Report No. 34154-AR Argentina Facing the Challenge of Ageing and Social Security January 15, 2007 Social Protection Unit, Human Development Department Argentina, Chile, Paraguay and Uruguay Country Management Unit Latin America and the Caribbean Regional Office Document of the World Bank NDC National DefinedContribution Pensions OECD Organizationfor Economic Cooperation and Development PAP Prestacih Adicional por Permanencia (DB earnings-related pension inpublic secondpillar) PAYGO Pay-as-you-go financing inwhich pay-roll taxes from current workers finance pensions of current retirees PBU Prestaci6n Bhsica Universal (Basic old age pensioncovering affiliates to SIJyP with 30 years contributions at 60 (women) and 65 (men)) PC Prestacih Complementaria (Pension recognizingrights acquired up to 1993 reform) PEA Prestacihpor EdadAvanzada (Advanced age pension for SIJyP affiliates with 10 years contributions at age 70) PMG PrestacihMinima Garantizada (Guaranteed MinimumProvision) PNC Pensi6nNo Contributiva (Non Contributory Pension) PTm Prestaci6nTotal Minima (MinimumTotal Provision) RRP RCgimen de Reparto P~blico(SIJyP public secondpillar) SAFJP Superintendencia de Administradoras de Fondos de Jubilaciones y Pensiones (Regulatory authority over the commercial secondpillar) SIJyP Sistema Integrado de Jubilaciones y Pensiones(National old-age income security system) sss Secretaria de Seguridad Social (National Social Security Secretariat inMTEySS) UAFJP Unidn de Administradoras de Jubilaciones y Pensiones (Pension and Retirement Association Union) VAT Value Added Tax Vice President: Pamela Cox Country Director: Axel van Trotsenburg Human Development Director: EvangelineJavier Human Development Sector Leader: Jesko Hentschel Social Protection Sector Manager: Helena Ribe Human Development Lead Economist: Jennie Litvack Task Team Leader: Truman Packard ... 111 Table of Contents Acknowledgements ............................................................................................................. VU Terminology Usedinthis Report ......................................................................................... ..... VIII Introduction, Structure. andMainFindingsof this Report 1 I Expandingcoverage-Argentina'skeychallengeinoldageincomesecurity . ........................................................ ....................... 1 I1. Purpose and Structure ofthis Report I11 Summary ofMainFindings . ................................................................................. 5 .............................................................................................................. .............................................................................................. 7 Part I. A Frameworkto GuidePension Policy and Analyze How Argentina Covers the Losses IV.LookingAhead 11 from Ageing ........................................................................................................... 13 I.a. TheProspective Lossesfrom Ageing and Objectives of Pension Policy .............................. 13 I.b.Argentina`sInstitutionsfor OldAgeIncomeSupport., ...................................................... 17 I.c. .......................................... I.d. Separate.Independent Pension Plans.............................................................................. TheMultiple Pillars of Old Age Income Security in Argentina 19 Part I1. A Closer Look at Coverage o f the Elderly inArgentina: Empirical Findings .................21 22 II.a Patterns in Social Security Coverage in Latin American Countries . .................................. 23 II.b.AProfile of Argentina'sElderly .................................................................................... 25 I I c Do Argentina's Coverage OutcomesReflect Exclusion or Preferences? . ........................... 31 II.d.A Closer Look at Reporting. Contribution. and Evasion .................................................. 33 II.e.IsContinuedLaborinOldAgea ViableIncomeSubstitutefor SocialSecurity?................ 37 Part I11. Closing the Coverage Gap ...................................................................................... 42 III.a, TheDistinctions Between"Pillar 0" and "Pillar 1 ......................................................... 42 III.b.AdvantagesandDrawbacksof AlternativePublicPoolingStructures ............................. 43 III.c.FiscalCostsof DirectMeasurestoCloseArgentina'sCoverageGap ............................. 45 III.d. Limited Deployment of Non-Contributory Pensions in Argentina., .................................. Part IV.Principal Challenges for Social Security and PrivatePensions inArgentina ................51 55 IV,a Adequacy of the Commercial SecondPillar: Portfolio Risk. Contribution Rate and the . Cost of Insurance .................................................................................................... 55 IV,b TheDeteriorating Value of Pensions and Fiscal Risks of Ad-Hoc Benefit Adjustments I V ,c TheMultiplicity of Pension Plans 64 I K d SIJyP Regime Choice and the Competing SecondPillars ... .....62 ................................................................................. ............................................... 67 IV,e Remaining Vulnerability to Demographic Changes . Part V Concluding Insights:Issues and Opportunities inArgentina's PensionReform Debate 74 . ........................................................ 71 . V,a Restoring an Adequate Commercial SecondPillar . .......................................................... 75 V,b A Purposeful and Predictable Policyfor Benefit Adjustment 77 V,c Argentina's Multiple Pension Regimes .. ............................................ ........................................................................... 77 E d. SIJyP Second-Pillar Choices ........................................................................................ 78 Y e TheRetirement Age and Future Demographic Risk . ......................................................... 79 V,$ TheDeployment of Non-Contributory Benefits ................................................................ 79 V,g.Looking Even Further Ahead ......................................................................................... 81 References .......................................................................................................................... 82 Technical Annex.Data Quality Assessment of the ETEEP ..................................................... 87 iv Tables Table 1. Poverty among the Elderly .............................................................................................. 17 ..................... 20 Table 4. Characterization of Contribution Density.by Individual and Household Characteristics Table 3. Summary Statistics for Covered andNon-Covered Elderly,ETEEP 2003 Table 2. Argentina's National Pension Systemthroughthe Multi-Pillar Prism............................ 26 Table 5.Characterization of Contribution Density, by Size o f the Firmof ReportedEmployment o f ReportedEmployment Episodes............................................................................... 36 ......................................................................................................................... Table 6. Sensitivity Analysis of Changes inSimulation Parameterson the Public PensionDeficit Episodes 37 .Fiscal (Percentageof GDP)...................................................................................................... Table 8. Portfolio Allocations inMandatory Commercial Second Pillars PensionPlans inLatin Table 7 Costs of Alternative Poverty-Prevention, Non-Contributory Pension Programs..48 50 Table 9.FundManagement and Insurance Fees inLatinAmerica's America and EasternEurope......................................................................................... 57 ......................................................................... Table 10.MinimumRetirement Ages and Life Expectancy for Menand Women. Selected Commercial SecondPillars, June 2004 59 LatinAmerican and OECD Countries. circa 2000........................................................ 71 Figures Figure 2.Contributors to Social Security inArgentina's Labor Force Figure 1.Changes inthe Shareof the Elderly Receiving Social Security Benefits ........................ 1 . Expected ............................................ ......................................... 2 Figure4.Trend inIncome Inequality amongthe Elderly.............................................................. Figure 3 Coverage o f Contributory Social Security Benefits 5 Figure5.PopulationAgeing inArgentina and Other LatinAmerican Countries ......................... 12 . . ....................................... 14 Figure 7.Per Capita Income andthe Shareof the Elderly Receiving Benefits, SelectedLatin Figure6 Povertyamong the Elderly vs Total PopulationinArgentina 22 .............................................................................. Figure8.Public Spending on Social Security andthe Shareo fthe Elderly ReceivingBenefits, American Countries, 2000 -2002 24 Figure9 Shareofthe ElderlyReceivingBenefits among ETEEP Respondents Figure10 Age at Which Benefits were Awarded, Reported by CoveredMenandWomen .......27 30 Figure11 Reasonfor ReceivingBenefits, ReportedbyRecipients of OldAge and Survivor ... Selected LatinAmerican Countries, 2000 - 2002 ...................................................... 24 .......................... Benefits, 2003 30 Figure 12 .ReportedReasons ............................................................................................................ Figure 13. Principal Reasons for Failing to Contributeto Social Security, Reportedby ETEEP for Not Receiving Benefits............................................................. 31 ............................................................................................................... Figure 14.Reportingand ContributionBehavior, by Sex ofRespondent..................................... Respondents 33 Figure15.Labor Market Participation of Poor andNon-Poor Elderly ......................................... 35 38 Figure16a.PercentageofActive RespondentsineachDecile of Predicted Earnings...................40 Figure 16.b.Percentageof all Active RespondentsinDeciles of Predicted Earnings ..............40 Figure 17.Fiscal Costs o f FullCoverage: Current SIJyP & ComplementaryBenefits .................................................................................................. 47 Figure19.Reported PensionIncome as a Multipleofthe Poverty Line., Figure 18. Argentina's NationalNon-Contributory PensionProgram.......................................... (Percentage o f GDP) ..................................... 52 Figure20.Changes inSelectedIndicesand Public Pensions, 2001-2004 .................................. 62 63 Figure21 ReportedRetirement Age and Average BenefitAmount of ETEEPRetirees, by . ............................................................................................................... Figure 22. Second-Pillar Choices of New Affiliates to the SIJyP, 1994-2004 Pension Plan ........................... 64 Figure 23. Distribution of SIJyP Affiliates between Public and Commercial Second Pillar, 69 byAge Cohort, 2004 .................................................................................................. 70 V Text Boxes Box 1 Definingthe Coverage Problem: What i s it andWhy do we Care? . Box 2. The ComprehensiveInsuranceFramework ............................................................................ 15 ..................................... 3 Box 3. Surveys of the Elderly inArgentina and Other Middle Income Countries Box 4.The Cuotu Social inMexico Commercial SecondPillar ................................................... ........................ 28 44 Box 5. Costing ................................................................ 46 Box 7.FiscalFederalism, Public Finance and Social Security...................................................... Box 6. Benefit Adjustment Policies Around the World Current Proposals for FullCoverage .................................................................... 65 Box 8. The Barr-Diamond Mortality Adjusted Retirement Age ................................................... 68 73 vi Acknowledgements This report is part o f the World Bank's analytical and advisory program with the Government o f Argentina. The report was prepared by a team led by Truman Packard (Senior Economist, Social ProtectionUnit inthe HumanDevelopment Department o f the Latin America and Caribbean Regional Office-LCSHD). The team consisted o f Paula Giovagnoli, Evelina Bertranou (JPA Economists, Human Development Sector for Argentina, Chile, Paraguay and Uruguay) and Leonard0 Lucchetti (JPA Economist, Social Protection Unit-LCSHD). The consultants working with the team were Marcelo A16s, Raul Oromi, Abigail Barr, Armando Barrientos, Carlos Grushka, and Marcelo Muiii6s. Rafael Rofman (Senior Economist, Social Protection Unit- LCSHD) initiated the survey o f the elderly on which many o f the findings inthis report are based, provided critical inputs to the report, and was instrumental inmaintaining the Bank's dialogue with authorities from Buenos Aires. The survey firm TNS Gallup Argentina assisted in the design o f the survey and collected the data. Invaluable research and administrative assistance was provided by Tania Gomez, Febe Mackey, Martha P.Vargas, and Natalia Moncada (LCSHD). Truman Packard and Evelina Bertranou were the main authors. The Bank's team worked under the overall guidance o f Axel van Trotsenburg (Country Director for Argentina, Chile, Paraguay and Uruguay), Evangeline Javier (Sector Director -LCSHD), Jesko Hentschel (Country Sector Leader for Argentina, Chile, Paraguay and Uruguay-LCSHD), Helena Ribe (Sector Manager for Social Protection- LCSHD), and Ariel Fiszbein (Advisor, DECRG). The report benefited immensely from the active contributions, suggestions and comments o f reviewers. The peer reviewers inside the Bank appointed for this task were Anita Schwarz (Lead Economist, Human Development for Europe and Central Asia Region (ECSCHD), Rafael Rofman (LCSHD). Carmelo Mesa-Lago (Emeritus Professor o f Economics and Social Security, University o f Pittsburg), and Albert0 R. Musalem (Chief Economist, Centre for Financial Stability, CEF) acted as external reviewers. Each o f the reviewers provided detailed comments that significantly increased the quality and value o fthis report. The report team is grateful for their advice and direction. The Bank would like to thank the Ministry o f Labor, Employment, and Social Security (MTEySS) and many o f its staff for the information, many useful discussions and collaboration in the preparation o f the report by the Bank team. However, the conclusions and assertions made in this report do not necessarily represent their views and are solely the responsibility o fthe World Bank report team. vii TerminologyUsedinthis Report This report categorizes the components o f Argentina's national social security system for old-age income support-sometimes referred to as "pillars" as in Averting the Old Age Crisis (World Bank, 1994), and sometimes as "tiers" as in The Economics of the Welfare State (Barr, 2004)-by their objective. This is in contrast to other publications that categorize the branches of a pension system by who administers them (the public or private sector); how benefits are structured (final-salary defined benefit formula, or defined contributions); or their financing mechanism (pay-as-you-go or full funding). Thus, the report uses the term "jkstpillar" or "pillar one" to refer to the parto f a pension system intended to keep the elderly out o f poverty; "second pillar" to that part intended to help individuals smooth consumption over their life-cycle to prevent a dramatic fall in income at the time o f retirement; "thirdpillar" to the financial instruments available on a voluntary basis for workers to increase their income in old age; and `tfourth pillar" in reference to non-financial elements o f welfare in old age, such as health care, housing andfamily support (Holzmann and Hinz,2005). Further, the report addresses an emerging distinction between "contributory" and "non- contributory" benefits. In some circles, the latter have come to be called "zero pillar" pensions, to distinguish them fiom minimumbenefit guarantees conditioned on a history o fpayroll contributions. ... Vlll Introduction,Structure,and MainFindingsof this Report I.Expandingcoverage-Argentina'skeychallengeinoldageincomesecurity 1. The share o f Argentina's elderly that receive social security benefits is falling. In 1992, about 80 percent o f the elderly