Report No. 27618-PE Peru Restoring the Multiple Pillars of Old Age Income Security January 26, 2004 Human Development Department Bolivia, Ecuador, Peru and Venezuela Country Management Unit Latin America and the Caribbean Region Document of the World Bank - CurrencyEquivalents - US$ 1.00 :S/- 3.48 - FiscalYear - January 1-December 31 - Glossary of Acronyms & Special Terms - - AFPs Administradoras de Fondos de Pensiones, the private pension fund managers charged with managing workers savings in individualretirement accounts AUM "Assets under management"fee structurecommon in financial management industry Cedula Viva PAYGO second pillar regime for civil servants and workers in public enterprises, created in 1974by DL 20530 DB "Defined benefit" type pension plan where pensions are based on a final- or average- salary formula, typically financed on a PAYGO basis DC "Defined contribution'type pension plan where benefits are based on accumulated savingsand the returns earned from their investment DI "Defined incentives" for voluntary retirement savings largely in the form of preferential tax treatmenton contributions into a pension plan, or matching contributions IPSS Instituto Peruano de Seguridad Social, the Government agency which administered the PAYGO regime created by DL 19990 ONP Ojicinade Nomalizacion Previsional, Government which administers PAYGO regime created by DL 19990(SNP) since 1992pension reform PAYGO Pay-as-you-goretirement security system where pay-roll taxes from current workers finance the pension benefits of current retirees POI "Percentageof income" fee structure common in mandatory private pension systems PROST Pension Reform Options Simulation Toolkit SBS Superintendencia de Bancosy de Seguros, regulatory body for the private pension system SNP Sistema National de Pensiones, the public PAYGO branch of the pension system, combining first and secondpillar benefits SPP Sistema Privado de Pensiones, the private second pillar branch of the pension system, based on individual accounts, created by the 1992pension reform Vice President: David de Ferranti Country Director: Marcelo Giugale Human Development Director: Ana Maria Arriagada Human Development Sector Leader: Daniel Cotlear Lead Economist Human Development: Ariel Fiszbein Social Protection Sector Manager: Christopher Chamberlin Task Team Leader: Truman Packard Acknowledgements This report was prepared as part of the World Bank's program of analytical and advisory assistanceto the Government of Peru. The Bank would like to thank Marisol Giulfo and Maria del Carmen Rivera at the Ministerio de Economia y Finamas, Javier Penny, Jose Quiiiones and the actuarial staff at Oficina de Nurmalizacidn Previsional, Lorena Masias and Eli0 Sanchez at Superintendencia de AFP, Armando Caceres at Superintendencia de Seguros, and Henry Barclay and his team at Banco Central de la Resewa for their insights and suggestions, and for providing much of the data used inthiS report. The report was prepared by a World Bank team led by Truman G. Packard (Senior Economist, Social Protection). The team consisted of John Pollner (Lead Financial Sector Specialist), Montserrat Pallares (Social Protection Specialist), Gillette Hall (Economist, Social Protection) and Ana Karina Rozas (Junior Professional Associate). The Bank's team worked under the overall guidance of Marcel0 Giugale (Country Director for Bolivia, Ecuador, Peru and Venezuela), Ana-Maria Arriagada (Sector Director of the Human Development Department, Latin America and Caribbean Region), Ariel Fiszbein (Lead Economist for the Human Development Department, Latin America and Caribbean Region), Daniel Cotlear (Human Development Sector Leader for Bolivia, Ecuador, Peru and Venezuela), and Christopher Chamberlin (Sector Manager for Social Protection in the Latin America and Caribbean Region). Invaluable assistance and support throughout the preparation of the report were provided b y ChristinaAlquinta, and LuisaMariaYesquen. The report summarizes the findings and conclusions drawn from four background papers and additional analytical work prepared by members of the team and external consultants, including Manuel Lasaga (Columbia University) and Abigail Barr (Department of Economics, University of Oxford). Moises Ventocilla and Pedro Llontop at Instituto Cuanto, also provided valuable inputs and advice. These background papers are cited in the references section and are available upon request. The report has benefited immensely from the comments of senior reviewers inside the World Bank. The reviewers for this task were Herman Von Gersdorff (Lead Economist, Human Development) Pablo Gottret (Senior Economist, Health) and Rafael Rofman (Senior Economist, Social Protection) who provided detailed comments, and to whom the report team owes a debt of gratitude. ... 111 Contents I.Introduction:RestoringtheMultiplePillarsofOldAgeIncomeSecurityinPeru 1 11. The Risksto Incomefrom Ageing, andPeru's Vulnerability to Poverty inOldAge 5 ii.a. Risks to Income from Ageing and the Objectives of Retirement Security Policy 5 ii.b. Coverageof Peru'sFormalRetirementSecurity Institutions 10 111. Reformsto Social Security inPeruinthe 1990'sandRemaining Weaknesses inthe 18 Pension System iii.a. Pensionreforms of the 1990's and their impact 18 iii.b. From an affiliate's perspective,the new private second pillar i s the most expensive inthe region 21 and amongthe poorestperforming iii.c. Governmentsubsidiesto a minority of the elderly still pose athreat to fiscal stability, are 32 inequitable, and leave little room for apoverty prevention pillar lv. Facing Problems and Exploring Policy Solutions 39 iv.a. Perushouldrestorethe private second pillar by bringingAFP costs down and improving 39 investmentperformance iv.b. Perushouldrestorethe voluntary third pillar by encouragingfurther development of savings 46 and insuranceinstruments iv.c. Most urgently, Peru's shouldrestoreformal poverty prevention pillars to cover the riskof 47 poverty in old age V. Charting the Way Forward: Policies Consistent with Covering Peru's Households Against 59 Poverty inOldAge References 66 iv Tables Table 1 Share of the labor force 14-65 contributing to a pension system in urban areas, 2000 12 Table 2 Share of the labor force 14-65 affiliated with a pension system in urban areas, 1999- 13 2001 Table 3 Share of the labor force 14-65 contributing to a pension system in urban areas, 1999 - 14 2001 Table 4 Share of elderly (65+) receiving a pension benefit 14 Table 5 Share of elderly (65+) receiving a pension in urban areas, 1999 - 2001 15 Table 6 SPP selected indicators 23 Table 7 AFP pension fund portfolios as of 31December 2002, in 000's New Soles 28 Table 8 Contribution rates and ceiling on wages subject to mandate, December 2000 31 Table 9 a. Comparative Statistics: AFPs, banks, and insurance companies & b. insurance 32 companies balances as of December 2001 Table 10 Parameters of Peru's public pension regimes 33 Table 11 Weaknessesof Peru's private secondpillar, and options the Government should 40 consider and explore further Table 12 Policy options that the Government should consider and explore, that are consistent 65 with meeting objective of increasing coverage against (at least) the risk of poverty in old age in a way that is fiscally sustainable Figures Figure 1 Simulatedchanges inlife-expectancyamong menat 65 in selectedLatin Americancountries 6 Figure2a Headcountpovertyamong the elderly in selectedLatin Americancountries, usingdata from 1 1998- Peruviandata are from 2000 Figure2b Headcountpovertyamongdifferent age groups inPeru, 2000 8 Figure3 ReportedValue of Assets andAnnual EamedPensionsIncome 9 Figure4 Share of economicallyactive populationthat contributes to aformal pensionsystem, andshare 10 of elderly receivinga formal pensionin 1999/2000 Figure5 Active workers not contributingand elderly not receivingbenefits,selectedLatinAmerican 16 countries Figure6 Secondpillar choices of new affiliates since 1992 22 Figure7 AFP retums, costs and fees inPeru's SPP 25 Figure8 Returnon equityandfeednet contributionsin Latin America's private secondpillars 25 Figure9 Average earnings of affiliates to Peru's pensionplans) 34 Figure 10 Average replacement rates for men retiring from Cedula Viva and SNP 35 Figure 11 January 2002 reforms on net govemmentexpenditure in SNP, percentageof GDP 36 Figure 12 January 2002 reformson net governmentexpenditure in CedulaViva, percentageof GDP 37 Figure 13 Cost of minimumpensionguarantee with no other reform, net govemment expenditure SNP, 48 Percentage of GDP V Figure 14 Cost of closing SNP to new affiliates, net govemment expenditure SNP, percentageof GDP 49 Figure 15 Cost of different options for closing SNP, net government expenditures SNP, percentage of 50 GDP Figure 16 Additional cost of MPGin each reformscenario, percentage of GDP 51 Figure 17 Impactof reforms on net governmentexpendituresCedulaViva, percentageof GDP 52 Figure 18 Fiscal cost of MPG for all SPP affiliates 53 Figure 19 Reported contribution density of affiliates to Peru's pension system 54 Figure 20 Projected fiscal costs of flat universal pension benefit, percentage of GDP 56 Figure21 Projected fiscal costs of a targeted benefit to elderly poor, percentageof GDP 57 Figure22 Estimated proceeds from a contribution requirement from Cedula Viva beneficiaries, 63 percentage of GDP Text Boxes Box One The "Multi-Pillar" Model:Diversifying the Risks to IncomeSecurity in Old Age 2 Box Two Defining the Coverage Problem: What is it andWhy do We Care? 11 Box Three The Cost to SPP Affiliates of Opaque andOutdatedProvisionof InsuranceServices 26 Box Four Pointsto KeepinMindWhen Comparing AFP Portfolio Risk inLatin America 29 Box Five A Single Clearinghouse to Lower the Costs of Account Management 41 Box Six ExploringAlternative Fee Structures for Mandatory Retirement Savings 43 - Box Seven Reapingthe Benefits from Foreign Investment While Managing Currency Risk 45 a vi Terminology In this report, we categorize the components of a national retirement security system - sometimes referred to as "pillars" as inAverting the Old Age Crisis (World Bank, 1994), andsometimes as "tiers" as in TheEconomics of the WelfareState (Barr, 1992)- by their objective. This i s in marked 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, we use the term "first pillar" or "pillar one" to refer to the part of a pension system intended to keep elderly out o f poverty; "second pillar" or "pillar two" to that part intended to help individuals smooth consumption over their life-cycle, that i s to prevent a dramatic fall in income at the time of retirement; and "third pillar" or "pillar three" to the instruments and institutions available on a voluntary basis for workers to increase their income in old age. Since a large part of this report i s dedicated to making a case for the establishment of a formal pension institution to cover Peru's households against the risk of poverty in old age, we necessarily address 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 from minimumbenefit guarantees conditioned on a history of pay-roll contributions. vii I. Introduction:Restoring the MultiplePillars of Old Age Income Security inPeru 1. By most measurespoverty among the elderly inPeruis high, even when compared to neighboring countries in Latin America. Further, the relatively low incidence of formal pension benefits among the old, and low rates o f participation in the country's formal retirement income security system among the work force, indicate that Peru's population may be particularly vulnerable to the risksto income that accompany old age. 2. High rates of poverty among the elderly and only meager coverage of Peru's workforce are at first puzzling given the number of formal pension institutions in the country and the highlevels of public spending on retirement income subsidies. But while Peru's retirement subsidies are high, they are directed to a small, relatively well-off minority. In 2000 only 23 percent of the elderly were receiving pensions, yet Government transfers to cover the deficits of Peru's public pension regimes in the same year were equivalent to 69 percent of the amount set aside in 2002 to service Peru's external debt; 63 percent of public spending on education; 1.8 times the Government's expenditure on healthcare; and 1.4 times the amount set aside to alleviate poverty. 3. The objective of this report i s to explore and present policy options - this i s to say, options that are realistic given Peru's current circumstances - to extend formal protection against (at least) the risk of poverty in old age to the greatest number of households at a fiscally sustainable cost. These policy options are aimed at restoring the multiple pillars of formal old age income security in Peru, that have suffered from administrative abuse, opportunistic behavior and even neglect, in what has been an environment of admittedly severe fiscal constraints for the Government. 4. Why do we argue that the multi-pillar model needs to be restored? Since structural reforms that introduced a branch o f mandatory individual retirement accounts in December 1992, and despite the considerable progress achieved with these reforms, the different pillars of Peru's retirement security system have grown weak, and largely fail to adequately diversify the risks to old-age income, as envisioned by proponents of the multi-pillar pension model (see Box 1). The public branch of the "second pillar" still threatens the Government's fiscal stance and constrains management of the economy. The private branch is costly, risky and administered by a private oligopoly. The "third pillar" of voluntary savings and insurance instruments is anemic, costly, lacks transparency and fails to adequately complement benefits from the mandatory pillars. 5. These problems are compounded by the all-but-lack of a "first pillar" minimum pension guarantee for the few who participate in the reformed pension system, but even more dramatically by a missingsocial assistance benefit (sometimes referred to as a "zero pillar") targetedat the elderly poor for the majority of Peruvians who do not participate. 1 2 6. To meet its objective, this report takes a comprehensive approach in its analysis of Peru's retirement security institutions. Although there are several separate pension plans operating in parallel for different segments of the formal work force - each with its own attendant problems - our analysis will show how the weaknesses of one can hinder the rest. In searching for solutions to Peru's vulnerability to old age poverty, the parallel retirement security institutions have to be analyzed as components of a single old age income security infrastructure. 7. This report is divided into five sections. Following this introduction, Section I1 presents the dimensions of Peru's vulnerability to poverty in old age, by examining the nature of the risks to income from ageing in Peru relative to some of its neighbors. The section continues with a look at how well Government administered and/or mandated pension plans are covering these risks, again comparing Peru to other countries in the region. Section I11provides institutional background, reviewing reforms to formal social security institutions in the 1990's andthe progress they achieved. The Section goes on to examine the serious problems that remain. Section lV presents an analysis of some of the proposals for reforms to each branch of the retirement security system. Section V concludes by presenting a menu o f policy options - some very straight forward and less debated measures, and other, deeper, more controversial reforms - for the Government to explore at greater length, that are consistent with meeting the stated objective of extending protection against poverty in old age to the largest possible number o f households in a way that can be fiscally sustained. 8. We find that the private second pillar introduced in 1992 has developed rapidly. In just ten years since its inception, what was once only an infant industry in the financial sector has become the largest branch of Peru's retirement security system, providing services to 80 percent of affiliated workers, and accumulating a substantial pool of new capital for investment. Recent survey data show that a growing majority of new entrants into formal employment since 1992 choose the private branch over the public branch; that affiliates to the private plan are more likely to contribute than those to the public branch; and that workers with less tolerance for risk choose private accounts over the public PAYGO option. 9. However, by most measures, the new private pension fund managers appear to be colluding to bring down their operating costs, and increase their profit margins. Despite gains in efficiency and lower system costs, the private fund managers fail to pass savings on to their affiliates by lowering the price of their services. Furthermore, high prices in the private second pillar are accompanied by volatile investment performance that reduces the likelihood that affiliates will earn adequate benefits. We propose various measures that could bring costs down further; ensure these savings are translated into lower AFP fees by increasingcompetition; and further improve investment performance. 10. We also find that Peru's voluntary pillar has not been given a fair chance. The voluntary pillar fails to offer affordable savings and insurance instruments to households that are still primarily saving and pooling against the risks to income from old age, using what may be less than fully efficient informal instruments. We argue that certain features of the current legal framework - such as taxing workers contributions to retirement plans; 3 the lack of a ceiling on workers' wages subject to the private second-pillar mandate; and restrictions on withdrawals of voluntary savings - not only hinder the development of voluntary instruments, but may provide disincentives for workers to comply with the obligation to save inthe second pillar. 