+%120 Country DepiL Latin Almerica Region Economic Notes Reforming Social Security: Lessons from International Experience and .. .. .. . . . Priorities for Brazil* Indermit 5. i.. May /997 The World Bank Country Dept. I Latin America and the Caribbean Region Economic Notes REFORMING SOCIAL SECURITY: LESSONS FROM INTERNATIONAL EXPERIENCE AND PRIORITIES FOR BRAZIL* Indermit S. Gill May 1997 The World Bank * This note is based on a presentation by the author at the International Seminar on Social Security in Brasilia on April 24th, 1997, which summarized the proceedings of the first day of the seminar. Angela Furtado provided invaluable assistance in preparing both English and Portuguese versions of this summary. Useful discussions with Lajos Bokros, Estelle James, Cheikh Kane, David Lindeman, Antonio Magalhaes, Olivia Mitchell, and Anita Schwarz are gratefully acknowledged. I. INTRODUCTION 1. This note presents, in summary form, Constitution Committee, consisted of a panel the main points of the first two sessions of a discussion by Senator Beni Veras, the seminar on social security held in April, relator of the social security reform bill, 1997, at the Senado Federal in Brasilia, senior officials from the Ministry of Social organized jointly by the Senate team on Security in Brazil and Argentina, Brazilian Social Security Reforms, and Instituto de researchers, and World Bank staff. Pesquisa Economica Aplicada (IPEA). The seminar had three sessions. The first session, 2. This note is not a complete summary which was chaired by Mr. Antonio Carlos of the presentations and discussions during Magalhaes, President of the Senate, the seminar. Also, it does not discuss the consisted of an overview of international merits of the current senate proposals, and experience with social security reforms and their adequacy in responding to the problems the analyses of the rationale for and nature of facing Brazil. It also does not discuss the these reforms in three countries (Argentina, problem of company or "closed" pension Hungary, and USA), which are at different funds, since this was not discussed in detail stages in reform. The second session, which at the seminar. Nevertheless, it does provide was chaired by Mr. Reinhold Stephanes, the main messages emerging from the Minister for Social Security, consisted of seminar, both in terms of the urgency of the diagnoses of Brazil's own private and public social security crisis in Brazil, and the ways pension systems, from the viewpoint of in which other countries are dealing with actuarial disequilibria, labor market similar problems. We hope this will be inefficiency, and inequality. The third useful to policy makers and other observers session, chaired by Mr. Bernardo Cabral, in discussing and designing remedies. President of the Senate Justice and Reforming Social Security: Lessons from International Experience and Priorities for Brazil 2 II. INTERNATIONAL EXPERIENCE 11.1 Overview of Social Security Reforms 3. Countries all over the world are RLa The Latin American Model reforming their social security systems. The old systems were generally publicly 6. Countries such as Chile (1981), Peru managed, payroll tax financed, defined (1993), Argentina (1994), and Uruguay benefit, pay-as-you-go schemes. The main (1996) have adopted multi-pillar systems, factors prompting these reforms are and several others such as Hungary are unsustainable fiscal imbalances, the need to currently implementing such reforms. There reduce distortions and evasion, the desire to are important differences in the nature of increase savings rates, and the desire to these reforms, though. In Chile, the old mitigate generational or socioeconomic system is being phased out by making it conflicts. The overall objectives of the compulsory for new entrants to the labor reforms are to protect the old and force to enter the new system, while older unfortunate and promote growth. workers have a choice of system; in Argentina and Peru, the old system 4. Instead of relying solely on a (consisting of the first pillar) is optional for intergenerational redistribution scheme for all; in Uruguay the old system was providing for old age, countries are "repartitioned", being mandatory for workers diversifying their old age and disability under 40 years, and optional for those above pension provision by providing some public 40. In all cases, however, when the benefits (e.g., as a fixed universal benefit, or transition is complete, the new system will means-tested transfers to ensure a minimum have three pillars: a defined contribution, pension level) but supplementing them with capitalized, privately managed second pillar, mandatory privately managed savings some form of minimum pension guarantee scheme ("sistema de capitalizacao"). Most which can be thought of as the first pillar, countries already have voluntary pension and a voluntary third private savings pillar. plans managed either by companies or individuals themselves. In pension 7. There can be considerable costs of terminology, these different sources of transition in the shift from an unfunded pay- pension finance and provision are termed as-you-go system to a fully capitalized "pillars of old age income security". A scheme. The problem arises because of the multi-pillar system has three pillars: a tax- need to set aside enough funds to finance the financed mandatory publicly-managed pillar, pension obligations that have been already a mandatory, fully-funded, regulated accumulated, i.e., for current retirees, who privately-managed pillar, and a voluntary receive benefits but do not contribute to the privately-managed fully-funded pillar (see system anymore, and current workers who Figure 1). have contributed and are still contributing. The cost of transition depends on the 5. From the experience of reforming magnitude of these liabilities of the old countries, three models of social security system, which in turn depends on how reform can be broadly identified (at the risk generous the defined benefits were, and how of oversimplifying a more complex pattern): old the population is. Chile, Argentina, and Peru had relatively young populations at the Reforming Social Security: Lessons from International Experience and Priorities for Brazil 3 time of reforms, and not overly generous contributions. The actual transfer of funds to defined benefits, so the