population was receiving old age and survivor pensions. However, by 2004 the portion o f covered elderly had fallen to 68 percent. Although still high relative to other countries in Latin America, Argentina's contraction in coverage over the past decade has been more persistent and accelerated than changes in coverage among its immediate neighbors in the region. Whereas coverage o f the elderly inArgentina contracted by almost fifteen percent from 1992 to 2004, the share o f elderly receiving benefits in Chile fell by only five percent during roughly the same period (from 1992 to 2003), inUruguay the fall incoverage was barely two percent (from 1995 to 2004), while inBrazil the share o f elderly receivingbenefits actually roseby over 7 percent (1992 to 2002). Furthermore, poorer households have borne the brunt o f Argentina's coverage contraction. The share o f the elderly receiving benefits fell by 30 percent in the poorest quintile o f the income distribution, by six percent in the second quintile, and by eight percent in the third quintile (see Figure 1, showing beneficiaries among the elderly, and the yearly percentage change incoverage by income groups). Figure 1. Changesinthe Share ofthe ElderlyReceivingSocial Security Benefits m Q 1 -Q2 Q4 - I Q5 -Beneficiaries Armng the Ederiy (%) right-hand scale 10 - -- 80 8 - -- 70 I -- 60 50 40 30 20 -4 10 -6 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 ource: R o h a n and Lucchetti(2006) based on successive waves of the EPH. 2. Without sustainable changes to Argentina's social protection system, the current trend is likely to persist. The most reliable indicator o f the portion o fthe elderly who will receive benefits in the future is the share o f the labor force that i s contributing to retirement security plans today (see Box 1). This share has also fallen. According to survey data for urban salaried workers, in 1992 about 47 percent o f the labor force contributed to a pension plan.' By 2004 the share o f contributors had fallen to 35 percent. As with the fall in coverage outcomes, this overall contraction in participating workers o f roughly 25 percent has been sharpest among lower income households. Participation among workers from households in the poorest quintile o f the income distribution fell by 96 percent, in the second quintile by 72 percent, and in the third quintile by 20 percent (see Figure 2, showing contributors to social security in Argentina's labor force, and yearly percentage change incontributors by income groups). 20 I 50 *o 4 3tc .-E 10 - Us Eo n e 0 - * o h e -10- 5 8 1ii 5 2 'E 0 -20- g E z gQ -30 - o s -40 - U $; 5= 4 e =o -50 - 1 -60 ' L O 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 Source: Rohan and Lucchetti (2006) based on successive waves ofthe EPH. 3. Among Argentina's population there have been two principal drivers o f the downward trend in participation. First, the repeated financial crises and economic recessions the country suffered in the latter half o f the 199Os, which led to an unprecedented downturn in 2001 and a debt default in early 2002. Second, institutional factors that encouraged the growth o f unregulated forms of employment as the economy recovered from each o f these episodes (although recovery since 2003 has been accompanied by a notable increase inthe share o f workers who contribute). Data from the Permanent Household Survey cannot be used to show whether self-employed workers contribute to the pension system. Since the level o f compliance observed inregisters is very low, this data assumes that coverage for that group is negligible. 2 Box 1.Definingthe CoverageProblem:What is it andwhy do we Care? At the turno f the twenty-first century, fewer than 15 percent of the world's 6 billionpeople had access to a formal system o f retirement income support (Holzmann, Packard and Cuesta, 2000). The majority o f this population that goes without formal cover lives and works in developing countries. To the extent that households inthese countries can still rely on traditional arrangements to provide income security in old age, there may be little cause for worry. However, with the rapid aging o f populations, urbanization and economic development, many Governments are increasingly concerned that the elderly will have reduced access to traditional safety nets. Thus, extending the coverage o f social insurance has become a policy priority. Retirement pensions are typically the largest component o f the set o f public interventions that make up a social insurance system. Low coverage o f the formal old-age pensions system, therefore, usually mirrors low coverage o f other forms o f social insurance such as income support during unemployment, disability, access to family allowances, and inmany cases health care. Low coverage o f social insurance presents several problems. First and foremost is the problem for the individual and by extension the household. Workers who do not contribute to formal social insurance, either by choice or because o f market or institutional barriers, are not accumulating rights toward the receipt o f benefits should they become unemployed, disabled, or when they loose their ability to work in old age. Nor are their dependents covered should they suffer an untimely death. Household members may find it difficult to cope with losses due to these risks, and can be forced into poverty should the losses from an adverse shock be great. Second is the problem for society. An individual's failure to save or insure imposes an externality. If he chooses to make no provisions for the risks to income, the costs o f his decision fall on others. In countries where a significant number fail to insure, governments face a "Samaritan's dilemma", in that politicians cannot credibly refuse to come to the aid o f a large number o fpeople who suffer a loss, and the burden o f these lossescanrapidly mount on current and future tax payers. Finally, low coverage poses a problem for the social insurance institutions themselves. Low coverage can weaken a traditional, PAYGO pension system if not enough active workers and employers contribute to finance the benefits o f the inactive retired, disabled or unemployed. If a substantial share o f the population is not contributing, the system cannot efficiently pool risks and can quickly become financially unviable. Similarly, where social insurance includes individual retirement accounts, the savings that can arise from scale in fund-management are difficult to obtain when a large number o f workers do not participate, implyingpersistently high administration costs that eat into the savings o f those that do. The growing "informal", unregulated sector in many developing countries is important to the analysis o f coverage to the extent that informal employment opportunities lift the constraint on choices by allowing individuals to avoid Government mandatesto pool risks or save inthe formal retirement security system. "Coverage" can be defined as both as a "stock" and a "flow" concept. The "stock" o f the population that is covered includes all those o f retirement age and older who are receiving a formal retirement pension. The "flow" refers to those individuals o f working age who are members o f the workforce and are currently accumulating rights toward a retirement pension, either by contributing under the parameters of a PAYGO benefit formula, or by regularly depositing savings into a private individual retirement account. Thus, the "coverage gap" also has stock and flow dimensions. The stock consists o f the current mass o f elderly (most worryingly the elderly poor and those living close to the poverty line), with no formal income protection. The flow consists o f the likely stream o f current active workers that would fall into the former category year after year. Source: Barr (2004 and 2000), Gill, Packardand Yermo (2004) and Rofman (2005). 4. In recent years, a worker's history of contributions has become increasingly important to determining whether they and their dependents will eventually receive benefits. Contribution requirements are not new. Argentina's principal social security institutions have always been "contributory", where at least by design the main sources o f revenue are mandatory premiums paid by employers and workers which are collected as pay-roll taxes. However, as part o f a package o fparametric and structural reforms passed inOctober 1993 to containa growing trend inspending, contribution and other eligibility requirements were not only raisedbut more strictly enforced. 5. The reforms passed in 1993 were aimed at halting an acute financial crisis in the social security system. Quite apart from the strain created by demographic changes, the retirement age, benefit, and contribution parameters o f the system had been set with little regard for actuarial balance; certain groups had enjoyed special exemptions from full contribution requirements and had won more generous treatment under the system; and a sizeable portion o f benefits had been awarded discretionally for years. Each o f these factors, and several others, had endangered the long-term financial viability o f the system. Inthe decade prior to the 1993 reform, the annual cash deficits between benefit payments and contribution revenue had grown rapidly. Transfers from general revenue had been used to make up the shortfall. In the mid and late 1 9 8 0 ~approximately 25 ~ percent o f annual spending on pensions had been financed with transfers from the national treasury. But despite the injection o f subsidies, deficits in the social security system had grown to be unsustainable. The Government's inability to honor its obligations spawned a rash o f legal action which led to the declaration o f a "state o f emergency" duringwhich payment o f settlements were suspended. The pension deficit at the national level, along with the financial short-falls o f separate pension plans for provincial civil servants, had grown to 3 percent o f GDP by 1993 (Rofman, 2003; MTEySS, 2005). 6. The changes in parameters and benefit cuts introduced with the 1993 reform put Argentina's social security system on a more financially sustainable path. Long-term projections o f the cash deficit .showthat from 2000 when the deficit stood at a little under 3 percent o f GDP, future deficits were expected to fall steadily to 0.9 percent by 2025 as a result o f reforms (Grushka, 2000). 7. The reforms passed in 1993-in particular the new tougher contribution requirements, the higher retirement age, and less generous benefit formula-averted what would have been a much had they not been passed. Of course, financial forecasts o f the long term savings from Feater long-term financial crisis in the social security system the 1993 reform have to be interpreted with caution, and even the most comprehensive estimates would not have completely captured many o f the changes to the system that have been made during and since the 2001-2002 crisis. But as a whole, the cautious long term financial outlook for Argentina's national pension system is better, although spending on social security at the national level still makes up roughly 29 percent o f total spending by the national government and only 30 to 35 percent o f this spending i s financed from worker and employer contributions (MTEySS, 2005). 8. This said, the 1993 reforms introduced a tradeoff between greater fiscal solvency and the share of elderly receiving benefits, at least for a very long period o f transition. * Estimates o f the growth inpublic pension liabilities had the reform not taken place, indicate that pension expenditure would have grown by about 3 percent a year requiring an additional annual injection o f 1 percent o f GDP for the Government to cover these liabilities (Rofman, 2003). 4 Most projections clearly show this trade-off, with the system's contribution revenue and benefit payments expected to balance by the year 2030, as the share of elderly receiving contributory pensions falls below 45 percent by 2025 (Rofman et al, 1997; Gmshka, 2001 and 2002; MTEySS, 2005). Figure 3. Expected Coverage of Contributory Social Security Benefits 100% ae v 5E c, v) 90% di? e8e 80% U 0 .-.-e> cn 70% P p! 60% 5% w E rc 5Q) 50% 0 2005 2010 2015 2020 2025 2030 2035 2040 2045 2050 Source: MTEySS(2005). 9. However, the trade-off introduced between expected fiscal solvency and the breadth o f coverage is deceptive. The simulated expectations o f long-term financial sustainability do not capture the likely repercussions o f Argentina's other social security crisis-namely the pressure that will be created to alter a national retirement security system that in 10years i s expected to cover less than halfo f the elderly and which will be "national" inname only. 11.PurposeandStructureof this Report 10. This report focuses on the reach o f Argentina's pension system today and in the future - the low coverage of the system that has not been addressed by the 1990sreforms. It examines and discusses policy options that can help solve the problem o f low coverage directly-such as improved deployment o f social assistance transfers to the elderly, and a more proportional link between workers' contributions and public pension benefits-and options that could help improve the reach o f the system indirectly-such as improvements to social security, private pensions and related institutions that might lower the perceived costs o f compliance with the mandate to contribute, restore credibility, and 5 thus provide employers and households with improved incentives to participate in the system. 11. Much has already been written about Argentina's old-age income security institutions and reformed pension system in recent years (see MTEySS, 2003 and 2005, ILO, 2003a, 2003b and 2006, Rofman, 2003). The abundance o f recent analytical work on the state o f social security and pension institutions has been motivated by the economic and social crisis o f 2001-2002, but also by concerns for the viability o f these institutions long before the crisis. For this reason, this report gives special emphasis to individual and household decisions, and the outcomes o f these decisions given the current set o f social security institutions. Several sources o f information are exploited, including administrative data, Argentina's regularly deployed labor and household surveys, and a special survey o f the elderly designed and deployed by the World Bank for this report, the Encuesta a la Tercera Edad sobre Estrategias Previsionales (ETEEP), which was conducted inlate 2003. 12. It is important, from the outset, for readers to note that none o f these sources o f data can be used to paint a complete picture o f the role social security and private pensions play in securing household welfare. Argentina still lacks a fully representative survey sample frame that captures both urban and rural areas and collects compliance information for the self-employed. Thus the coverage and contribution indicators presented earlier (in Figures 1 and 2) which were drawn from Argentina's urban sample frame, are likely to over-estimate coverage. The ETEEP survey uses the same sample frame o f urban areas covered by Argentina's Permanent Household Survey (EPH) and Living Conditions Survey (ECV). For this reason, the analysis reflected in this report suffers from an urbanbias that i s common to much o f household-level empirical work in Argentina, and should be interpreted with this bias inmind. 13. Two additional caveats about the ETEEP have to be disclosed up front. The first is with regard to the timing o f the survey. Interviews were conducted in late 2003 as Argentina was just beginning to recover from the crisis-this is likely to have affected the behavior and outlook of respondents significantly. The second i s the limited extent to which lessons can be drawn from behavior reported by the elderly today to guide policy for income security o f the elderly in the future. Social security institutions and Argentina's labor market have undergone profound structural changes in the last two decades that have to be kept inmindwhen considering the implications o f the results. 14. This policy report summarizes the findings o f four pieces o f background analysis that exploit the new survey data (all available online at www.bancomundial.org.ar, along with details about the ETEEP and the database); highlights the problems and challenges faced by policy makers in strikinga balance between the objective o f extending coverage and that o f maintaining fiscal stability; and presents some o f the issues and options that could be considered ina national dialogue on the future o f the social security system. 