11. Finally, we find that remaining government subsidies to public pension plans still pose a fiscal threat, and crowd out the creation of a poverty prevention pillar with wider coverage. Subsidies from the treasury - financed mainly through value added taxes - to the remaining public pension plans are still large, especially considering that only 23 percent of Peru's elderly receive pensions. O f these, 76 percent are in the higher income quintiles. One of the arguments often put forward against restoring a "first pillar" for younger workers who choose the private branch of the system, i s that the Government cannot afford to offer any minimum guarantees, and may not even be able to afford a modest targeted non-contributory "zero pillar" benefit to the elderly poor. We do not pretend to offer the Government a fully-vetted proposal for a new poverty prevention pillar. The options for setting up a poverty prevention pillar that need to be explored extend beyond the scope of this report. Considerable thought and analysis needs to be dedicated to resolving this institutional failure in the short term. However, we present arguments for and against options taken by some of Peru's neighbors in the region, and our simulations of the fiscal costs of several reform scenarios show that the real issue that has to be considered i s not the affordability of a poverty prevention pillar, but rather how Peru chooses to distribute the public resources it i s already dedicating to subsidize old age. 12. However, few of the policy options suggested in this report can be effectively explored or eventually implemented without stronger leadership in pensions policy than currently exists. Old age income security i s a complex arena of public policy, spanning issues from social assistance to financial sector regulation. Several government agencies in Peru have direct and indirect interests and roles in the functioning of the pension system, but these can often clash. Currently, no single institution has a clear mandate to investigate, formulate and coordinate pension policy. Although among the strongest of Peru's public institutions the Ministerio de Economia y Finanzas (MEF) i s conducting this leadership role by default rather than by design. While MEFcurrently benefits from the presence of eminently qualified and experienced staff to carry out the leadership role inpensions, this is afortunate circumstance rather than apermanent institutional feature. 13. The reform options presented in this report will require more formalized leadership, with a widely recognizable mandate over the agencies involved, and will have to be guided by a medium and long term vision. More important than any single reform measure among the options presented in this report, i s the need for Peru's Government to invest energy and resources in the establishment of this long-term vision for restoring the multiple pillars of its old age income security system. The World Bank will continue to work closely with the Government to develop this vision and to take the more immediate steps requiredfor its achievement. 4 11. The Risksto IncomefromAgeing, andPeru's Vulnerability to Poverty inOldAge ii.a. Risksto Incomefrom Ageing and the Objectives of Retirement Security Policy 14. Retirement security policy has two separate but complementary objectives: (1) to help individuals (and their dependents in the household) to smooth life-time consumption by replacing income when they lose the ability to earn with the onset of old age; and (2) to prevent the elderly from falling into poverty. Therefore, the part of the social protection system that "covers" old age, i s designed to mitigate the risk of two possible losses with very different characteristics. The first, i s loss of earnings ability (that i s consumption during a period of life that mustbe financed without the ability to work) and the second i s poverty in old age (Gill, Packard and Yermo, 2003). 15. The risk of losing earnings ability in Peru is increasing, as in much of Latin America, with risinglongevity. But longevity i s not a risk, per se. Inmost societies, it i s considered a blessing to live a long life, and advances in technology and healthcare have brought longevity to a greater share of the population. However, as life-expectancy increases, the probability that most people will face a period of life in which they will need to consume but be unable to work can also rise. This i s to say that an increasing share of the population faces a relatively predictable disability as the incidence of old age becomes more frequent (see Packard 2002, andGill, Packardand Yermo, 2003). 16. Figure 1 shows how much longer a man can expect to live having reached age 65 and projected changes in his life expectancy in selected Latin American countries. Although it i s important to emphasize that these projected increases are based on certain assumptions, as shown in the figure, life expectancy in Peru i s increasing roughly at the same rate as in Ecuador, Bolivia and Mexico, but i s slow relative to the rate of increase in Colombia, Uruguay and Argentina. A similar illustration for life expectancy of women, would show an even greater projected increase. 17. Trends in longevity are immediately relevant to this discussion, because they indicate the sort o f policies and institutions a Government can afford to have in place to secure adequate retirement income for its citizens. All else equal, a retirement security system that seeks to pool the risk of losing earnings ability, financed on a pay-as-you-go (PAYGO) basis, will become more expensive as the probability of long-life and the "frequency" of old age (relative to working age) inthe population increases. This i s often because a pension system's statutory retirement age is politically difficult to adjust upward with changes in life-expectancy and the increased capacity to work into old age that accompanies development. The need to raise pay-roll taxes to finance PAYGO systems in countries with aging populations (and ever-larger transfers from Government general revenues to pay public pensions where contributions fall short), are strong 5 indications of the increasing cost of offering PAYGO arrangements relative to pension plans based primarily on individual savings to cover the loss of earnings ability.' Figure 1.The loss of earningsability with longevity is increasinglyfrequent (Projected changesin life-expectancy among men at 65 in selected Latin American countries) 19 18 (0 v) I 17 E 5 u- * 0 em 16 4- 8P f W 15 14 13 2000 2010 2020 2030 2040 2050 -- -Chile Argentina I Bolivia +Colombia +Ecuador +Mexico *Peru - - - Uruguay ---+-Venezuela Source: CELADE (2003) 18. On the other hand, the risk of poverty in old age, while related to loss of earnings ability and longevity, i s somewhat different. In the context of growth and economic development, a1 else equal, the opportunities to accumulate wealth over the life-cycle should increase. Each generation should reach old age with increasing accumulated wealth. Thus, although increasing longevity can make the loss of earnings ability more frequent, the incidence of poverty in old age should, in most countries, become increasingly rare.2 This said, by many conventional measures we find that poverty among the elderly in Peru (relative to other age groups and to other countries at similar 'Although contributions from payroll and/or the age of retirement also have to be adjusted in a defined contribution system to accommodate rising longevity, this adjustment is made automatically. An important caveat to this argument goes as follows. Although each generation should reach old age with increasing wealth, as each will live longer in old age, the wealth necessaryto stay out of poverty also increases, thus there is no guarantee that the incidence o f poverty will decline. 6 levels of development) i s only slightly lower than poverty among the population of working age. Figure 2.a. Poverty among the elderly inPeruis high relative to its neighbors,and even relative to other lower-middle-income countries (Headcountpoverty among the elderly in selected Latin American countries, usingdata from 1998. Peruviandata are from 2000) a, 1 45 c P z a, a 0 40 sa, - Ba, 35 m 30 0 i -0 C a 25 v) (0 P e, 20 .-0 C -m2 I 15 Q a 0 10 c 0 E c m 5 v) 0 Source: Wodon, Lee and Saens (2002), Peruviancalculation provided by Instituto Cuanto 7 Figure 2.b. Poverty among the elderly inPeru,although lower than other age groups stands at 35%, only slightly lower than poverty among those of working age. (Headcount poverty among different age groups in Peru, 2000) 70 60 - a, 1 c $ 5 0 - a 0 6 a, - 2 40 - m a, 'zc -0 30 - 3 0 a a 5 2 20 - r m v) 10 - Total 0-14 15-39 40-64 65+ Peru's Population Dividedby Age Group Source: Instituto Cuanto usingENAHO2000 - Fourth Quarter Survey 19. The data in Figures 2a and 2b show head count poverty rates usingincome reported inhousehold surveys. While the incidence of poverty among the elderly by this measure i s often similar to that in other age groups, the numbers fail to capture ~ e a l t h .When the ~ value of accumulated assets are taken into account (see Figure 3), Peru's elderly are well off compared to those of working age. Poverty statistics are typically calculated using current income from employment, pensions or public and private transfers reported inrepresentative household and labor market surveys. Older respondents, many of whom are likely to be retired, will report lower levels of income, which will naturally place them in the lower tail of the income distribution 8 Figure3. While incomedeclineswith age, the elderly holdtheir wealthinlarger stocks of accumulatedassets (ReportedValue of Assets and Annual EarnedPensionsIncome) 30000 14000 "Value of accumulated assets 25000 12000 E 10000 !$ 2 20000 3 n 0 b fn 2 8000 J '5S ->" 15000 -82-. 3 3 6000 3afn I 2 0 10000 6 4000 3 5000 2000 0 0 14-25 26 - 39 40 - 64 65+ Source: Barr and Packard(2003), usingdata from the PRIESOconductedin Lima, Peru, inMay 2002. The sample for the PRIESO survey was selectedfrom the pool of working respondentsto the ENAHO survey taken inthe third quarter of 2001. 20. In the context of economic development, all else equal, average life-time incomes should rise, the opportunities to save and accumulate assets should increase, and thus poverty among the elderly, measured in wealth (accumulated assets), should become even less frequently occurring. In a review of poverty among the elderly in 44 countries - largely wealthier, developed countries - Whitehouse (2002) finds that although the incomes of the elderly are 80 percent of incomes o f the population as a whole, they are not typically poor. The old are either represented proportionately or under-represented among the poor. Again, as with the first risk the statistics showing the nature of the second risk - poverty in old age - are important in this policy discussion. To the extent that poverty among the old i s increasingly rare, and effective targeting instruments are in place to accurately measure household wealth, a social insurance mechanism to keep the elderly out of poverty should become more affordable. 21. Obviously, in makingthis argument, the legal and financial institutions that protect property rights and allow households to convert illiquid assets into income for 9 consumption in old age are critically important. The existence and efficiency of such institutions cannot be taken for granted inPeru or elsewhere inthe region. ii.b. Coverage of Peru's FormalRetirement Security Institutions4 22. By most commonly employed measures, Peru's population i s poorly coveredby the formal old age income security system (defined broadly to include all formal pension plans in this section of the report). With respect to coverage, Peru's system i s ranked among the lowest in Latin America, and by some measures i s even lower than countries with lower income per capita (see Figure 4). These dismal coverage rates reveal the extent of vulnerability to the risks to income from old age, both in the aggregate and at the household level. This places efforts to close the coverage gap among the Governments highest policy priorities (see Box 2). Figure 4. Peru's old age income security system covers only a minority of the current elderly and even fewer among the current work force. Coverage inPeruis low, even by regional standards. (Share of economically active population that contributes to a formal pension system, and share of elderly receiving a formal pension in 1999/2000) 100 90 0 0 Q 8 70 .-c z E -mxa 60 n 50 - --9 a 40 0 0 cn a 30 2 n 0 20 10 0 Source: National Social Security Authorities and household surveys in selected countries. This section draws primarily from Hall and Rozas (2003) analyzing data provided by Instituto Cuanto. 10 11 23. The majority of the workforce are employed informally and/or self-employed especially in service and commercial activities. According to the latest available data with national coverage (2000), only 13 percent of Peru's workforce are contributing members of a formal pension plan.5 In addition, Metropolitan Lima shows the highest participation rate (19.8 percent), with lower rates of participation among the workforce in all other regions in the country, as follows: coast (13.5 percent), mountain (8.3 percent) andjungle (8.9 percent). Table 1.Participation rates rates of contribution among populations of interest - - vary greatly with incomeand region Peru: Share of labor force aged 14-65contributing to a pensionsysteminurbanareas, 2000 Total Quintile Q1-poorest Q2 Q3 Q4 Q5-richest Total 13.0 1.6 6.1 10.6 17.0 25.1 MetropolitanLima 19.8 5.5 14.6 18.7 21.7 30.6 Coast 13.5 0.9 6.4 13.7 19.0 23.4 Mountain 8.3 0.0 0.0 2.5 12.4 23.5 Jungle 8.9 0.2 2.2 7.8 13.7 17.5 Source: INEI.EmploymentSurvey. ENAHO 2000-111Trimester 24. Of greatest concern, however, are the data on participation by income level of the workforce, which may largely be explained by a possible correlation between poverty and informal employment status. Only 1.6 percent of the workforce in the poorest quintile are contributing members of a pension system, while this rate rises to 25.1 percent in the wealthiest quintile. Similar patterns hold across quintiles of the labor force ineach of Peru's major regions (coast, mountain, andjungle). 25. Participation in Peru's formal retirement institutions shows a highly regressive pattern. Rates of participation are skewed toward wealthier segments o f the labor force, with very limited access among economically disadvantaged households which are arguably the most vulnerable to the risks to income that accompany ageing. I t i s worth noting that unlike the remaining sections of this report, the coverage statistics for Peru cited here are comprehensive. Since these data are drawn from representative household surveys, in Peru's case they include ALL of the formal pension systems, including the special regime for military and police personnel for which the data needed to perform the sort of actuarial simulations presented in later sections of this report are not yet available. However, this may not be the case for some o f the other countries whose coverage statistics are shown in Figure 4. The breadth of the population coverage and pension institutions reflected in the statistics is limited by the quality of the household surveys available. To illustrate, the data from Argentina only show the contributors to the national pension system. If contributors to special pension regimes for provincial civil servants were to be included, the ratio would increase to around 40%. 12 26. In order to assess recent trends in coverage of the current workforce only data for urban areas are available.6 According to these figures, affiliation rates (that is, the rates showing workers signing up with the system - in many ways, a measure of access to formal cover) have fallen between 1999 and 2001 by 3.1 percent among the urban workforce, but displays divergingtrends by region and income level. Affiliation rates fell inby 7.5 inMetropolitan Lima, but rose by 9.4% inthe rest of urban Peru. This increase, however, occurred among the upper quintiles of the workforce - affiliation rates actually fell quite sharply among the poorest quintile outside of Lima (by 42.5%). This negative trend i s worrying to the extent that it shows fewer opportunities for poorer workers to gain access to ~overage.~ Table 2. Rates of affiliation to the pensionsystem among Peru's workers fell most dramatically among lower income groups Peru: Share of labor force aged 14-65 affiliated to a pensionsysteminurban areas, 1999-2001 Total Quintile Q1-poorest Q2 Q3 Q4 Q5-richest 1999 22.7 - - 21.5 7.2 12.7 - - 24.7 39.0 Metropolitan Lima 26.7 11.0 15.9 25.3 24.4 46.8 Rest of urbanPeru 18.0 8.0 12.2 17.3 17.3 29.4 2000 22.4 - 14.1 6.1 22.3 25.9 - 3.6 Metropolitan Lima 24.2 8.4 16.7 23.3 26.3 37.0 Rest of urbanPeru 21.0 4.4 12.0 21.5 25.6 35.1 2001 22.0 - 13.9 6.9 21.5 26.8 35.6 Metropolitan Lima 24.7 9.5 16.8 22.5 30.3 39.0 Rest of urbanPeru 19.7 4.6 11.6 20.6 24.0 32.7 1999/2001 -3.1 - - - - - -4.2 9.4 0.0 8.5 -8.7 Metropolitan Lima -7.5 -13.6 5.7 -11.1 24.2 -16.7 Restof urban Peru 9.4 -42.5 -4.9 19.1 38.7 11.2 Source: INEI. Employment Survey. ENAHO 1999-2001.I11Trimester 27. The share of the workforce not only affiliated but actually contributing to a pension system (actual contribution better proxying future coverage outcomes) has fallen almost imperceptibly, b y 0.6 percent in 1999-2001. However, the distribution across income quintiles reveals a sharp decline in the share o f contributing workers (-25.6 percent) in the bottom quintile, which has been most severe (-33 percent) inLima. Thus in 2001, the latest year for which data are currently available, only 5 percent of workers in the poorest quintile were contributing members of the social security system, and this figure falls to 1 percent inurban areas outside o f Lima. Data for the urban areas omit an important component of the total workforce ( in Peru 34.3 percent of the people live inrural areas) but cover the bulk of participants in social security systems (only 9.4 percent of 'theIti rural workforce contribute to public or private pension systems). s important to point out that these are not panel data. We are using three separate waves of the third- quarter ENAHO survey. This point is particularly important in analyzing affiliation, since a worker affiliates once in their life, and cannot "disaffiliate". The statistics shown have to be interpreted with care, given the possibility of differences in sampling strategies that could confound attempts to compare statistics from one year to the next. This said, Instituto Cuanto gives assurances that the statistics are comparable. 