costs of transition their accounts would come later. In Peru's were manageable. To further ease the case, the total cost of recognition bonds was transition, Chile and Peru issued "recognition about 3% of GDP. Brazil's relatively young bonds" to currently employed workers who population would imply low costs of switched from the old to the new system transition, but the generosity of defined which formally acknowledged their previous benefits potentially raises these costs. Figure 1 The Pillars of Old Age Income Security Objective Nm ..,. . . . .. . . . . . . . . ... Financing omx4 tdFyud Pillar 1: Pillar 2: Pillar 3: Mandatory Mandatory Voluntary publicly managed privately managed II.1.b The OECD Model "OECD model" of social security reform. The reforming countries are Australia, 8. Countries such as Germany and Switzerland, Denmark and the UK. France have opted so far to maintain the old system (which consisted of the first and third 9. The main differences between the pillars). Reforms are under consideration in OECD and Latin America models of reform some OECD countries such as the US. is that in case of the former - because it was However, a few industrialized countries have small and affordable in the first place - the already made major structural changes, and first pillar remains essentially the same, i.e., their experience may be considered the universal, pay-as-you-go and defined benefit. Reforming Social Security: Lessons from International Experience and Priorities for Brazil 4 In Australia the first pillar is a means-tested approach was pioneered by Sweden in the pension paid from general revenues, in the 1990s. Faced by a growing deficit and an UK it is a small flat benefit, and in increasingly distorted labor market, but Switzerland it is a compressed earnings- unwilling to bear the high costs of transition related benefit. The original contribution of moving to a capitalized system, Sweden rate was also small, so a second pillar could devised a system of "notional accounting". be added on with additional contributions. Countries such as Italy, Poland, and Latvia And because the second pillar was an add-on are considering such a system as well. rather than a substitution, there were no transition costs. For historical reasons, the 12. In this system, contributors have investment decisions of funds of the second notional or virtual accounts that earn a pillar are entrusted to employers and/or labor notional interest rate, which is based on unions, instead of individual workers. economic and demographic projections. But it is still a pay-as-you-go system: while contributors are aware of their account 10. OECD countries such as France and balances, their contributions are in fact used Germany that already have large pay-as-you- to pay current retirees: no pension fund go pension systems and relatively old exists except to bridge the gap between populations, would face large costs of current contributions and benefits. Thus the transition in adopting the Latin American government is not required, as in the Latin (capitalized) model of reform. But these American model, to contribute large sums of countries have also managed to avoid large money to start a fund with which accrued scale evasion of payroll taxes that fund these pension rights are met. The notional link systems. While real pension benefits in these between contributions and benefits is countries have been eroded over time due to believed to eliminate some disincentive unanticipated increases in retirement effects associated with traditional pay-as- duration, being wealthy has enabled these you-go systems: the somewhat tighter link countries to prop up their social security between contributions and benefits may deficits. For all these reasons - none of reduce evasion and annuity calculations can which resemble conditions in Brazil today - be used to help workers understand why these countries have been exceptional in that retiring early will result in low replacement they have not moved away from defined rates. But the system remains pay-as-you- benefit system that characterized older social go, so it does not offer the benefits of the security schemes towards a system relying on Latin American model: there is no increase in defined contribution. As a result, while saving and financial market development and these countries have postponed the transition contribution rates will have to increase as the to a capitalized pillar, they retain the population ages so intergenerational transfers potential distortions of the old system. stay large. HLic The Swedish "Notional Accounts" 13. In summary, international experience Model shows that the common goals of social security reforms are: 11. The third approach seeks to combine the more attractive features (from a political economy standpoint) of the OECD model, * a closer link between benefits and viz., low transition costs, with those of the contributions to minimize distortions; Latin American model, viz., low levels of labor and capital market distortions. This Reforming Social Security: Lessons from International Experience and Priorities for Brazil 5 * including a funded element to generate 1.2 Argentina's Recent Reforms more savings and help growth; 16. The Argentinean social security * long-term fiscal sustainability and reform, which was initiated in 1994, was minimizing intergenerational conflicts by done against the backdrop of a large fiscal eliminating scope for manipulation of the deficit which reached 1.5% of GDP, system for short-term redistributive widespread evasion of taxes that funded the gains, system, and charges that the system was unfair. The successful stabilization plan 14. The only countries that are not initiated in 1990 had exposed the financial currently reforming are some OECD weaknesses of the old system. There was countries, because they have avoided large consensus that the system was financially scale evasion of taxes that fund their unsustainable, inequitable, contained serious systems, because the age distribution of their distortions that led to strong incentives to populations would imply very high costs of evade, and that change was needed. transition, and because they are wealthy However, there was little or no consensus on enough to fund deficits of the system from what the shape and speed of this change general revenues. These circumstances