15. This report does not provide specific suggestions o f what Argentina needs to reform and how. The report is intended as a contribution to the public debate. However, the findings4rawn from a review o f recent policy and academic literature; interviews 6 with stakeholders in and outside o f the government; analysis o f existing sources o f data; as well as from a special survey o f the elderly-motivate a discussion o f some o f the principal challenges for Argentina's current social security and private pension institutions. 16. The report i s structured in five parts that follow this introduction and summary o f main findings. Part Ipresents a conceptual framework for decision makers charged with crafting policies for old age income security, and describes the evolution and structure o f Argentina's pension institutions. Part I1 presents the findings o f the household empirical analysis using existing surveys and the new data from the ETEEP, including the importance of pensions to household income and poverty, the micro- determinants o f coverage, and the viability o f other income security strategies for the most vulnerable. Part I11 focuses solely on policies and programs designed to ensure minimumincome and cover poverty inold age, presenting estimates o f the fiscal cost o f proposals made by the principal actors and stake-holders in the pension reform debate. Part IV examines the main pension policy issues inthis debate and places these issues in international context. Part V presents some options for the Government and stake- holders to consider as Argentina enters a more detailed and deliberate phase o f defining pension reform. 111.Summary ofMainFindings 17. The main findings and recommendations made in the final part o f this report are summarized inthe remainder o fthis section. 18. Strengtheningpoverty-preventionpensionsis important, but is just one item on the larger reform agenda. Overall, while improvements inthe deployment o f non- contributory, social assistance pensions and a greater proportionality in the eligibility requirements for minimum contributory pensions are important policies for Argentina to improve pension coverage, they are just one item o f a much broader reform agenda. Considered singularly and with no other changes to the system, these policies could be fiscally costly, could displace forms o f social protection targeted to other age groups, and undermine the projected long-term fiscal sustainability gained from the 1993 reform. The fiscal analysis conducted by A16s and Muiiios (2006) for this report, shows that increasing the coverage o f poverty-prevention pensions could cost Argentina up to 0.8 percentage points o f GDP and that as a portion o f GDP these costs could decline over time. The report intentionally reports on a conservative, high-cost scenario - not as an endorsement o f any specific set o f proposed measures, but to act as a base line estimate and to motivate a closer consideration and more exact costing o f options. The likely cost o f non-contributory benefits and proportional contributory benefits will o f course vary depending on a wide range o f specific design features. For example, targeting non- contributory pensions only to the elderly poor with no other source o f revenue could lower fiscal costs to 0.5 percent o f GDP. Further measures, such as raising the age at which non-contributory pensions are available can lower these cost further. But the real key to making new poverty prevention pensions fiscally sustainable in the long-run i s to increase participation inthe earnings-related pension pillars. 7 19. The lack of credibility of formal institutions may be one factor for low participation rates in social security. The current participation rate in contributory, earnings related plans i s too low, and if the current share o f the work-force that contributes does not increase, the cost estimates o f non-contributory and proportional minimum contributory benefits quickly increase. A large portion o f the analytical work conducted usingthe ETEEP seeks to answer whether the low level o f participation among workers in Argentina i s a reflection o f social exclusion or household choice. Although far from conclusive, the evidence presented here suggests that exclusion is a relatively more important determinant o f low levels o f participation, and that employers are playing a critical "gate keeping" role. But also apparent in the results is that the labor market only offers a viable alternative to the pension system for those who have always had and continue to enjoy good labor prospects. Revealingly, those who continue to work in old age are individuals who report the longest employment histories. Furthermore, the working elderly in the ETEEP are not only those whose predicted labor incomes are relatively high,but also those who have contributed to the pension system for fewer years throughout their working lives (than those who are inactive and receiving benefits). However, whether exclusion i s more important than household choice in determining participation, incentives are still likely to matter: the incentives employers have to offer formal contracts to their workers, and the incentives workers have to rely on the pension system as an important tool for securing income and welfare in old age. Employers will find it easier to play a gate keeping role if the social security system is lacking in credibility and workers do not value the coverage it offers. 20. The lack of a credible, systematic, predictable adjustment rule for pension benefits creates a vacuum. As in the mid 1980s, this vacuum has been filled with numerous lawsuits brought by beneficiaries, which have made Argentina a country with one o f the highest levels o f legal action related to pensions. Argentina's National Constitution presents constraints on the options that the National Government has in formulating an adjustment rule. However, there are examples o f indexing policies from other countries that would allow the Government to do much more even within these constraints. A credible and predictable adjustment policy couid help restore the credibility o f public pensionbenefits inthe eyes o f prospective contributors. 21. The commercial branch of the pension system would need to be given a fair chance as, at current rates, it will provide inadequate benefits. The contribution rate i s a policy variable that the Government has not restored to the values intended by the architects o f the SIJyP. Further, the high concentration o f pension hnd portfolios in public fixed-income securities comes at a high opportunity cost in forgone investment in the wider economy and could expose workers' retirement savings to political risks. For this reason, diversification o fAFJP holdings could be important not only to the economic and financial performance o f the commercial second pillar, but also for its sustainability. Regulatory, industrial and political factors have raised and kept the cost o f insurance too high. Problems with second-pillar insurance lie inArgentina's highbenefit expectations; generous eligibility requirements; regulation o f reserve requirements; rules that index insured capital to share value; structures for determining eligibility and processing benefits; the way AFJP portfolios are valued; and a market structure that has resulted in a close relationship between the fund managers and insurance providers that hinders 8 healthy competition. Examples o f public provision o f insurance are not encouraging. But inorder for privateprovision to once again be viable, specific changes would have to be underpinned by consumer protection mechanisms; clearer rules on what and how information i s disclosed by insurance companies; the development o f risk-focused supervision; and enhanced corporate governance standards. 22. The lack of a credible, systematic, predictable adjustment rule for pension benefits creates a vacuum. As in the mid 1980s, this vacuum has been filled with numerous lawsuits brought by beneficiaries, which have made Argentina a country with one o f the highest levels o f legal action related to pensions. Argentina's National Constitution presents constraints on the options that the National Government has in formulating an adjustment rule. However, there are examples o f indexing policies from other countries that would allow the Government to do much more even within these constraints. A credible and predictable adjustment policy could help restore the credibility o f public pensionbenefits inthe eyes o f prospective contributors. 23. The multiplicity of pension plans is a source of inequity, undermines fiscal stability, and hurts the efficiency of policies at the national level. There are important disparities between the pension parameters o f the SIJyP, and the public pension plans for provincial and municipal civil servants that remain separate from the system, as well as special regimes for other select groups (members o f the legislature, judges, the military and others). The multiplicity o f substitutive pension plans has a negative impact on equity and efficiency, and poses a risk to long-term fiscal sustainability. Recent efforts to restore plans that were integrated with the SIJyP pose a risk to equity and fiscal sustainability as well. Further, the fact that the professional pension plans are not subject to government regulation or non-government independent scrutiny, yet benefit from government mandates which could entail implicit guarantees, i s a risk that remains largely un-quantified. 24. The remaining choice between the public and commercially administered branches of the SIJyP second pillar poses several problems but to the extent that affiliates to the public branch are fewer and older, these problems are relatively minor and could disappear in time. The continued existence o f a public, defined- benefit, earnings-related pension plan, leaves open the option o f return which could be appealing inthe short-run but would need to be assessedcarefully. Inthe past four years, calls to allow AFJP affiliates to return to the public branch o f the second pillar have become widespread. In countries where a periodic choice to switch has been offered, it often has led to uncertainty in calculating public pension liabilities and to strategic behavior, that raises the cost for the public system. For the sake o f long-term stability and credibility o f the system, there might be benefits in addressing legitimate grievances that may exist among SIJyP affiliates who found themselves affiliated to the commercial branch o f the second pillar against there will. But the fiscal-cost implications o f any offer to choose again between the public and commercial branches would need to be carefully evaluated, especially ifgoing beyond a one-off, conclusive, and time-bound offer. 25. Ageing will again threaten fiscal sustainability; however, there is an opportunity to mitigate this risk early on. Life expectancy at birth has increased by 9 more than 10 years since the middle o f the twentieth century. Argentina can expect further population ageing, albeit at a less accelerated rate than in recent decades which will eventually raise demographic risks faced by the pension system yet again. While a demographic bonus is expected in the form o f a temporary and limited increase in the share o f population aged 15 to 64, policy makers might want to take advantage o f this demographic window o f opportunity. In the medium term, policymakers may need to either equalize retirement ages for men and women, raise the retirement ages, and probably both. 26. There is no policy rationale for the gruciuble pensions granted by members of Congress, and expansion of targeted non-contributory pensions could be financed with expenditures currently used for the graciable pensions. With respect to existing non-contributory pension programs, the current PNC program is critically important, and on the whole, the elements o f the program that are designed to be targeted to the poor perform reasonably well. However, the graciable pensions granted by members o f Congress are an exception. Although these sort o f discretionary benefit programs exist in other countries, they rarely are as large as the program inArgentina. If further measures to raise or expand targeted non-contributory benefits are considered, they could be accompanied with the stepwise reduction o f the graciables. This would make more fiscal space for non-contributory benefits that are targeted according to clear poverty criteria, and strengthen the credibility o f the PNC program as a whole. 27. Since Argentina introduced full statutory coverage of poverty-prevention pensions, an evaluation of how the government implements this program would be important. Argentina has already taken a big step towards eliminating the risk o f poverty in old age. When it lifted the ration on targeted social assistance pensions in 2003, Argentina fundamentally changed the nature o f at least the targeted benefits in the PNC program, creating a form o f coverage that i s closer to a right than a standard social program. From the moment this decision was taken, at least intheory and with respect to the risk o f poverty in old age, Argentina introduced full statutory coverage. Interms o f preventing poverty in old age, whatever coverage gap still exists between the poor and non-poor elderly since 2004 reflects implementation or take-up problems. Clarifying how the PNC are targeted and providing public information on coverage expansion would be important to reduce the current low levels o f coverage. The shift from a mere social assistance program to a universal right to protection against poverty in old age raises the question o f where best to `house' the PNC institutionally in order to improve implementation and attain optimal impact, given the incentive structure o f eligibility and benefit parameters inthe broader social security system. 28. With full coverage against the risk of poverty in old age, the Government could reconsider the role of other minimum-pension structures. A remaining policy question i s whether Argentina needs to retain a state-subsidized, minimum contributory benefit with an un-rationed social assistance benefit targeted to the elderly in place. A subsidized contributory benefit-r matching contribution scheme-may indeed still be useful, not as a guarantor o f minimum old age income, but as a device for improving incentives to participate among lower-income workers. However, any such incentive instrument would need to be followed closely, for if it were not to lead to increased 10 participation, it could result in perverse regressive transfers from uncovered to covered households. IV. LookingAhead 29. Looking beyond these important areas that are in need o f improvement in the medium term, there are fundamental, long-term issues and questions that would need to be answered before a clear understanding and consensus on the social contract can be achieved. For generations, the expectation held by Argentineans o f a "decent pension" was about 80 percent o f a relevant multiple o f earnings in the years leading up to retirement. This expectation did not change with the 1993 reforms, but the institutions did change and were no longer designed to meet these expectations. With respect to income security in old age, what have Argentine households come to expect from the State since the 1993 reforms? Even after the fundamental re-distribution o f risk brought about by these reforms, have household expectations been altered? What are the rights households are entitled to, if they meet the responsibility to comply with the mandate to participate? To be covered against the risk o f poverty in old age, or something more? If households expect more from the State than just to prevent poverty in old age, manage the risk o f inflation, and regulate financial risks, can that expectation be made more explicit? When Argentineans call for the State to take a role in helping to smooth consumption from old age through a mandatory, earnings-related system, what do they expect (measured inaverage replacement rates) from their compliance with the mandate? 30. Whatever the particular reform model that i s put inplace to address medium term challenges o f the current system, it is important that it be implemented and guided by a central policy makingbody with undisputed authority over the system and a clear idea o f the long-term objective. The need for a strong, centralizedinstitution to guide policyhas been particularly apparent in the years since Argentina's 2001/2002 crisis. For the past years, the first and foremost challenge o f national pension policy has been to extend coverage and thus improve the equity, efficacy and efficiency o f the national old-age income security system. Yet despite numerous decrees, regulatory changes and legislative measures initiated by various stakeholders since 2001, the coverage problem remains first and foremost. And the inequitable impact of the pension system has grown worse (see Figure 4). 