3. EXTENDING THE FA COVERAGE AND NON-CONTRIBUTORY PENSIONS 139. The authorities could also implement reforms to extend coverage of income transfer programs. The previous subsection discussed the impact that eliminating the citizen income program and redefining benefits for the family allowances program would have on different indicators. These reforms would require legal changes, and would be relatively simple to implement. On the other hand, not all children that qualify under the current rules of the family allowances program receive a benefit. In fact, nearly 125,000 children, or fifty percent of those currently in the program, should be included to achieve full coverage of the system. Of course, enrolling these children may be difficult, as proven by the fact that their parents have not applied until now. 140. Failure to reach these children in the past has been attributed to several causes, which could be reversed if appropriate measures are taken. These include the low benefit level (that creates insufficient incentives for some households), lack of information from parents, difficult legal situations (lack of personal documentation, irregular family structure), and the exclusion of some households produced by the school attendance condition. Some of these difficulties might be overcome thanks to the increase in benefits (as participating in the program becomes more attractive, families would be more active to enroll), and the availability of PANES' records, as many children that were not receiving family allowances were identified as their families registered to become PANES beneficiaries. Even so, it is important to note that an expansion in coverage would require a very proactive approach from the Government, and that, in any case, full success is not guaranteed. 141. Extending the FA coverage to all children under 18 and living in households with income below 3 BPC would have a significant impact on poverty and income 54 inequality. Figure 33 shows that this extension would result in a reduction in inequity, a sharp decline in extreme poverty incidence (that would decline to less than half of the baseline value), and reductions in poverty and the coverage gap of the poorest quintile. 142. Extending the FA coverage results in a high cost/benefit relationship. Paying new benefits at a higher unit value would inevitably increase costs, but the marginal impact would be higher than with any other scenario. Aggregate expenditures would rise by UR$2,724 million with respect to the previous scenario and UR$3,802 million if compared with the baseline value. This means that each point of poverty reduction would cost UR$ 3,395 million a year, the highest cost/benefit relationship in the simulated options. Even though the cost of reducing extreme poverty by one percentage point is less than the cost for poverty, the values are still high in terms of the other simulated alternatives (see table 7). Figure 33. Simulation results: Scenarios with Revised FA Benefit Structure and Extending Coverage to All Qualifying Children Scenarios FA inc FA inc FA Extreme poverty p/school p/school ext.coverage 1.5 FA ext.coverage Extreme 2.16% 1.52% poverty 1 Annual cost Poverty Poverty 26.91% 26.25% 0.5 0 Poverty 9.07% 8.52% gap Coverage 28.01% 16.19% gap Gini Poverty gap Gini 0.4484 0.4447 Annual U$ million 45,976 48,700 Cost GDP% 10.7% 11.3% Coverage gap Source: Bank's staff calculations based on ECHA 143. An additional proposal discussed in Uruguay regarding extending coverage of the income transfer programs involves the non-contributory pension scheme. As part of the "exit strategy" for PANES, the Ministry of Social Development started a process in 2006 to provide a non-contributory pension for individuals that, while qualified for it, were receiving citizen income benefits. Furthermore, the proposal could also include a reduction in the minimum age required to qualify for this benefit, from 70 to 65 years. 144. The results of a simulation that includes the extension of non-contributory pension coverage indicate that the impact of this measure would be limited. Nearly 38,000 new beneficiaries would be added to the system including those aged 65 or more and living in households with total income below 3 BPC, with a monthly benefit of UR$2,930. Neither poverty nor extreme poverty varies significantly with respect to the previous scenario. Changes of .05 and .35 percentage points, in extreme poverty and poverty, respectively and a reduction of 0.001 in the Gini coefficient make this the 55 scenario with the weakest impacts so far. The coverage gap for the poorest quintile would decline by 1.3 percentage points. Since the cost of this reform would be around UR$1,158 million, this is the worst scenario in terms of cost/benefit relationship. For a 1 percentage point decrease in the poverty rate, expenditures must increase by UR$3,375 millions. Figure 34. Simulation Results: Scenarios with Extended Family Allowances Coverage and Extended FA and Non-Contributory Pensions Coverage Scenarios FA Extreme poverty ext.coverag FA+NC Pension FA coverage e 1.5 coverage FA and NC pensions Extreme 1 1.52% 1.47% poverty Annual cost Poverty 0.5 Poverty 26.25% 25.90% 0 Poverty 8.52% 8.37% gap Gini Poverty gap Coverage 16.19% 14.90% gap Gini 0.4447 0.4433 Coverage gap Annual U$ million 48,700 49,858 Cost GDP% 11.3% 11.6% Source: Bank's staff calculations based on ECHA 4.4. COMPARING ALTERNATIVE TARGETING CRITERIA 145. This section discusses the impact that a change in targeting criteria could have on poverty, equity and costs. So far, the simulated scenarios adopted the approach currently in use by the family allowances and non-contributory pension systems as targeting criteria16. While using labor earnings to target benefits has several advantages (including simplicity and data availability), it would be interesting to explore the impact that a change of criteria could have. 146. Most social programs in Uruguay have used labor income as the targeting tool, but a proxy for poverty was first used when PANES was introduced in 2005. The traditional approach used to compare individual or household income with a threshold (usually a multiple of the minimum wage or, more recently, the BPC.) When PANES was implemented in 2005, a new targeting method was adopted, and a proxy means test was applied to identify households that were below the extreme poverty line. 16To be eligible for a non-contributory FA benefit, the household needs to earn less than 3 BPC. For the informal sector, this means test cannot be done through the administrative records from BPS. Hence, a sworn statement is presented as proof of that. For the simulations, only labor income has been taken into consideration, instead of total family income. The latter was considered inferior to the former given that certain items of income are estimated rather than declared as perceived, i.e. rent value. 56 For this, detailed home visits and interviews had to be organized, a process that took significant time and resources and in some cases became the object of political discussions. 147. Traditional programs considered the ratio of income to minimum wages. More recently, the basis for these calculations was replaced by an index that is regularly updated ("Base de prestaciones contributivas"- BCP). In the case of FA, a threshold of 10 BPC has been used to determine who perceives benefits17 in the formal sector (while those earning less than 6 BPC qualify for double benefits). For households with informal workers, if earnings are below 3 BPC, the children may qualify for family allowances. To be eligible for a (means-tested) non-contributory pension, the claimant has to be at least 70 years old and the household income must be below 2 or 3 BPC, depending on their marital status18. 