ought to be. By 1994, some general clearly do not fit those of Brazil, where there principles for the new system were agreed are serious concerns about evasion of payroll upon. These were: the system should be taxes (of which social security taxes are the universal, equitable, redistributive, most important) and the resulting high levels transparent and credible, of informality of employment, the population is relatively young (though aging rapidly), The main features of the reform because propping up the existing system were: would drain resources from investments in 0 The new scheme is multi-pillar, funded primary education and health and by a 11% payroll tax on workers, 16% infrastructure, and make it impossible to on employers, and 27% for self- raise saving rates. Even the experience of employed workers. richer, more aged countries - those * The first pillar consists of a universal flat concerned about competing in a global pension (prestacion basica universal - marketplace - shows an increasing emphasis which was a defined benefit of 2.5 times on reductions of distortions that raise the the average contribution (about cost of labor and discourages saving. US$200), upon retirement at age 60/65 Economic concerns about social security are (female/male) after contributing for 30 no longer politically incorrect. years in contrast to Chile, where the 15. The cases of Argentina, Hungary, sppln benits f thendefined- and the USA (which respectively has contributnait pr to enur implemented, is beginning to implement, and everio caiimum level. is still deliberating the course of reforms) illustrate the rationale for and nature of In the new second pillar, which is social security reforms. capitalized, workers have a choice between a public-managed and privately managed system, in contrast to Chile where the second pillar is entirely privately managed. Reforming Social Security: Lessons from International Experience and Priorities for Brazil 6 * Compensatory pension for past much. In Brazil, where high evasion rates are contributions, in contrast to Chile, where a serious concern, bolder measures may be recognition bonds were issued to required. Hungary reforms may contain switchers to the new system, and useful lessons in reducing evasion and * A defined-contribution, privately- informality. provided, regulated second pillar, similar H.3 Hungary's Current Reforms to Chile's. 17. The system is believed to fulfill four abou 3 abou th a e ata t of of the five rules. The flat pension makes it azil), is bin to ipment o both universal and redistributive. The secuity reforms.nHngarysiplt of system is transparent: the AFJP publishes about 10om s dning, ain an system accounts periodically and people have acing deteriorat ing alth and full information of their individual accounts. epcnc e dbyion syse The privately managed system is credible: widely se diabse s part r the number of participants in the public and maret resu ng to deat g r private second pillar was about 2 million unelyet,ung al used rly each in February, 1995; since then rtemetsmes in the sad early participation in the private system has risen which hav e in the old ae pens to 2.8 million, and that in the public system syste desie low and d eni l fell to less than 1.5 million. The average rate e Mae lif an cairt is of return on accounts has risen from 13 to 22 percnt ver he ame erid. Bt wile only 64 years, and that of women is about 75 percent over the same period. But while yas progress in making the system equitable has been made, some inequities remain: e.g., the 20. Hungary's old system is essentially a highest pension in the first pillar is about 15 single-pillar system (with universal times the lowest. entitlement), with a minimal third pillar. The 18. or razl, he eperenc of retirement age is low: 60 for men and 55 for 18. For Brazil, the experience ofg Argentina has the following lessons. First, T he e ceewere a high because of political reasons, Argentina's de endec rto f a two peser reformers could not make it mandatory for eery t otrbut Pyo teso anyone, including new entrants to the labor finne the penintand.heal te ar force, to join the new, capitalized system. thereoe vehih in 14all tare But almost all new entrants are going into the second pillar. Because they cannot54% (44 on employers and 10 on switch back to the old system, this implies widespread andsincreasingesion,eand t that in the long term everyone will have both syse a bec eain dain In 196 a private account and the basic minimum scial reiabout 15% from the government. So while this o GDPuwito u more aonthird transition will take longer than in Chile, it of ceta go penturs,iad has the benefit of being completely cntrt toaeficit ofpabou a3 o voluntary. Second, Argentina has . nt pit is at Hungry' maintained relative high payroll tax rates (of Pble ital scit ste t from about 40-45%), and because only about one- an a ing populato ut rom orm fourth of these are linked directly to future benefits, evasion rates may not come down distortions. Reforming Social Security: Lessons from International Experience and Priorities for Brazil 7 21. Faced by these problems, Hungary is 22. While the reforms of the first and launching a set of stringent reforms: second pillar are only now being implemented, these reforms and other changes have already encouraged the growth * As part of pre-reform measures, the of the third voluntary pillar. In the three government reduced the official years since 1994, the number of voluntary replacement ratio and, through pension funds has grown from 80 to 270, indexation rules (in a situation where and the number of members has increased annual inflation rates are about 25%), cut seven-fold from 50,000 to 350,000. the real benefits even further. Hungary's reforms aim at achieving two broad goals: striking a balance between * Falling real pensions have fueled the inter- and within-generation solidarity growing dissatisfaction with the system, (redistribution) on the one hand and strengthening public support for deeper individual insurance on the other; and reforms. The reforms, being initiated in encouraging fiscal sustainability and 1997, aim to modernize the first pillar, increasing saving. In designing these reforms, introduce a second pillar, give a boost to Hungary has sought to learn from the the third pillar, and tighten labor and experiences of countries