11 Figure4. TrendinIncomeInequalityamongthe Elderly with andwithout Pensions (Gini Coefficients among Specific Population Groups, 1992 -2004) .*.... 0.65 0.60 0 0 . 0.55 0 0 0.50 0 0 0 0.45 a 0 5 1 0.40 m e * c 3 * 0.35 a * * 0.30 0.25 0.20 92 93 94 95 96 97 98 99 00 01 02 03 03-11 04-1 Receiving Wnsions (all) Receiving bnsions & Older than 65 0 FupulationOlder than 65, Receivingand Not ReceivingPensions Source: CEDLASusingEPH2004, second semester. 12 Part I.A Framework to Guide PensionPolicy and Analyze How Argentina Covers the Losses from Ageing 31. Social insurance i s intended to be an instrument (or more precisely a package o f instruments) made available by government to help households manage the risks to their consumption posed by a wide array o f contingencies. Because it i s designed to cover the losses that private insurance cannot cover-and to augment private coverage where this fails or falls short-social insurance instruments are not expected to strictly conform to the actuarial rules that govern the market for risk. This said, policy-relevant insights on the nature o f a particular loss, how it could best be covered, and the degree to which the private sector should be expected to help, can be drawn from the classical economic insurance models on which these rules are based. La. TheProspectiveLossesfrom Ageing and Objectivesof Pension Policy 32. How policy makers approach population ageing presents a case in point. "Old age" is not a bad state, per se. In most societies, long life is considered a blessing, and even a goal o f development. However, from an individual's perspective, ageing i s also accompanied by at least three distinct (although related) prospective losses: the loss of earnings ability, as the body and even the mind succumb to natural deterioration over time and an individual i s no longer able to count on his labor as a source o f income; unanticipated longevity that increases the period o f life when an individual has to consume but cannot earn income from his work; and derived from these, the prospect o f poverty inold age. 33. Covering these losses lies at the core o f broadly accepted objectives o f pensions policy, which are (i)to prevent a dramatic fall in household means when income from labor i s no longer an option-the "consumption smoothing" objective; (ii) ensure that to these means are at least o f a level adequate to sustain individuals' needs for the remainder o f their life-the "income adequacy" objective; and (iii)to ensure that household consumption will never fall below an absolute minimum, socially-acceptable level-the "poverty prevention" objective. Principles drawn from the classical insurance model can help guide policy-makers on how best to pursue these objectives and cover the three prospective losses from ageing. 34. As a population ages and average life expectancy improves, the first prospective loss becomes increasingly predictable. That is, a greater share o f the population i s likely to reach old age, typically marked by a threshold set at 60 or 65.3 The first national social insurance plan ever established in Germany in 1889 set a similar threshold when average life expectancy for men was only 45.4 At the time, the probability o f reaching old age Population ageing is actually an outcome o f several forces, including people living longer and having fewer children. The logic presented here places greater emphasis on the former, but is not invalidated by the latter: the outcome o f both demographic forces is an increasing share o f older people in the population. From an individual's perspective, this should make the prospective loss o f earning ability from ageing more predictable, while from a policy-maker's perspective it becomes more systemic. When Otto von Bismarck proposed the creation o f the system in 1881, the intended retirement age was 70. This was then lowered inthe political negotiations prior to establishment o f the system. 13 and incurring this loss was very low. Like other forms o f insurance, this system, and those that were established later in other countries, was based on a risk-pooling model that defined the benefits which would be received by the relatively few that reached old- age. 35. But as with other prospective losses, as old-age becomes more frequent it is increasingly difficult to cover with a risk-pooling, defined-benefit model. This is becoming more apparent across the OECD and in developing countries, where governments are concerned with rapidly growing public pension liabilities. Thus with increasing longevity (see Figure 5) and a greater share o f elderly in the population, it i s reasonable to expect individuals to shoulder. a greater share o f this burden, and to rely more on individual savings to smooth consumption between working life and old age (see Box 2). Figure5. PopulationAgeinginArgentinaand OtherLatinAmericanCountries (Projected change inlife expectancy of average male at age 65) 19 1 8 - .a 8 1 7 - 1 1 6 - t3 9 3 15- 14 - 13 1 Source: CELADE (2004). Boletin Demografico, numero 74. Santiago, Chile 14 Box 2. The ComprehensiveInsuranceFramework Individualsand societies can respondina variety of ways when faced with the prospect of economic losses from shocks like natural disasters, sickness, sudden death, disability and unemployment. Social risk management is a conceptual framework encompassingthree broad categorieso f responses to shocks: ex-ante prevention, mitigation, and ex-post coping (Holzmann and Jorgensen, 2000, and World Bank, 2001). The comprehensive insurance approach offers a tool for determining which mitigationinstrumentsandpreventive measures will be most effective giventhe size, fiequency, and the extent o f externality of a range ofpossibleeconomic losses(Ehrlich, and Becker, 1972, and Gill and Ilahi, 2000).' The comprehensive insurance problem for individuals, households, or g o v m e n t s (as the risk-mitigating agent o f interest) is to determine the optimal mix of "market insurance", "self- insurance", and "self-protection". Both market- and self-insurance transfer income from "good" states to the "bad" states o f the world. Market insurancepools risks across individuals. Where it is available, it canbe purchasedat a pic+-the insurance premiumwhich, in classic insurance, should be set according to the size of the prospective loss and the probability o f the bad state coming about. (Conventionally, the price o f market insurance 7r i s said to be "actuarially fair" if xi = (1 +a)p, L ,where p,L is the expected loss, Le., size of the loss L, weighted by probability p o f the loss coming about in the bad state, and a is a "loading" charged by the market- insurance provider to cover administrative costs and profit, as inEhrlick and Becker, 1972, Barr, 2001). Self-insurancs-essentially individualsaving-does not involve risk pooling. While it has no explicit price, its cost can be imputed fiom the expense people incur to save, say in forgone consumption. Individualswithout market or self-insurancemust copewithwhatever lossesbefall them. They can, however, mitigate riskby "self-protecting" i.e., taking preventive measures. Preventionreduces the probability that losseswilloccur, butdoesnotreducethe size o fa loss should one occur. Individualsand householdsthat are unable (or choose not) to take preventive measures or insure by saving or through risk-sharing structures, are forced to cope with the full costs o f losses in the wake o f shocks. To reflect the growing sophistication and changes in the lexicon used in the literature we have replaced the terms originally used as follows. We refer to market-insurance as riskpooling; to "self insurance", as saving; and "self protection" as prevention. Thus, risk pooling redistributes consumption opportunities toward the bad states o f the world at a price. Saving redistributes income similarly-cash balances reduce fluctuations inconsumption-but does not pool risks. Prevention lowers the probability o f the bad state occumng. According to the framework, individuals or households smooth consumption over good and bad states o f the world. Where riskpooling is missing, the individual is forced to smooth consumption using only savings and prevention. In a world where the option o f both risk pooling and saving exist, the individual sees these as substitutes. Risk pooling-available at or near actuarially fair prices-reduces saving. However, greater coverage o f risk pooling does not inevitably result in individuals spending less on prevention (what economists call "moral hazard"). Ifprevention leads to a lower likelihood that the bad state will occur, and if this i s rewarded by the market in the form o f lower premia, risk pooling and prevention can become complements-individuals can be encouraged to take up more prevention in return for cheaper risk-pooling instruments. Figure 2.1 illustrates the prescriptions o f the comprehensive insurance framework on two axes, each representing different dimensions o f prospective losses: size (amount o f the loss) andfiequency (probability o f occurrence) o f the loss. I t is more efficient for individuals to cope with rather than try to insure against small, rarely-occurring losses (the bottom-most, left-hand corner o f Figure 2.1). However as prospective losses become more frequent, it is relatively more efficient to engage in prevention and savings to mitigate (that is, to lower the probability, and cover the costs of) the loss. As a prospective loss becomes less frequent but increases in size, it is more efficient to engage in risk pooling. For many o f these large, rare losses, households will have incentives to engage inprevention to lower the probability they will occur even further. However, for losses that are frequently occurring and catastrophic insize (the right-hand, upper-most comer o fFigure 2.1) there is little that individuals, households or even markets can do on their own. (continued on next page) 15 (continuedfrom previouspage) Figure 2. 1.Mitigation Instrument According to Size and Frequency of Potential Losses A I I I I I I I I I I I I I F r e q u e n c y o f L o s s ( p r o b a b i l i t y o f o c c u r r e n c e ) Source: Ehrlich andBecker, (1972), Gill and Ilahi(2000) and Gill et al. (2004). 36. With respect to the second prospective loss, unanticipated longevity, within a given generation, there will always be some who are longer lived than others. Even if these individuals can form reasonable expectations about their longevity based on their own health and on that o f their parents, many will still underestimate the period o f life without work that they will need to finance. Thus, by definition, this prospective loss is relatively rare and readily insurable. Indeed the private sector inmany countries provides defined-benefit, longevity insurance to households inthe form o f annuity contract^.^ 37. Finally, with respect to the third prospective loss, in the context o f long term economic development, as each generation's labor market prospects and opportunities for investment and capital accumulation improve, the share o f each cohort that arrives at old age without the means to finance at least a minimum level o f consumption should shrink. Somewhat less sanguinely, given the strong links between income and life-expectancy, The sound and efficient private provision o f annuities, like other forms o f insurance and contractual savings, cannot, o f course, be taken for granted. A safe, strong and efficient insurance sector demands careful regulation in even the most sophisticated economies with the strongest institutions. This issue is covered ingreater detailinPart 111. 16 there is a "non-random" attrition among the elderly that lowers the share o f elderly poor relative to the share o f poor in younger age groups (Barrientos, 2003). All else equal, poverty in old age should be increasingly rare and, therefore, also a loss that can efficiently be pooled (see Table 1)6. But given the social definition o f the loss, "poverty", markets cannot be expected to respond with risk-pooling instruments, and the State often steps into fill this gap. Table 1.Povertyamongthe Elderly Per capita income equal or Per capita income equal or Equivalent income equal or greater than $2 greater than 50 per cent of greater than 50 per cent of median median* AII 60+ 65+ All 60+ 65+ All 60+ 65+ Argentina 16.6 5.3 4.5 20.7 6.4 5.2 18.6 9.5 8.5 Bolivia 47.0 48.0 50.2 28.4 30.2 31.4 27.9 36.6 38.4 Brazil 22.2 6.3 5.4 25.6 7.5 6.3 23.5 10.3 9.4 Chile 8.9 4.4 3.8 13.1 6.6 5.7 11.8 7.9 7.3 Colombia 30.4 29.9 31.2 26.6 26.4 27.6 25.8 29.2 30.8 Costa Rica 11.8 16.8 19.1 18.1 22.2 24.2 17.3 30.4 33.6 DominicanRep. 13.8 19.8 21.8 25.2 34.1 36.4 23.8 38.1 40.9 Ecuador 40.9 47.3 49.4 26.5 33.5 34.7 26.0 38.4 40.0 El Salvador 36.9 33.7 33.6 30.5 28.5 28.4 30.1 30.4 37.7 Guatemala 36.6 44.7 36.6 44.7 34.1 45.9 Honduras 36.6 28.9 30.3 27.5 19.9 20.9 26.8 24.3 25.7 Jamaica 17.5 17.5 18.2 26.2 24.6 26.4 25.0 29.7 32.3 Mexico 27.2 30.2 31.O 22.9 27.4 28.0 21.9 30.5 31.3 Nicaragua 38.3 31.8 32.6 20.3 15.8 16.2 19.7 18.3 19.0 Panama 21.4 18.1 18.0 22.2 18.8 18.7 20.6 21.7 21.7 Paraguay 30.2 23.3 23.2 25.9 19.4 18.6 24.5 24.4 23.9 Peru 31.8 21.9 21.6 24.7 15.8 15.2 24.2 18.3 18.7 Uruguay 1.I 0.1 0.2 22.0 6.5 5.4 17.8 8.7 7.5 Venezuela 29.5 25.0 25.3 20.0 17.4 17.5 19.7 20.9 21.2 ~ Source: "Alleviating Old Age Poverty in Latin America'' presentation given by J.J. Dethier, June 2004, Bogoth, Colombia. * number of adults + 0,3 x number of children (aged 16 or less)]. Income equal or below 50 per cent of median income using OECD equivalence scale [equal to 0,5 + 0,5 x Lb.Argentina's Institutionsfor OldAge Income Support 38. Argentina's social security institutions are among the oldest in Latin America. The earliest retirement pension plans offered in the country were created just over a century ago. Up to the late 1950s the social security "system" was in fact a diverse array o f heterogeneous pension plans. However, the legislation o f a uniform national benefit level with little follow-up to ensure an adequate level o f contributions, led to growing financial problems. To address these problems, a new law was approved in the late- Table 1 shows poverty rates among different age groups in the population for a selection o f Latin American countries. In some notable cases, the assertion that poverty among the elderly should be lower than among other age groups, is not borne out by the data. However, these data do not capture asset wealth. Typically, although their income is lower, the elderly hold a greater share o f non-liquid assets. And with economic development, all else equal, the total life-time opportunities to buildassets should increase, again sustaining the assertion that poverty inold age should become relatively rare. 17 1960s, which consolidated most o f the previously existing funds into a national, defined benefit, pay-as-you-go (PAYGO) system covering most o f the formal labor force with three separate plans: one for civil servants, another for the self employed, and a third for private sector workers. Separate pension plans offered by the national government for the military and still others run by the provincial governments for local civil servants and specific professional activities continued operating along side the three principal plans with variable financial outcomes (Rofman, 2003). 39. This national social security model, which segregated coverage by sector o f employment, remained in place for more than 25 years, when growing financial difficulties again became apparent. By the early 199Os, the deficits between the benefits defined by the system and contribution revenue grew and the social security system was unable to meet its obligations. The Government consolidated the three national plans under a single social insurance institute and began to consider deeper parametric and structural changes. A package o f reforms passed in October 1993 introduced a new social security system that started operating in July 1994. The reform package raised the retirement age by five years to 65 for men and 60 for women; phased in changes to the benefit formula, contribution rates, and eligibility requirements o f defined-benefit plans that tightened the link between benefits and contributions; and introduced a new branch o f individually owned, commercially managed retirement savings accounts. 