13 Table 3. Rates of contribution show a dramatic fall inthe share of lower income workers covered by the system Peru: Share of employed labor force aged 14-65 contributing to a pension system in urbanareas, 1999-2001 Total Quintile Q1-poorest Q2 Q3 Q4 Q5-richest 1999 16.1 4.3 - 15.8 9.7 18.6 26.3 Metropolitan Lima 18.0 8.0 12.2 17.3 17.3 29.4 Rest of urban Peru 14.5 1.6 7.6 14.6 19.7 23.6 2000 17.9 - - - - - 3.2 11.9 17.8 21.3 28.9 Metropolitan Lima 19.8 5.5 14.6 18.7 21.7 30.6 Rest of urban Peru 16.4 1.5 9.8 17.0 21.0 27.3 2001 16.0 - - 15.7 3.2 9.6 20.6 26.6 Metropolitan Lima 18.8 5.3 12.9 16.9 24.1 30.4 Rest of urban Peru 13.6 1.3 7.0 14.6 17.7 23.4 1999/2001 -0.6 - -25.6 _. -1.0 - - - -0.6 10.8 1.1 Metropolitan Lima 4.4 -33.8 5.7 -2.3 39.3 3.4 Rest of urbanPeru -6.2 -18.8 -7.9 0.0 -10.2 -0.8 Source: INEI.Employment Survey. ENAHO 1999-2001, I11Trimester 28. Turning to the incidence of pensions among today's elderly - the "coverage outcome" if you will - we find that the share of Peru's retirement-age population actually receiving pensions (23.5% in 2000) i s significantly higher than the share of the workforce currently participating in formal social security systems. However, there may be important selection issues reflected in the data that need to be considered: those workers who reach retirement age and live for a substantial period thereafter, are likely to have been those better off, holding jobs covered by the pension system in the past; additionally, elderly people are concentrated in urban areas (64 percent of the population over age 65 live in urban areas) where pensions are relatively easier to acquire - in rural areas, only 10percent of the elderly receive pensions. 29. The data on receipt of public pension benefits show the same regressive pattern as participation rates proxied by contribution. The unequal distribution of pension coverage rates i s most marked outside of Lima, and in particular, in the jungle and mountain regions, where pension incidence rises from 0% in the bottom quintile to 22% and 38% respectively inthe top quintiles. Table 4. The coverage of public pensions among the elderly is regressive Peru: Share of elderly (65+) receiving a pension benefit, 2000 Total Quintile Q1-poorest 4 2 Q3 Q4 Q5-richest Total 23.5 8.2 17.5 26.2 34.4 41.8 Metropolitan Lima 39.6 23.1 39.0 37.2 50.3 52.0 Coast 26.5 7.4 23.5 29.6 40.0 36.8 Mountain 10.9 0.0 1.3 18.4 16.2 38.3 Jungle 8.9 0.0 1.8 12.9 15.3 22.2 Source: INEI.Employment Survey. ENAHO 2000-111Trimester 14 30. The most worrying finding in our analysis of benefit receipts, is that from 1999 to 2001, the distribution of benefits has become even more regressive. The share of elderly receiving pensions in the poorest quintiles dropped dramatically. In Lima alone, the share of elderly receiving pensions in the first quintile dropped by 32 percent. This was contrasted by an increase in share of elderly in quintile 4 and 5 receiving pensions of 32 percent and 11.5 percent, respectively.* Table 5. The share of elderly receiving pensions in the poorest quintiles fell from 1990 -2001 Peru: Share o f elderly (65+) receiving a pension benefit in urban areas, 1999-2001 Total Quintile Q1-poorest 4 2 43 4 4 QS-richest 1999 32.2 21.5 32.6 34.3 - 41.3 34 Metropolitan Lima 39.8 39.9 25.3 44.1 36.2 47.1 Rest of Urban Peru 25.0 9.2 37.9 23.2 31.1 35.0 2000 33.0 14.4 35.0 32.9 43.9 44.5 Metropolitan Lima 39.6 23.1 39.0 37.2 50.3 52.0 Rest of Urban Peru 26.7 5.7 30.9 29.6 37.6 36.8 2001 31.9 14.6 24.0 35.0 41.7 44.2 Metropolitan Lima 40.4 26.7 32.5 35.5 48.0 52.5 Rest of Urban Peru 22.4 5.7 14.6 44.4 21.5 31.5 1999l2001 -0.9 - -26.4 -32.1 - - - 2.0 22.6 7.O Metropolitan Lima 1.5 -33.1 28.5 -19.5 32.6 11.5 Rest of Urban Peru -10.4 -38.0 -61.5 91.4 -30.9 -10.0 Source: INEI. Employment Survey. ENAHO 1999-2001. I11Trimester 31. The low incidence of benefit receipt and low rates of participation presented in this section confirm that Peru's population i s especially vulnerable to the risks to income that arise in old age. Continued lack of access among the working poor i s particularly worrisome and will likely aggravate the regressive distribution of benefits shown above among future generations. This vulnerability i s compounded by the lack (until recently) o f a minimum income guarantee for some of the few who participate in the formal pension system, but more importantly, a missing social assistance benefit targeted at the elderly that could cover the risk of poverty in old age for the majority of Peruvians who do not participate. * Aswith the data presentedin previous tables, these are not panel data, but the results of the thirdquarter ENAHO surveys in three consecutiveyears. Again, the changes inbenefit receipts among different groups have to be interpretedwith care, given the risk of differences in the sampling strategy/methodology that could confoundattempts to compare these statistics from year to year. As mentionedearlier, Instituto Cuanto has given assurances that the statistics are comparable. 15 Figure 5. By all measures,Peru's coveragegap is wide, and particularly worrying amongpoorer households (Active workers not contributing and elderly not receiving benefits, selectedLatin American countries) 100 Working but not contributing Elderly but not receiving pensions 90 80 .. L 0 2m a, 40 P a, n a, 30 20 10 0 Inthe Population Poorest Quintile2 Quintile 3 Quintile4 Richest Source: Hall and Rozas (2003) with ENAHO data 32. Although Peru's composite set of targeted social programs designed to reduce poverty - the "safety net" - does not include a program targeting benefits specifically to the elderly poor, one might argue that other safety net programs are indirectly providing benefits that reach the elderly and reduce their vulnerability to poverty. The safety net includes a wide range of other programs and benefits. The largest both in terms of budget and coverage are food programs. Expenditures on food programs total roughly $US 350 million per year- or more than one-third of all targeted poverty reduction spending- andbenefits reachover 40% of the Peruvian population, and are progressively distributed,reaching 60% of the poor.g 33. Given their extensive size and coverage, it i s reasonable at first glance to ask whether these programs are providing a kindof safety net for the elderly. However, food programs are largely targeted to households with children, and lactating mothers, and not directly to the elderly. The elderly would benefitonly indirectly, by living in a household Budget data: MEF, as presented in World Bank, 2002 (Memorandum of the President on Proposed Programmatic Social Reform Loan 11, August 19, 2002), report number P7514-PE; data on coverage o f food programs: ENAHO 2002-IV household survey data, as presented in Herrera, Javier. 2003. "Governance, corruption, citizen participation and poverty in Peru, 2002". Institut de Recherche pour le Developpement, Paris, and Instituto Nacional de Estadistica e Informatica, Lima. (mimeo). 16 receiving benefits under one of Peru's food programs. However, while many among Peru's population over age 65 indeed live with younger relatives, and are potentially beneficiaries of these programs, outcome indicators on poverty rates among the elderly indicate that whatever indirect benefits may be being received, they are not nearly sufficient to reduce poverty among the elderly. 17 111.Reforms to Social Security inPeruinthe 1990's and RemainingWeaknesses inthe Pension System iii.a. Pensionreformsof the 1990's andtheir impact 34. Peru has had separate PAYGO retirement security regimes for workers in the private sector, civil servants and the armed services (including police personnel) since the early 197"s.'' These publicly administered pension regimes suffered from miss- management, ran deficits, and accumulated enormous contingent liabilities that hindered Peru's macroeconomic stability and aggravated inequality. Peru's changing demographic structure - and increasing share of elderly in the population - only aggravated these problems. 35. In the face of unsustainable fiscal costs, and as part of a structural adjustment program supported by the World Bank (see World Bank, 1992 and 1998), in December 1992 the Peruvian Government legislated structural reforms to the largest of its retirement security plans - the regime created in 1973 by DL 19990 for workers in the private sector and administered by the Instituto Peruano de Seguridad Social (IPSS). 36. The December 1992 reforms (DL 25897) allowed covered workers to redirect their social security contributions into privately managed, individual retirement savings accounts. The Oficina de Normalizacidn Previsional (ONP) - a new public institution created by the 1992 reforms to replace IPSS and to manage public pension policy - issued bonds to workers who opted for private accounts to recognize their past contributions to the public system. 37. Unlike similar reforms elsewhere in Latin America, every new cohort of entrants to formal (that is, regulated) employment inPeru since December 1992 i s allowed to choose between a down-sized PAYGO regime administered b y ONP (and renamed Sistema Nacional de Pensiones, or SNP), and a new branch of individual retirement accounts (Sistema Privado de Pensiones, or SPP) managed by "dedicated" (that is, specialized, single-service) private fund managers, the Adminstradoras de Fondos de Pensiones (AFPs). 38. Even after the December 1992 reform, the SNP was actuarially unbalanced, and its contingent liabilities were difficult for the Government to predict. For this reason, the Government sought to phase out the earnings-related PAYGO system in the medium term, and to let the private system become the mainstay o f formal retirement security in Peru. Thus since 1992 if new workers initially choose the public SNPPAYGO, they can always move to the private SPP/AFP branch at a later date. However, workers that loPrior to this, a Government organized retirement security regime existed since 1965 (created by law DS 04-65-TR)for workers in the fishing industry (Kane, 1995) 18 choose branch.1Private accounts cannot subsequently choose to move back to the PAYGO 39. In the private branch of the system, the specialized AFPs manage workers' private accounts and invest their accumulated savings in tightly regulated portfolios. A portion of workers' contributions to the private branch pays for the services provided by the fund managers and covers the cost of premia for mandatory private disability and life insurance policies. Workers are allowed to choose their fund manager from among the limited number in the dedicated industry, and those workers that have been with a fund manager for some minimum period of time are allowed to switch their individual accounts to another. Upon retirement, workers can either negotiate a gradual draw down of their savings, or use the accumulated balance in their individual accounts to purchase private annuities. 40. The rights of workers affiliated to the public pension regime for the civil service (created in 1974 by DL 20530) popularly known as "Cedula Viva"'2 were not affected by the 1992 reforms. However, in the years following the reform, entry into the special regime for civil servants and public sector workers was tightly restricted to judges and public prosecutors, and administration of the regime was given to ONP. Thus, in theory every new cohort o f entrants into public sector jobs since 1992 should face the same pension choices as entrants to formal employment in the private sector.13 Public sector workers already affiliated with the special regime who wish to join the new private pension system are allowed to do so. The special pension plan for the armed forces and the police was unaffected by the 1992 reforms. 41. SPP to create incentives for workers to open individual account^.'^ The Government has made successive changes to the parameters of the SNP and In 1992 when the private system was introduced, a worker's contribution to the SNPPAYGO system was 9 A similar choice is offered to workers inArgentina. Colombia also allows workers to choose, but allows them to alternate their choice every three years - a feature of he Colombian reformthat has severely weakened the retirement security system inthat country. Which also covered workers inpublicly owned enterprises, such as Peru's nationalizedpetroleum company. l3Only judges and magistrates are still legally allowed to enter the Cedula Viva, however, the generous parameters of the regime are difficult for organized public sector employment groups to ignore. Numerous such groups have managed to successfully argue in Peruvian courts to be allowed entry into the regime, making very difficult for the Government to actually close. l4The manner in which workers past contributions to the SNP are recognized in the switching process can also explain some of workers' reluctance to move into the new system. There have been specific, discrete issues of "recognition bonds": the first issue of recognition bonds was made to recognize affiliates' accumulated contributions up to when the new system began to operate in 1993. However, no new recognition bonds were promised to cover contributions to the PAYGOsystem after 1993, and migration to the new system slowed. The Government then issued a new set of bonds to recognize contributions made from 1993 to 1996. Another drop in the rate of affiliation to the private system lead the Government to announce a third issue of recognition bonds in January 2001. 19 percent of wages, while the contribution to the new private system was 13.9 percent. The retirement age in the SNPPAYGO was also initially lower (at 60) than that in the new private system (65). Although there was a surge of affiliation to the private system shortly after its inception, especially among younger workers, those earning relatively higher salaries, and with higher levels of education (see Palacios and Whitehouse 1999, and Barr and Packard 2003), the higher contribution rate and higher retirement age in the private system created a disincentive for workers to switch out of the SNP (Kane, 1995). In 1995 the Government raised the contribution rate to the SNPPAYGO to 11percent and the retirement age to 65 (with Law 26504). The contribution rate was raised again in 1996 to 13 percent (DL 874). Contribution rates to the private system fell to between 11.6 and 12percent in 1997 and have remained at roughly this level ever since.15 42. The December 1992 reforms (and subsequent adjustments to the contribution and retirement age parameters of the SNP) contributed to fiscal stability, eliminated a source of inequity and lead to the deepening of Peru's capital markets and the development of the financial sector. Simulations using the World Bank's Pension Reform Options Simulation Toolkit (PROST) presented in Zviniene and Packard (2002), as well as earlier work (see Kane 1995), show substantial reductions in the Government's contingent liabilities from the SNP stemming from the 1992 reform.16 The reform succeeded in dramatically slowing the growth of public pension liabilities, as well as the rate at which the total public debt for pensions - that is, the sum of public expenditure on pensions, net of revenue from contributions - accumulates . 43. The 1992 reforms also improved the impact of formal pension institutions on income equality, substantially altering distribution between covered workers of different income levels. Analysis of the percentage point difference in the internal rate of return earned b y a representative wealthier-than-average worker covered by the system, and that earned by a poorer-than-average worker, shows that the regressive impact o f the SNP was lowered by the 1992 and follow-on reforms (Zviniene and Packard 2002, elaborated in Gill, Packard and Yermo, 2003, and Pallares, 2003 for this report). 44. Finally, the introduction of the private branch of individual accounts has lead to deepening of Peru's capital markets, and contributed to the development of the financial sector. Pension assets managed b y the private pension fund managers accounted to nearly 5.9% of GDP in Peru by mid 2002 (Lasaga and Pollner, 2003 for this report). In the area of financial market development, the reform can be credited with setting up a new financial industry that, in terms o f regulatory oversight, has been a role model for other financial industries. There i s little doubt that the new branch of mandatory private l5Only an approximate contribution rate can be given for the private AFP system, since a portion of the contribution rate consists of variable insurance premia and AFP service fees. AFP fees have hovered at about 2.5% of wages, and an additional 1.5% covers the premia on group life and disability insurance. Of workers total salary deduction 8% accumulates as savings in their individual account. l6Details on the methodology and assumptions used for these simulations can be found in Zviniene and Packard (2002). Growth in the covered wage bill is 3%; growth in GDP i s set at 3% and the interest rate in the economy i s set at 5%. These assumptions differ from those presented in Pallares (2003) for this report. 20 pensions, like its counterparts in other countries that have reformed, has achieved some of the highest standards inrisk-ratingand disclosure (Yermo, 2002). 45. However, despite the considerable progress achieved by reforms in the 1990's, Peru's formal retirement security system remains weak. Each of the parallel pension plans suffers from its own problems. The failure of the formal retirement security system to reach even a third of the population - discussed at length in Section I- i s the system's most serious short coming. In each of the sub-sections that follow, we show how seemingly unrelated problems ineach branch of the system may contribute to this failure. In Section 111, we propose options for policy reforms to each branch of the system, that could extend coverage against (at least) the risk of poverty inold age. iii.b. From an affiliate's perspective, the new private second pillar is the most expensive inthe region and among the poorest performing." 46. The SPP/AFP system has developed rapidly since it was founded in 1992 and in just ten years since its inception, what was once only an infant industry in the financial sector has become the largest branch of Peru's retirement security system, providing services to 80 percent of affiliated workers, and accumulating a substantial pool of new capital for investment. Recent survey data show that a growing majority of new entrants into formal employment since 1992 choose the private branch over the SNPPAYGO; that affiliates to the private branch are more likely to contribute than those to the public branch; and that workers with less tolerance for risk choose private accounts over the public PAYGO (Barr andPackard, 2003).'