all over the world. disability regulations. Hungary's attempt to return to an official economy from a shadow economy by * The first pillar will be reformed by a lowering the payroll tax rate and establishing gradual increase in the retirement age to a tight link between today's contributions 62 years for all, linking replacement rate and future benefits, while retaining a to time of service, which will be defined redistributive component, should be closely strictly (only military service and studied by Brazil which has similar problems maternity leave will be included other and aspirations. than work), and Swiss indexation of pensions (indexed to both wages and H.4 The Social Security Debate in the price indices). United States * The second pillar will be a fully-funded 23. The social security debate in the US defined contribution scheme, with has heated up because of several factors, the mandatory participation by workers most important of which is that the system is below 40 years (as in Uruguay), freedom believed to have registered an actuarial to choose among pension funds and deficit, which occurs when system assets fall individual account management (as in short of accumulated expected liabilities. Chile), and a minimum of 10 years of Unlike fiscal deficits, which occur when contribution. The contribution rate will current revenues fall short of current be almost halved to 30% (20% payments, these deficits are difficult to employers and 10% employees). calculate since they rely on projections of work force, growth of income and social * Regulation of these pension funds is security taxes, and the average benefit level separated from schemes that belong to and duration of retired life. What is clear, the third pillar, and carry a guarantee by however, is that the number of workers per the state. beneficiary has fallen from 14 in the 1950s to less than 3 today. The problems facing social security have in part been the result of an Reforming Social Security: Lessons from International Experience and Priorities for Brazil 8 unanticipated increase in life expectancy, and 0 Legislators have the mandate to keep the because benefits and coverage were boosted system in good health for the next 75 sharply. The future does not look rosy, even years. In contrast, Brazil's constitution if immigration of young workers is requires lawmakers to ensure that accounted for. The population above 65 benefits are paid, but does not ensure years is expected to rise from 12% today to that the system is financially viable. more than 20% by the year 2030. As a result, the ratio of the present value of benefits to The US system will not go into a current present value of taxes were about 7 for men deficit (i.e., when social security and 9 for women in 1960. These have fallen payments exceed contributions) for to about 1.5 in 1995. approximately another 25 years, but the situation is already being treated as a 24. Replacement rates - which differ by crisis. In contrast, Brazil's INSS system wage level, being high for low-wage workers and vice versa - have already been reduced from an average of about 55% of average Social security benefit levels - for both earnings to less than 40%. The situation private and government workers - are calls for either cutting benefits further or calculated using individual indexed raising taxes. The current proposals include earnings over the last 35 years of work, three plans: in essence the entire working life of contributors. In contrast, INSS pensions * The first proposal would maintain the and publichensions3useontheast system, but increase the coverage of (usul teis m only alary. taxes on social security income from 50- 85% to 100% of benefits, and invest 40% of the assets in equities to raise the rate of return. 26. Some important points that emerge * The second proposal would add a system countisins an t eupe, an of individual account funded by 1.6% of Nor in Aeria aE wages and managed by the government, raise the retirement age, and lower the replacement rate even more for rich * Long-term thinking required The very retirees. nature of the problem of social security implies that long-term thinking, which * The third proposal would establish a goes beyond the concerns of one or even multi-pillar system with a flat pension of two generations, is required for two-thirds of a poverty line, and a sustainable solutions. second pillar of individual accounts funded by a 5% contribution, with * Reforms will be difficult to implement. pension funds managed privately. The reasons are, first, politicians are shy 25. The design of the US system, and the to be borenow while bnfts discussion surrounding it, is notable for come later and second, the potential long-term thinking: losers from reform (e.g., old workers and retirees) are often well-organized and Reforming Social Security: Lessons from International Experience and Priorities for Brazil 9 vocal, while the gainers (young workers) are unaware of the benefits or disinterested because the gains are so far in the future. * But reforms are being implemented successfully: Nevertheless, many countries, even those that were welfare states (e.g., Sweden) or socialist economies (e.g., Poland and Hungary) are implementing reforms. * Solutions are unique, but problems are the same: The solutions are often unique, but the problems that prompt these reforms are the same: financial unsustainability, inefficiency, and inequality. Reforming Social Security: Lessons from International Experience and Priorities for Brazil 10 III. THE PROBLEMS FACING BRAZIL 27. All these problems - fiscal But, unfortunately, there is little evidence of unsustainability, inefficiency, and inequality - this. In this section we examine these that have prompted countries to reform their concerns in turn. social security systems confront Brazil today. Using one of the measures of fiscal M.1 The System is Financially sustainability - the ratio of estimated pension Unsustainable debt to current GDP - Brazil's ratio of 200% is about the same as that of Hungary (but 29. The actuarial imbalances in the about three times that of China). As in the Brazilian pension system for the private case of the US, the ratio of contributors to sector (INSS) and the public sector (state beneficiaries has fallen sharply. But, unlike and federal pension systems) are examined in the US where the