INTERNATIONALBANK FOR WORLD BANK R E T C N O E N STRUCTION PM AND DEVELO September 2004 No. 53 A regular series of notes highlighting recent lessons emerging from the operational and analytical program of the World Bank`s Latin America and Caribbean Region KEEPING THE PROMISE OF OLD AGE INCOME SECURITY IN LATIN AMERICA1 Indermit S. Gill, Truman G. Packard, Juan Yermo and Todd Pugatch Many Latin American governments have radically restruc- 1. Overall, reforms improved fiscal sustainability, shown by tured their old age income security systems, starting with simulations of governments' pension debt, and a much lower Chile's 1981 pension reform. The reforms move national rate of accumulation of pension obligations with reforms, pension system from purely social to largely individual especially in Peru, Bolivia, Uruguay, Chile and El Salvador. responsibility; from risk pooling to individual savings ac- Some transitions were more costly than expected, however. In counts. They maintain reduced public risk pooling features, Bolivia, for instance, the actual pension-related deficit in- combined with mandatory and voluntary individual private creased as a percentage of GDP instead of falling, due to savings to finance pensions and diversify the risks to ad- insufficient regulation and fraudulent claims; lax interpreta- equate income in old age--the "multi-pillar" approach. tion of retirement rules; indexation linked to the exchange rate (recently reversed); and a high minimum pension (nearly How has the new approach to pensions in Latin America twice the minimum salary). In Argentina, policies introduced fared? Are workers and their families better off? after the reform - to integrate provincial civil servants into the new pension system, and to encourage formal job creation A Brief Assessment of Reforms by lowering employer contributions­ aggravated the transi- tion costs of reform.. Increased coverage, a major reform objective, would open 2. Growth of pension funds and other institutional investors access to protection under pension 70.0% systems more widely and equita- o Figure 1 bly. Changing from defined benefit sect pay-as-you-go (PAYG) systems-- al 60.0% rm where payroll taxes paid by today's fo Affiliates/Contributors as Percentage of Economically Active in Population r workers finance pension benefits o 50.0% for today's retirees--to individual system savings, defined contribution sys- n 40.0% tems where workers contribute to ios ne p their own pension accounts, was thi w expected to increase coverage, of- 30.0% fering workers stronger incentives. Yet despite an initial improvement affiliated P 20.0% A in incentives, coverage remains Ef o persistently low in most countries, n oitr 10.0% even long after the introduction of o p or individual accounts (Figure 1). P 0.0% Colombia Argentina El Salvador Costa Rica Nicaragua Chile 2000 Peru 1999 Mexico 2001 Bolivia 2000 There has been greater, although 1999 2002 1998 2000 1999 Male 63.6% 12.8% 19.8% 35.2% 45.1% 11.8% 24.2% 23.2% 8.9% qualified success, in meeting other Female 60.9% 9.1% 26.9% 37.2% 46.6% 8.3% 27.6% 22.5% 15.3% objectives of pension reforms: Total 62.7% 11.2% 22.3% 36.0% 45.7% 10.3% 25.5% 22.9% 11.0% 1 can make capital markets more resilient and dynamic. In turn, many Latin American countries have made eligibility for the capital market development improves the efficiency of minimum pension (the poverty prevention pillar) conditional resource mobilization and investment in the economy, con- on contributions to individual savings accounts, rather than tributing to economic growth. Latin America's new pension keeping the two pillars separate, as prescribed by the fund administration industries are beginning to dominate economics of insurance. To the extent that pooling-arrange- financial systems. Asset growth has been rapid in countries ments are financed through general taxation, this can perpetu- that have reformed pensions. In Chile, assets managed by ate regressive transfers from those without coverage, to those pension funds are over 50 percent of GDP. In all Latin covered by the system. America, pension fund assets as a share of GDP have almost doubled in five years. However, many pension funds invest Why Hasn't Coverage Increased? heavily in government debt, curtailing their contribution to private investment, and leaving them insufficiently diversi- fied and excessively exposed to sovereign risk. Moreover, the Despite improvements, the failure of reforms to improve fund management industries in many reform countries fail to coverage significantly has frustrated reformers. There is fully pass along their gains in efficiency and savings to emerging evidence that stagnating coverage may result not from investors, reneging on one of the principal promised benefits worker myopia or lack of information, but from rational choice of privatization brought by reforms. by workers and their households. 