40. Today Argentina's Government makes an array o f instruments available to help households manage the losses that arise from ageing, both directly and through regulation o f private provision. The 1993 reform established a new "integrated" national social security system, SIJyP (Sistema Integrado de Jubilaciones y Pensiones), along the lines o fthe "multi-pillar" model that was shaping pensionreforms elsewhere inLatin America. 41. All economically active individuals over 18 are required to participate in the national system, whether employed or self e m p l ~ y e d . ~Since the establishment o f the SIJyP new entrants to the labor market who find jobs in Argentina's formal sector (including regulated self employment) have chosen between individual accounts (managed by Administradoras de Fondos de Jubilaciones y Pensiones, "AFJP"), and a down-sized defined benefit plan (managed by the public Social Security Agency "ANSeS") to finance the earnings-related portion o f their retirement pensions. The earnings-related benefit from the branch o f individual accounts, the Jubilacidn Ordinaria or "JO", can either be received in the form o f regular withdrawals o f the affiliates' accumulated savings or as an annuity, and inthe public branch, the Prestacidn Adicional por Permanencia or "PAP", is paid directly by ANSeS.' Underpinningearnings-related benefits i s a "universal" flat pension (Prestacidn Bcisica Universal, "PBU") that sets a lower bound on the income covered workers who have contributed to the system for at 7 Those covered by independent, separately administered plans-discussed below-are exempted from rarticipation inthe SIJyP. The 1993 law also established clear transition arrangements. Workers that were contributing to the system prior to the 1993 reform can stay in the RRP or migrate to an AFJP. Those who stay in the RRP contribute toward a PAP, and receive a PC (Prestacidn Cornpensatoria)that recognizes rights acquired through contributions to the system prior to 1994. Those who choose to open a private AFJP account also receive a PC from the public pillar, and the JO financed from accumulated savings and returns from investment. 18 least 30 years can expect to earn when they retire. Workers that have at least 10 years o f contributions can claim a lower flat pension, the Prestucibn p o r Edud Avunzada or "PEA", but not untilthey reach 70 years o f age.' 42. Financed and administered separately from the SIJyP, the Government offers targeted social assistance pensions-Pensiones No Contributivas or "PNC". There i s no contribution requirement to receive the PNC, nor are benefits under this non-contributory program strictly limited to covering losses from ageing." The non-contributory program also pays special pensions granted at the discretion o f members o f Congress (gruciubles), as well as other non-contributory pensions granted by special laws to several groups other thanthe poor andvulnerable. l1 Lc. TheMultiple Pillars of OldAge Income Security in Argentina 43. Since the 1993 reform, and with some important exceptions, the instruments that Argentina makes available to help households manage the risks to income from ageing are better aligned to cover these losses. The Government requires individuals to shoulder a greater portion o f the risk o f the increasingly predictable loss of earnings ability, either by requiring they contribute to savings accounts managed by the financial sector, or to a restructured defined-benefit planthat tightens the link between benefits and contributions. It has developed strict rules on how these savings can be withdrawn through guidelines for "programmed withdrawals" and enforcement o f minimum annuitization requirements through privately provided insurance to cover the risk o f individuals outliving their retirement savings. It guarantees a minimum level o f retirement income for individuals who have met minimumparticipation requirements in either the commercial or the public earnings-related plans, and provides a modest benefit targeted to the elderly indigent who were too poor or otherwise unable to participate in the system. In addition, the Government maintains tax incentives to encourage voluntary retirement savings, and regulates the provision o f contractual savings instruments by the financial sector. 44. Inthe widely accepted pensions lexicon, Argentina's multi-pillar pension system (i) two"first pillar"instruments, (thePBUandthePEA), aswellastargetedsocial has assistance pensions (PNCFthat pool the risk o f poverty in old age; (ii) mandates the demand and regulates the supply o f "second pillar" savings (the JO through AFJP individual capitalization accounts) and annuities to ensure that individuals engage in and benefit from a minimum level o f consumption smoothing, while allowing the option for workers to retain a defined benefit plan for their earnings replacement (the PAP administered by ANSeS); and (iii) encourages additional, voluntary "third pillar" savings through regulated private instruments. ~~~~ ~ The SIJyP receives statutory contributions from employers and workers. Minimum and maximum earnings subject to mandatory contribution are calculated using the MOPRE (discussed in greater detail in Part 111). Self employed workers-also mandated to participate inthe SIJyP-contribute as both employers and employee on a presumed income roughly equal to the employee earnings subject to the mandate. lo Non-contributory pensions are granted to the elderly, the disabled, mothers o f seven or more children, Malvinas war veterans, and the relatives o f the political victims who "disappeared" during the military government. " The PNC program is examined ingreater detail inPart 111, which includes description of the various benefits paidunder the program as well as administrative and budget details. 19 Affiliateschoosebetween.., Pillars(or "tiers") Public (RPP) Commercial(AFJP) Administeredby... Pillar One PEA & PBU PEA & PBU ANSeS PAP ANSeS Contributory I PillarTwo I I JO r insurancecompanies Transition Generation PC PC ANSeS Contributory I Pillar Zero PNC (social assistance) MDS I I I I Source: Adaptedfrom Bertranou, Rohan and Grushka(2003). 45. The SIJyP public second pillar RPP/PAP benefit remains a risk-pooling, defined benefit instrument. However, Gill et al., (2004) argue that the key feature o f sustainable instruments intended to meet the "consumptions smoothing" objective as the population ages, is greater "individualization". Reform to national social security systems in Latin America have introduced this individualization in the form o f defined-contribution, commercially-managed retirement savings accounts. Elsewhere, notably in several countries inEurope and the former Soviet Block, "individualization" has been introduced through individual accounts with a "notional" defined-contribution (NDC), accumulation and rate o f return that retain PAYGO tax financing and remain under the direct administration o f the government. In Brazil, greater individualization was achieved in 1999 by altering the public pension benefit formula to tie pensions closely with contributions and life expectancy. The common outcome o f this individualization is a shift in how demographic risk is borne, away fiom the government and onto the individual, albeit to very different degrees. While the RPP/PAP remains a defined benefit pension plan and lacks many of the desirable characteristics o f the NDC instrument, the 1993 reforms introduced a much closer link between contributions and benefits that shifted a greater burden for consumption smoothing onto individuals. 46. Apart fiom a better alignment o f different instruments (risk-pooling and individual savings) to more efficiently cover the losses from ageing, a multi-pillar structure i s important for diversifymg and distributingthe risks to adequate income inold age between government, employers, and households. Instead o f government-and by extension, tax payers-primarily bearingthe risk, as in a single pillar system, the multi- pillar approach's mix o f government guarantees, mandated individual savings, and voluntary pension arrangements-including those between employers and their workers-spread the demographic (longevity), macroeconomic (inflation and recessions) and investment risks (low or negative financial returns) to adequate retirement income. The diversification o f risk inArgentina's multi-pillar system, although not yet optimal, is regarded as an improvement over that which prevailed under the single-pillar, defined benefit, PAYGO system that existed prior to the 1993 reform. 20 I.d Separate, Independent Pension Plans 47. However, separate military, provincial and professional pension plans remain untouched by the institutional changes o f the last 15 years. Currently there are 141 pension plans in Argentina, operating independently from the SIJyP, each with its own eligibility requirements, minimumretirement age, benefit levels and average replacement rates.l2 48. Inthe public realm, in addition to special regimes for federal judges, the military and police, numerous separate defined benefit plans for subnational public sector workers, sometimes subsidized with public finds, continue to operate. In 1994 and 1996 selected provincial pension plans were integrated with the SIJyP. Provinces that chose to integrate their plans were required by the National Government to relinquish their right to provide pensions for their civil servants, and to adjust the eligibility and benefit parameters o f the plans being transferred to the parameters o f the national system. The fiscal and efficiency results have been mixed (World Bank, 1998). But by the end o f the 199Os, the National Government abandoned the policy o f integration which had been applied to only 11o f 24 provincial pension plans and 1 o f 24 municipal pension plans. 49. Inaddition to public plans for sub-national civil servants, a wide array o f separate plans are offered by professional associations for accredited members o f their profession, such as notaries, medical doctors and others. These plans receive mandatory contributions from their particular professional group, and in turn offer benefits ranging from old-age pensions to maternity pay and even credit. Access i s strictly limited to members o f a particular profession and mandatory contribution i s enforced through the professional accreditation process. Although created and sanctioned by provincial and municipal laws, the professional plans are unregulated, data on their solvency i s scarce, and monitoring o f their activities is difficult. All these institutions are examined in greater depth inlater parts o f this report. l2This figure includes the various benefit programs under the PNC. Ifthese are not counted, the numbero f pension plans that operate separately from the SIJyP falls to 136 (ILO, 2006). 21 50. ~ o in ~~~ ~ e~nist lessn ~a ye ~amonge the~ elderly than a ~ o n gother age ~~ ~ t groups: by the end of 2004, p ~ v ~ -mong the ctdcrly was 19 percent, c o ~ ~ to~ 40 d t y e percent a~~~~~the total p ~ p ~ ~ aExtreme .povcrty is also lowcr at 5 percent a ~ # n ~ t ~ ~ n 15 percent the total popu~at~onAs.~shown inTabfc 3, a ~ ~~~~~ hwith the crisis in 2002. Figure5. Poverty ~ rthe Etdcrly~vs. Total P ~~ p ~inArgentina~ ~ ~ n ~ l ~ t 50 45 40 35 30 $ 25 20 15 10 5 0 Total 654 Total 65+ Below poverty line Below extrema poverty line of poverty a ~ theoeldcrly is ~stro ~ also associated with rn d U ~ ~ the extent ~ ~ ~ ~ - 2002). In A r ~and~ ~ ~ i ~ elscwhere in the region, social security benefits ~ ~ oor ~ i~o n - ~ o ~ t~~play~at o ~~y ~ t c ~ ~ ~ very l ~ ~ o roleain nr c d u ~ i npoverty. According to cstirnatcs using the EPW, in tfzc ~ ~ ~ 22 absence o f social security benefits, poverty among the elderly would rise to over 50 percent. Similarly, in the absence o f social security benefits, extreme poverty among the old would be as highas 30 percent. Furthermore, the regularity and reliability o f pension benefits provide effective buffers against the kindo f income risk generated by changes in macroeconomic conditions and crises. Ina study o f income risk and poverty inArgentina straddling the 2001-2002 crisis, Cruces and Wodon (2002) conclude that households with elderly members were less exposed to income risk compared to households with inactive or unemployed members. 52. The importance of social security for keeping households out o f poverty i s observed elsewhere inthe region as well. Social security benefits constitute a significant proportion o f total household income: 65 percent inUruguay, 89 percent in Brazil and 87 percent in Argentina (Rofinan, 2005). In Brazil, Barrientos (2003) finds that having a pensionrecipient inthe household lowers the probability that the household i s poor by 21 percent. Thus, pension benefit levels and the extension o f coverage explain not only the incidence o f poverty among the elderly, but the differential between this group and the total population (CELADE, 2003). Ila. Patterns in Social Security Coveragein Latin American Countries 53. The level o f social security coverage among the elderly in Argentina i s relatively highwhen compared to other countries in Latin America. Gill et al. (2004) show that among 10 selected Latin American countries, Argentina has among the highest proportion o f pension recipients among the population aged 65 and over. Rofinan (2005) includes seven additional countries in the analysis and finds that Argentina is only surpassed in coverage o f the elderly by Uruguay and Brazil, and by Chile and Bolivia if non-contributory benefits are also taken into account. But as explained in the introduction, what is particularly worrying about the coverage o f pension benefits in Argentina is that, despite these relatively high levels, there i s a notable downward trend inthe last decade. This sharp fall inthe portion ofelderly receivingbenefits hasnotbeen observed inother LatinAmerican countries. 54. Among selected Latin American countries, social security coverage for the elderly is positively associated with national income. However, coverage in Argentina, which in 2000-2002, prior to the full unfolding o f the crisis, had the highest GDP per capita, appears relatively low than would have been assumed for its GDP per capita (see Figure 7). This outcome i s likely to reflect not only to the new strict eligibility requirements in the contributory system since 1993, but also to the relatively limited deployment o f non- contributory benefits. Further, Argentina seems to achieve lower coverage outcomes, again measured by the total share o f beneficiaries among the elderly, than do its neighbors, when compared to public spending (see Figure 8). Total levels o f coverage are highly correlated with social security expenditure. Both Chile and Brazil achieve higher levels o f coverage o f the elderly for a lower per capita expenditure on social security. While pension spending i s highly correlated with the share o f the elderly in the population, and per capita spending in Chile and Brazil is lower partly because their populations are younger, coverage in Argentina is still significantly lower even afier controlling for these demographic differences (Grushka, 2006). Although this 23 benchmarking uses data from prior to the crisis, and GDP per capita fell considerably and only recovered after several years, low and declining coverage represents a long-term trend, as shown inthe introduction. Figure7. Per Capita Incomeandthe Share of the ElderlyReceivingBenefits, SelectedLatinAmericanCountries, 2000 -2002 100 I I 90 - ur 80 - br 70 - bo 60 - 50 - 40 - 30 - 20 - 10 - O S 1 Gross Domestic Product per capita (constant 1995 US$) Figure8. Public Spendingon Social Securityandthe Share of the Elderly ReceivingBenefits, SelectedLatinAmerican Countries,2000 - 2002 s + n 100 f 90 ur ' 80 v) 70 v A Q) 60 %! 50 0 0 .E" 40 30 JJ 20 30 10 v0 ) o 0 200 400 600 800 1000 1200 Public Social Security Expenditure per capita (constant 1997 US$) ource: Grushka(2006), basedon datafrom CEPAL (2005)andRofman(2005). Nore: The spendingdata for Boliviain the figures abovemay omit spending on non-contributorybenefitswhile coveragedatareflect these benefits. 