social security system still turn. enjoys a current surplus (i.e., annual contributions are greater than annual social security benefit payments), in Brazil benefit payments have exceeded contributions by a 30 Imba in the ndt te c growingratio of contributors to beneficiaries 28. Unless this trend is reversed, this ("ativos" to "inativos"), contribution rates widening gap will have to be met by reducing for employers and employees, and the trends other expenditures. There is the danger that in annual contributions and benefits. the government will have to reduce critical investments in primary education, health care and infrastructure. Another problem that has the ratio of contributors to beneficiaries arisen because of declining faith in the fiscal has fallen from about 30 in 1938 to about viability of the system and the weak link 2.5 today, and will fall to about 1.5 by between contributions and benefits, is that the year 2020, increasing number of workers are opting to be informal, thus being neither contributors the rate of contribution for workers to nor beneficiaries of the social security earning between 1 and 3 minimum system. The system, as currently designed, salaries has risen from 3% to 8% inadvertently distorts work decisions: the between 1938 and 1996, but degree of informality of work is higher because of the weak links between while receipts have grown from R$32 contributions and benefits, and otherwise billion in 1988 to R$42 billion in 1996, productive workers retire from the work benefits have more than doubled from force earlier because of overly generous R$18 to R$43 billion over the same eligibility rules and replacement rates (the period. Thus while receipts were roughly ratio of individual monthly pension to twice the level of benefits less than nine salary). In a country with high income years ago, they are smaller than benefits inequality by international standards, some now. inefficiency may be acceptable if the social security system promotes equality of welfare. Reforming Social Security: Lessons from International Experience and Priorities for Brazil 11 31. Demographic changes have played an regions, and is about 55% in the North, important part in this sharp downturn of the Northeast, and Central East, and more fiscal health of the system. than .40% in the Southeast and South. * Estimates of evasion showed an increase * Life expectancy at birth has risen from from about 14% in 1979 to about 31% a about 50 (48 for men and 52 for women) decade later. to about 68 years today, and is expected to rise to about 75 (70 for men and 77 33. At the current, very generous, level for women) by 2030. of benefits compared to the contributions, efforts to extend the system to cover those * For social security considerations, a more currently outside the system may actually relevant indicator is the expected worsen the actuarial imbalances. For fiscal additional life at retirement. At 55 years, reasons, efforts to increase coverage should men are now expected to live until 72 therefore come after reform of the system. If years, and women until 77; at 65 years, the reforms imply tighter links between men are expected to live until 75, and contributions and benefits, it would also be women until about 80 years. easier to persuade workers to contribute to the INSS system. * In 1940, only about 10% of Brazil's population was older than 50 years; by 34. The INSS system comprises of four 1996 this fraction was about 15%, and major subsystems - time of service, disability, by 2020 it will be about 27%. old age, and special programs. Table 1 lists the shares of each. The time of service program takes half of all benefits, although it 32. But these patterns, which explain the accounts for one-third of all beneficiaries. growing ratio of inativos to ativos, are only The old age and the disability programs one side of the story. The other, perhaps together account for one-third of all benefits more important, side is the pattern of but have more than half of all beneficiaries. contributions. The average level of INSS benefits has doubled in real terms from R$100 in 1991 to * The annual growth of receipts has R$200 in 1996. In the time of service dropped from about 10% in the 1970s to program, the average benefit rose from about 0% in the 1990s. R$250 to R$450 over this period. The number of beneficiaries in this programs has * The share of the formal sector in national also risen from less than 50,000 in 1988 to employment, defined as those paying more than 300,000 in 1996. In all programs, payroll taxes, has fallen since 1988. women - because of their lower age and higher life expectancies - will receive benefits * The rate of urban informality grew for about 50% longer than men. between 1990 and 1993 for all five Reforming Social Security: Lessons from International Experience and Priorities for Brazil 12 Table 1: Characteristics of INSS Program Participants, 1995 Program Share of total: Average Expected Duration: Number Value Age* Men Women Time of service 33.2 51.6 54.0 17.6 22.4 Disability 32.3 16.6 51.2 19.2 24.9 Old Age 25.4 16.2 51.5 16.0 22.6 Special programs 9.1 15.5 63.5 10.8 16.5 Source: de Oliveira and Beltrao, presentation. * The average age is lower for women in all four programs 35. Without any changes, the fiscal tend to be higher, especially for relatively imbalance in the INSS is expected to grow. high wage (skilled) workers. While contributions are projected to rise from 5% in 1995 to 6% of GDP until 2030, * Weak benefit-contribution links. These benefits are projected to increase from 5% to result in high rates of evasion, but with 13%. These are relatively optimistic the irony that increasing coverage of the estimates: other estimates indicate a deficit current system will worsen, not improve of 2% of GDP as early as the year 2000. matters. While demographic trends cannot easily be changed by policy makers, and overall * Time of service rather than economic growth depends on other factors contribution. Reliance on time of such as monetary and fiscal policy, the fiscal service for determining eligibility rather imbalances faced by the INSS system can be than length of contribution period, and effectively tackled by reforming the social the overly generous rules for calculating security system. The main system-design time of service, have led to unsustainable issues are: high