* 47. However, most standard and measures indicate that the new private pension fund managers may be operating as an oligopoly, arguably colluding to bring down their operating costs, increasing their profit margins, and failing to pass savings from efficiency on to their affiliates. Furthermore, high costs are accompanied by volatile investment performance that, as shown by simulations conducted for this report (Pallares, 2003) reduce the likelihood that the affiliates to the private branch will earn replacement rates above 45 percent (and this only under optimistic assumptions about contribution history -40 years- as well as AFP returns - 6 percent real). 48. Initially there were five AFPs: El Roble, Horizonte, Integra, Profuturo, and Union. After an initial period, three more were established: Megafondo, Nueva Vida, and Providencia. Starting in 1994, several AFPs merged: Horizonte purchased Megafondo, and Nueva Vidapurchased Providencia. In 1996 Profuturo merged with El Roble; and in l7Our analysis of the SPP/AFP branch draws on Lasaga and Pollner (2003). Readers will find a detailed discussion of every aspect of the private secondpillar in Peru inthe background paper. 18Householddata show a marked shift out of the SNPPAYGO system at all income levels inLima, with workers in the poorest and wealthiest deciles shifting out of the public branch at similar rates (a decline on the order of 30%) (Hall and Rozas, 2003). 21 January 2000, Union and Nueva Vida merged, leavingjust four AFPs: Horizonte, Integra, Profuturo, and Union Vida." Figure 6. More new entrants into formal employment, choose the private second pillar over the publicPAYGO (Second pillar choices of new affiliates since 1992) 100,o Yo to SPP/AFP Yoto SNP 90,o 80,O - 70,O h 8 60,O u) c a = 50,O m .e L' a 40,O z 30,O 20,o 10,o 010 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 Source: PRIES0 Peru 2002, in Barr and Packard, 2003 49. Initially the magnitude of start-up costs for the AFPs resulted in significant losses for the industry during the first several years. Nevertheless, an up-front, pre-payment fee structure based on a percentage of affiliates' income (POI), assured the AFPs a substantial increase in revenues during the start-upphase of the new private pillar. Since 1995, the AFPs have made considerable gains in both market size and profitability. In dollar terms, funds under AFP management have grown at an average annual rate of 29.7 percent during 1996-2002. At the same time, profitability has surged, with a return on equity (ROE) of 59.6 percent in 2002, and a period average of 38.8 percent. Compared to other AFP systems in Latin America, the Peruvian system has achieved consistently higher profitability. InChile, where a mandatory private system has been inoperation for the longest period, AFPs have averaged 26.9 percent ROE.20 l9 The remaining four AFPs each account for approximately 25 percent of market share (affiliates). The Herfinhdal Index calculated using data from July 2002 on the number of affiliates per AFP is 2,499. 2o Since the Chilean system dates back to 1981, the rates of return in that market would appear to approximate "normal" returns in the industry. If so, the Chilean figures also indicate an unusually high rate o f profitability in pension industry compared to returns in other areas of the financial system. 22 Table 6. Peru's AFPs haveincreasedtheir efficiencyand profitabilityby bringing their administrationandmarketingcosts down (SPP selected indicators, percent, except where indicated) 1996 1997 1998 1999 2000 2001 2002 Earnings ROE XA 21 I 29 0 22 I 38 0 63 1 59 6 ROA NA 1.3 13 8 16 8 30 3 45 1 43.1 Efficiency 0per:itionnl Expenses i lnconie 120.6 95.9 92.2 74.5 55.1 48.1 47.8 Administrative Exp./ Income 54.8 45.1 40.3 36.6 29.7 27.6 28.5 Sales Exp./ Income 65.9 50.7 51.9 37.9 25.4 20.5 19.3 Operational Expenses i AffXares (V.) (75.17 194.40 185.98 155.87 109.26 88.96 8`4.89 Operational Expenses / Fund Assets NA 10.2 7.7 5.0 3.0 2.2 I.R Memo Items: Fund Assets ( Mill 9.) 2,478 4,114 5,473 8,45 1 9,721 12,475 15,956 Fund Assets ( Mill US$) $ 953 $ 1,507 $ 1,732 $ 2,408 $ 2,756 $ 3,626 $ 4,541 Notes: ROEis return on equity and ROA is return on assets, where assets and equity are the averages of end of period balances Source:Lasaga and Pollner (2003) drawing on data from PensionIntendencyinthe Intendencyof Banking and Insurance, SBS 50. Administrative costs and operational expenses in Peru's private pillar-two have fallen considerably.21 Operational expenses as percent of income of the AFPs has fallen from 120.6 percent in 1996 to 47.8 percent in 2002. Lower sales costs have accounted for most of the improvement, following legislation that restricted affiliates' ability to switch between AFPs, and subsequent lower pressures for the AFPs to defend their market share. As a proportion of funds under management, operational expenses have fallen from 10.2percent in 1997to 1.8 percent in 2002.22 21 Administrative costs include record-keeping, investment management, and amortization of start-up cots. Operational expensesare largely marketing and sales costs. 22 Peru has shiftedfrom a relatively flexible to a rigid process of transfer of affiliate accounts between AFPs. The number of transfers as percent of affiliates fell from 3.6 percent in 1998 (comparable to the regional average in 1998 of 4.0 percent) to 0.3 percent in June 2002, (substantially below the regional average of 2.6 percent in 2002). The reduction in transfers was achieved by making the process more onerous to the affiliates, by requiringthem to go in person to their original AFP to sign a special request for the transfer of their account, and allowing the process to proceed on a slower time path.' While the cutback in transfers has partially alleviated the high marketingcosts associated with a high frequency of switching, there is certainly no evidence of savings being passedon to affiliates. The measurestaken to slow transfers 23 51. However, despite this progress in terms of system productivity, the SPP i s still considered a high cost system when compared to other Latin American countries. The cost of Peru's private second pillar have historically been higher than in other private pension systems in the region. Whether in terms of total operational costs, assets under management per employee, or other cost indicators, Peru is considered to be at the upper range of the scale (see Prim America, 1999, Fernandez-Baca, 2000, and Yermo, 2002). In 2002, operational expenses of the AFPs as percent of net contributions by affiliates were 14.5 percent, compared to a regional average of 9.5 percent.23 Argentina and El Salvador were the highest, with 22.1 percent and 23.2 percent respectively; while Chile and Mexico were the lower cost systems, with 7.9 percent and 6.5 percent re~pectively.~~ 52. But while the profit marginsfor the fund managers are growing with efficiency, the price of the private pillar paid by affiliates has remained stubbornly high. Fees charged by AFPs in Peru were the highest - 30 percent of affiliates' net contributions (that is, contributions net of fees and premia for insurance) - considerably above the regional average of 15 percent. This relatively high average cost to affiliates has persisted for almost five years. Since 1998, AFP fees have averaged consistently about 30 percent of affiliates' net contributions. The minimal variation in that percentage across all four AFPs during the past five years i s another indication of a lack o f competitive market forces,in the industry. In Chile and Mexico the fund managers in the private pillar-two charge much lower fees as a percentage of net contributions, at 15.0 percent and 12.0 percent respectively. may have lead to the opposite extreme, and perhapsjust as detrimental an outcome, further diminished competition and the incentives for AFPs to cut their fees by reducing affiliates choices. 23This is based on AIOS June 2002 data. Net contributions are gross contributions net of premiums for insurance coverage. 24The comparison of the Peruvian SPP/AFPpillar with those in other countries i s made difficult by differences in institutional arrangements; by the stage o f development of each private pillar; by the scale of the markets in each country; as well as currency and other macroeconomic factors. This said, several analysts have concluded independently that Peru's private pillar is relatively costly, and the comparison to the private pension pillars in other countries i s helpfulin establishing feasible targets for improved performance 24 Figure 7. Peru's AFPs are increasingly efficient and profitable, but do their affiliates benefit? (AFP returns,costs and fees in Peru's SPP) 100,o 90,o 111Fees/NetContributions 80,O Operational Expenses/NetF n 8 70,O *Return on Equity v z3 60,O U 5 50,O E 40,O c 3 30,O 20,o 10,o 070 1998 1999 2000 2001 2002 Source: LasagaandPollner (2003) Figure 8. Peru's AFPs are the mostprofitableinthe region, but the most expensive to affiliates (Return on equity and feednet contributionsinLatinAmerica's private secondpillars) i FeeIncomelAffiliateNetConBibuCons AFP Argentina Bolivia Chile Costa Rica El Mexico Peru Uruguay Region Salvador Average 25 26 53. Lasaga and Pollner (2003) attribute persistently high AFP fees to a number of structural factors that quash competition between the fund managers, and may have contributed to the formation of an oligopoly. These include (i) the high start-up costs of entering the industry of dedicated private pension fund administration; (ii) difficulty the of comparing AFP fees (assessedon a pre-paid, percentage o f income, or POI, basis) with the charges of similar service providers inthe private financial sector (often assessedas a percentage of assets under management, AUM); (iii) opaque and outdated provision of insurance services tied to mandatory retirement savings; (iv) the hightransactions costs to affiliates of switching from one AFP to another; and finally, (v) the model of "dedicated provision" itself and the implied exclusion o f other regulated financial sector actors - some, like commercial banks with established, farther-reaching supply networks - from managing mandated retirement savings. 54. Furthermore, the authors point out that an increasingly outdated, decentralized structure of account management and contribution collection leads to needless and expensive duplication of efforts, impedes the AFP industry from making further efficiency gains, and keeps the threshold of profitability for potential new entrants to the dedicated industry unnecessarily high. 55. The highprice of the private second pillar to affiliates i s accompanied by relatively volatile investment performance. Although among the better in the region, the investment portfolio managed by the AFPs has achieved only limited diversification. As in other Latin American countries, AFP portfolio choices in Peru are severely limited by the relatively small supply and lack of diversity o f debt and equity instruments available in the domestic market. The structure of the investment portfolios managed by the AFPs i s quite similar. Their investments in financial firms, leasing companies, the industrial, miningand agricultural sectors are matched even by corporate names. This may indicate a lack of diversity of available investment grade issuers of securities in the domestic market. This constraint has also resulted in a relatively high concentration of the investment portfolio in the banking sector, approximately 30 percent of the total. This allocation i s inefficient as it reveals an unnecessary level of intermediation, since AFP funds investedin commercial banks are usually placed as certificates of deposit (CDs) or other liability products that the banks offer directly to their individual customers. 56. The high degree o f parallel investment portfolios is also reflected in the rate of return. The annualized real rate of return on the investment portfolios of the four AFPs during 1994- 2002 has averaged as follows: Horizonte, 5.9 percent; Integra, 6.1 percent; Union-Vida, 5.5 percent; and Profuturo, 5.7 percent; thus the spread between the highest and lowest returns i s a modest 60 basis points, makingthem practically identical. From the point of view o f the affiliate, with almost equal returns, there i s little to distinguish one AFP from another. 27 Table 7. With littleincentiveto improveinvestmentperformanceor to take risks, andlittleroomto offer differentiatedproducts,the four AFPs holdalmostidentical investmentportfolios (AFPpensionfund portfolios as of 31 December 2002, in000'sNew Soles) - Horizonte Integra Profuturo Union Vida Total SPP Value % Value % Value % Value YO Value % ~ ~~ I.DomesticInvestments 3 767 357 93.0 4712922 92.7 2 267 342 92.9 4 007 419 92.6 14 755 040 92.8 1. PublicSector 598 025 14.8 604312 11.9 385 076 15.8 473 074 10.9 2 060 487 13.0 BCRP Certificates 34 344 0.8 16 788 0.3 119530 4.9 255 050 5.9 425 711 2.7 Central Government Bonds 493 939 12.2 488476 9.6 232 050 9.5 158522 3.7 1 372 988 8.6 Brady Bonds 69 742 1.7 99 048 1.9 33496 1 59 502 1.4 261 788 1.6 2. FinancialSector 1 516 139 37.4 2089710 41.1 917 189 37.6 1 887 091 43.6 6 410 130 40.3 CDs and TDs 955 399 23.6 1329950 26.2 563 119 23.1 1191709 27.5 4 040 177 25.4 Financial Leasing Bonds 162 694 4.0 185220 3.6 68 041 2.8 256 159 5.9 672 113 4.2 Subordinated Bonds 34 356 0.8 147422 2.9 42 635 1.7 64 319 1.5 288 731 1.8 Other Bonds 94 188 2.3 80 391 1.6 59 952 2.5 29 946 0.7 264477 1.7 Mortgage Notes 500 0.0 137 0.0 92 0.0 2 310 0.1 3 039 0.0 Mortgage Bonds 12 083 0.3 12083 0.1 Equity Investments 269 001 6.6 346591 6.8 183351 7.5 330 566 7.6 1 129 509 7.1 3. Non-FinancialCorporations 1 609 110 39.7 1879603 37.0 854 703 35.0 1568 102 36.2 5 911517 37.2 Commercial Paper 50 888 1.3 51 562 1.0 92 703 3.8 66 061 1.5 261 215 1.6 Non-Financial Cop Bonds 521 013 12.9 571 366 11.2 190952 7.8 443 773 10.3 1 727 103 10.9 New Projects Bonds 29 685 0.7 35 148 0.7 18 964 0.8 7 088 0.2 90 885 0.6 Equity Investments 1007 525 24.9 1221 526 24.0 552 084 22.6 1051 180 24.3 3 832 314 24.1 4. InvestmentFunds 7 361 0.2 45 169 0.9 37 138 1.5 33 994 0.8 123662 0.8 5. Asset-backedSecurities 36 723 0.9 94129 1.9 73 236 3.0 45 158 1.o 249 245 1.6 Firms II.ForeignInvestments 286 077 7.1 371 055 7.3 171419 7.0 316 657 7.3 1145207 7.2 1.PublicSector 64 288 1.6 138973 2.7 82 568 3.4 154448 3.6 440 276 2.8 2. FinancialSector 4 481 0.1 20594 0.4 25 075 0.2 3. Non-FinancialCorporations 0.0 6444 0.1 6 444 0.0 4. MutualFunds 217 308 5.4 205 044 4.0 88 851 3.6 162209 3.7 673 412 4.2 111. In-TransitFunds (949) (0.0) 75 0.0 1 727 0.1 5 385 0.1 6 238 0.0 TOTAL 4052485 100.0 5084052 100.0 2440487 100.0 4329461 100.0 15906485 100.0 Source: Lasaga and Pollner(2003) using data from Pensions lntendency in the Superintendencyof Banking and Insurance SBS 57. Limited diversification combined with Government imposed constraints on the share of pension funds the AFPs can invest in foreign securities, contributes to the volatility of the private pillar's returns. From 1993 to 2000, Peru's AFPs generated returns that averaged 0.46 percent on a monthly basis, but the standard deviation was one of the highestin Latin America at 1.2. Hence, the ratio of return to standard deviation (in other words, the return per unit of risk) was by far the lowest in the region (0.36) (Yermo, 2002). Although this measure of investment risk must be considered along with some important caveats (see Box 4), it shows Peru's private pillar i s among the riskiest in the region. This volatility could lead to significant differences inpensions across cohorts 28 29 58. Based on data for June 2002, AFPs in Peru invested 15 percent of the funds in Government debt securities, as compared to a regional average o f 57 percent; they invested 33 percent in securities issued by financial intermediaries, compared to a regional average of 18 percent; and 28 percent in equities, significantly higher that the regional average of 7 percent. 59. In addition to the limiteddiversification, the reported value of the AFPs portfolios may not be a real measure of an affiliate's assets. One of the principal risks facing affiliated workers i s the possibility of unfavorable price fluctuations that could reduce the value of their initial investment. Affiliated investors rely on periodic valuation of an investment portfolio based on the marking-to-market principle for effective risk management. Currently there i s no consistent methodology for determining the value of debt securities, which comprise a major component of investment portfolios. Infact, one o f the deterrents to investment in the capital market i s the uncertainty investors have regarding the true value of the securities they acquire. The reportedreal rate of return on portfolios i s a combination of realized income and unrealized capital gains. The low liquidity of many of the financial instruments casts some doubt on the true value of those assets for which the fund i s reporting capital gains. 60. Lasaga and Pollner (2003) attribute volatile investment performance of the AFPs to a number of structural problems including: (i) a lack of investment grade securities - equity in medium sized enterprise, mortgage backed securities, and even Government bonds - that would allow the AFPs to further diversify their portfolios; (ii) a minimum return requirement that discourages innovation and leads to herd-behavior in the allocation of portfolios; related to the two previous (iii) constraints on the fund managers ability to offer affiliates differentiated (with respect to riskheturn characteristics) investment products; (iv) the lack of appropriate valuation techniques to accurately value portfolios; and (iv) a low and bindinglimit on AFP investment inforeign securities. 61. In addition to the factors that keep costs high and increase the volatility of AFPs' investment performance, there are two additional features that set Peru's second pillar apart from others inthe region that may be detrimental: there i s no ceiling on the portion of affiliates' income subject to the mandate to save, and retirement savings are taxed up front (conventionally, the taxation of pensions in Peru i s tax contributions-exempt returns-exempt pensions, or "TEE" rather than the recommended "EET"). 