148. PANES participants, including those receiving citizen income benefits, were selected through a more sophisticated mechanism. Home visits by MIDES staff and volunteers were used to compile data on household characteristics, which in turn was used to build a weighted index and assign a score value to the household. A cut-off value on that score determined eligibility. Although neither the perception of other transfers nor belonging to the formal sector were reasons to automatically reject applications, beneficiaries are largely over-represented by the labor informal sector. 149. In short, "traditional" non-contributory transfers use an income test to target beneficiaries, while PANES (and IC) used a more complex proxy means test. While the income test is simpler to manage and very accurate among formal employees, it is subject to moral hazard problems usually associated with self reporting of income or other characteristics. 150. These different targeting criteria produce clearly different results. Table 5 presents the estimated number of children below 18 years old in 2006, according to perception of FA benefits, total labor income in the household and poverty19. Out of 867,000 children, 44 percent were living in households with total income below 3 BCP, and 48 percent were living in poor households. While these figures are similar, the overlapping between both definitions is not perfect: nearly one third of poor children were living in households with total income over 3 BPC, and nearly one third of children living in households with income below 3 BPC were not poor. 17The perception of FA benefits is bounded above (10 BPC) for households with 2 dependent children. For larger households (more children) the boundary is an increasing function that depends on the number of dependent children (see Section 2) 18For the non-contributory pensions, the formal requirements establish that the members of the household, responsible for the elder, cannot earn more than 3 BPC if they are married and 2 BPC if they are single. For the sake of simplicity and homogeneity, household labor income below 3 BPC has been established as the threshold to be considered eligible. 19In this section on the discussion about targeting criteria, poverty was estimated without considering the benefit coming from Citizen Income. It is assumed that at the time of implementing PES, the IC will not be in force, thereby increasing the number of beneficiaries who would qualify for the new program. 57 Table 5. Targeting Criteria: Children under 18 Years Old by Participation in the FA Program and Poverty Total Poor Non poor Total 867,011 417,726 449,285 Receiving FA 557,336 327,739 229,598 Household labor income >= 3 BPC 299,610 113,906 185,703 Household labor income < 3 BPC 257,726 213,832 43,894 Not receiving 309,675 89,988 219,687 Household labor income >= 3 BPC 182,565 25,467 157,099 Household labor income < 3 BPC 127,109 64,521 62,589 Attending school 101,012 45,990 55,022 Not attending 26,097 18,531 7,567 Source: Bank's staff calculations based on ECHA 151. A similar problem affects the elderly. Table 6 shows that out of 456,000 Uruguayans aged 65 and more, slightly more than 37,000 are poor. Among them, 10,800 are not receiving a pension, even if they could qualify for a non-contributory benefit. On the other hand, there are 330,000 elderly Uruguayans living in households that could qualify for a non-contributory pension benefit. Eighty eight percent of this group is already covered by the pension system, but 34,000 low income elderly are currently uncovered. Table 6. Targeting Criterion: Individuals Older than 65 years by Perception of Pensions and Poverty Total Poor Non poor Total 456,359 37,130 419,229 Receiving Pensions 391,240 26,341 364,899 Household labor income >= 3 BPC 95,166 4,921 90,245 Household labor income < 3 BPC 296,074 21,419 274,655 Not receiving Pensions 65,119 10,790 54,330 Household labor income >= 3 BPC 30,924 2,525 28,399 Household labor income < 3 BPC 34,195 8,264 25,931 Source: Bank's staff calculations based on ECHA 152. Changing the targeting criteria would imply important changes in the covered population. If Uruguay's Government decides to implement a new targeting strategy under PES and provide benefits under family allowances and non-contributory pensions to children and elderly living in poor households, it would have to include 90,000 new children, and exclude many beneficiaries that are currently receiving non- 58 contributory family allowances but would not qualify under the new rule. Similarly, 10,800 elderly would be included in the non-contributory pension system, but many others would have to be excluded. 153. Of course, excluding current beneficiaries from the programs would be difficult due to social, political, and legal reasons. Thus, the final scenario, presented in figure 35, compares the impacts of the reform presented in figure 34 (where family allowances benefits were increased with sliding scales and coverage of both FA and non- contributory pensions expanded) with a similar scenario where the extension of coverage is focused on the poor instead of those with low total household income. While in the first case almost 685,000 children would receive family allowances and 435,500 elderly a pension, in the second scenario the number of family allowances beneficiaries would be 644,000 and the number of pension recipients would be 402,300. In other words, there would be less beneficiaries, but benefits would be better targeted at the poor. Figure 35. Simulation Results: Scenarios with Extended Family Allowances Coverage and Extended FA and Non-Contributory Pensions Coverage Scenarios Full (Income Extreme poverty Full (Income Full (Poverty criterion) 1.5 criterion) criterion) Full (Poverty criterion) Extreme poverty 1.47% 1.44% 1 Annual cost Poverty 0.5 Poverty 25.90% 25.28% 0 Poverty gap 8.37% 8.11% Coverage gap 14.90% 15.29% Gini Poverty gap Gini 0.4433 0.4437 U$ million 49,858 46,918 Annual Cost Coverage gap GDP% 11.6% 10.9% Source: Bank's staff calculations based on ECHA 154. The results of this final simulation indicate that targeting based on income is more expensive than doing so using poverty status, as it includes more beneficiaries, but also more effective, for the same reason. While targeting those people with household labor income below a threshold of 3 BPC would cost UR$4,960 millions more than the baseline scenario (1.2 percent of GDP), expanding coverage to the poor would raise expenditure by UR$2,020 millions, about 0.5 percent of GDP. Poverty would decline by 1.47 percentage points using the first criterion, and by 2.09 points in the second case, if compared with the baseline. A strategy for the Plan de Equidad Social that targets beneficiaries using the income criterion would have a more important impact on income distribution than using poverty as a targeting strategy. Finally, the incidence on extreme poverty is only marginally different. Although the total impact on outcome indicators would be smaller using poverty to target benefits, this approach would be more efficient than using household income. Reducing the poverty rate by 1 percentage point 59 would cost UR$1,211 million if income were used to target benefits, and UR$1,027 million if poverty status is used as a targeting principle. 