benefits. SHigh replacement ratio. Brazil hasminimum age at which retirees replacement rat atio r a as begin receiving benefits. The above among the highest in the world. While these ratios seldom exceed 60% in other stipulated age at which individual start receiving benefits, can result in countries, the ratio is 100% for workers "retirement" beginning at 40 years of with earnings less than 8 minimum age, and continuing for more than 30 salaries. years. * End-loaded replacement formula. The 36. Since the actuarial problems faced by problem of high replacement ratios is the rNSS system arise in large part because worsened by calculating benefits using a of the generosity of benefits, changes in formula that is too "end-loaded viz., is these features are necessary for solving these based only on monthly salary levels in the problems. last three years of service when earnings Reforming Social Security: Lessons from International Experience and Priorities for Brazil 13 III.1.b Pensions for Public Employees have often led to public sector employment being viewed as a place of 37. The main indicators of the privilege rather than an organization that seriousness of the situation facing pension efficiently delivers public services. systems for federal, state and municipal workers are: 38. The solution to bringing pension- related spending by states under control lies Rising pension expenses. The ratio of in dealing with the following problems: pension expenditures to salary bill for ativos is about 50% in Minas Gerais, * Lenient eligibility rules. Eligibility almost 100% in Rio de Janeiro, and conditions are even more lenient than for 120% in Rio Grande do Sul. The INSS pensions, which are quite generous situation is equally grave in many other by international standards. Due to states and even the federal government generous rules for calculating years of has seen the share of inativos in service (e.g., programs of study count as personnel related expenses rise from 23% service) and special retirement schemes, in 1987 to 42% in 1995. the average age of inativos is only about 60 years, and civil servants are known to * Generous pension levels. This is largely begin receiving pensions as early as 40 the result of very high average pensions years. relative to average wages. This ratio ranges from 150 to 180% for these three * Generous benefits. The public sector states. It is striking that the average plan benefits are even more generous pension relative to average wage is four than pensions for private sector workers. times higher for civil servants compared There is no upper limit on pensions to workers in the private sector. (which cannot exceed 8 minimum salaries for the INSS system), and pensions are * Expensive retirement privileges for based only on the salary in the last month some groups Overly generous of service. Actuarial calculations retirement schemes for politically indicate that if raising employee powerful groups, e.g., for teachers and contributions were the only way to bring politicians who can retire with full state systems into balance, even pensions after 25/30 years for contribution rates as high as 75% of women/men, contribute substantially to salaries would be insufficient to cover the burden on state governments. In current benefits in some states. The Minas Gerais, for example, teachers are solution, therefore, must include more than 75% of ativos and inativos. lowering the level of benefits. * Distortion of incentives. Government * Zero contributions. While private sector pensions are more generous, and salaries workers contribute up to 10% of their and conditions of work are no worse salaries towards pensions, civil servants than in the private sector. Private do not contribute anything (with the employees who join the government can exception of Rio Grande do Sul, which carry their years of service into the public introduced a 2% contribution rate in system, and many join the civil service 1996). late in their careers to avail themselves of the higher pensions. Such distortions Reforming Social Security: Lessons from International Experience and Priorities for Brazil 14 * Parity rules that raise fiscal burden of costs in this way have two choices: go administrative reforms. Pension out of business, or evade payroll taxes by adjustment rules are structured so that becoming "informal". The latter option benefits for inativos goes up when is preferred by most employers and many salaries of ativos in public administration workers. Informality of employment go up, implying high cost of thus arises as the result of collaboration administrative reforms that often between workers and employers. There necessitate a reduction of the work force is mounting evidence that informality in and increases in the salaries of some Brazil is largely a tax-related worker categories. phenomenon- compliance with minimum wage increases, hours of work, and 39. Without effective measures to tackle modes of pay are similar for formal and these problems, it will be difficult to informal sector workers, and only implement administrative reforms. Just as noncompliance with payroll tax the consequence of an unreformed INSS obligations distinguishes informal from system is an uncompetitive private sector formal employment. labor market and strained treasury, the result of not reforming the public pension system is 0 The absence of a minimum age at an inefficient civil service and a severe strain which retirees can begin collecting on state and federal budgets. benefits. Combined with other attributes such as the high replacement rations, 1I.2 The System is Distortionary weak link between contributions and benefits and generous eligibility 40. The distortionary effects of social conditions, this leads to persons retiring security in the Brazilian labor market are due while relatively young. The distorted to four design-related factors: labor-leisure choice results in wasted " Th wek lnk etwen cntrbutonshuman capital and excessive reliance on * The weak link between contributions and benefits. This results in government transfers. contributions for social security being 0 The "parity" rule for ativos and viewed largely as a tax by workers and inativos. The practice of indexing employers. For all but the last three benefits of those currently retired to the years of an individual's working life, the salaries of