62. Since there i s no ceiling on wages subject to the SPP/AFP contribution rate - and consequently to AFP fees - Peru's mandate for private savings i s much greater inrelative terms than Chile's or Argentina's even though the statutory rate of contribution to the private pillar in each of those countries i s higher (see Table 7). Taxing affiliates' contributions instead of pensions (as best practice recommends, discussed in Whitehouse 1999) i s inefficient and inconsistent with widespread preferences for current consumption. Both features increase the costs of complying with the mandate to save for those in the population with other, perhaps more urgent and pressing, demands on their limited income. 30 Table 8. Although the statutory rate of contribution is lower inPeru's SPP, the lack of a ceiling on wages to which it is applied, makes the mandate to save more onerous (Contribution rates and ceiling on wages subject to mandate, December 2000) Country Contribution intofund / Maximum taxable eamings / individual's salary average national earnings Argentina 7.72 5.8 Bolivia 10.0 12.5 Chile 10.0 3.1 Colombia 10.0 Costa Rica 4.25 No ceiling Dominican Republic 10.0 14.0 ElSalvador 8.53 5.8 Mexico 12.07 11.0 Peru 8.0 No ceiling Uruguay 12.32 5.7 Source: Gill, Packard and Yermo (2003) Note: the total contribution rate (for capitalization, insurance and commission) was cut to 5 percent in Argentina in December 2001. The contribution that i s channeled into the individual's account i s therefore less than two percent since that date. The contribution rate in the Dominican Republic i s rising gradually untilreaching 10percent in2008. 63. Finally, there i s not much offered by Peru's financial sector (beyond the mandatory private retirement instruments) with which to top up retirement income from the mandatory branches of the system, should affiliates wish to save and invest on a voluntary basis. Peru's private voluntary pillar three i s anemic, and highly dependent on the private pillar two. As mentionedpreviously, roughly 80 percent of insurance policies for disability and death can be attributed to the private second pillar. The financial sector i s offering only a feeble supply of savings and insurance instruments independently of those policies demanded by AFPs. 64. This i s not necessarily evidence that the mandatory second pillar somehow displaces the voluntary third pillar. In fact many o f the problems that weaken the mandatory private second pillar hinder the development of a strong voluntary third pillar, particularly since the AFPs are allowed to manage additional voluntary retirement savings. These include the lack of a ceiling on the wages subject to mandatory savings; up-front taxation of these savings; pre-payment structure o f fund management fees; and limits on investment abroad. 31 Table 9. Peru's voluntary pillar of savings and insurance instrumentsis still small and weak relative to the private second pillar (a. ComparativeStatistics: AFPs, Banks, and InsuranceCompanies) 1999 2000 2001 2002 Total Assets (mill W.): AFPs 8,451 9,721 12,475 15,956 Commercial banks 66,224 61,428 61,812 61,822 Insurance companies 3,353 3,857 4,282 4,957 ReturnonEquity (%): AFPs 22.1 38.0 63.1 59.6 Commercial banks 3.8 3.0 4.5 8.4 Insurance companies 8.7 (1.0) 18.0 12.1 (b. Insurance Companies Balances as of December 2001) Number Assets Capital of Firms Amount Amount ( Millions S/. ) % (Millions S/. ) % Generaland Life Insurance 4 1487 34.7 344 27.7 GeneralInsurance 5 1106 25.8 646 52.1 Life Insurance 7 1689 39.4 251 20.2 Total insurance companies - 16 4282 100.0 1241 100.0 PSource: Lasaga and Pollner (2003) using data from Superintendency of Bankingand Insurance, P iii.c. Government subsidies to a minority of the elderly still pose a threat to fiscal stability, are inequitable, and leave little roomfor a poverty preventionpillar.25 65. The Government is still spending a large portion of its budget on subsidizing public PAYGO pension plans. However, a large part of public transfers to the SNPPAYGO are a direct result of the decision to allow contributing workers to open private individual accounts in December 1992 - a policy directly supported the World Bank, that accords well with fundamental principles of risk management. Since these workers' contributions are accumulating as private savings rather than paying the pensions of retired workers, the PAYGO financing mechanism of the SNP has to be subsidized during a period of transition until it can return to balance (between income from contributions o f those workers who did not switch to the private branch, and payments to retirees). 66. However, budget transfers to subsidize SNP benefits made up only 24 percent of the total amount spent by the Government on pensions in 2000. Furthermore, the per capita actuarial deficit (that is, the Government's contingent liability) to workers covered 25This section draws primarily from Pallares (2003) for this report. 32 by the SNP - considerably reduced by the 1992 reforms - i s only 18 percent of the deficit for workers covered by Cedula Viva, the separateretirement regime for civil servants that was all but closed by structural reforms inthe early 1990's. 67. While the transitions cost of losing contributors to the SNPPAYGO pose a challenge, this i s largely manageable. The immediate threat to the Government's fiscal stance lies in the Cedula Viva. Although the regime i s closed to all entrants into public sector employment since 1992 except judges and public prosecutors, the remaining threat to Peru's fiscal position i s largely due to the generous benefit parameters of the regime, even when compared to the parameters of SNPPAYGO and to PAYGO systems in other countries. The generosity of the Cedula Viva even surpasses that of civil service regimes inmost of the region, and is matched inthis regardonly by Brazil's RPPS. Table 10. The parameters of the Cedula Viva are notoriously overgenerous and matched inthe Region only by Brazil's RPPS (Parameters of Peru's public pension regimes) Parameter SNPPAYGO CedulaViva (private & public sector, not in (judges, public prosecutors and AFP) civil servants prior to 1993) MinimumRetirement Age 9 65 for menand women None MinimumYears of 9 20 12.5 for women, 15 for Contributions men, and 20 for full indexation (see bullet below) Replacement Rate 9 Less than 100%of average 100%of last salary salary in last 60,48 or 36 months Replacement Rate Survivors 9 50% of deceased'spension 100%of deceased's 9 Additional benefit to pension orphans, up to 18 or 21 if Additional 100%for each orphan studies full time orphan, and no age limit on benefits to orphaned girls Indexation of Benefits 9 To inflationif funds To current salaries of available from the Treasury position from which retired Benefit Ceiling 9 857 soles None ~ Source: Material provided by ONP 68. Separate, more generous retirement security regimes for workers in the public sector are often justified by the relatively lower levels of remuneration in government work. However, while civil servants and public sector workers in developed countries may earn far less than workers in the private sector, the public sector in Latin American countries often offers not only better pay but a greater degree of job security than private sector jobs. This comparison i s even more stark when public sector jobs are compared to employment inthe private, unregulated sector where most Peruvian's work. 33 69. Figure 9 present a comparison of wages between workers affiliated to Cedula Viva, and those affiliated to the SNPPAYGO and SPP/AFP system. While affiliates to the Cedula Viva on average earn significantly less than affiliates to Peru's private second pillar, their earnings are comparable to affiliates to the SNP, and significantly greater than SNP earnings among older workers. The relative comparability of earnings o f Cedula Viva affiliates with that of SNP affiliates provides little justification of the staggering difference average pension benefits shown in Figure 10 (as average replacement rates).26 Figure 9. The average earningsof affiliates to Peru's Cedula Viva, are comparable to earningsof affiliates to the SNP (Average earningsof affiliates to Peru's pension plans) 3,000 2 HAveragewageof contributors inAFPs t l Average wage of contributors in SNP OAverage wage of contributors in Cedula Viva 2 iz 1,000 500 1 1 r <21 21 -2! 26 3( - 31 -35 36 4( - 41 - 4 46-50 51 -55 56-6( 61 - 6 > 65 Age Source: ONP 70. Conscious of the potentially ballooning fiscal costs of the regime, the Government has continued its efforts to tightly restrict entry into Cedula Viva. However clauses in Peru's Constitution protecting "acquired rights" and interpreted to extend these rights to enshrine almost every aspect of the regime, make it all but impossible to undertake further structural reforms. Furthermore, since its inception the generous benefits attract rent-seekers among groups of workers able to exercise influence and make a case in Peru's courts for inclusion under the regime. 26Readers should note that these replacement rates are not those earned by specific individuals, Le. pension benefivlast salary or an average o f last salaries. Rather the replacement rates show the average pension divided by the earnings received by an active individual o f the same gender and age. 34 71. The ruling in almost every recent case that has come before the Constitutional Court has extended rather than constrained the generosity of the regime. Ina landmark decision in 1998, the Court undidthe progress of earlier reforms at curbing Cedula Viva liabilities by taking responsibility for administration of the regime away from ONP and giving it back to each of the ministries, Government agencies and parastatal companies whose employees still could claim a right to benefits. MEF and ONP have struggled to keep control of Cedula Viva spending, however, this task has been made almost impossible by uncooperative public sector bodies unwilling to relinquish data on beneficiaries and active workers acquiring further rights. Therefore the latest complete dataset of contributors and beneficiaries available for actuarial analysis i s from 1997. Figure 10. Cedula Viva pensions are substantially greater than SNP benefits, yet contribution and eligibility criteria are more lenient (Average replacementrates for menretiring from CedulaViva and SNP) 180,I 120 t 60 40 20 E Age Source: ONP 72. Seeking to slow the growth of public pension liabilities, while encouraging a greater number of workers to open individual retirement accounts in the private AFP system, the Government initiated a second major reform of its public pension plans in 2001. The objectives of this second reform were to: eliminate incentives for rent-seeking behavior by bringing greater parity between the parameters of the Cedula Viva and the SNPPAYGO; and to make the SPP/AFP system a more attractive option for workers who still remainedin SNP as well as new entrants to the formal sector. 35 73. The Government introduced a reform proposal that was passed by the legislature, and came into effect on January 1, 2002 (D.L. 21617). The reform did not change the pensions of current retirees, and included special provisions for workers aged 55 and older who are closer to retirement. The new law: (i) substantially lowered survivor benefits paid by Cedula Viva to the spouses and orphans of affiliates, while introducing a floor on survivor benefits equal to the minimumwage; (ii) lowered retirement pensions in SNP through changes to the benefit formula that tied pensions more closely to workers' contributions, while increasing the minimum pension slightly higher than the minimum wage, and financing this increase by absorbing FONAHPU to finance the increase; (iii) made it easier for workers to switch to the AFP system, and especially attractive to workers 55 and older; and (iv) introduced a minimum pension guarantee (Bono de Compensacion), but only for AFP affiliates born prior to 1945 (at the time of the reform, 55 years of age and older) who have contributed to a retirement security system (SNP or and AFP account) for at least 20 years, similar to the minimumpension guarantee (MPG) offered to all affiliates to the private second pillar in Chile. Figures 11and 12 show the impact of this second major reform on the Government's total subsidies to the public branches of the pension system (i.e. benefit expenditures, net o f contributions from active affiliates, and including liabilities represented by recognition bonds to workers who switched to the SPP/AFPbranch). Figure 11. The January 2002 reformplacedthe SNP/PAYGO on a morestable fiscalpath 1.20% ,Net Government Expenditure (Benefits Contributions, and Recognition Bonds) - 1.OO% 9n 0.80% u) tiS +SNP(1 990s Reforms) n +SNP(2002 Reforms) c 0 0.60% P) m m CI E 8 n $ 0.40% 0.20% Source: PROST Simulations in Pallares (2003) 36 Figure 12. The January 2002 reformloweredliabilitiesof Cedula Viva, but due to missingdata, the actual impactcanonly be guessed Net Government Expenditure (Benefits Contributionsand Recognition Bonds) - 1.60% 1.40% -s 1.20% n 0 c3 1.OO% u, $ *Cedula Viva (2002 Reforms) .c 0.80% 0 E Q 0.60% Q n i 0.40% 0.20% 0.00% zQQeQ 3 2 z % 8 89zQz320620.+ z&% z&g?zob3 ebz&\ zQ6% 2069 3. 2062 20qz zQ2% 2 Source: PROST Simulations in Pallares (2003) 74. The January 2002 reform had a more dramatic, positive impact on the finances of the SNP/PAYGO than on Cedula Viva. Our simulations show that the total net Government expenditures on the SNP/PAYGO (including recognition bonds to workers who affiliate to the SPP/AFP branch) would have passed 0.8 percent of GDP in 2007 and reached 1percent by 2055 had it not been for the changes inthe benefit formula passed in January 2002. Instead, the Government's reforms succeeded in stabilizing net expenditures at between 0.2 and 0.6 percent o f GDP from 2016. 75. However, a critical assumption inthese simulations that must be kept inmindi s that a majority of new formal sector workers will continue to choose the private second pillar over the SNPPAYGO. We have assumed that the rate of affiliation to each branch remains constant in time. However, the January 2002 reforms may have increased the appeal of the SNP/PAYGO by increasing the minimum pension guarantee in the public branch. Readers are reminded that only those affiliates in the SPP/AFP system born before 1945 have a similar minimum pension guarantee. This asymmetry between the private and public branches could not only invalidate the simulations presented in this report, but more importantly imperil the Government's objectives of weaning the work- 37 force off of publicly provided second-pillar pensions (that is, pensions designed to replace income, rather than just to keep the elderly out of poverty). 76. Our simulation of the impact of the January 2002 reforms on Government's net expenditure on Cedula Viva pensions show a substantial but relatively modest impact. The reforms are likely to save the Government about 0.10 percent of GDP in the next ten years. However, even after the January 2002 reforms, public outlays to retired public sector workers and their survivors are still substantial - about double what the Government spends on the SNP and recognition bonds to AFP affiliates until 2008. Since entry into the regime i s nominally restricted to judges and public prosecutors, our simulations show the net spending quickly falls to 0.05 percent of GDP but only by 2040. Our simulations, of course, assume that the eligibility criteria for Cedula Viva pensions are applied as stipulated in the law, and will underestimate the actual public cost of regime to the extent that these criteria are loosely applied, or new groups of public sector workers win access to generous benefits. 77. Even after the reforms of January 2002, subsidies from the treasury - financed mainly through value added taxes - to the SNPPAYGO and to Cedula Viva pensions are still large, especially considering that only 23 percent of Peru's elderly receive public pensions. O f these, 76 percent are in the higher income quintiles. One o f the principal arguments put forward a,winst restoring a "first pillar" for younger workers who choose the SPP/AFP branch of the system, i s that the Government cannot afford to offer any minimum guarantees, and may not even be able to afford a modest targeted non- contributory benefit to the elderly poor. Our simulations in this section and in the next, show that the real issue i s not the affordability of a poverty prevention pillar - especially a non-contributory, targeted benefit to the elderly poor - but rather how Peru chooses to distribute the public resources it i s already dedicating to subsidize old age. 38 IV.FacingProblems andExploringPolicy Solutions iv.a. Perushouldrestorethe privatesecond pillar by bringingAFP costs downand improvinginvestmentperformance 78. The system costs of the SPP/AFP second pillar have been falling. However, if the private second pillar i s to be the mainstay of retirement income security system, these costs should be lowered further. But more importantly, the price paid by affiliates in commissions and fees has to be lowered and investment performance improved if the private system i s to become more marketable to future affiliates. Bringing the system costs and price of the private second pillar down will be critical for meeting the over- arching objective of increasing coverage, whether Peru fully restores a poverty preventionpillar or not. 79. What i s the connection between the price and performance of the private pillar and the goal of increasing effective coverage? As new entrants to formal employment join the SPP/AFP pillar in ever greater numbers (as shown in Barr and Packard, 2003), the bulk of retirement pensions in Peru will be increasingly determined by the assets that affiliates to the SPP/AFP branch are able to accumulate before they reach retirement age. High AFP fees and volatile investment returns, lower the likely asset accumulation. Without a poverty prevention pillar, this state of affairs leaves the majority o f affiliates highly vulnerable to the risks to income from old age. But for the same reasons, even with a poverty prevention pillar, the size of the Government's new contingent liability to keep the elderly out of poverty (or at least that portion attributable to the elderly who participated in the reformed pension system, should the Government opt for a contributory minimum guarantee) will be directly related to the performance of the SPP/AFP branch. 