24 II.b.A Profile of Argentina's Elderly 55. Given the importance o f social security in preventing poverty among the elderly, the country-level analysis o f coverage in Argentina and how this compares with other countries in the region, the remainder o f Part I1i s dedicated to an examination o f social security and other instruments for securing old-age welfare at the household level, and exploits existing sources o f micro-data (such as the national EPH and ECV surveys) as well as the new survey o f the elderly (ETEEP) commissioned by the World Bank for this report (see Box 3). As mentioned inthe introductory section, it is important to note that none o f these sources o f data can be used to paint a complete picture o f the role social security and private pensions play in securing household welfare. The analysis suffers from an urban bias that pervades much of household-level empirical work in Argentina, and should be interpreted with this bias in mind. Two additional caveats about the ETEEP have to be made up front. The first has to do with the timing o f the survey. Interviews were conducted in late 2003 as Argentina was just beginning to recover from the crisis-the magnitude of which is likely to have affected the behavior and outlook o f respondents significantly. The second i s the limitedextent to which lessons can be drawn from behavior reported by the elderly today to guide policy for the elderly in the future. Social security institutions and Argentina's labor market have undergone profound structural changes in the last two decades that have to be kept in mind when considering the implications o fthe results.16 56. The survey o f the elderly was designed to provide answers to the following set of questions. If the formal pension system is covering fewer of Argentina's elderly every year, how are the old financing consumption? What are the insurance and coping strategies o f the elderly that the system does not reach? Are the elderly that go without cover worse o f f or just as well-off as those covered? Is lack o f coverage mostly determined by exclusion or individual and household choice? What are the institutional, labor market and household factors that determine whether an individual o f retirement age receives a formal pension? And finally, what conclusions can be drawn from the past and present behavior of the elderly that would need to be considered by Government as it undertakes reforms? 57. Before turning to formal empirical analysis, simple descriptive statistics provide an interesting snap-shot o f the elderly in Argentina (readers are reminded o f the limited coverage o f the ETEEP sample which largely explains statistics that may diverge with those drawn from the EPH or ECV, see the Technical Annex o f this report). Table 3 presents differences between covered households and those excluded from coverage under the system. 58. O fthe ETEEP sample aged 65 and older, roughly 29 percent live without any type o f social security benefit. Those without coverage are relatively younger, averaging 71 years o f age compared to a mean age of 75 of those receiving benefits. The uncovered l6Having said this, a sizeable share o f the sample worked, contributed and evaded under Argentina's current pension institutions. About 50 percent of the ETEEP sample was below the retirement age in 1994 when the SIJyP was introduced. In2003 when the survey was taken, 10 percent of respondents were below the retirement age. 25 elderly are also predominantly women (67 percent o f those without coverage); only 70 percent have withdrawn from the labor market (compared to 90 percent inactivity among the covered); 22 percent keep working (compared to just 8 percent o f benefit recipients who work); and 8.3 percent seek but fail to find employment. 59. On average, those without benefits have one year less o f schooling (5.5 years as opposed to 6.6 years of education among those covered); live in larger households (2.9 members as compared to 2.5 members in covered households); and have lower incomes ($310 compared to $444). Most of the sample owns the home inwhich they live. Home ownership i s highest among the covered elderly (at 79 percent) and among these, highest among the elderly receiving retirement pensions (85 percent). However, home ownership among the non covered (72 percent) i s also significant. Only 62 percent o f recipients o f the mainly targeted non-contributory pensions own their home. Few elderly households report holding long term savings or investments. Savings and investments are reported more frequently (8.5 percent) among recipients o f contributory benefits (and most frequently among recipients of retirement benefits, at 10.4 percent). Among the elderly who are not receiving any kind of benefit, 6.4 percent report holding long term investments. Table 3. Summary Statistics for Covered and Non-Covered Elderly, ETEEP2003 Retirement Disability Survivor- Contribu- Non con- With any ship tory tributory benefit Nobenefit Total 1 2 3 4 = 1+2+3 5 6 = 4+5 7 8 = 6+7 Weighted cases 843 98 543 1,484 140 1,644 665 2,309 Distribution(%) 36.5 4.2 23.5 64.3 6.1 71.2 28.8 100.0 Meanage 75.1 72.1 75.3 75.0 75.3 75.0 71.3 74.0 Proportion female (%) 44.0 34.4 95.6 62.3 69.6 62.2 67.2 64.1 Privatehealthcoverage (%) 12.5 8.5 5.8 9.8 5.1 9.3 8.5 9.1 Employed(%) 9.4 4.0 6.6 8.0 5.9 7.8 22.1 11.9 Inactive(%) 86.8 96.0 92.6 89.5 91.0 89.7 69.6 83.9 Unemployed (%) 3.8 0.0 0.8 2.5 3.1 2.5 8.3 4.2 Meanyears ofschooling 7.9 6.6 5.3 6.9 4.0 6.6 5.5 6.3 Household size 2.41 2.75 2.55 2.48 2.76 2.51 2.93 2.63 Meanhouseholdincome 549 443 348 468 209 444 310 406 Working years 37.1 30.8 11.8 27.5 24.9 27.3 22.8 26.0 Contributingyears 31.0 23.9 3.4 20.4 5.0 19.1 9.9 18.9 Ownhouse (%) 84.9 80.3 73.5 80.4 61.7 18.8 72.7 77.0 Sewerageinthe block (%) 66.2 51.4 60.8 63.6 37.5 61.4 49.4 58.1 Long-terminvestments (%) 10.4 7.3 5.5 8.5 1.5 8.0 6.4 7.5 Source: Grushka(2006). 60. As already discussed in the introduction and examined in greater detail in later sections, a history o f labor market activity and contribution to the social security system i s a critical determinant of coverage. The elderly who do not receive benefits averaged fewer years o f work (23 vs. 27). Those without benefits have 10 years of contribution on average, while those covered by the system averaged 19 years o f contributions, which reflect both shorter contribution requirements in the past and less stringent enforcement o f these requirements. 61. The share o f the ETEEP sample that receives benefits rises with age among both men and women, partlyreflecting the age and contribution requirements for contributory 26 benefits, but also the receipt o f partial benefits (the PEA), and non-contributory benefits (PNC) which become available at latter ages. From the age o f 85 (25 years after the current retirement age o f women, 20 years after the retirement age for men, and 15 years after the eligibility age for non-contributory benefits) coverage among ETEEP respondents i s consistently above 90 percent among men, but i s lower among women (although the sharp drop inbenefit receipt among very elderly women may be an artifact o f the ETEEPs sample). Figure9. Share of the ElderlyReceivingBenefits amongETEEPRespondents . 0 2r,0.5 - . 8 0 0 0 z .. 0.4 - W c 0 0 0 0 5 0.3 - O . 0. f 0.2 - 0 0 o female male 0.1- n 55 60 65 70 75 80 85 90 95 100 Age Source: Staffestimatesusing ETEEP 2003. 27 Box 3. Surveys of the ElderlyinArgentina andOther MiddleIncome Countries Most o f the analysis in this report relies on data collected inOctober-November 2003 inthe ETEEP, Encuesta a la TerceraEdad sobre Estrategias Previsionales, a special survey o f the elderly. The survey targeted households with a member aged 60 and over inurbanareas of Argentina, excluding the Patagonia Region. I t interviewed one individual aged 60 or over in 3014 households selected under a stratified probabilistic sampling. The survey instrument contained questions on individual and household characteristics, household income, assets, expenditure, and livelihood strategies; the labor and pension history o f the respondent, and summary information on the economic status and characteristics o f other household members. Micro-data focusing on the elderly is very rare. Two other recent surveys in Brazil and South Africa collected under the Non-Contributory Pensions And Poverty Prevention research project (www.idum,man,ac.uWncuus) provide usehl cross-country comparisons. There are important differences inthe sampled population that needto be taken into account. The ETEEP (2003) Survey targeted urban areas in Argentina, whereas the Brazil and South Africa datasets (2002) included both urban and rural areas. While the ETEEP has wide coverage, the latter are also restricted to specific regions (Western and Eastern Cape in South Africa, Rio and Ilheus in Brazil), and inthe caseofSouthAfrica the survey excludes whites. Table 3.1 provides some information on work and pension receipt among people aged 60 and over inurban areas across the three samples. Retirement ages vary across pension plan and country. In Argentina the target retirement age for contributory plans is 60 for women and 65 for men (same as in the non-contributory plan in South Africa), but early retirement is available inthe contributory program (same as in the contributory plans in Brazil) and no specific retirement age in the non-contributory plan, except in the old age non-contributory benefit. InBrazil, non-contributory plans inurban areas have an age o f entitlement at 67, and inthe rural plan at 55 for women and 65 for men. The Table suggests that Argentina has the lowest incidence o f pension coverage among the elderly, and Brazil the highest (note, a different base is used inTable 2). By contrast, activity rates are highest inArgentina. Table 3.1. Work and pensionreceipt amongpeople aged 60 and over in urban areas: Argentina (2004), Brazil (2002) and SouthAfrica (2002) Argentina Brazil South Africa n=3014 n=98 1 ~ 6 5 9 Work andpensions %with pension 61.0 89.0 63.6 % without pension 39.0 11.0 36.4 % active 23.5 15.1 9.9 % inactive 76.5 84.9 90.1 % active with pension 6.9 10.9 2.1 % inactive without pension 22.4 6.8 28.7 Source: Bamentos (2006); samples are: for Argentina is all urban areas except Patagonia region, for Brazil is Metropolitan Rio and Ilheus, and for South Africa i s Western and EasternCaperegion. The incidence o f pension recipients active in the labor market i s highest in Brazil, where early retirement followed by continued employment is common among those with contributory plan coverage, and lowest in South Africa where high unemployment rates for the population as a whole, let alone older workers, significantly restrict employment opportunities at later ages. I t is likely that households with older members who are inactive are vulnerable. The incidence o f older people inactive and without a pension is highest in South Africa and lowest inBrazil. Table 3.2 contains some information on the livelihood protection instruments available to households with older people. The data for Brazil and South Africa are directly comparable in that the responses are to the same question. In the case o f Argentina, the question is different, and the possible responses more restrictive. However, some information can be extracted from the comparison. Family and community networks are an important source o f livelihood protection in all three countries. The banking and financial systems appear to be less important, certainly much less important than consumer credit. The impact o f financial difficulties on food consumption is clearly observable for Brazil and South Africa, and reflects a much greater exposure o f households with older people in South Africa. (continued on nextpage) 28 (continuedfrom previouspage) Table 3.2. Livelihoodprotectionamong people aged 60 and over in urban areas: Argentina (2003), Brazil (2002) and South Africa (2002) Argentina Brazil South Africa n=3014 n=98 1 s 6 5 9 I n the last three Whenyour Whenyour household months hasyour household is in is infinancial household had financial diflculty dificulty do you ...? to ...? do you...? %selecting the %selecting the %selecting the category category category Ask fiends andrelatives for help 12.7 42.8 49.8 Ask employer for help n.a. 1.3 7.7 Ask NGOKhurchfor help n.a. 1.4 4.7 Borrow money from the bank 3.5 9.9 9.1 Cut down on food n.a. 11.2 38.5 Findextra work n.a. 5.0 25.5 Buyfoodgoods on credit 13.5 4.7 14.3 n.a. is not available Barrientos (2006); samples for Argentina i s all urban areas except Patagonia region, for Brazil is Metropolitan Rio and Ilheus, and for South Africa i s Western and Eastern Cape region. 62. The reported retirement ages o f elderly men and women receiving contributory pensions peak at 55, 60 and 65. This may reflect the differences in regime under which participating respondents inthe ETEEP sample worked, gained rights and retired, and the increase in the national retirement age in 1993. These peaks also show the relevance o f the SIJyP reform and the institutions it set in place to the sample. However, there i s considerable dispersion around these peaks. Lower reported retirement ages can reflect the relatively more generous parameters o f pension plans for sub-national civil servants (a matter taken up in detail in Part IV o f this report), disability benefits, survivors' benefits, special regimes, and potential fraud. 63. Respondents' causes for becoming a beneficiary are presented in Figure 11. There are important, if expected, differences in the responses between elderly men and elderly women. Most covered men report retiring because they had fulfilled age requirements. The most frequent response given by covered women is the death o f a spouse, thus qualifyrng for a survivor benefit. A significant portion o f elderly men and women reported chronic illness as the main reason for retiring. Less than halfa percent o f covered men and only 1percent o f covered women reported claiming benefits because o f a lack o f employment. A higher percentage o f covered elderly reported early retirement. 29 Figure IO. Age at by co lcflts Awa orted and en, 2 I 4 - 65 12 60 20 8 6 55 - 3s do 44 48 52 55 60 $4 aga af retrewnt rn male when asked the most i ~ ~ preason u.hy~they t o ~ ~ ~ as never having co social security, 67. The ETEEP data allow us to examine whether there are significant differences in the determinants o f receiving contributory and non-contributory benefits. Because Argentina's regularly deployed surveys do not distinguish between benefit-type, this refinement o f the analysis was previously not possible. The analysis shows that while the pattern o f significance is similar, the relative importance and magnitude o f the determinants o f coverage changes depending on type o f benefit, contributory or non- contributory. Men are 60 percent more likely than women to receive contributory pensions which reflects their greater past participation in the labor market and probably in "formal" jobs. Those aged over 70 are three times more likely than younger (65-69) cohorts to receive contributory benefits. This may reflect the more stringently enforced eligibility requirements since 1993 lowering the relative rates o f contributory pensions among younger cohorts o f the elderly. Each additional person inthe household decreases the odds o f having a contributory pension by 10 percent. Although this result could be taken as evidence o f a preference for a larger family as an income security strategy, it could also reflect the larger average size o f poorer households. 68. Completing primary school increases the likelihood o f having contributory coverage. Each additional year o f work increases the likelihood o f contributory cover by 6 percent. Work in larger firms increases the likelihood o f contributory cover by four times. And as inthe previous results, having worked inthe construction sector lowers the likelihood o f contributory cover. Home ownership and access to sewerage increase the likelihood o f receiving a contributory pension. The likelihood a respondent receives a non-contributory pension i s greater for older respondents, those living alone, and widows. These results are largely expected, given the population targeted by Argentina's non- contributory pension programs, which has specific benefits for surviving spouses and for people older than 70. 69. The findings thus far privilege the argument that the lack o f social security coverage among today's elderly i s a manifestation o f exclusion during working life. Obstacles and difficulties in making contributions to a pension system when working, and problems in design and implementation o f Argentina's non-contributory pension system, are the primary explanations for low and declining household coverage. These arguments are to a large extent supported by the reasons respondents to the ETEEP reported for failing to contribute to social security, presented in Figure 13. While the answers to this sort o f question (that elicits respondents to take responsibility or to apportion blame) have to be interpreted with caution, the largest segment o f responses point to the employer as the "gate-keeper" to coverage. 