current workers raises the cost worker's reported earnings do not of effective civil service reform and determine pension levels, giving both compromises the effectiveness of employers and workers a strong minimum wage policies. State and incentive to under-report earnings for all federal governments have to consider the except these three years. enormous fiscal burden of paying * The high payroll tax to finance the itatios wher ii foun d t b social security system Social security secto w ere unda Simi contributions, which total 22% of wages, te feder goenetwud haeyt raise the cost of labor. This has two tefdrlgvrmn ol aet rais th cot o labr. hishastwoconsider the fiscal implications (i.e., effects: first, for those employers who raising pension levels for many retirees) can switch to less labor-using methods of of raising minimum wages even if it is production, this reduces the demand for found that this would help some labor. Those who cannot save labor disadvantaged workers. The parity rule Reforming Social Security: Lessons from International Experience and Priorities for Brazil 15 may thus compromise the effectiveness earnings also translate into much higher of both efficiency- and equity-enhancing overall pension benefits than private sector policies. employees. As mentioned above, the ratio of average pension to average wage is four 1I.3 The System is Not Very Equitable times higher for civil servants compared to workers in the private sector. Pension rules 41. In its current state, Brazil's old age exacerbate the private-public difference in system has three main inequities: earnings: * Civil servants and employees of public * Contribution rates are lower (usually enterprises that have closed funds get zero) for civil servants. Private sector more from the system than INSS workers contribute, together with their pensioners. employers, about 20% of their monthly salaries toward pensions. * Within each of the public and private systems, the system is regressive because * While monthly INSS pensions are capped of the use of final month's or last three at 8 minimum salaries (or about years' salary to compute pension levels. R$1,000), government employees receive up to 120% of their final salary, which * Due to differences in mortality, there is can be higher than $10,000 per month. In some inequality in net social security Rio Grande do Sul, for example, about benefits across groups differentiated by 15% of current civil servants would gender, region, and skill (or wage) levels. receive pensions greater than $1,000 despite earning the same amount 42. We briefly take these up in turn. In (controlling for education, age, race, addition to these inequalities, there are many gender, and tenure) contributing only 2% workers who, because of relatively short of earnings towards pensions; in 1995, formal sector length of service (i.e., less than more than 6,500 workers would receive 25 years for women, and less than 30 for pensions above R$5,000 per month. men), have contributed to the system but will not receive benefits from the time of service * While salaries for the final 36 months program. before retirement are used to calculate pension levels for private sector workers, III.3.a Public-private differentials. pensions for public servants are calculated using the final monthly salary. 43. Estimates of private-public earnings Even if age-earnings profiles are identical differentials from the 1995 PNAD survey for an estatutario and a CLTista, the indicates that employees in public public employee would get a higher administration earn almost twice as much as pension due to greater "end-loading" in those in the private sector. While a large the formula for calculating pension part of this difference is due to their higher levels. education levels and age, significant differentials remain for federal 44. Recent analysis shows that public administration, and legislative and judicial employees have higher or similar earnings employees even when these attributes are than similarly qualified private sector controlled for. For all estatutarios, higher employees, and have greater job security. Reforming Social Security: Lessons from International Experience and Priorities for Brazil 16 Pension differences between these groups 46. The example is hypothetical, but add to, rather than compensate for, this reality in Brazil is not very different. And inequality in labor market rewards. when we consider that less educated workers usually work for more years and live shorter III.3.b Inequality due to skill differentials. lives, the differences in pension benefits relative to contributions are exacerbated. 45. In general, higher levels of education are associated both with higher levels of III 3.c Inequality due to mortality earnings, and faster growth of earnings with differentials age. Put another way, more educated workers start with earnings levels that are 47. The inequality in pension benefits due similar to their less educated counterparts to differences in mortality arises because but, over much of their working lives, this groups with longer retired lives can expect to earnings gap widens. In the case of the receive pension benefits for longer, even Brazilian pension system, where only the last though their contributions may be over month (in the case of government working lives that are similar in length. The employees) or the final three years (in the matter is complicated somewhat if eligibility case of private sector workers) of an periods for these groups differ as well. A individual's earnings history is used to methodology that uses "rates of return" (in compute pensions, this pattern implies that terms of expected pension benefits to each less skilled (poorer) workers will not receive real of contribution) has been used to as much from the pension system as would compute some aspects of inequality due to more skilled (richer) workers. Consider the mortality differences. This methodology was following example. applied to the US and Argentina, and found to be useful in determining the inequality- related effects of social security systems. * Assume that educated workers earn a constant amount R$200 per month in the 48. A 1991 study of differences in life first half of their working lives, and expectancy in Brazil found that the main uneducated workers earn R$100 per determinants are gender, region and income month. In the second half of their career, level. Using these three categories, and assume that these numbers are R$400 applying the "rate of return" methodology, and R$100 respectively. the following results stand out: * Thus, while the educated worker's Gender. Upon reaching 60 years, the lifetime contributions to social security average woman can expect to live for were based on average earnings of another 20 years, relative to 15 years for R$300, his pension level will be R$400. The less educated worker's contributions checks than men. Women also have and benefits will be based on an average lower time of service requirements: ofR$100. 