80. Furthermore, a cheaper, better performing and less taxing SPP/AFP pillar i s more likely to attract participants from among the large portion of the workforce for whom the mandate does not bind and who still may only save for old age informally (see Barr and Packard, 2003), as well as retain new participants. Finally, those who are not free to choose whether to participate - workers in the formal sector for whom the mandate to save does bind (again, see Barr and Packard, 2003) - would see a better return on their contributions and be better served by a mandated retirement-savings instrument. 81. The measures recommended to bring down operational costs; ensure that these savings are passed on to affiliates as reductions in AFP fees through the restoration of competitive market forces; and to further improve investment performance, are detailed at length in Lasaga and Pollner (2003), and summarized in Table 10 overleaf. A fuller discussion of the costs and benefits of some of these proposals, such as exploiting economies of scale by centralizing account management operations, and converting from a percentage of salary (POI) to an assets under management (AUM)fee structure (to ease performance comparisons between the AFPs and other fund managers), can be found in Boxes 5 through 7. 39 \ A A A 4 A A A A 2 A A A 0d Y L A A A A A \ A A A A . e, Lo i BEk dv1 P 2 z Y E m e 8 EM c 2M 41 82. Given their financial interests in the current structure of the SPP, it i s fair to expect the AFPs to resist even minor changes to the present rules and regulations governing the private second pillar, much less some of the more radical, longer term proposals made by Lasaga and Pollner (2003). Market participants interviewed for this report presented clear reasons why many of the proposed changes - even the most modest - could make the current situation even worse for affiliates. Many of these arguments merit closer examination. While it i s possible to go point by counter-point through Table 10, the AFPs position against any changes can be illustratedby their arguments against measures to simplify the account switching process. 83. Making it easier for affiliates to switch their accounts between AFPs could encourage competition between the fund managers based on fees and performance. However, the AFPs argue the contrary: that makingthe process easier could actually raise costs and fees, since there i s no evidence of price elasticity in workers demand for one AFP's services over another, and indeed the only determinant of which fund manager an affiliate chooses i s the size of their direct-marketing sales forces. Prior to changes in the private second pillar that severely restricted account switching, the costs incurred by the fund managers in protecting their market share with these armies of sales staff, were substantial. The Government imposed restrictions on account transfers expecting that the AFP savings on marketing staff would be passedon to affiliates in the form of lower fees. 84. However, the recommendation to make account transfers easier, i s not to repeal all the current restrictions and return to the days o f affiliates switching several times a year. Rather, the recommendationi s to make the switching process less cumbersome. Whether this is done by harnessing the internet and other technology solutions, or by setting up a single clearinghouse to centralize account management functions (or both, as described in Box 4), the actual frequency of permitted transfers need not be changed. Infact, reliance on technology solutions could eventually make the direct-marketing staff redundant. Although several studies in Peru and other countries have shown low price-elasticity of demand in the choice between fund managers, and the observed frequency of account transfers i s below the limit, the current restrictions make it much easier for the fund managers to collude. 85. Finally, the increase in AFP profit margins since the restrictions on account transfers were introduced has been substantial, and fees have remained stubbornly high. Were additional marketing costs indeed unavoidable to protect their market share, the AFPs should be able to find sufficient room to cover these additional expenses without expecting affiliates to pay the difference. The Government could ensure that affiliates benefit from whatever additional competition i s introduced with easier account switching, by monitoring and publishing AFP costs and profits to increase transparency (and more importantly) awareness prior to and after any regulatory changes. Whereas there may have been little attention paid to AFP fees in the past, the fund managers super-normal profits in recent years have come under scrutiny, and the issue has entered the public debate appearing regularly inthe popular news media. 42 43 86. It would be unfair to argue that the AFPs are the only problem. It i s important to point out that it i s not only the AFPs who are likely to resist change. Other institutions, even in the Government itself, argue against measures that could increase investment performance and leadto more effective diversification of affiliates' portfolios. Increasing the effective limit on the share of AFP pension funds that can be invested abroad presents a case inpoint. 87. While all observers agree that Peru's capital market does not offer a sufficient number of securities for AFP investment and adequate diversification of risks, many argue against lifting the limit on how much of private forced savings should be invested outside the country. Many of these arguments present an outdated, nationalistic logic that has largely been discredited. However, other legitimate concerns for the safety and stability not only o f workers' savings, but also for Peru's currency market deserve closer attention. 88. Under the current law, Peru's Central Bank determines the composition of AFP portfolios, by setting ceilings b y class of investment instrument, including foreign securities. While the nominal limit can be changed by law in Congress, the effective limit on investment abroad is strictly set by the Central Bank. To be fair to the Central Bank's position, they are concerned that increasing the limit of investment abroad could expose workers' savings to volatility of international markets, and given the potential movement of large amounts of capital in and out of the country - especially if the current structure of the industry continues to lead the fund managers to mimic each others' investment allocations so closely - could even threaten the stability of Peru's economy. 89. The Central Bank's position i s not that controversial in the region. Similar argumentshave been presented inevery country with a private second pillar. Chile only allowed its AFPs to investing abroad years after the introduction of individual accounts in 1981. Mexico and Uruguay still do not allow the managers of mandatory retirement savings to invest abroad. Nevertheless, most policy makers would agree that further diversification away from the Peru country risk would be beneficial to protect affiliates' assets in the long term. Lasaga and Pollner (2003) provide an example of how the objective of diversifying the risks to affiliates' savings can be met, without necessarily exposing the economy to undue currency risk and volatility, as presented inBox 7. 44 45 iv.b. Peru should restore the voluntary third pillar by encouraging further developmentof savings and insuranceinstruments 90. As discussed earlier, the pillar of voluntary retirement savings and insurance instruments in Perui s very small and constrained from developing further by many of the features that provide strong disincentives for workers to join the SPP/AFP mandatory pillar. Since the AFPs are allowed to manage voluntary savings along with mandatory savings, many of the suggested changes in Table 10 could increase voluntary participation in the formal retirement security system. Most important among these, a change in tax treatment of contributions to provide workers an incentive to save for retirement, and the eventual imposition of a ceiling on affiliates salaries subject to the mandate. 91. Despitethe weaknesses of the voluntary pillar, relatively attractive substitutes to the products offered by the AFPs are already made available by Peru's financial sector and enjoy potential demand from the segment o f the workforce not subject to the mandate. Under the current law, the self employed are exempt from the mandate to contribute to either the public or private branch of Peru's second pillar. Professional self employed, which represent a lucrative market niche, tend to opt for savings plans offered by insurance companies instead of affiliating voluntarily with an AFP. Some of these savings programs offer 8 percent real rate of return or higher, based on fixed income investments (Lasaga and Pollner, 2003, Barr and Packard, 2003). 92. The choices of this relatively well informed clientele, may indicate the need for additional financial instruments with certain "preferred" characteristics to be introduced both by the AFPs and other financial sector actors, to increase the share of households that take up formal savings and insurance instruments as part o f their portfolios. Insurance industry products are currently more attractive to independent workers because among other reasons: (i) the funds channeled to AFPs are denominated in domestic currency, while the insurance products can be denominated in foreign currency; (ii) contributions can be invested in different types of vehicles, as compared to the SPP, which allows only one fund; (iii) contributions are voluntary, they are not mandated as a fixed percent of workers' entire wages; (iv) the fee structure of voluntary instrument providers i s usually the more transparent and preference-consistent AUM structure; and (v) there is greater flexibility in terms of how insurance benefits are designated and paid to beneficiaries, while in the SPP these are controlled by more rigid rules. Finally, another savings incentive would be to allow individuals to make early withdrawals from the voluntary component of their account under certain circumstances, but with an income tax penalty. 93. When asked about voluntary savings and insurance policies, many point to the current low take-up of voluntary instruments as evidence that a third pillar i s not really viable due to the average low incomes in Peru. However, low take-up of voluntary instruments under the current set of laws and incentives i s hardly evidence that the average Peruvian household would not save voluntarily in a formal instrument if given the opportunity and the incentives to do so. Barr and Packard (2003) show that workers who are free to choose whether and how to insure, secure income in old age in a number 46 of ways including the formal pension system but additionally in property and in their children. iv.c. Most urgently, Peru's should restore formal poverty prevention pillars to cover the risk of poverty inold age 94. Inearlier sections we have argued that Peru is particularly vulnerable to the risks to income that arise in old age, and that the lack of a first pillar for the majority of affiliates to the retirement security system since 1992, as well as no social assistance benefit to the elderly, increases this vulnerability. Currently there are a number o f structures designed to establish a floor below which affiliates' benefits will not fall, including: a contributory minimum pension for affiliates to the SNP/PAYGO; a contributory minimum pension guarantee for affiliates to the SPP/AFP born prior to 1945; and a minimum return requirement that the AFPs are obligated to post to affiliates' accounts. However, these structures do not provide a solution to the problem, for they either cover only a minority of affiliates to the retirement security system, or as in the case of the minimum return requirement, distort the capital market and compound AFPs' incentives to hold virtually identical investment portfolios (Lasaga and Pollner, 2003). 95. There are also strong equity arguments for restoring a poverty prevention pillar. First,the fiscal costs of subsidies to Peru's remainingpublic second pillars, as well as the first pillar guarantee covering a minority of affiliates in the SNP/PAYGO and older affiliates in the SPP/AFP branch, are financed by all current and future tax-payers, including those without any rights to protection. Second, a poverty prevention pillar i s a critical component of the new social contract implied by the multi-pillar model, as poverty inold age i s a risk that only the Government can cover. 96. The arguments in favor of a poverty prevention pillar are not only based on greater equity. There are also efficiency arguments for a poverty prevention pillar. First, as argued by proponents of the multi-pillar model, a poverty prevention pillar will lead to a more efficient diversification of investment and longevity risk. Second, a minimum guarantee would provide greater incentives to workers at the lower end of the income distribution to participate in the retirement security system. Currently, these workers have little to gain from contributing to a costly system. Third, along with formal cover against the risks that arise with old age are bundled insurance policies that cover more immediate risks such as disability and untimely death which can prove more catastrophic to poorer households. Fourth, a minimum pension guarantee would allow the Government to eliminate the current minimumreturn requirement inthe SPP/AFP branch that has a similar objective, but which has lead to costly capital market distortions. And finally, by clearly defining a poverty prevention promise ex-ante before the first substantial wave of affiliates to the SPP/AFP begin to retire, the Government can shield itself from the fiscal costs of bowing to populist demands ex-post from a wave of retirees receiving lower pensions than they expected to earn from their individual retirement accounts. 97. However, public guarantees come at a cost, and Peru's Government must operate under very tight fiscal constraints. The fiscal cost of offering a minimum pension guarantee (MPG) similar to the guarantee in Chile to all SPP/AFP affiliates (assuming no 47 other change to the system since the January 2002 reforms), i s shown in Figure 13 as the same indicator used in previous sections - net Government expenditure on pensions (in this case, adding minimum pensions to SPP/AFP affiliates) and recognition bonds. As shown, the additional cost of the minimum guarantee grows steadily reaching a peak of 0.78 percent of GDPin 2064. Figure 13. With noadditionalreformsto the pensionsystem, the cost of extending the minimumpensionguaranteeto all SPP/AFPaffiliatesis considerable Net Government Expenditure (Benefits Contributions +Recognition Bonds) - 1.20% 0Difference +With 1.OO% Mffi -a- WithoutMffi h 8 Y n 0.80% n 0 v) J n 5 'c 0.60% w m CI : C 0 n 0.40% 0.20% 0.00% 7 - 7 7 , ,I , ,I I I , , , -I- 98. Where can the Government find the resources to finance a minimum pension guarantee? Since the introduction o f the private second pillar in December 1992, observers have called for the SNPPAYGO second pillar to be closed (World Bank, 1992, 1998 and Kane, 1995). The principal argument for closing the public second pillar to new entrants i s that it remains an uncertain contingent liability that i s difficult for the Government to predict and thus threatens Peru's fiscal stability. However, clauses in Peru's Constitution have foiled every effort to close the SNPPAYGO to date.27 Rather 27 Further, there is some reluctanceto close the public second pillar forfear of turning away relatively more risk averse affiliates who mightprefer a PAYGO defined benefitplanover defined contributionsinprivate individual accounts. Barr and Packard(2003) show that in fact risk averse affiliates to the pensionsystem are more likely to chose the SPP/AFPindividual accounts over the SNPPAYGO - an outcome the authors attributeto lack of confidenceinpublic administrationin Peru. 48 than persist in pushing a Constitutionally impossible reform, the Government opted in 2002 for aligning SNP benefits more closely with affiliates' history of contributions, and cutting Cedula Viva benefits. As shown in earlier sections, while this has lead to some savings, a substantial subsidy to Peru's public second pillars remain. As the Government struggles to find sources of financing to restore the poverty prevention pillar, it i s important to analyze the fiscal savings that could be had by all options for reform, even those prevented by the current Constitution. Figure 14 shows the fiscal impact of closing the SNP to all new entrants to formal sector employment, makingthe SPP/AFP the only second pillar option (the Figure does not yet show the cost of a minimum pension guarantee), Figure 14. Closingthe SNP to new entrants would lead to some savingsinthe very long term, but increasecurrent annual Government expenditure considerably Net Government Expenditure (Benefits Contributions + Recognition Bonds) - 1.20% 1.OO% 0.80% 0.60% 0.40% 0.20% 99. As shown inthe figure, although there would be some long-term fiscal savings to be had from closing the SNPPAYGO to new affiliates, these are not realized until 2066. Since the Government would no longer be able to count on the contributions of new entrants to help finance the benefits of SNP retirees and the acquired rights of current 49 affiliates, this measure would entail a significant increase in up-front annual expenditure on benefits and recognition bonds by as much as 0.30 percent of GDP. 100. In Figure 15, we show the fiscal impact (same indicator) of progressively radical reform proposals in addition to closing the SNPPAYGO to new affiliates, including forcing current affiliates 55 and younger to switch to the private pillar, and forcing all affiliates regardless of their age to take up individual accounts. The fiscal cost of each of additional measurer i s shown cumulatively to provide an idea of the marginal gains from each specific reform. Figure 15. Closingthe SNP completely has different fiscal implications, depending on how the transition to a single second pillar is timed. However the up-front costs are substantial. Net GovernmentExpenditure(Benefits Contributions Recognition Bonds) - + 140% +1990's Reforms 43- 2002 Reforms 120% +All