155. Whilst using the poverty targeting criteria is most effective in the reduction of indicators associated with poverty (indigence, poverty and the poverty gap), using an income criteria has a higher incidence of reducing income distribution inequality, at the same time improving the coverage of income transfer programs. However, this second targeting criteria is more burdensome given that it includes people who qualify as beneficiaries that live in homes whose income is below the threshold considered: 3 BPC. The reduction by 1 percentage point of poverty index would have a cost of UR$3,375 million if the benefits were based on income, whilst if the selection of beneficiaries were based on the state of poverty the cost would be UR$967 million. 156. The simulations results, summarized in table 7, show that PES could improve the impact on welfare indicators, but at a cost. Scenario 6 presents the simulation that is probably closest to what is being considered by Uruguay's authorities. This alternative assumes: (i) a sharp increase in family allowances benefits for the older child in households with total income below 3 BPC; (ii) the introduction sliding scales to reduce benefits for additional children and increase them to with years of schooling, (iii) an expansion of coverage to all qualified children, and (iv) an expansion of non-contributory pension coverage to all population older than 65 that qualify under the current rules. Implementing such policy could result in an important decline in extreme poverty to almost negligible levels, a small improvement in income distribution, a reduction by one third of the coverage gap for the poorest quintile, and a relatively minor improvement in poverty indexes. On the other hand, this policy would cost nearly 1.2 percent of GDP above that is currently spent on income transfer programs. Table 7. Simulation Scenarios: Impacts on Coverage, Social Indicators and Costs Marginal Cost per pp Coverage* Social Indicators Costs (in U$ million) Scenario Family Extreme Poverty Coverag $U Extreme Pensions Poverty Gini % GDP US$ mill Poverty Allowances Poverty gap e gap million Poverty 1.- Baseline 64.3% 85.7% 27.4% 2.87% 9.64% 0.4519 22.9% 44,898 10.4% 1,833 2.- No IC 64.3% 85.7% 27.8% 4.26% 9.13% 0.4564 27.2% 43,625 10.1% 1,781 2,893 916 Increasing FA benefits 3.- 64.3% 85.7% 26.9% 2.23% 9.13% 0.4487 28.0% 45,870 10.6% 1,872 2,260 1,519 (sliding scale per child) Adding benefits per year 4.- 64.3% 85.7% 26.9% 2.16% 9.07% 0.4484 28.0% 45,976 10.7% 1,877 2,345 1,520 of schooling Expanding FA coverage 5.- 78.9% 85.7% 26.3% 1.52% 8.52% 0.4447 16.2% 48,700 11.3% 1,988 3,395 2,817 under current rules Expanding Non Contrib. 6.- 78.9% 93.2% 25.9% 1.47% 8.37% 0.4433 14.9% 49,858 11.6% 2,035 3,375 3,543 pensions to 65+ Switching targeting tool 7.- 74.7% 88.1% 25.3% 1.44% 8.11% 0.4437 15.3% 46,918 10.9% 1,915 967 1,413 to poverty Note: (*) Coverage of family allowances indicates the percentage of children younger than 18 receiving a benefit, coverage of pensions indicate the percentage of elderly aged 65 and more receiving a benefit. Source: Bank's staff calculations based on ECHA 157. If authorities aim at having a stronger impact on poverty incidence, the most effective policy would be to further increase the family allowances benefits. Table 8 shows that doubling the expected benefits would result in a reduction of poverty 60 incidence by another 2.5 percentage points. If the goal were to bring poverty below 10 percentage points (as compared to the current level), then family allowances benefits would have to be quadrupled. Of course, these policies are unfeasible, due to their high fiscal costs and potential negative impacts on labor markets, but represent an interesting exercise to assess how far Uruguay is from a sharp reduction in poverty incidence. Table 8. Additional Simulation: Impact of Further Increases in Family Allowances Benefits on Poverty and Spending Poverty current after Simulation: Simulation: Households Persons figures simulation FA X2 FA X4 Total 100.00% 100.00% 27.37% 25.90% 23.38% 16.94% Hh with children 54.75% 38.73% 41.26% 39.32% 35.42% 24.81% Hh with elder 17.52% 27.34% 5.19% 3.98% 3.98% 3.98% Hh with children and elder 7.62% 4.80% 33.89% 31.10% 25.79% 16.93% Other 20.11% 29.12% 7.17% 7.23% 7.23% 7.23% Cost $U 44,898 $U 49,858 $U 54,934 $U 65,086 % of GDP 10.42% 11.57% 12.75% 15.11% Additional cost $U 4,961 $U 10,036 $U 20,188 %GDP 1.15% 2.33% 4.69% Note: (*) Coverage of family allowances indicates the percentage of children younger than 18 receiving a benefit, coverage of pensions indicate the percentage of elderly aged 65 and more receiving a benefit. Source: Bank's staff calculations based on ECHA 158. Some final observations can be made about these simulations. First, all scenarios presented assume that by extending legal coverage and increasing benefits all potential beneficiaries will join the programs. However, the experience in Uruguay and other countries clearly indicates that even the best designed program has problems reaching all potential beneficiaries, and that some of them will never join it. A very proactive strategy by the Government to reach out and include these potential beneficiaries would be necessary to achieve coverage levels close to the 100 percent simulated here. 159. Second, while poverty is relatively easy to estimate and measure through surveys at the aggregate level, designing a consistent system to define the poverty status of individual households is difficult. Information on household composition, location, total income and other data used to define poverty status is not simple to collect, and may have significant quality shortcomings. This is a common problem, and the usual solution is to develop a proxy means test, such as the one used to select PANES beneficiaries. However, this approach also has some complications, in particular regarding the costs involved in maintaining a permanent structure to assess new applicants and regularly reassess existing beneficiaries. 160. Third, neither inclusion nor exclusion errors are fully avoidable. The analysis presented here assumed that these errors (know as "type I" and "type II") are avoided, but in practice that is nearly impossible. In other words, the simulation results presented here assume that no eligible person is rejected, and nobody who is non-eligible is accepted. However, in real life programs usually suffer these problems, affecting its costs, effectiveness, and political support. When selecting targeting tools it is important to 61 consider these problems, since stricter tools tend to increase type II errors, and simpler tools may increase type I errors. 161. Finally, the simulations assume that implementation of the policies is "instant", when in reality it is only possible (and in some cases desirable) to advance progressively. Due to the short-term fiscal restrictions and limitations in institutional capacity, it is difficult to imagine that a scheme can be fully implemented just a few months after being approved. As a result, it is reasonable to work from the beginning knowing that implementation is progressive, in order to manage the advancement according to an adequate set of priorities. This will reduce fiscal impact in the short term, and will permit a more organized program launch. 