20/25 years to 30/35 years. These " Soeve thughskiled orkrs am nlydifferences translate into about 20-30% * So even though skilled workers earn onlySince three times more than less skilled women have lower earnings than men, workers, their pensions are four times as these higher returns may actually be much. equality- enhancing. On the other hand, since female labor force participation will Reforming Social Security: Lessons from International Experience and Priorities for Brazil 17 continue to rise in Brazil, this pattern 0 Earnings levels. Since wealthier persons implies an increasing fiscal burden. live longer, they may have higher returns * Region. Life expectancy levels are since they can collect benefits for longer. higher in the South and the East than in In case of women, this "advantage" the poorer Northeast, and these translate appears to be offset by other factors. into higher rates of return for these But wealthier males in both the old age regions in case of the old-age program. program and length of service programs The old-age program therefore does not appear to receive higher rates of return. help reduce regional disparities in Both programs appear to worsen income. In the length of service inequality. program, regional mortality differentials appear to be offset by other factors. Table 2: Rates of return (pereBnt), by groups with different mortalit rates Group pOld Age Progral*ength of Service Females te18.3 r Males 14.7i. Region:_________ South Old Amo East 18.3 9.7 Frontier 17.0 7.3 Northeast 16.2 7.9 Earnings: <2 min. wages, Males 11.1 >4 min. wages, Males 13.5 7.2 Source: Rofman, presentation. The assumptions are 12 years of contributions in the case of the old age program, and 25/30 years in the case of the length of service program Reforming Social Security: Lessons from International Experience and Priorities for Brazil 18 IV. SUMMARY AND IMPORTANCE OF REFORM 49. The main lesson of international Uruguay, Peru, and Bolivia, have experience appears to be that faced by already launched far-reaching reforms. problems of insolvency, economic distortions and social equality considerations, many * Formerly communist countries such as countries are moving towards multi-pillar old Poland, Hungary, and China have begun age social security systems. The new reforming their social security systems. systems tend to have in-built mechanisms that do not distort individual incentives and * Even rich countries such as Sweden, the cannot be tampered with to obtain short term US, and Japan, who face high transition gains for selected groups. Principally, this costs due to their older populations, are involves installing a capitalized pillar that has contemplating deep reforms or have a tighter link between contributions and already begun reforming their systems. benefits to encourage savings and reduce labor market distortions, while restructuring 51. Social security reform is critical for the redistributive pillar that most countries the development and functioning of labor have already to ensure a universal minimum and capital, for efficiency of public pension. Countries that face very high costs expenditures, and hence for economic and in transiting to a fully capitalized system are social welfare: trying to find ways around this: Sweden's notional accounts system is one such * Capital market functioning. The innovation. President of the Senate, Mr. Antonio Carlos Magalhaes, has correctly warned 50. In light of this experience, it is clear policy makers that "Social security will that Brazil's problems are neither unique nor undermine every financial system in beyond repair. As compared to countries Brazil if not reformed." such as the US, Sweden, and Uruguay, * Labor market efficiency. Besides its Brazil has a young population. This means better prospects for moving to a multi-pillar importanciorcal mats cial security is critical for smooth functioning system, since it implies lower transition costs of labor markets. Without reform of to a capitalized system. But the window of social security, it will be impossible for opportunity is small. The population, while Brazil to have an efficient civil service young, is aging quickly; the rate of increase ofpnso db i ih eaueofhg and an efficient, competitive private labor of pension debt is high because of highmakt benefit levels; the absolute number of market. retirees, and the ratio of inativos to ativos is * Allocation of public expenditures. If rising rapidly; and Brazil is facing strong left unreformed, the current social pressures to increase labor market security system will divert resources competitiveness. from much-needed investments in health, education and infrastructure, and hence jeopardize not only current economic * Other Latin American countries that face progress, but economic prosperity and similar pressures, e.g., Chile, Argentina, social welfare for many generations to come. Reforming Social Security: Lessons from International Experience and Priorities for Brazil 19 SUGGESTED READINGS James, Estelle. 1995. "Evitando a crise nos sistemas de amparo a velhice," Financas & Desenvolvimento, junho de 1995, Washington, DC: publicacao trimestral do Fundo Monetario Internacional e do Banco Mundial. Ministerio da Previdencia e Assistencia Social. 1997. Livro Branco da Previdencia Social. World Bank. 1994. Averting the Old Age Crisis: Policies to Protect the Old and Promote Growth. Oxford University Press: Published for the World Bank. World Bank. 1995. Brazil: Social Insurance and Private Pensions, Report number 12336-BR, Washington, DC. World Bank. 1996. Rio Grande do Sul State Reform Program: Pension and Health Reforms. Latin America and Caribbean Region Department I report, Washington, DC.