new entrantsto SPP/AFP *All new entrants & SNP affiliates under55 to SPP/AFP All new entrants & SNP affiliates to SPP/AFP 100% -: 8 -L .m 2 o 80% 0 3 % 2 060% 0 40% 0 20% 0 00% 88$8,Q"@~',d.",d?' @''@@@%Q@$ $@,..',..",..",..' %Q",$'&',$'+$ @@@@@"&6@4,Q5'&'@6'$'& &'&'&&' Source: PROST Simulations in Pallares (2003) 101. The most controversial reform option simulated in Figure 15 i s a complete closure of the SNPPAYGO that would force all affiliates into the private second pillar (light blue trend line with white markers). While this measure would immediately double transfers to the SNPPAYGO, the difference with current expenditure would fall to roughly a third in 2007 and disappear by 2016 after which the Government would rapidly enjoy substantial annual savings. However, what makes this avenue fiscally difficult for the Government, i s that it would have to still face these up-front costs while paying subsidies of roughly the same annual amount to Cedula Viva. 50 102. This said, the different options for closing the SNP cannot be dismissed outright. This is because in introducing a minimumpension guarantee, the Government would be re-defining its responsibilities for covering the risks to income from old age. Thus rather than simply closing the SNP second pillar, the Government would essentially be further scaling the SNPPAYGO system down into a true contributory first pillar for all affiliates. Financing the relatively modest promise of a first pillar minimum pension (and the transition away from financing second pillar pensions) changes the simulated picture of the fiscal costs of each option. Figure 16 shows the additional combined Government expense of each of the options presented in the previous Figure 15 (Le. in each scenario, the additional expenditure with respect to current net expenditures after 2002 reforms without a minimum pension guarantee), adding the costs of a contributory minimum pension guarantee for all affiliates to the SPP/AFP pillar. Put simply, each trend-line shows the additional fiscal cost of each of the reform measures (closing SNP to new entrants, etc.) and a minimumpension guarantee. Figure 16. Simulations of the additional expenditure necessaryto offer a minimum pensionguarantee ineachof the reformscenarios, indicate that closingSNP to new entrants and shifting affiliates under 55 to SPP/AFP would be the leastexpensive course Additional Cost of MPG in Each Scenario 1.OO% 0.90% 0.80% h E$? v 0.70% 8 In 5 0.60% n r 0 & 0.50% CI 0 Q 0.40% n 0.30% , 0.20% 0.10% 0.00% Source: PROST Simulations in Pallares (2003) 103. Are the fiscal savings needed to restore a poverty prevention pillar to be found by further cuts to Cedula Viva benefits? Figure 17 shows the impact of various measures on the size of Government subsidies to the special retirement regime, including raising the 51 affiliates' contribution rate from the current 6 percent, to 13 percent (the rate paid by affiliates to the SNP/PAYGO, under the rationale that contributions to the remaining public second pillars should be the same), and to 20% (the rate of contribution our simulations show would be the required to balance the SNP/PAYGO for the entire simulation horizon). The most radical measure would be for the Government to close the Cedula Viva entirely, forcing all contributing affiliates into the SPP/AFPbranch (the only "structural" reform we have simulated). Figure 17. Parametric and even structural reforms to the Cedula Viva lead to only minor savings Net Government Expenditure (Benefits Contributions + Recognition Bonds) - 1.80% 1.60% -Cedula Viva (2002 Reforms) 1.40% -All CV Affiliates move to SPP/AFP h v8 Increase CV contributionsto 13% n 1.20% aIn n Increase CV contributionsto 20% ti = 1.00% -Taxing CV Pensions n c f& 0 0.80% Q) 2 0.60% 0.40% 0.20% 0.00% 104. Pallares (2003) presents the simulated impact on Government subsidies in greater detail, the effect of each of these measure on the distributional structure of the SNPPAYGO and Cedula Viva, and the subsequent cost of providing a contributory minimumpension guarantee. 105. However, providing a minimum pension guarantee would represent a step in the right direction, only if accompanied by aggressive reforms to bringdown the costs of the SPP/AFP branch. To illustrate how critical SPP/AFP performance can be, Figure 18 shows the cost o f providing a contributory MPG using different assumptions about AFP returns from investment. Assuming returns o f 4 percent instead of 6 percent doubles the long term costs of the guarantee. 52 Figure 18. Fiscalcosts of a minimumpensionguaranteewill depend criticallyon the performanceof the privatesecond pillar (Fiscal cost of MPG for all SPP affiliates, top line assumes AFP returns of 4%, and bottom line 6%) 0.00% ~ I I I I I I I I I I I I ' 1 1 1 1 I I I I I I ' I l l l l l> I I I 1I I I l l I I 1 I I I I 1 I II l l 1 1 , I I I 1 II I 1I Source: PROST Simulations in Pallares (2003) 106. But a contributory minimum pension guarantee similar to that found in Chile, Colombia and Argentina may not be the best use of Peru's limited public resources. Given current rates of participation in Peru's pension system, conditioning eligibility on some minimumlevel of contribution may exclude too many from cover -even among the current number of affiliated workers. 107. Figure 19 shows the declared combined contribution density of men and women affiliated to all branches of Peru's retirement security system. The horizontal axis represents the contribution density required to be eligible for the minimum pension currently guaranteed to older affiliates in the private pillar, i.e. 20 years of contributions. Thus the percentage of affiliates represented by the bars lying below the horizontal axis are those men and women who are not likely to have the contribution history required to qualify for minimumguarantee structured along the same lines. 53 Figure 19. Given current contribution patters, 30% of affiliated men and 40% of affiliated women are not likely to qualify for a contributory minimumpension guarantee2' (Reported contribution density of affiliates to Peru's pension system) 1,oo 0,90 Affiliated h n w .-a 0Affiliated C 0,80 . 5 In P C 0,70 5 In P C 0,60 .-C P) -a=I CI 0,50 CI L C 0 0,40 .- -2- v) 8 C 0,30 .-0 Y C c 3 0,20 g c L C 0 0 0,lO 0,oo 10 20 30 40 50 60 70 80 90 100 %of Affiliated Workers (Deciles by Contribution Density) Source: Barr and Packard (2003) 108. A contributory minimumpensionguarantee is unlikely to cover even a substantial number of affiliated workers. Given the relatively large number of worker who are not even affiliated with the system, financing the cost of the guarantee out of taxes like Peru's VAT raises some of the same equity issues the Government is already grappling with in financing the large subsidies the SNPPAYGO and the Cedula Viva. Given the large share of the workforce that are not participating in any branch of the system, and the substantial additional public cost of a minimum pension guarantee shown in our simulations, it would be more equitable for the Government to opt for a non-contributory benefit to meet the objective of covering at least the risk of poverty inold age. 28The contribution densities shown are a cross section. Affiliates of different ages are grouped together by contribution density. However, it is likely that workers may increase their contribution density as they age, and even as they approach retirement age. However, plotting contribution densities by age does not reveal a clearly increasing pattern, and regression analysis shows that there affiliates may even perceive diminishing marginal returns from contributions, leading many to slow their contributions well before retirement age (Barr and Packard, 2003). 54 109. Policy makers often hesitate to consider non-contributory pensions because of the very real dangers of moral hazard and potentially ballooning fiscal costs. However, the danger of moral hazard - that is, that providing a non-contributory benefit for the elderly would eliminate household incentives to contribute to the pension system or to save outside the system - can be loweredby: (i) offering a much lower benefit that the average contributory pension; (ii)making the benefit available at a much later age than the retirement age for the contributory pension; and (iii) accurately targeting the benefit to the elderly poor, or making the benefit taxable at source to claw back expenditure from all but the most needy. An additional factor that will determine the fiscal costs of a non- contributory benefit that lie within a Government's control, i s how the benefit i s indexed, whether to the growth in current wages or to changes inprices. 110. In Figure 20 we show the simulated fiscal cost of offering a non-contributory pension, equal to approximately half of the minimum pension in the SNPPAYGO (roughly equivalent to Peru's poverty line) to the entire population over a certain threshold age (80, 75, 70, and finally 65). The top half of the figure shows the cost of indexing the benefit to wages, while the bottom half shows the considerably cheaper option of indexing the benefit to prices. It i s worth noting in our simulations that the annual cost of a non-contributory benefit, generously paid to the entire population over 65 (assuming such a benefit were indexed to inflation), grows only `slightly higher than current annual subsidies to the Cedula Viva. 55 Figure 20. There are many optionsfor structuring a non-contributory benefit, with different fiscal imdications (Projected fiscal costs of flat universal pension benefit, percentage o f GDP) I (a. indexed to changes in wages) (b. indexed to changes inprices) 2,0% 1 -cost for 6% +cost for 70t 1,8% - -cost :: costfor7.5 for 80t 1,6% 1,4% 1,2% 1,O% O,8% 56 111. Inconsidering the fiscal costs of a non-contributory guarantee simulatedinFigure 20, readers should remember that our simulations show the costs of providing a flat benefit to everybody above a minimum age. We have shown the cost of this very generous policy to provide a conservative upper-bound cost estimate of a non- contributory poverty prevention benefit. However, the costs would be considerably lowered by targeting the benefit to the elderly poor, currently 40 percent of the population over 65 in Peru. The cost of a targeted benefit to the poor over 65 would thus be 40 percent of that shown in Figure 20 for the population over 65. Infact, this exercise can be done for each o f the age thresholds provided, as the latest household income and expenditure data show that the poverty rate does not vary much from 40 percent in the higher age cohorts of the elderly. The estimated cost of a targeted benefit are shown in Figure 21. Figure 21. Targeting a non-contributory poverty prevention pensionto the elderly poor, would lower the estimatedfiscal cost of the new pillar substantially 1.00% , (Cost of Targeting Non-ContributoryBenefitto Poor) 0 90% +Poor 65+ 0.80% Poor 75+ Poor80+ h 0.70% v n 0 C3 0.60% n 0.50~~ 5 r 0 3 Q 0.40% C 22 0.30% 0.20% I 0.10% Source: PROST Simulations in Pallares (2003) 112. However, means testing to target the benefit efficiently comes with a host of complications and costs. Means tests increase administrative costs and provide opportunities for corrupt behavior on the part of public officials (World Bank, 1994, Willmore, 2001a). 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 (Hubbard, Skinner and Zeldes, 1993 and 1994) as well as continued work in 57 old age. Finally, means tested benefits are often regarded as charity, which reduces their political appeal, makes the benefits vulnerable to budget cuts, especially in economic downturns (Snyder and Yackovlev, 2000), and may discourage eligible applicants dissuaded by social stigma (Barr, 1992, Willmore, 2001b). 113. There are costs and benefits to different program structures that the Government will have to consider in establishing a fiscally sustainable poverty prevention pillar. As discussed in greater detail in recent review of non-contributory pensions by the L O (LO, 2002), these poverty pension programs have had an important, positive impact. Although there i s still room for improvement inextending coverage and controlling fiscal costs, the case studies compiles by the L O show that in 2000 and 2001, non-contributory pensions lowered the rate of poverty among the elderly by 67% in Argentina, 95% in Brazil, 69% in Chile and 21% in Costa Rica. Given the relatively high rates of poverty among the elderly in Peru, and low rates of coverage of contributory programs, a non- contributory poverty prevention pension pillar would be a welcome addition to the country's safety net. 58 V. Charting the Way Forward: Policies Consistent with Covering Peru's HouseholdsAgainst Poverty inOld Age 114. The Government's present policy course - making individual savings the mainstay of the old age income security system, covering the ever-frequent loss of earnings ability - i s sound and accords well with risk management principles. However, if individual savings in private retirement accounts is to continue as the largest component of Peru's formal retirement security system, the private second pillar i s in urgent need of deep structural reforms to bring down the price paid by affiliated workers and to further improve investment performance. 115. The urgency of reforms to Peru's private second pillar i s difficult to exaggerate. The current high price and volatile performance of the SPP/AFP pillar, not only increase the likelihood of affiliates earning disappointing pensions, but leave the substantial progress achieved since 1992 dangerously vulnerable to political attack. The course taken by many pension reforms in Latin America i s being scrutinized- even reconsidered-in countries within the region. The principal concern, if unfairly placed, i s the alleged failure of the new model to increase coverage of formal retirement security, which was considered an important selling point of reforms when they were initiated (Gill, Packard and Yermo, 2003). 116. In late 2002, in the process of rewriting Peru's Constitution, articles that would have allowed affiliates to the private second pillar to return to the public PAYGO branch, and lowered the retirement age from 65 back to 60 years, were only narrowly defeated. The Government i s still fighting the prospect of reform legislation that could lead to a return of affiliates to the SNP en masse. This fiscal threat i s surpassed only by the constant attempts of organized workers in the public sector to gain entry to the still restrictedbut highly attractive and tempting parameters of the Cedula Viva. 117. But the justification of reforms to the SPP/AFP pillar extend well beyond responding to current political challenges and the fiscal risks that they pose. The retirement security system i s not doing its job, and Peru's vulnerability to the risks that arise with ageing seems to be increasing. Our simulations show that those coveredby the private second pillar can expect replacement rates just over 40 percent, under optimistic assumptions of contribution history and investment performance. And the coverage gap seems to be widening. Affiliation rates have fallen between 1999 and 2001 b y 3.1 among the urban workforce, and even more sharply among the poorest quintiles (by 42.5%). Further, there has been a sharp decline in the share of contributing workers (-25.6 percent) in the bottom quintile, which has been most severe (-33 percent) inLima. From 1999 to 2001 the distribution of benefits has become more regressive. The share of elderly receiving pensions in the poorest quintiles dropped dramatically. In Lima alone, the share of elderly receiving pensions in the first quintile dropped by 32 percent. This was contrasted by an increase in share of elderly in quintiles 4 and 5 receiving pensions of 32 percent and 11.5 percent, respectively. While coverage indicators vary with changes that lie beyond changes to the pension system - even beyond the control of policy makers - a number of the reform options presented in this report could make the system cheaper and better performing, and ultimately more attractive to workers. 59 118. Encouragingly, steps are already being taken to improve the performance of the private second pillar along some of the lines recommended in this report (see again, Table 10). While not initiated by the Government, a legislative proposal, recently approved by Congress, would allow Peru's AFPs to offer multiple funds with differing riskheturn characteristics, as well as increase the statutory limit on AFP investment abroad from the current 10 percent to 20 percent of portfolio^.^^ The approved measure still needs to be implemented with the necessary accompanying regulation, and the Central Bank - still wary of allowing the AFP's to invest a larger share of their funds abroad - has yet to announce any change in the effective portfolio limit. Additionally, recent changes to the process by which SPP affiliates are covered against the risk of untimely death and disability - abandonment of the "regimen temporal" - i s likely to increase choices and spur competition in the insurance indu~try.~' Furthermore, the Government i s preparing a Presidential Decree that would make the recognition bonds for the past contributions of affiliates who switch from the SNP to the SPP tradable. This would not only increase the set of investment instruments available in the capital market and help the AFPs to better diversify their portfolios, but also act as a bench mark for the develo ment of a yield curve in a market that i s still lacking a sufficient issue of treasury bonds. P, 119. The measures already taken by the Government should have a lasting positive impact on the SPP system, by increasing competition, and hopefully lowering AFP fees. New rules for pricing insurance contracts designed to provide coverage against the risk of untimely death and disability have had an impact, with premium reductions of approximately 30 per cent. Given that there i s still room to increase competition, the Government i s encouraged to continue implementing market friendly measures to bring system costs and service fees down. However, if the private pillar continues to exhibit characteristics of an oligopoly, and if some of the "easier", less controversial options for reforms Peru's private second pillar presented in Table 10 fail to lower prices, serious consideration must be given to stronger measures. Preference should be given to market friendly measures, but in the face of widening AFP profit margins and persistently high fees; alternative pricing mechanisms may be adopted. 