162. In short, the simulations show that PES could have an important impact on extreme poverty incidence, reducing it by more than 50 percent from the current levels, or more than 70 percent compared to the pre-PANES situation, and on the coverage gap, but not as much in other indicators. On the other hand, the impact on poverty incidence would be limited (a reduction of between 5 and 12 percent of the current level), and the effect on income distribution would be even smaller, since the Gini coefficient would decline by less than 2 percent. Finally, the gap in coverage in the income transfer programs for the first quintile would be reduced by 65 percent from the baseline, leaving only 15 percent of homes without benefits. 163. To affect these indicators, authorities would have to significantly increase expenditures on these programs in the near future. PES would have to spend approximately three times the actual amount spent in 2006 by IC, or nearly US$ 249 million, a figure that represents almost 1.2 percent of GDP if income is used as the targeting criteria. If the program is targeted to the poor population, the cost will be significantly less, subject to the design of an efficient targeting mechanism. Even though such an expansion in expenditure in the social sector seems inaccessible in the short term, the debate over the social policy model for the medium and long term will require the cost of these options and will need to be able to compare them with other public policy priorities. 62 5. CONCLUSIONS AND LOOKING FORWARD: THE MAIN CHALLENGES FOR INCOME TRANSFER POLICIES 164. Income transfer policies in Uruguay have a long history, starting more than a century ago. During those years, social insurance programs provided a solid protection to formal workers and their families against different social risks, including poverty. However, the traditional programs confronted two growing problems in recent years, as they were not able to provide adequate protection to individuals and households affected by long-term unemployment or informality, and their cost represented a growing burden for the State that crowded out the fiscal space for alternative measures. 165. The economic crisis of the late 1990s and early 2000s brought the discussion about limitations of income transfer programs to the forefront of the social policy debate, and new policies were implemented. Uruguay's authorities reacted by implementing reforms and introducing new programs, that included many previously uncovered households in the social protection system. The extension of the family allowances system in 1999 and 2004 and the creation of the emergency plan PANES were the two most relevant steps in this direction. 166. The income transfer programs introduced in the last few years have helped to reduce extreme poverty, but more efforts are necessary to affect total poverty incidence. As of 2006, the non-contributory family allowances impact on poverty incidence was of less than one percentage point, and the citizen income program (part of PANES), was responsible for a reduction of 1.2 percentage points. While the latter had a more important impact on extreme poverty incidence and the poverty gap, it is clear that there is significant space to improve the effectiveness of these policies. 167. Reforms since the mid 1990s also had the effect of blurring the limit between social insurance (contributory) and social assistance (non-contributory). Nowadays, no income transfer program is purely contributory, as they all receive funds from taxes or other sources apart from participants' contributions. Furthermore, in many cases those who are receiving benefits have had a limited contribution history, and some of those who are currently contributing to these programs will most likely be unable to qualify for benefits in the future. 168. As of early 2007, Uruguay had four main income transfer programs, which were designed independently and have insufficient coordination in targeting, financing, and management. First, the traditional pension scheme provides protection to formal workers after retirement (and to some poor elderly, through the non-contributory scheme) through an income substitution scheme that receives a substantial subsidy from general revenues. Second, the family allowances program provides a small income transfer to children under 18 of mid and low income households. Third, the unemployment insurance program provides temporary income support to a small number of unemployed workers. Finally, the fourth program is the citizen income program, which provides, since 2005, an income transfer to some poor households. These programs were created independently, and no explicit coordination exists among them in terms of 63 population covered, financing or targeting tools. This fragmentation reduces the programs effectiveness, both in terms of coverage, impacts, and sustainability. 169. The core challenges for the medium term are related to coverage, sustainability, and system integration. Income transfer programs coverage of at-risk population is not complete, as some groups and risks are not the focus of any program. Second, overall fiscal and social sustainability is at risk, as different programs have to compete for funding, resulting in suboptimal resource allocation and excessive pressure to find additional funds, and lower effectiveness may result in less political support. Finally, the fragmentation in design is replicated at the operational level. The participation of multiple institutions in the management of these policies, sometimes without enough coordination, exposes them to possible conflicts. This section discusses each of these challenges in detail. 170. While the overall performance of Uruguay's income transfer programs can be considered satisfactory, there are several challenges that remain regarding coverage and sustainability in the medium and long term. As in other countries in the region, Uruguay's income transfer programs' performance is affected by several problems that could be addressed by the authorities. The reforms in the early 2000s and the expected introduction of PES represent an important step in the right direction, but the construction of an integrated social protection model that consistently provides protection to vulnerable groups in a fiscally and socially sustainable way will continue to be a major policy concern for the medium and long term. 5.1. COVERAGE GAPS 171. Coverage gaps have significantly declined since the early 2000s, but they remain a challenge, among young adults and, in the future, the elderly. As labor informality and unemployment grew in Uruguay in the late 1990s and early 2000s, the magnitude of coverage gaps became a critical problem. As discussed in section 3, more than fifty percent of households in the poorest quintile were excluded from the main income transfer programs by 2003. The reforms in family allowances and the introduction of the citizen income program had a dramatic impact on this figure, as the percentage was rapidly reduced and, by early 2006, represented less than 25 percent. 172. The simulations of section 4 show that the introduction of PES would reduce the coverage gap among the poorest quintile of the population by almost 50 percent, but will have a limited effect on poverty incidence. While the gap among the poorest Uruguayans would be reduced to around 14 percent (a major accomplishment, considering that it was over 50 percent at the beginning of the decade), the impact on poverty would be limited to a reduction of 3 percentage points compared to the current situation. 