32 71. Access to ~ ~ n s ne^^^ i s cfoscfy related to the history o f ~ i ~ n # ~ ~to ~ ~ ~ ~ ~ social security system and linkedto ~ n d ~ ~labor history,~ Several prior studies show i d ~ ~ s ~ 33 all, where individuals encounter social insurance systems and join them-in a decision made sometimes on their own, others in conjunction, and often under the compulsion o f their employer. In the interest of fiscal sustainability o f social assistance programs, policy makers prefer individuals to at least reach the minimum threshold o f contribution required for contributory coverage (in Argentina, 30 years at age 60 for women and 65 for men to obtain the PBU, and 10 years at age 70 to get the PEA), which makes individuals' history o f contribution and evasion behavior o f particular interest (see Bertranou, 2006). l8 72. Inthis section, labor histories and contribution behavior collected inthe ETEEP are examined. Given that pension coverage requires cumulative years o f contribution, contribution density-Le., the share o f anindividual's working life duringwhich he made contributions to the social security system-best represents the probability o f being covered by pension benefits. However, since the number o f jobs for which respondents to the ETEEP failed to report their contribution behavior i s very high, attention needs to be given to reporting behavior and, accordingly, the analysis needs to control for the determinants o f contribution density. The distinction between non-reporters and non- contributors (who, by definition, reported) is important to the extent that non-reporting could proxy for purposeful intent and reflect the exercise o f personal preference to evade the pension system during a particular employment spell. 73. Reporting and contribution behavior are considerably different depending on the sex o f the respondent. Men failed to report in 27 percent o fjob episodes. Further, when they did report, men said they did not contribute in 13 percent o f their jobs (employment episodes), but made contributions in 60 percent o f employment spells. In 51 percent o f jobs, men contributed for as many years as they worked (i-e., reported the maximum contribution density), and in 9 percent o f the jobs they did not contribute for as many years as they worked, but for a shorter period (i-e., contribution density was positive but less than one). Inthe case o f women, the reported contribution scenario looks worse. In 31 percent o f their jobs women did not report, and for the jobs in which they reported, women failed to contribute in 27 percent. However they contributed in the remaining 42 percent o fjobs. Women had the maximum contribution history in only 36 percent o f their employment spells. l8Despite recent additions, studies on individual contribution behavior are still scarce, mainly because specific data collection is required. Arenas de Mesa, Bravo and Mitchell (2006) study the determinants o f contributions to the pension system in Chile, based on a specific household survey o f labor and social security issues. They find that contribution history from self-reported data is considerably lower than is typically assumed. Quintanilla (2004) uses the same survey to study the transitions between contributing- jobs, non-contributing jobs, unemployment and inactivity. Buchelli (2004) studies contributions to the social security system in Uruguay and finds an increasing correlation between contribution behavior and education, particularly at high levels o f education, and attributes that link to institutional factors. In a regional study for Latin America, Auerbach, Genoni and Pages (2005) analyze contribution patterns for different labor categories and find that, to a large extent, low rates o f contribution are determinednot only by the low willingness to participate in pension programs but also by the inability o f governments to enforce the mandate to contribute. These studies and how the results compare with similar estimates from Argentina are reviewed inBertranou (2006). l9The longitudinal analysis takes each employment episode reported by respondents as a separate observation. This not only increases the sample size, but allows us to control for a host o f non-observable characteristics by conducting the estimationusing fixed-effects techniques on a panel o fjobs. 34 35 service than non-reporters. Thus, while non-reporters and non-contributors are both mainly classified as employees in small size firm or self-employed workers, they are quite different in terms o f other characteristics, mainly the features that distinguish the jobs typically held by women. 77. Contribution density varies with individual and household characteristics (see Table 3). Men, married individuals and heads o f households show higher densities. Contribution density increases together with education levels. Individuals who own their dwellings tend to have higher contribution densities, as well as those who made investments to secure old age income. When the regional dimension i s considered, contribution densities vary little between 0.61 and 0.64-except inGreater Buenos Aires, where contribution density rises to 0.74. Contribution behavior differs significantly with the size o f firm (Table 4). Jobs in small firms, with up to 5 employees, have the lowest contribution density (0.40). Jobs inmedium size firms show a much higher density value and a substantial share o f jobs with maximum contribution density. Finally, large firms concentrate more than 38 percent o f the jobs and have a very high contribution density (0.90). Table4. Characterizationof ContributionDensity,by IndividualandHousehold Characteristics of ReportedEmploymentEpisodes Variable Employment Contributiondensity episodes Mean Std. Dev. Variable Employment Contributiondensity episodes Mean Std. Dev. Sex Household Female 1771 0.56 0.48 Headofhh 3078 0.72 0.43 Male 2216 0.77 0.39 Nonheadofhh 904 0.55 0.48 Civil Status Dwelling Single 319 0.69 0.45 Yes 2980 0.70 0.44 Manied 2027 0.71 0.43 No 905 0.61 0.46 Widow 1291 0.63 0.46 Literacy Investment Literate 3806 0.69 0.44 Yes 352 0.75 0.41 Non-literate 181 0.37 0.47 No 3635 0.67 0.45 EducationLevel Region IncompletePrimary 1507 0.58 0.47 cuyo 182 0.63 0.47 Complete Primary 1276 0.67 0.45 Amba 1766 0.74 0.42 Incomplete Secondary 400 0.75 0.41 NEA 365 0.61 0.47 Complete Secondary 448 0.81 0.37 NOA 430 0.63 0.46 IncompleteHigherEducation 107 0.88 0.30 Pampeana 1244 0.64 0.45 Complete HigherEducation 239 0.88 0.31 Source:Bertranou(2006). 78. The contribution densities o f women are not significantly different across sectors, except inthe following two cases: finance, that has a negative effect with respect to commerce, and education and health services, that has a positive effect possibly because many of the jobs inthis sector are performed inthe public sector, where evasion i s relatively more difficult. In contrast, there are significant differences from sector to sector in the contribution behavior o f men. Men working in primary activities have slightly lower contribution densities, while jobs in utilities, finance, public administration, and education and health services have significantly higher contribution densities than those in commerce. The duration of a particular employment episode has a positive effect on contribution density, especially for men. Labor category appears to be 36 important for men but not for women. Self-employed male workers have considerably lower contribution densities than salaried men. Finally, both in the case o f men and women, there are strong positive effects on contribution density with an increase infirm- size. Table 5. Characterizationof ContributionDensity,by Sizeof the Firmof Reported EmploymentEpisodes Size of firm Employmentepisodes Contributiondensity # YO Mean Std.Dev. p25 p50 p75 1to 5 1,266 34.4 0.40 0.46 0 0 1 6 to 39 1,002 27.2 0.77 0.40 0.69 1 1 40 or more 1,417 38.5 0.90 0.28 1 1 1 Total 3,685 100.0 0.70 0.44 0 1 1 Source: Bertranou (2006). ILe. I s Continued Labor in OldAge a Viable Income Substitutefor Social Security? 79. The focus i s now shifted to a closer examination o f the current labor market participation o f older people and the extent to which employment constitutes an effective livelihood strategy as a substitute and/or a complement to formal pensions. Declining coverage rates will undoubtedly reduce the social protection afforded by pensions. An important policy question i s whether continued participation in the labor market can compensate for declining pension coverage. In OECD countries, withdrawal from the labor market is largely explained by the presence and design o f employment-based pension systems and public assistance programs (Gruber and Wise, 1999). In low- income countries, the absence o f pension systems makes continued work a necessity for older people and health factors become dominant in the decision to continue working (Benjamin, Brandt et al, 2003). Inmiddle-income countries, like Argentina, it i s likely that a mix o f factors can explain labor supply at later ages: the influence o f pension systems on labor market withdrawal i s stronger for workers in formal employment, but considerably attenuated for the rest (Clark, York et al. 1999). 80. Labor market participation o f the elderly in Argentina compares reasonably well with other countries in Latin America at a similar stage o f economic development and coverage o f pension provision. Across Latin America, earlier withdrawal from the labor market is related to economic development-a pattern that is consistent among the elderly in OECD countries. Aggregate data on participation fail to show any direct relationship between participation and the economic need o f older people. Figure 16 compares the participation o f poor and non-poor older people. Only in Uruguay, El Salvador and Brazil are participation rates higher among poor than among the non-poor elderly. For the majority o f countries in the region, the labor market participation rate o f the non-poor elderly exceeds that o f the poor. This "gap" in participation is not directly related to the share o f older people receiving pensions ineach country. Argentina has the largest gap in participation between the poor and the non-poor. Thus, while labor force participation o f the elderly, at around 23 percent, i s comparable with other countries at a similar stage o f economic development and pension provision in the region, it is especially low among the poor. 37 pation o nd Y (Aged 60 an urban a e l I 5tr h&? u dY 38 83. Turning to occupational characteristics, a significantly higher share o f active respondents i s employed inconstruction, transport, utilities and services, compared to the last job o f inactive groups.2o The share o f active respondents reporting self-employment in their current job is twice as high as among the last reported occupations o f the currently inactive elderly. This may indicate that self-employment constitutes a strategy for older people to continue in employment later in life because the opportunities for dependent employment are restricted for older workers.21 84. Active respondents appear to have, on average, reasonably long spells in their current employment compared with the length o f the last job for those who have withdrawn from the labor market and are receiving pensions. They also have, on average, longer engagement inthe labor market and morejobs intheir lifetime than those older people who are currently inactive: nearly one third more jobs than inactive non- pensioners. Their contribution density is significantly lower than inactive pensioner respondents, although higher than among the inactive non-pensioner respondents. Taken together, these variables suggest that active older people have worked longer but have had a greater share o f informal work duringtheir working lives. 85. The elderly who work also appear to be healthier. Active respondents are significantly less likely to report poor or very poor health, and are more likely to report good or very good health. By comparison, inactive non-pensioner respondents are the most likely to report poor health and the least likely to report good health. 86. Moving on from these simple comparisons, Figure 16.a shows the percentages o f respondents within each decile of estimated potential earnings that are active. Figure 16.b shows the percentages o f all active respondents to be found in each o f the deciles o f estimated potential earnings. Current participants in the labor market are concentrated in the higher deciles o f predicted earnings. Few o f those with predicted earnings inthe low deciles are active. The vast majority o f respondents predicted to have earnings in the top decile are active (91.7 percent) and this also applies to the ninthdecile (67.4 percent). In Figure 16.b, over three quarters o f active respondents can be found in the top three deciles when the full sample i s considered. Ifonly men are considered, over one half o f active males can be found inthe top three deciles o f predicted earnings. 87. This analysis shows that the elderly who are currently active in the labor market are also those with higherpotential productivity, i.e., those with higher potential earnings. This suggests that in Argentina's labor market less productive older workers are filtered out through voluntary or involuntary withdrawals and/or pension incentives. While the urbanbias of the data is an important driver o f these results and has to be kept inmind, the main conclusion that can be made from these finding i s that only elderly men and women with higher predicted earnings are likely to be in employment in later life, 2o Where the comparison between the two groups i s on labor-market variables, the last reportedjob of the respondents who are not currently working is taken. 2'This could result either becausefirms expel workers at the retirement age (`push'), or becausedependent employment pay and conditions are unattractive for older workers (`pull'). These generate compositional effects, where dependent employees retire leaving the self-employed a higher proportion of those active at later ages. 39 suggesting that continued employment i s unlikely to be an effective alternative livelihood strategy for older people inlow income andvulnerable households. Figure 16a. Percentageof Active Respondents ineachDecile of PredictedEarnings Figure 16.b. Percentageof all Active RespondentsinDeciles of PredictedEarnings 40.0 4eI20.0 ii 3 E 0.0 0.0 0.1 0.0 0.8 2.4 4.4 9.8 15.4 27.7 37.6 0.5 II 1.9 1j 3.0 II 5.6 II 6.5 1 9.1 11 13.1 II 14.3 II 19.9 II 23.6 Daclh of predictedearning8 :ource:Bamentos (2006). 88. Labor market participation does not appear to compensate for the decline in pension coverage in Argentina. While participation in the labor market can be an important livelihood protection strategy for older people with highpotential earnings, it is unlikely to be an effective strategy for older people with low potential productivity. The opportunities and benefits from "active" ageing appear to be skewed towards the healthier and wealthier among the old, at least as far as remunerated work i s concerned. 40 89. However, these results seem to also shed more light on the discussion o f exclusion and household preferences in the earlier section, and require that the earlier findings be further nuanced. Those who continue to work inold age and are not covered by the pension system, are not only those with the best prospects inthe labor market: they are also those who have the longest history o f work; the highest number o f jobs; the largest number o f other household members who work; and lower contribution densities, although not as low as the inactive and uncovered elderly. 90. These results identify a group within the uncovered elderly whose current status outside the social security system has little to do with vulnerability or historically poor labor prospects, and which may have chosen to avoid forms o f employment where the mandate to participate was difficult to evade. Having said this, the analysis also identifies a group o f uncovered elderly whose employment prospects have always been poor, with less involvement in the labor market in the past, and greater reported health problems. For this second group, the exclusion explanation for low coverage may hold, andthe current lack o f coverage is an immediatepolicy concern. 