3. Increased equity : Single-pillar public pension systems in Individuals find alternative savings instruments more developing countries, particularly in Latin America, tend to attractive than the reformed pension systems.. The returns generate regressive transfers from poorer workers to the to investment in the new funded pillars, although relatively relatively small number of higher income workers covered by high, have been volatile. Fees for private management of the systems. Moving from PAYG systems to multi-pillar retirement savings have been high, and early affiliates have systems with a large funded component can reduce regressive borne the brunt of new system start-up costs (Figures 3a and transfers and increases equity among covered workers (figure 3b). Management fees ­ in particular, flat fees - have been 2). But retaining a reduced risk pooling component is essential: poorly-run risk-pooling systems are bad for equity, especially burdensome on workers with lower incomes. but well-run risk-pooling components with an explicit pov- Finally, many reforms pay benefits as an annuity, costly for erty-prevention function are good for equity. Fearing misuse, disadvantaged groups with low life expectancy, since premiums Figure 2 2 (based on average life expectancy of the whole population) are saving to smooth consumption. Policy makers may not have above the actuarially fair rate for these groups. Annuities fully realized the importance of this preference to contribu- can also introduce interest rate risk, as the amount of an tion behavior and the incentives to participate in the formal pension system. affiliate's retirement benefit will depend on the real rate of interest at the moment the annuity is purchased. These Workers make rational decisions about securing ad- factors can raise the relative cost and risk of the new funded equate retirement income. If reformed pension systems pillar compared to other formal and informal instruments fail to make workers better off, they may rationally available to secure adequate retirement income, perhaps choose to avoid contributing. They may also be averse to explaining why so many Latin American workers still ignore the policy risk inherent even in privately managed public formal pension systems, despite the improvements in incentives pension systems, of which the Argentine crisis is a pointed brought about by reforms. reminder. Heavy investment in government bonds makes pension funds vulnerable to government's fiscal problems Workers reveal a preference for government to provide a and risk of default. poverty-reduction instrument,. Most reforms have essen- tially split the government's role in pension provision in two: How Can Pension Systems Be Improved in Latin providing (or regulating) a savings vehicle, through indi- America? vidual savings accounts, and a poverty-reduction vehicle, through a risk pooling mechanism (for example, minimum pension guarantees and social assistance pensions). In Chile The multi-pillar approach to pension reform is based on the contribution behavior suggest that workers want insurance belief that government has two fundamental roles in promot- against poverty in old age from government. Workers tend to ing old age income security: 1) preventing poverty in old age; contribute to the just enough to qualify for a minimum and 2) facilitating consumption smoothing, to prepare for the pension. This is observed among workers earning average risk of losing earnings-ability that increases with ageing. As incomes and higher­those least likely to suffer poverty in old life expectancies increase, loss of earnings ability in old age is age­revealing a preference for government-provided instru- becoming more frequent in Latin America. Thus savings ments for pooling against the risk of poverty, over forced mechanisms are better than pooling for consumption-smooth- ing. In contrast, risk pooling is cost-effective for relatively rare losses. With economic development, the incidence of poverty in old age should be increasingly rare. Thus, Figure 3a governments still have a risk pooling role: insuring against what should be a relatively rare losses represented by the incidence of old age poverty. 1. Improving Old Age Poverty Prevention A well-run risk pooling component to mitigate old age poverty is within reach of most Latin American governments. As a very generous illustration of this point, the cost of providing pensions equal and indexed to minimum wages to all citizens over 65 regardless of their contribution history or income, is typically comparable to the level of current government transfers to subsidize public pensions. This illustration is not intended to advocate that poverty prevention pensions should be set as high as minimum wages. Given the typical distribution of wages seen in Latin America, Figure 3b setting the public poverty pension equal to the minimum wage would provide a strong disincentive for workers to save privately. The example simply serves to address a strong fiscal concern often faced by policy makers when trying to determine the appropriate size of their poverty-prevention pillar. These are cost simulations of what is a likely upper- bound benefit--that which even the more ardent populist politicians might be willing to support as the acceptable level of the minimum pension in their country--in order to arrive at conservative estimates of the cost of restructuring the pooling pillar in different ways. In this spirit, the simulations also generously index the minimum benefit to current wages. Significant cost saving can be had from