173. The most serious problem regarding coverage affects young adults with labor market problems. Children and most of the elderly would be reasonably covered through the combined action of family allowances and PES (for children), and contributory and non-contributory pensions (for the elderly). However, young adults affected by persistent unemployment or high informality (linked to low productivity-low income jobs) are not protected by any specific program. The unemployment insurance 64 scheme provides assistance to some individuals in this group, but access restrictions are important and most of those affected by these problems do not qualify for its benefits. 174. By redesigning or expanding the unemployment insurance scheme to provide income support to unemployed adults regardless of their past job history, Uruguay could get nearer to closing this coverage gap. However, neither family allowances, PES, nor non-contributory pensions offer protection to young adults with labor market problems, If authorities advance with the idea of introducing an "employment insurance" scheme for long term unemployed and informal workers, then it will be important to consider several risks with regards to labor supply incentives. These benefits would be temporary, designed to provide incentives to workers to return to the labor market as soon as possible and strongly linked with other programs, such as training and job placement services20. Workfare programs implemented in Uruguay in recent years (including PANES, through its workfare program "Trabajo por Uruguay", and the training components included in "Rutas de Salida") were interesting attempts to cover this population. In recent months, the Ministry of Labor and Social Security began analyzing possible actions to close this coverage gap in a consistent and sustainable way. Also, monetary benefits could be granted not only to unemployed workers, but also to include poor formal workers, to eliminate the incentive to stay unemployed or in the informal market. Also, benefit amounts would be calibrated to ensure that the program does not compete with the lower end of the formal labor market. 175. Most of the elderly are currently covered by the contributory and non- contributory pension system, but mid-term projections show a declining trend in coverage of this group. The combination of declining labor market formality rates in recent decades and strengthening of requirements to obtain a contributory pension, might result in a slow but sustained decline in coverage among the elderly, leaving some individuals unprotected, as they would not qualify for contributory pensions, and could also be excluded from the non-contributory scheme, targeted at the poorest groups. 176. A more effective income transfer system would need to include multiple programs, targeted at different ages and risk and designed as an integrated system. Different groups are exposed to different risks, thus a mix of programs, well articulated and rationally financed could be the best policy. Table 9 presents the structure that an integrated income transfers system could have in Uruguay to provide adequate coverage to all. The table includes the existing programs (such as family allowances, unemployment insurance, and pensions), the planned PES, and the possible "employment insurance" that would provide protection to long-term unemployed and informal workers. The table also indicates the potential need to offer protection to a group of the retired workers that had intermittent formal jobs, as they might become an important proportion of the elderly in the future. 20Uruguay has had training and placement services since the early 1990s, run by the Ministry of Labor and Social Security through the National Employment Directorate, but their effectiveness was limited, partly due to the same access restrictions affecting the traditional unemployment insurance scheme. 65 Table 9. An Integrated Income Transfers System to Provide Adequate Coverage to All Group at risk Targeted population Programs Program status /risk Children of formal workers Family allowances Existing Children / Children of poor and poverty risks PES Expected for 2008 informal workers Recently unemployed Unemployment Existing Young Adults / formal workers Insurance Labor market Long term unemployed, Employment Under preliminary risks underemployed, informal insurance discussion workers Poor, long term informal Non-contributory Existing, to be workers pensions expanded under PES Elderly / poverty Elderly with long term Contributory Existing risks formal jobs pensions Elderly with intermittent ? Not covered formal jobs 5.2. FISCAL AND SOCIAL SUSTAINABILITY 177. To ensure effectiveness, income transfer programs must be able to deliver, consistently and across time, the benefits promised to vulnerable groups. As with any other public policy, fiscal sustainability is critical to ensure that the programs' goals are achieved and that improvements in social indicators are sustained. Funding income transfer programs is always a challenge, as many other competing demands require support from public funds, and these transfers are sometimes seen as "unproductive spending", as compared with public investment or other spending areas. Uruguay has dedicated an important proportion of its GDP to income transfers in recent years. As discussed in Section 3, total expenditures in these programs reached an average of approximately 13 percent of GDP from 1993 to 2002, and then declined to 10-11 percent of GDP since 2004. This level is high, probably the highest in Latin America (together with Brazil), and fiscal space to expand it is limited. 178. While the programs appear to be fiscally and socially sustainable, there is little margin for spending expansion, limiting the space for policy reforms. Expenditure projections prepared by BPS technical teams indicate that spending should decline over time, ensuring that the policies are fiscally sustainable. However, these same projections predict a decline in coverage that could be socially unsustainable, thus resulting in renewed fiscal pressures. Also, other social programs, including the new PES and a possible revamped unemployment insurance scheme would also require additional funding. The combination of these factors creates a potential risk that should be closely monitored by Uruguay's authorities, to ensure that the medium term sustainability of the programs is not affected. 179. Overall spending has been high, and there is an important imbalance that limits the ability of policy makers to attend to those in need of social protection. Since the late 1990s the pension system has represented between 90 and 95 percent of total expenditures in these programs, leaving little space for other programs to expand. The pension system is expensive due to its high coverage and the demographic profile of 66 Uruguay, and there is little that can be done to reduce its costs, but it is important to prevent this program from crowding out other important transfers. Due to its magnitude, minor policy decisions may have a dramatic impact in other areas. For example, an increase of 4 percent in pension expenditures would have the same costs as all transfers under citizen income in 2006, and the expected cost of PES, assuming it is fully implemented as discussed in section 4 in the most ambitious scenario, it would only cost 12 percent of what is currently spent in pensions. 