29 Law No. 27988, published on June 4, 2003, requires that the fund managers in the private second pillar offer affiliates at least two different investment-fundoptions for their mandatory savings, and gives the AFP's the freedom to offer a larger selection of investment funds for affiliates' voluntary contributions. 30 ResolutionNo. 900-2003 implements a new set of rules for the administration of the private second pillar's disability and life insurance. Prior to these new rules, affiliates could only receive insurance from a private provider selected by their AFP. This new set of rules gives AFP affiliates greater choice in the selection of their insurance provider, and forces private insurance providers to compete to offer the best price and insurance products (in Soles or in dollars). This change has already lead to a fall in insurance premia in the private second pillar. From November 1, 2003 the premia charged to affiliates of AFP Integra is 35% less. 3'Provided affiliates are fully informed of the risk trade-off from "cashing in" their inflation indexed recognition bonds - essentially exchanging policy risk for marketrisk- this should be a positive change. There are many who would benefit from holding the inflationindexed bonds. 60 120. However, strengthening the private second pillar through reforms cannot dominate the attention of policy-makers and the pension specialists who advise them. If Peru's vulnerability to the risks to income that accompany ageing is to be lowered, all of the pillars of old age income security must be restored. The demands of designing and regulating a new private second pillar may have distracted authorities from attending to the remaining pillars of old age income security. As argued in this report many of the options we present for reforms to the private second pillar, will give a healthy boost to the voluntary third pillar, but only if two critical changes come about. First the mandate to save privately should be lowered with a ceiling on wages subject to SPP/AFP contributions - a feature of most private second pillars in the region. Second, households have to be given incentives to save (both as incentives to comply with second-pillar mandates, as well as to save voluntarily) through preferential tax treatment for the money they set aside for retirement, inline with international best practice. 121. When the development of voluntary savings and insurance is proposed as a critical policy for lowering vulnerability to the risks to income from old age, most policy makers point to the current low take-up of voluntary instruments as evidence that a third pillar is not really viable due to averagelow incomes inPeru and elsewhereinthe region. Many argue that most Peruvian's are too poor to save voluntarily. However, this i s a false argument. Low take-up of voluntary instruments under the current set of institutions i s hardly evidence that the average Peruvian household would not save voluntarily if given the opportunity and the incentives to do so. In this report we show that workers who are free to choose whether and how to insure, secure income in old age in a number of ways, including the formal pension system, but additionally by investing inproperty andintheir children. 122. Furthermore, with respect to voluntary savings for old age, policy makers need to consider the following question: when in recent history has the voluntary pillar really been given a chance? Prior to reforms in 1992, the unsustainablepension promises of a large public pillar were sufficient for most workers in the formal sector, providing little incentive to save more. Since the 1992 reforms, the large portion of income the Government mandates workers to save, the relatively large portion spent on management fees to an oligopoly, and the unfavorable tax treatment for both mandatory and voluntary savings, make saving for retirement in a regulated instrument incredibly unappealing. This leaves workers with few other options but to evade what is still in many ways a taxing formal pension system, to perhaps save and insure for old age with less efficient informal instruments. The Government has an enormous fiscal incentive to give formal voluntary arrangements every chance to succeed, for every dollar Peruvian households save and invest voluntarily may lower their vulnerability to the risks to income that accompany old age, andlower the eventual fiscal costs of apoverty prevention pillar. 123. This last point brings our discussion to what is probably the single most important policy reform the Government will need to consider - restoring a poverty prevention pillar for all affiliated to the retirement security system, and/or one that covers all Peruvian's against the risk of poverty in old age, regardless of whether they have participated inthe system or not. 61 124. This report has presented arguments for restoring a poverty prevention pillar, pointing to the benefits of such an institution from both an equity and an efficiency perspective. While a contributory "pillar one" would be a step in the right direction (if eventually followed by steady increases in the share of the work force acquiring rights toward the benefit), a non-contributory "pillar zero" i s probably more appropriate. Many workers affiliated to the retirement security system would not qualify for a contributory guarantee along the current parameters, and most workers are not even affiliates. Thus a modest, non-contributory social assistance benefit may not only be the solution that i s fiscally viable, but also the only solution that will really protect the largest number of Peru's households from the more serious of the two risks policy makers seek to cover - old age poverty. But, as our simulations and the history of pensions policy in the region show, when it comes to designing a sustainable poverty prevention pillar, the proverbial "devil" i s in the detail. An important step would be for the Government to immediately invest analysis inthe design and implementation of a poverty preventionpillar. 125. Inthis report, we have initiated this process of looking for sources of financing to restore a poverty prevention pillar, by simulating the fiscal savings from further structural reforms to Peru's remaining public pension plans. We have shown that the simulated fiscal savings from closing SNP (either just to new affiliates or to current contributors as well) entail daunting up-front costs, and would be politically (to be more specific, Constitutionally) very difficult. The difficulty o f meeting these up-front costs cannot be ignored, however the possible long-term savings from closing the SNP in some degree (and with an eye toward sequencing to contain the annual expenditure on pensions in any given year) should also not be disregarded outright. 126. Inaddition to closure of the SNP, the Government can look to the Cedula Viva as a possible source of financing for a poverty prevention pillar. Since a Constitutional Tribunal made up of beneficiaries of the special regime i s likely to rule on changes to Cedula Viva benefits, any change will be difficult, but not impossible. An idea being considered by the Government i s to tax Cedula Viva pensions above some threshold - say S/. 1,000 per month - and use the proceeds of this tax to constitute a properly managed and regulated fund to finance a new poverty prevention benefit. Investment o f the fund could be out-sourced to a private entity, and the risks to old age income security might even be further diversified by requiring that the fund be invested largely inforeign securities. These ideas need considerable more thought to be properly and fully developed, but their objective is laudable. 127. Although controversial, taxing Cedula Viva pensions could be justified by the generosity of the benefit parameters (particularly benefit indexing to current civil service salaries), and the less-than-fully-transparent manner in which many in the regime acquired rights. Further, this initiative need not be presented as an "attack" on the beneficiaries of Cedula Viva alone, and its impact could be spread over a wider group, by following the earlier recommendation of altering the tax treatment o f pensions, from taxing contributions (TEE) to taxing pension benefits (EET). Figure 22 shows the simulated fiscal gains from taxing Cedula Viva pensions (compared with raising the contribution rate). 62 128. As shown in Figure 22, the savings from increasing contributions from active Cedula Viva affiliates are meager. The resources that could be clawed back by the Government by taxing Cedula Viva pensions are also meager when compared to the cost of providing a non-contributory benefit targeted to the elderly poor. The resources from a tax on Cedula Viva pensions would only be sufficient to finance a non-contributory pension to the poor 80 years and older. Although this specific proposal i s far from ideal, and would leave too many elderly inpoverty, the Government's initiative ininvestigating options for a poverty prevention pillar i s very welcome. Figure 22. Savingsfrom taxing Cedula Viva pensionsabove S/.l,OOO will not cover the likely costs of a poverty preventionpillar, but could be part of a solution Source: PROST Simulations in Pallares (2003) 129. In any case, provided that a poverty prevention pillar is restored in Peru, the Government should move toward closing its public second-pillar systems. This may be difficult fiscally, but it may be critical from a politicalperspective. Our recommendation extends to closing Cedula Viva completely, even to judges and public prosecutors. For while the public second pillar systems remain, the exact nature of the Government's responsibilities toward income security in old age will continue to be ambiguous, and vulnerable to expensive populist interpretations. The attempts to restore a full public second pillar late in 2002 are an illustration of this risk. The abusive rent seekingthat has extended Cedula Viva rights in spite of the system's "closure" in 1992 i s another illustration. 63 130. Clearly defining the Government's responsibilities to income security in old age - that is, providing insurance against poverty and helping households to smooth consumption through the regulation of safe investment and insurance instruments - and cementing this "social contract" by restoring Peru's multi-pillar system, would go far in protecting the sustainability of the system, and toward covering a larger share of Peruvian households. Table 11lists the policies options that are consistent with this objective, and that the Government should explore in greater depth. The Word Bank is prepared to work closely with the Government to explore these options further. 131. However, few of the policy options presented can be effectively explored or eventually implemented without stronger leadership in pensions policy than currently exists. Old age income security i s a complex arena of public policy, spanning issues from social assistance to financial sector regulation. Several government agencies in Peru have direct and indirect interests and roles in the functioning of the pension system, but these can often clash. Currently, no single institution has a clear mandate to investigate, formulate and coordinate pension policy. This lack of strong leadership may have contributed to the weaknesses ineach pillar identified inthis report. 132. Although among the strongest of Peru's public institutions the Ministerio de Economia y Finanzas (MEF) i s conducting this leadership role by default rather than by design. While MEF currently benefits from the presence of eminently qualified and experienced staff to carry out the leadership role in pensions, this i s a fortunate circumstance rather than a permanent institutional feature. 133. The reform options presented in this report will require more formalized leadership, with a widely recognizable mandate over the agencies involved, and will need to be guided by a medium and long term vision of the multi-pillar pension system that would best serve Peru's households. More important than any single reform measure among the options presented in this report, is the need for Peru's Government to invest energy and resources in the establishment o f this long-term vision for restoring the multiple pillars o f its old age income security system. The World Bank will continue to work closely with the Government to develop this vision and to take the more immediate steps required for its achievement. 64 lA References AFP Horizonte,CincoArios del SistemaPrivado de Pensiones,June, 1998,Peru Asociacidn Interamericanade Organismosde Supervisidn de Fondosde Pensiones(AIOS). Boletin (1999- 2002). BARR, Abigail & Truman Packard , (2002) "Preferences, Constraints and Substitutes for Coverage Under Peru's Pension System" Background Paper for Regional Study on Social Security Reform, Office of the ChiefEconomist, Latin America and CaribbeanRegionalOffice, World Bank BARR, Nicolas(1998), TheEconomics ofthe Welfare State,3`d Edition, Oxford University Press (2000), "Reforming Pensions: Myths, Truths, and Policy Choices", IMF Working Paper WP/OO/139, DevelopmentIssuesDivision,Washington DC (2001), The Welfare State as Pinay Bank, Oxford UniversityPress CELADE (2003) BoletinDemografico.Santiago. CONASEV, Acceso a Financiamiento a Trave`sdel Mercado de Valores, Resultadode Consulta Ciudadana,Peru, May 25,2001. FERNANDEZ-BACA, Jorge, B. Seminario, R Urrunaga,P. Pollarolo,La Reduccion de 10s Costosy el Incremento de la Competencia en el SistemaPrivado de Pensiones,Centro de Investigacih, Universidad delPacifico,Lima, February 29, 2000. GILL, Indermit, Truman Packard and Juan Yermo, (2003) "Keeping the Promise of Old Age Income Security in Latin America" Regional Study on Social Security Reform, Office of the Chief Economist, LatinAmericaand CaribbeanRegion, The World Bank GrupoMacroconsult,Informe Final de la ConsultoriaPara el Desarrollo del Mercado de Capitales Peruano, Peru,August 2000. HERRERA and Javier (2003) "Governance, corruption, citizen participationand poverty in Peru, 2002". Institut de Recherche pour le Developpement, Paris, and Instituto Nacional de Estadistica e Informhtica, Lima. (mimeo). HOLZMANN, Robert, Truman Packard & Jose Cuesta, (2000), "Extending Coverage in Multi Pillar Pension Systems: Constraints and Hypotheses, Preliminary Evidence and Future Research Agenda" in R. Holzmann& J. Stiglitz,Eds. New IdeasAbout Old Age Security, The World Bank ILO, (2002), Pensiones No-Contributivas y Asistenciales :Argentina, Brazil, Chile, Costa Rica y Uruguay. Eds.Fabio Bertranou, C. SolorioandW. Van Ginneken, Santiago JAMES, Estelle, J. Smalhout, and D. Vittas,Administrative Costsand the Organization of Individual Retirement Account Systems, Policy ResearchWorking Paper 2554, The World Bank, February 2001. KANE, Cheikh, T. (1995) "Peru: Reforming the Pension System", ESP Discussion Paper Series, No. 17361,World Bank LASAGA, Manual and John Pollner, (2003) "Peru's Private Pension System: An Analysis Of Its Evolution, Current Market Structure And RecommendationsFor Reform" BackgroundPaper for Analytical and Advisory Assistance to the Government of Peru, Social Protection Sector Unit, Latin America and CaribbeanRegional Office, World Bank. LEFORT, F. and Walker, E. (2000a), Pension Reform and Capital Markets: Are There Any (Hard) Links?, Working Paper,UniversidadCatdlicade Chile. 66 PACKARD, Truman, (2002a) "Pooling, Savings and Prevention: Mitigating the Risk of Old Age Poverty inChile" BackgroundPaper for RegionalStudy on Social Security Reform, Office of the ChiefEconomist, Latin America and CaribbeanRegional Office, World Bank (2002b) "Reforming Social Insuranceand the Incentivesto Self Insure:Essays on Coverage of Old Age Income Security Systems InLatinAmerica" Ph.D. Thesis, Departmentof Economics, Universityof Oxford. (2001), "Is There a Positive IncentiveEffect from Privatizing SocialSecurity?Evidencefrom PensionReformsin Latin America", World BankPolicy ResearchWorking Paper, No. 2719, Washington, DC PACKARD, Truman, Naoko Shinkai and Ricardo Fuentes, (2002) "The Reachof Social Security inLatin America and the Caribbean", Background Paper for RegionalStudy on Social Security Reform, Office of the Chief Economist, Latin America and CaribbeanRegionalOffice, World Bank PALLARES, Monsterrat(2003) "Simulating Options for Reforms to Public Pensionsin Peru" Background Paper for Analytical and Advisory Assistance to the Government of Peru, Social Protection Sector Unit, LatinAmerica and CaribbeanRegionalOffice,World Bank. PrimAmericaConsultores (1999a), Diagndstico Respecto a las Regulaciones de la Superintendenciade AFP y de la Prdcticas Intemas de las AFP que Inciden Negativamente sobre 10s Costos Operacionales del SistemaPrivado de Pensiones,Lima, November 1999. PrimAmericaConsultores (1999b),Resumende las Regulaciones Vigentesde la Superintendenciade Administradoras de Fondos de Pensiones en Peru queMayor Incidencia TienenSobre 10s Costsode Operacidn, November 1999,Peru. Superintendenciade Bancay Seguros, MemoriaAnual2001, Peru. WHITEHOUSE, Edward(2000a)Administrative Charlesfor Funded Pensions: An International Comparison and Assessment, Social ProteccihDiscusih Paper Series, No. 0016, The World Bank, June (2000b), "How Poor are the Old?A Survey of Evidence from 44 Countries" Social Protection DiscussionPaper No. 0017, WashingtonDC. WORLD BANK (1992) Memorandumof the PresidentIBRD: StructuralAdjustment Loanto the Republic of Peru, ReportNo. P-5714-PE WORLD BANK (1994) "Averting the Old Age Crisis" Oxford University Press WORLD BANK (1998) ImplementationCompletionReport (ICR) for a PensionReformAdjustment Loan to Peru, ReportNo. 18048 WORLD BANK, (2002) Memorandumof the Presidenton ProposedProgrammatic Social ReformLoan 11, August 19,2002, report numberP7514-PE; YERMO, Juan, (2002) "The Performance Of Funded Pension Systems In Latin America", Background Paper for Regional Study on Social Security Reform, Office of the Chief Economist, Latin America and CaribbeanRegionalOffice, World Bank ZVINIENE, Asta & Truman Packard, (2002) "A SimulationOf Social Security Reforms in LatinAmerica: What Has Been Gained?", BackgroundPaper for RegionalStudy on Social Security Reform,Office of the Chief Economist, Latin America and CaribbeanRegionalOffice, World Bank 67