indexing benefits to 3 inflation or to a combination of prices and wages. Preferably, voluntary savings should play an increasing role in the consumption-smoothing function. Governments Further in the future, longer life expectancy makes such should increase voluntary savings options and ensure that generous policies very costly indeed. But a lower universal even low-income workers can take advantage of tax incen- benefit or benefit available at later ages in line with longevity; tives to save, through innovative schemes like matching taxed at source to claw back resources from the better off; or contributions or earned income credits. targeted to the poor, would be plausible options to contain costs . Conclusion There is much that could be done to improve these cost estimates in a country-specific study of viable options. The The Latin American experience with pension reform is still in projected costs do not take account of: (i) the portion of the progress, but already yields important lessons: current transfers to the elderly that are still financing the transitions cost of introducing individual retirement accounts; 1. The poverty prevention pillar needs more attention. This (ii) the cost of excluding workers without a contribution government role becomes more important with economic history from receiving benefits; (iii) the revenue that the development--as the likelihood of old age poverty declines, government could "claw back" from setting a surcharge on pooling this risk becomes more appropriate and affordable. pension payments for workers above certain income levels, or Private insurance markets will not provide this coverage by simply making the universal flat benefit taxable; (iv) the which is critical to ensuring adequate income security in old administrative costs of means tests to efficiently target a age. benefit to the elderly poor; and (v) the costs of corruption and leakage to the non-poor that, given the track record of public 2. The mainstay for earnings replacement during old age pension administration in the region, must be kept in mind. should be individual saving. For most workers, savings Each of these considerations can change the projected costs schemes should involve no redistribution of benefits or of poverty-prevention benefits considerably, and should be pooling of risk across generations. carefully examined in country specific analysis 3. Mandatory saving schemes are not always necessary, but 2. Improving the Savings Component may be useful for transitioning from overly generous PAYG Forcing workers to participate in the savings component to be systems and in providing an initial boost to capital and eligible for benefits has been advocated to increase coverage. insurance markets. However, evidence indicates that savings mandates won't increase coverage unless savings instruments are sufficiently 4. Countries with mandatory savings schemes should lower attractive. Furthermore, a near-universal minimum benefit costs to affiliates by increasing competition among pension would largely solve the coverage problem, at least against the fund managers. This will lead to more attractive savings risk of old age poverty.Savings mandates may still be desir- options for workers; most other solutions appear ineffective. able to: 1) mitigate moral hazard inherent in pooling mecha- nisms; 2) ensure political support for transitioning from By fundamentally remaking pension systems that were PAYG systems; and 3) protect the "infant industry" of private bloated and inequitable, structural pension reforms in Latin financial management for long-term saving. But each of these America and the Caribbean represent a major step forward for reasons also suggests that mandatory savings components the region. There is considerable scope for improvement in should be small or decline over time. Moral hazard can be the design and operation of the new systems, but mitigated with a much lower mandate to save than typically policymakers should direct their efforts at carefully assessing maintained in Latin American countries today. And long and building upon existing reforms, rather than hastily re- experience has shown the dangers of coddling infant indus- scinding them. tries for too long. Over time, mandates combined with dedicated industries, lead to market concentration and impose Notes welfare costs, denying workers the benefits of privatization. 1A conference to discuss the full report on which this ar- Much more competition based on performance is preferable. ticles was based was held June 22 and 23, 2004, in Bogota. Details can be found at http://www.worldbank.org/ Measures to improve the effectiveness of mandatory savings lacpensionsconf are needed, that would: About the Authors ˇ reduce administrative costs and commissions; ˇ improve risk management of savings to reduce volatil- Indermit S. Gill in an Economic Advisor in the office of the Vice-President, Poverty Reduction and Economic Manage- ity; and ment. Todd Pugatch is a Junior Professional Associate in the ˇ lower contributions for poor and young people to in- same office. Truman G. Packard is a Senior Economist in crease their participation. the Human Development Department of the Latin America and the Caribbean Region. 4