180. Considering the projected evolution of the pension system and the expected costs of PES, income transfer programs appear to be sustainable in the medium term. Spending projections for the pension system indicate a stable trend in terms of GDP. However, these projections also indicate a declining term in coverage. As future governments will likely look for policy alternatives to maintain the current high coverage levels, it will be critical to carefully assess the fiscal impacts of any reform proposal. The fragmentation in design, funding, and management of these systems adds an additional risk to the sustainability problem, since the space to define funding allocation strategically is limited. 181. Social and political sustainability are as important as fiscal sustainability. Income transfer programs are sometimes perceived as "unproductive spending", and therefore has a negative effect on labor markets and economic growth and development. This discussion has been common across most countries in the region and the world, especially (but not limited to) regarding non-contributory schemes. These concerns usually focus on the impact that these programs might have on labor supply, and the pressure that they generate on fiscal accounts, requiring higher taxes that could damage economic growth. 182. In short, concerns about the cost of income transfer programs and their effectiveness should be openly discussed by policy makers and authorities. Uruguay, as in many other countries, has many urgent needs that require public financing including other social programs, infrastructure, et cetera. Authorities could publicly discuss whether these programs deserve a particular space in the fiscal strategy, given their social value, contribution to population's welfare, and, through a virtuous circle of human capital development and higher productivity, to a faster economic growth, but also consider the potential trade offs on fiscal balance and labor markets. 5.3. SYSTEM INTEGRATION 183. Partly due to the traditional separation between social insurance and social assistance schemes, the income transfer policies in Uruguay are not fully integrated. Programs have been designed independently in most cases, with goals, targeted population, financing policies, and institutional structure that were adopted without considering whether other similar arrangements already existed, resulting in a multi- layered institutional and normative scheme that, in some cases, diminished the programs effectiveness. 184. The institutional fragmentation among those in charge of designing and managing the programs might also result in lower than expected effectiveness. The multiplicity of institutions and asymmetries among them requires strong political 67 coordination efforts to avoid conflicts of interest, miscommunications or other problems that might result in limited impact of the programs. Uruguay's authorities have been very effective in reducing this risk, by creating coordination mechanisms and promoting continuous dialogue between the different agencies. However, the risk is always present, and more formal coordination structures, including clearer lines of authority, could improve the environment for the institutional structure governing these programs. 185. Currently, there are at least three ministries, five government agencies and six independent institutions participating in the design, operation, and supervision of income transfer programs in Uruguay. Some of these institutions are focused exclusively on managing programs, others on the macro financial aspects, and others on the design and operation of the different programs. Some of them are part of the central administration; others have different degrees of autonomy. The role of each institution is defined in the legislation but, in some cases, political pressures or differences of approach might result in conflicting views about the programs and their management. 186. An efficient institutional structure is critical to maximize policy impact. Coordination problems among agencies in charge of income transfer programs have been common in the region. Conflicts between managing agencies and supervisory bodies are common21, as well as differences between agencies in charge of social assistance programs and those in charge of social assistance. To avoid these conflicts, governments have usually adopted two possible strategies, concentrating the authority in one institution or reinforcing coordinating mechanisms. 187. In recent years, Uruguay has clearly moved towards improving inter- institutional coordination efforts. This approach has not eliminated all possible disagreements22, but created some mechanisms to solve them. Soon after taking office, President Vazquez' government created an Intersectoral Commission of Social Security, with the participation of the Ministries and official agencies involved in the management of the social insurance schemes. Soon after that, a Social Cabinet, integrated by Ministers from social sectors, was also formed in early 2005, with the coordination of the Social Development Ministry. Collaboration between ministries and agencies to design and implement programs have been common, particularly between MIDES and BPS, in the enrolment process for PANES and during the design work for PES. 188. More recently, a project to integrate a single database of social program beneficiaries was started. This project, called Integrated Information System for the Social Area (Sistema de Información Integrado para el Area Social ­ SIIAS) will integrate information from BPS, MIDES, Ministry of Public Health, Ministry of Housing, Ministry of Labor, Interior Ministry and the National Institute of Statistics. By doing so, each institution will be able to verify whether current or potential beneficiaries of its 21For example in 2004, Ecuador's banking supervision, in charge of overseeing all pension funds (including the national schemes), declared that authorities at IESS (the national social security institute) were ignoring regulations, and requested their removal, which was rejected by IESS. In Uruguay, the Ministry of Economy has recently prepared a proposal to introduce reforms to the Bank Employees Pension Fund that was rejected by authorities of that fund. 22For example, in early 2007 the BPS formally complained about a technical report prepared by the Central Bank about the funded pension scheme, indicating that, in writing it, the BCU overstepped its responsibility as supervisor of the managing firms. 68 programs is already registered in other institution, and full information on beneficiaries of all programs will be available to policy makers and analysts. 189. In conclusion, Uruguay appears to have managed the institutional fragmentation risk until now, but risks for the future are still important. If political conflicts or other problems arise in the future, they could affect the operations of the different income transfer programs, reducing their effectiveness and impact. The efforts made by Uruguay's authorities to avoid these risks have been successful so far, but it is not possible to ensure that these problems will not arise in the medium term. 190. The next few months will be critical for the future of Uruguay's income protection policies. As the Government advances in its work to define the components and financing of PES, it will be essential to maximize the new program's impact while recognizing the existence of these remaining challenges and start addressing them. Income transfer policies have had an important role in Uruguay for providing poverty relief and improve income distribution during most of the twentieth century, and the main challenge for policy makers and Uruguay's society as a whole is to ensure that they continue to do so in an effective and sustainable way into the new century. 