41 Part 111.Closingthe CoverageGap 91. The summary inPart I1o f the analytical work usinghousehold data clearly shows contribution history and labor market factors that influence contribution behavior are important determinants o f coverage outcomes. The analysis identifies groups that may be choosing to evade the system - on their own or in collaboration with their employers. And there are also those whose lack o f contributions raises concerns about exclusion, and that this exclusion may take the form o f employers acting as gate keepers. Also apparent i s that the labor market only offers a viable alternative to the pension system for those who have always had and continue to enjoy good labor prospects. Revealingly, those who continue to work in old age are individuals who report the longest employment histories. Furthermore, the working elderly in the ETEEP are not only those whose predicted labor incomes are relatively high, but also those who have contributed to the pension system for fewer years throughout their working lives (than those who are inactive and receivingbenefits). 92. The analysis points to a policy priority among the three objectives o f social security discussed inPart Io f this report: preventing poverty in old age. While the other two objectives-consumption smoothing and income adequacy-are also important and deserve the attention o f policy makers, among the prospective losses from ageing, poverty inold age is a risk that households are least able to cope with on their own. IILa TheDistinctions between "Pillar 0" and "Pillar 197 93. Since the publication o f Averting the Old Age Crisis (World Bank, 1994) where the basic architecture o f the multi-pillar system o f public and private pension provision was described, a distinction has emerged between two types o f pensions still directly administered by government: "fzrstpillar" pensions paid to individuals with a history o f contributions to an earnings-related retirement security regime, and "zero pillar" pensions offered universally or targeted to the elderly poor regardless o f whether they have contributed to an earnings-related regime or not (Holzmann and Him, 2005). However, according to the analytical framework presented in Part I,when the policy objective i s to provide an absolute minimum income to prevent poverty among the old, the distinction between afirst pillar that conditions benefits on contributions, and a zero pillar that pays pensions regardless o f contribution history can be confusing. Neither i s structured as actuarial insurance, yet both pillars pool the risk o f poverty inold age; both can be structured to pool the risk o f other contingencies that markets are unable to manage or manage poorly, such as life-time low earnings and inflation; both pillars are directly administered by government; both pillars are typically financed on a PAYGO basis; and both pillars receive "contributions" either from payroll taxes or from other levies with a broader base such as taxes on income or on consumption. 94. Once the rationale for prescribing public PAYGO risk pooling to cover the relatively rare loss o f poverty in old age i s accepted, and the level o f pooled benefits are set with care not to upset household incentives to save in earnings-related plans, the distinction between a zero pillar from the first becomes less important. In fact, 42 perpetuating a distinction between pillar zero and pillar could potentially lead to undesired outcomes, and even increase the vulnerability o f the poorest insociety. 95. How? First, public PAYGO systems that deny minimum pensions to individuals without a history o f explicit contributions, but that pay benefits that are nevertheless guaranteed by government transfers, can redistribute income from all current and future tax payers to those who have accumulated pension rights. The benefits o f Argentina's "contributory" public pensions are financed mostly from general revenue transfers. But even where the contribution and benefit parameters o f pillar one are set to be "self financing", who pays for short-falls between benefits and contributions during economic downturns? All current and future tax payers "contribute" to maintain the benefits paid to a relatively smaller group o f "covered" workers. Second, separate, seemingly non- contributory transfer arrangement to the elderly poor are often perceived as charity rather thanjust another instrument with which households canmanagerisks to income, and are often only reluctantly considered in budget allocations. Small budget allocations to "social assistance" pensions separated from the first pillar are typically supported by small, relatively weak political constituencies, and have been historically vulnerable to budget cuts. 96. Does this implythat once zero-pillar "non-contributory" benefits are inplace, that first-pillar, contributory instruments are redundant? Not necessarily. Where a financially sustainable zero pillar exists, first pillar benefits, such as marginal-accrual minimum defined benefits, or a matching contribution scheme, can prove highly beneficial, but more as a means to increase incentives to participate in the earnings-related pillar, particularly among lower-income affiliates. Thus both pillar 0 and pillar 1 have an important role in closing the coverage gap when structured correctly: the former can cover the risk o f poverty in old age, and the latter can attract the participation and reward the savings effort o f lower income groups. However, even innovative incentive measures, such as Mexico's Cuota Social matching contribution scheme, are as yet untested for their incentive effects (see Box 4). Should gains in participation fail to materialize, the dangers o f regressive transfers and the political marginalization of poverty programs discussed above, remain. III.b.AdvantagesandDrawbacksof AlternativePublicPoolingStructures 97. There are at least three basic alternatives for structuring the public pooled component of a retirement security system. These are (i) a minimumpension guarantee, benefit top-up, or matching contribution scheme to workers who participate in a retirement security regime; (ii) a benefit targeted to the elderly poor; and (iii) universal a flat pension, sometimes called a "demogrant"-paid to all men and women over some threshold age, regardless of their means. Inseveral countries these structures overlap. 98. From the conceptual discussion in the pervious section, in a country where all individuals contributed to the earnings related pension system, a contributory minimum guarantee structured as a "top up" i s a satisfactory public pooling arrangement on its own: it guarantees a minimum level o f retirement income at a minimum cost to tax payers and encourages workers to save inthe earnings-related plan. However, as already pointed out, in countries where most workers will not have a sufficiently long history o f 43 contributions to the earnings related pension system, a top up conditioned on participation can not only exclude large segments o f the population, but also lead to undesirable transfers. Box 4. The CuotuSocial inMexico Commercial Second Pillar The Government o f Mexico introduced a flat contribution subsidy along with private individual accounts in 1997. The government makes a daily payment to the individual retirement accounts o f all workers affiliated with the new private defined contribution pillar whose contributions are up to date. This contribution, called the CuotaSocial, is commission-free and does not vary with workers' income. Cuota Social is calculated as a percentage o f the minimum wage. When the new AFORE pillar came into effect inJuly 1997,the CuotaSocial was set at 5.5 percent o f the monthly minimumwage and now equals 6.05 percent, reflecting adjustments for inflation. Because all contributing affiliates to the reformed pension system in Mexico receive the same amount o f Cuota Social contributions in their individual retirement accounts, its share o f total pension contributions is inversely proportional to workers' incomes. That is, the Cuota Social bolsters the value o f low-income workers' pension accounts more than that o f high income workers. In fact, Cuota Social contributions represent more than one-quarter o f the value o f retirement contributions for the 68.5 percent o f the Mexican workforce earning three minimum wages or less; it accounts for almost 55 percent o f the retirement contribution for a worker earning one minimumwage. The CuotaSocial introduces a minimumincome mechanism directly into the defined contribution pillar created by the Mexican pension reform. For workers earning up to three minimum wages, the Cuota Social is greater than commissions charged to their individual contributions taken from wages, allowing their fbture pension benefits to be larger than their individual contributions. This serves as an incentive for low-income workers to join or remain in the formal sector by contributing to the defined contribution system. I t is still too early to adequately estimate the effects o f the Cuota Social on workers incentives to join the AFORE system and to keep their contributions up to date. However, the contribution subsidy is not the only minimum income instrument in the Mexico's multi- pillar system. AFORE affiliates with at least 25 years o f contributions who have not accumulated savings sufficient to finance a determined minimum annuity, qualify for a minimumpension guarantee top-up similar to that inChile's AFP system. This contributory instrument is more targeted to the likely elderly poor than the Cuota Social, although a minimum history o f contributions is still required and workers with the lowest life-time earnings in the Mexican economy are largely failing to participate in the pension system. The fiscal burden o f the CuotaSocial to the Mexicangovernment will be substantial inthe first decade o f reform. At 0.33 percent o f GDP at the outset o f reform, it is the largest cost itemamong government contributions to the reformed pension system. However, as GDP and real wages grow, the relative burdeno fthe CuotaSocial is likely to decline, evenallowing for increases inpensioncoverage. Source: Gill, et a1(2004). 99. Targeting public pensions to the elderly poor is probably the public pooling arrangement that is closest to the risk pooling ideal to cover poverty inold age for middle income countries where many workers will fail to contribute regularly to an earnings- related pension pillar. This is especially true if the targeted benefit is financed with a broad based tax, such as VAT. However, means testing to target benefits efficiently also comes with complications and costs. Further, just as over-generous social assistance benefits can lead to moral hazard, so means tests can discourage private saving and wealth accumulation for retirement as well as continued work in old age. Finally, as mentioned inthe previous section, means tested benefits are often regarded as charity and makes the benefits vulnerable to budget cuts, especially ineconomic downturns. 44 100. A universal flat pension does not strictly comply with the conceptual ideal o f a public risk pooling mechanism to insure against poverty in old age, since all individuals above a specified age would receive some benefit, not only those suffering the "bad state" o f poverty. However, they are the simplest public poverty-prevention mechanism to administer, with the lowest transactions costs. These benefit structures are found in developed countries, such as the United Kingdom and New Zeeland, as well as developing countries, like Mauritius. Correctly structured, universal flat pensions prevent governments from creating the disincentives to work and save that are often inherent in means testing. The main concern with universal minimum benefits is for fiscal sustainability. While a large portion o f the transfer to the elderly can be regained by taxing universal pensions as any other form o f income (as inNew Zealand's universal flat pension) the financial sustainability o f the benefit will then depend on a government's administrative capacity to claw back benefits from the non-poor elderly through taxation channels. The universal old age benefit inNew Zealand cost 4 percent o f GDP in 2001 and i s forecast to grow to 9 percent o f GDP in 2050. Gill et al., (2004) argued that universal minimum pensions are fiscally unviable in most Latin American countries, particularly in light o f competing demands for social protection for other vulnerable groups. IIXc. Fiscal Costs of Direct Measuresto CloseArgentina's CoverageGap 101. Several proposals have emerged from stake-holders inthe current pension reform discussion to close the coverage gap: (i) the creation o f a universal old-age pension benefit available to all the elderly regardless o f their history o f participation; (ii)the availability o f this non-contributory benefits earlier (at age 65) than the current age requirement for social assistance pensions (age 70); and (iii) the "graduation" or creation o f a "proportional" minimum contributory benefit that would recognize a shorter history o f contributions at age 65 than the current 30 years required in the SIJyP (see MTEySS, 2003, UAFJP, 2003). This section reports on simulation results from A16s and Muiiioz (2006) which use incomes, labor and contribution histories inthe ETEEP to estimate the fiscal costs o f a new non-contributory pension and a more proportional contributory minimum benefit as lower-bound complementary benefits to shore up the current and expected future coverage o f the SIJyP. 102. The objective o f this exercise i s to quantify current proposals and provide a "base line" estimate o f the fiscal costs o f these direct policies to extend coverage that i s informed by the actual reported labor and contribution histories o f today's elderly. For this reason and with some important exceptions (detailed in Box 5), the simulation assumes no other changes are made to the pension system other than the elimination o f the PEA and PNC which would be replaced with new benefits available at earlier ages. Since the earnings-related contributory benefits o f the SIJyP are unchanged in these simulations, this estimate is a conservative scenario intended to stimulate further investigation o f options, and not as an endorsement o f any particular proposal. A wider consideration and costing o f options would need to include not only policies to cover poverty in old age, but also ways to increase participation and bring down the costs o f sustaining full coverage. 45 Box 5. Costing Current Proposalsfor FullCoverage Ina new complementary scheme, based on existing proposals to extend coverage, the current PEA and social assistancepensions would be replaced with a Minimum Guaranteed Benefit (PrestacidnMinima Garantizada,or PMG). The State would guarantee a minimum level o f Total Benefit (Benejcio total or BT) equivalent to the amount o f the PMG for all those people who have 10 years or less o f contributions to the social security system. A Total Minimum Benefit (Prestacidn Total Minima or PTm) would guarantee that everybody who has contributed between 10 and 30 years to the system will collect a PT as a function o f the years o f contributions that will inno case be less than the PTm. For individuals who have contributed for 30 years or more, pensions would be the same as under the current system, according to the rights granted by the reform law in 1993. The age to access old-age benefits is unified at 65 years for both men and women (one o f the important changes made to the parameters o f the SIJyP that is discussed ingreater depth as a policy option inPart IV). For those with less than 30 years o f contributions, a Basic Benefit (Prestacidn Busica - PB) would be paid to complement the amount in pesos o f the AFJP/JO and the RPPPAP, derived from the years o f contribution to the social security system until reaching the defined levels o f PMG or PTm. If the JO and/or PAP turn out to be higher than said levels, the PB is equal to zero. Finally, the complementary set o f benefits maintains the transitional Compensatory Benefit (Prestacidn Compensatoria-PC) that the SIJyP currently pays inrecognition for the years o f contributions prior to the effectiveness to the 1993 reform. The benefit is calculated exactly as indicated inthe reform law, as a defined percentage o f the remuneration o f reference, weighted by the number o f years paid into the previous regime. Current eligibility requirements for this benefit are also unchanged, requiring at least 30 years o f contributions to the system. Thus PT =PBU+ PC + (PAP + JO) + PB Where: PT = Total Benefit PBU=Basic Universal Benefit PC = Compensatory Benefit PAP = Additional Benefit for Permanence JO = Ordinary Retirement Benefit PB =Basic Benefit n=Years ofcontributions The amount o f the PB is defined as a function o f the following criteria: a) Ifn