69 6. REFERENCES - Arim R. and Vigorito A. (2006) "Las políticas de transferencias de ingresos y su rol en Uruguay. 2001-2006" Background Paper, The World Bank, Buenos Aires - Arriagada A. and Hall G.(2000) "Managing Social Risks in Argentina". Mimeo, The World Bank - Banco de Prevision Social (2005) "Proyección financiera del Sistema Previsional Contributivo administrado por el Banco de Previsión Social. Análisis Global". Comentarios de Seguridad Social, BPS, Montevideo - Banco de Previsión Social (2006) "Boletín Estadístico 2006". BPS, Montevideo - Banco de Prevision Social (2007) "Boletín Estadístico". BPS, Montevideo - Bertranou F. (2005) "Desarticulación o subordinación? Protección social y subordinación en América Latina" in Protección social y mercado laboral, Bertranou (ed.), ILO, Santiago de Chile - Buchelli, M. and Amarante, V. (2006) "El Seguro de Desempleo en Uruguay". Background paper, The World Bank, Buenos Aires - De Ferranti, D. et al (2000) Securing our Future in a Global Economy. The World Bank, Washington DC - ECLAC (2005) "La protección social de cara al futuro: acceso, financiamiento y solidaridad", Santiago de Chile - Ferreira-Coimbra A. and Corteza A. (2004). "Protección Social en Uruguay" ILO, Santiago - Fiszbein, A (2005) "Beyond truncated welfare states: Quo Vadis Latin America?" Mimeo, Washington DC - Flood, C. (Coordinadora), Grau, C. y Melgar, A. (2004): Análisis del gasto público social, Serie de Estudios económicos y sociales, Banco Interamericano de Desarrollo - Hennchen E. (2007) "The debate around Income Transfer Systems in Uruguay". Background paper for the World Bank - Holzmann, R. and Jorgensen, S. (2001) "Social protection as social risk management: conceptual underpinnings for the social protection sector strategy paper" Journal of International Development, vol. 11, issue 7. - INE (2006) "Pobreza y Desigualdad en Uruguay, 2006". - Lindert, P. (2003) "Why the welfare state looks like a free lunch". NBER Working Paper 9869. Cambridge, MA - Mesa Lago, C. (1978) "Social security in Latin America: pressure groups, stratification, and inequality". Pittsburgh University Press, Pittsburgh - MIDES-OPP (2006) "Identificación y análisis del Gasto Público Social en Uruguay 2002-2005" 70 - OECD (2006) Sickness, Disability and Work: Breaking the barriers. OEDC, Paris. - OPP (2004) Oficina de Planeamiento y Presupuesto ­OPP (2004): El Gasto Público Social en el Uruguay (1999-2003), OPP - Rawlings, L. and Rubio, G. (2005) "Evaluating the Impact of Conditional Cash Transfer Programs". The World Bank Research Observer, vol. 20, no. 1 - Rofman R. and Luchetti L. (2006) "Pension Systems in Latin America: Concepts and Measurements of Coverage". SP Discussion Papers, The World Bank - Saldain R. and Lorenzelli A. (2002) "Estudio del programa de pensiones no contributivas en el Uruguay", in Pensiones no contributivas y asistenciales. Argentina, Brasil, Chile, Costa Rica y Uruguay, Bertranou, Solorio and van Ginneken (eds.). ILO, Santiago de Chile - World Bank (2007) "Argentina: Facing the Challenge of Ageing and Social Security". World Bank, Washington DC - Zepeda, E. (2006) "Do CCTs Reduce Poverty?". International Poverty Centre, UNDP, Brasilia 71 Annex 1. A profile of income transfer programs beneficiaries 1. Section 2 of the report presented data on coverage levels for each individual income transfer program and discussed the relevance and magnitude of gaps and overlaps observed when all programs are considered together. In this annex, a more detailed analysis of the beneficiaries profile is presented, describing their demographic and socioeconomic characteristics. By using probit models, we are able to describe the population receiving each program's benefits, as well as their common profile and differences with those who do not receive transfers at all, using data from the first semester of 2006. Table A.1 reports the marginal effects of some variables on coverage, holding the other variables constant at their means. 2. Considering households as the unit of analysis, three sets of control variables have been included on the full specification to assess their impact on income transfer programs coverage: socio-demographic, socio-economic, and use of social services. The first set includes sex, age and race of the household head, size and age structure of the household. The second consists of controls for labor market participation by the household head, quintiles of total family income, use of domestic service, and home ownership. Finally, the third set contains controls for use of health, education and other social services. For discrete variables, the most frequent case is always the excluded group. 3. Gender of the head of household was not homogeneous across income transfer programs. Female headed households had a higher chance of receiving a transfer in all programs but unemployment insurance. Pensions seem to be the program where the marginal effect was the largest, possibly explained by the effect of survival pensions and the growing difference in life expectancy (between men and women) at older ages. On the other end, citizen income was the program with lower differences by sex. 72 Table A.1. Marginal Effects (Probit Models) of Socio-Demographic Variables on Income Transfers Received by Households, 2006 Family Contributory Non-contributory Unemployment PANES Pensions Allowances FA FA Insurance (1) (2) (3) (4) (5) (6) Female head 0.0351*** 0.0087*** 0.0166*** 0.0021*** -0.0012* 0.1310*** (5.9258) (2.9243) (4.7304) (3.2677) (1.8881) (16.3447) Head's age -0.0060*** -0.0014*** -0.0017** -0.0001 -0.0002* 0.0248*** (5.4899) (2.5979) (2.5136) (0.8201) (1.7799) (13.0297) Head's age (square) 0.0000 0.0000 -0.0000 -0.0000 0.0000 -0.0001*** (1.3991) (0.4731) (1.1629) (0.9587) (0.5008) (7.5506) Household size 0.0481*** 0.0148*** 0.0280*** -0.0003 0.0032*** 0.0776*** (17.0612) (10.6811) (17.2078) (0.9484) (11.9514) (19.9314) Number children (0-5) 0.1362*** 0.0286*** 0.0295*** 0.0045*** -0.0033*** -0.1035*** (24.7489) (11.5810) (10.1045) (8.7494) (5.9929) (11.8934) Number children (6-12) 0.1323*** 0.0371*** 0.0157*** 0.0029*** -0.0031*** -0.0838*** (28.7399) (17.8317) (6.2954) (6.4406) (6.7646) (12.0348) Number children (13-18) 0.1273*** 0.0441*** 0.0130*** 0.0026*** -0.0027*** -0.0569*** (25.8510) (19.3947) (4.6744) (5.0430) (5.2879) (7.9153) Number of people over 65 -0.0424*** -0.0166*** -0.0235*** -0.0028*** -0.0033*** 0.4095*** (6.5363) (5.0263) (5.8210) (3.3263) (4.4632) (45.1105) Head's afro-origin 0.0415*** 0.0157*** 0.0082* 0.0044*** -0.0006 0.0158 (4.8335) (3.8318) (1.7408) (4.9913) (0.6922) (1.2696) Head unemployed -0.0137 -0.0380*** 0.0425*** 0.0047*** 0.0142*** -0.2616*** (0.9763) (6.8421) (4.7851) (3.2587) (6.2482) (15.8310) Head formal 0.0106 -0.0468*** 0.0620*** 0.0015* -0.0001 -0.3622*** (1.2146) (12.6136) (11.0906) (1.7981) (0.1486) (37.3195) Head informal 0.0598*** 0.0598*** -0.0306*** -0.0036*** -0.0055*** -0.5757*** (6.6364) (13.1259) (5.8024) (3.6550) (5.9204) (56.7884) 2nd quintile -0.0251*** 0.0182*** -0.0370*** -0.0079*** 0.0018** 0.1378*** (3.9553) (5.6528) (10.6718) (16.2977) (2.0891) (12.9059) 3rd quintile -0.0949*** -0.0398*** -0.0177*** -0.0109*** 0.0001 0.1841*** (13.9209) (12.8832) (4.1327) (16.5240) (0.1295) (15.4335) 4th quintile -0.1767*** -0.0914*** -0.0053 -0.0126*** -0.0002 0.1776*** (24.6481) (30.1741) (1.0263) (12.1606) (0.2266) (13.3809) 5th quintile -0.2236*** -0.1120*** -0.0362*** -0.0143*** -0.0030*** 0.1721*** (26.7022) (24.6328) (5.9365) (5.4919) (2.8493) (11.2172) Domestic service (in the h'hold) -0.1133*** -0.0438*** -0.0145** -0.0005 -0.0013 -0.0893*** (11.0946) (7.9688) (2.2048) (0.2161) (1.0405) (6.1088) Tenant (rent) 0.0138** -0.0029 0.0078** -0.0023*** -0.0006 -0.0503*** (2.2270) (0.9907) (2.1125) (3.2585) (0.8046) (5.4312) Occupant 0.0141** -0.0060* 0.0108*** 0.0039*** 0.0006 -0.0497*** (2.1104) (1.9193) (2.8311) (5.5941) (0.8546) (4.9307) Health services (social 0.0311*** 0.0432*** -0.0427*** -0.0128*** 0.0053*** 0.1555*** insurance) (4.7841) (14.4293) (10.5931) (13.1675) (7.7470) (16.2431) Private health services -0.0459 0.0063 -0.0366** -0.0069** 0.0090 0.0224 (1.4837) (0.3581) (2.1504) (2.3192) (1.6443) (0.4832) No health coverage -0.1171*** -0.0228 -0.0599*** 0.0001 0.0025 0.0458 (2.8932) (0.9787) (2.7337) (0.0056) (0.4962) (0.7734) Complete primary school (head) 0.0197*** 0.0047 0.0079** -0.0045*** -0.0001 -0.0181** (3.0561) (1.4736) (2.0825) (6.9706) (0.1206) (1.9821) Complete secondary school -0.0175* -0.0067 0.0020 -0.0059*** 0.0005 -0.0739*** (head) (1.9293) (1.4911) (0.3614) (7.5749) (0.4681) (5.6765) Complete university or tertiary -0.0291*** -0.0097** 0.0005 -0.0067*** -0.0006 -0.0175 (3.1771) (2.1189) (0.0914) (7.7646) (0.5809) (1.3305) Children (5-18) not attending -0.1174*** -0.0582*** -0.0056 0.0009 -0.0005 0.0132 school (15.9738) (17.3100) (1.4397) (1.4270) (0.5824) (1.1955) Feeding programme attendants -0.0227*** -0.0143*** -0.0103*** 0.0026*** 0.0002 -0.0088 (5.1370) (8.6830) (5.1618) (9.2058) (0.3658) (1.4027) Receiving soup kitchen -0.0028 -0.0158** 0.0106 0.0021** -0.0000 -0.0183 (0.1437) (2.0716) (1.2734) (1.9830) (0.0067) (0.7908) Basic food supplies 0.0606*** -0.0046 0.0348*** 0.0080*** -0.0003 0.1456*** (8.3763) (1.5582) (10.0981) (16.9663) (0.4283) (14.5863) Observations 45962 45962 45962 45962 45962 45962 Pseudo R2 0.4293 0.3519 0.2569 0.4203 0.1287 0.5467 Lagrange Multiplier 24349.6740 14713.1297 9716.1271 9028.1574 629.8562 34358.3081 Absolute value of z statistics in parentheses * significant at 10%; ** significant at 5%; *** significant at 1% Source: Bank's staff calculations based on ECH and ECHA. 73 4. As expected, pensions and FA were well targeted at their respective age groups. In the first case, households with older heads were better covered. The higher the number of elderly members in a household, the higher the probability of being covered was. Family allowances were going to households with younger heads (as they have higher probabilities of including children), while the likelihood of receiving benefits from citizen income or unemployment insurance had no relation to the age of the head of household. 5. Family allowances were more likely to be received in households with many children, regardless of their age, and there was a similar (but weaker) effect for citizen income. There was an interesting difference between the likelihood of receiving contributory and non-contributory family allowances: while the probability of receiving the first benefits grew when older children were present in the household, the effect was the opposite with the latter, as the probability of receiving a non-contributory family allowance declined when the children in the household were older. The presence of people over 65 years old decreased the overall likelihood of receiving a benefit, except for pensions. Finally, households headed by individuals of afro origin had a higher likelihood of receiving a benefit from FA (especially contributory ones) and IC. The marginal effect for FA was 5 percent, but only 0.5 percent for citizen income. No significant differences were found for neither UI nor Pensions. 6. The second panel in table A.1 shows that both IC and FA were negatively correlated with income levels, UI was not strongly targeted and pensions concentrate at the middle of the household income distribution. The likelihood of participating in the FA program monotonically decreased by income quintile, a tendency that affected both contributory and non-contributory systems. While contributory beneficiaries were more densely concentrated within the second quintile, non-contributory FA concentrated on the first quintile. Citizen income, as well as total FA, was negatively correlated with income, and unemployment insurance had no significant correlation with income. In the case of pensions, the marginal effect was positive for all quintile coefficients with a peak at the third quintile. 7. Labor market participation parameters showed the expected results: pensions were more often found in households with economically inactive heads of household (that is, retired workers), while unemployment insurance and citizen income were found in households with unemployed heads. Households with heads in the informal sector had a 5 percent more chance of receiving a benefit from FA than those with an inactive head. 8. Other indicators of socio-economic level help to understand the profile of the households that participate in each program. The presence of domestic workers reduced the likelihood of participating, especially in the FA program. Homeowners were prevalent among households receiving most benefits, except for non-contributory family allowances, where households renting or occupying their homes were common. 9. Participants on the contributory programs (pensions, UI and FA) received medical attention through the health insurance system provided through BPS. This is a reasonable arrangement, considering that the same contributions that give them access to these benefits open access to the health program. On the other hand, those who received 74 benefits from non-contributory programs (FA and IC) received health services from public hospitals. 10. Systematic differences were found when controlling for educational attainment. For citizen income and pensions, the likelihood of participating in the program was negatively correlated to the education level. Non-contributory family allowances showed a positive and significant coefficient. On the other hand, households headed by someone with a tertiary education were less likely to receive contributory family allowances, probably because they had a higher income and were excluded from the system. In consequence, total family allowances presented a concave parabolic relationship between participation and education level, as households with heads that completed primary school had the highest chance of receiving a benefit. No systematic differences were found for UI in relation to educational level of the head of household. 11. The presence of dropouts among children younger than 18 reduced the chance of receiving family allowances in the contributory scheme, suggesting that conditionalities linked to this benefit were enforced by BPS or self-enforced by would-be beneficiaries. Interestingly, neither non-contributory FA nor IC reported significant impacts from this variable, indicating that there was no enforcement in these programs. Having the same result for pensions and UI is not surprising, since neither program requires school enrolment. 12. Most beneficiaries from IC received benefits from other social protection schemes, including food programs. Beneficiaries of the non-contributory FA did not participate in food programs but received basic food supplies. Households receiving contributory FA participated less in food programs and soup kitchen provisions, than those who did not benefit from this transfer program. Finally, being a pensioner increased the likelihood of also receiving basic food supplies, in comparison with those households that did not receive pension benefits. 75