Framing answers to these questions is not always easy, nor are answers always evident at the outset of a reform program. This is because a pension system is inherently a dynamic and long-lived institution, promising benefits to retiring workers and their dependents over a period that extends as much as a century into the future. For this reason, it is fair to say that the jury is still out on many aspects of Chile's social security reform. Nonetheless, excellent recent economic research helps clarify several aspects of the system's design and performance, making it possible to offer an interim assessment of the strengths and weaknesses of Chile's experiment with privatizing social security. ₁

Focusing first on who participates in the system, the Chilean reform initiated in 1981 established new privately managed defined contribution pensions run by groups known as afps (Asociacion de Fondos de Pensiones). Every new labor market entrant is required to chose one of several competing afps into which his contributions are deposited, and most old-system workers have migrated to this new system as well. Access to the old underfunded pay-as-you-go program was closed off as of 1985; of course, the fact that some older workers elected to remain under the old system means that both social security programs will run in parallel for an estimated 40 more years.

Turning next to financing, all covered Chilean employees are required to contribute 10 percent of their pay to an afp defined contribution account (up to an earnings ceiling of about US$2,000 per month). ₂ Covered persons are legally permitted to hold their retirement accumulations in only one afp at any given time, but may reallocate their holdings up to four times per year. The law also stipulates that a contributor switching afps must move not only new contributions but also past accumulations as well. Contributions are collected by individual afp agents from each company for each covered worker, resulting in sales as well as marketing costs that are perceived to be high.

In addition to what has come to be called the "second pillar" plan, or the defined contribution afp account, the Chilean system provides a minimum or "first pillar" pension. This is a publicly provided benefit payable to retirees with small afp accumulations (more on this below). This benefit is financed out of general revenues rather than from the earmarked social security payroll tax.

Under the new system, afp retirement benefits may commence at the normal retirement age, which is 65 for men and 60 for women, as long as 20 years of contributions have been achieved. ₃ As in any defined contribution plan, the eventual retirement payout depends on how much the worker has contributed over the years as well as the funds' investment returns. At retirement, the person may move his funds to an insurance company and purchase a retirement annuity; alternatively, a minimum withdrawal pattern can be chosen, or a partial lump sum--as long as the remaining fund can produce a retirement payment at least equal to 120 percent of the minimum government pension and 70 percent or more of the last decade's average pay (World Bank 1994).

This "second pillar" defined-contribution benefit is backstopped by a minimum welfare-oriented benefit, payable in the event that the afp sum is insufficient to generate a high enough retirement income. Specifically, the government guarantees that covered workers will receive a retirement payment worth at least 75 percent of the minimum wage, or one-quarter of the worker's average pay over the 10 years prior to retirement, whichever is more (Edwards 1996). If the retiree's afp amount is too small to generate the threshold benefit level, in effect the afp is taxed at a 100 percent rate, to help pay the retiree's minimum pension benefit.

It is not yet known how important this minimum pension benefit will turn out to be in practice, since very few people have retired under the afp system (most current recipients are survivors and dependents). The existence of this welfare or minimum pension has one potentially serious effect, which is that low-wage workers will be tempted to evade making afp contributions after they have contributed at the lowest level needed to become eligible for minimum pension coverage. After that, employees in the low-paid sector will have an incentive to work "off the books" or in the informal sector, because they are guaranteed 75 percent of the minimum wage in retirement without paying additional afp taxes. In effect, the retiree welfare benefit becomes an asset-tested payment, with the afp accumulation used to test eligibility.

Evidence that the new system faces problems of this nature is beginning to appear, in that most of the Chilean labor force appears to be covered by the new system but only about half of all workers are active contributors to afp's in any given year (Diamond and Valdes Prieto 1993). How this will play out in the future as more covered cohorts retire is unclear, though one simulation model concluded that half of all retirees could end up being entitled to more money under the welfare retirement benefit than under their own afp accumulation (Zurita 1994). Another concern that has yet to be clearly articulated is that the existence of the minimum guarantee will likely have unwanted labor supply and savings effects in Chile. For instance, some workers might not save as much in other forms or work as long as they might otherwise, since the minimum pension guarantee acts as a disincentive to devote personal resources to increase retiree well-being. ₄ In fairness, it must be recognized that all social insurance programs with a minimum benefit will suffer from these behavioral consequences. On the other hand, the as yet unanswered question in Chile is how large these undesirable behavioral consequences might be.

While the Chilean privatization effort has been praised as strengthening private capital markets, it must be recognized that the afp system is strictly regulated. For instance, there are minimum capital requirements that a fund manager must post to enter the afp market, and the government regulates the types of fees afps may charge. In addition, each fund must meet annual rate of return constraints: the minimum return earned may be no less than half the afp average, nor less than 2 points below the afp average. Any fund unable to hit the targets must resort to its own reserve fund to make up the difference, or be liquidated by the government if reserves are insufficient (the government also makes up the return shortfall in this event). Higher-than-average returns are also controlled, with amounts greater than 2 points over the afp average (or 50 percent more than average) being sent to the afp's reserve fund. Perhaps most critically, the government has set limits on the types of investments an afp can hold. Initially virtually all funds had to be held in Chilean government bonds, with a rising fraction in Chilean equities over time. Until recently the afps were prohibited from much foreign investment, and today their non-Chilean holdings are still minuscule.

How Chile's new social security system handled the transition obligation is in many ways the most important and perhaps least well-understood feature of the social security reform in that nation. Put simply, an unfunded pay-as-you-go pension system owes promised benefits to retirees and some current workers, yet by definition has insufficient funds on hand to pay on these promises. When social security experts worry about the transition obligation, therefore, they are asking how to pay the implicit promise made under the old defined-benefit rules, a promise that older people are counting on receiving.

Many believe that the Chilean approach to the transition problem was novel, and perhaps irreproducible. Whether or not this is precisely true is less important than the fact that Chile's afp reform was enacted at a time that the government budget was running a surplus. This eased the need to raise taxes to pay for existing retirees' benefits, while the new afp system was building reserves. Of course fiscal surpluses are not characteristic of other Latin American nations attempting to follow Chile's example, nor in more developed nations in Europe or the United States as they contemplate a similar move (Quinn and Mitchell 1996).

Another reason the Chilean reform is seen as a "special case" is that that country's government was a military dictatorship. Some argue that this enabled the nation to handle the transition problem by issuing Recognition Bonds to middle-aged workers that were worth less than benefits promised under the old programs (Bonos de Reconocimiento). These Bonds are credited with a 4 percent real rate of return by law, and are scheduled to be paid off over approximately 40 years (Cortazar 1995). On top of this, the national retirement age was raised for all future retirees, which in effect was a form of benefit cut. In these ways, Chile's transition obligation under the old system were spread across several different stakeholders. In particular, the transition costs were partly borne by reductions in the benefit promises to workers under the old system. The remainder of the obligation is yet to be paid, and will be covered by current and future workers who face taxes to cover government debt--including retiring the Bonos--in the future. Even after these adjustments, the transition obligation under the social security system remains relatively substantial large, estimated at upwards of 80 percent of gdp (Schmidtt-Hebbel, 1995).

To sum up, as the Chilean social security reform process has evolved, it has fashioned an interesting new structure for delivering old-age benefits. The reform's key features are as follows:

Evaluation Framework

To assess how well the Chilean model is working within its own borders, and to evaluate its relevance to other countries, it is necessary to ask how the reformed program is affecting behavior and whether these changes are judged positively by well-informed observers. Following the evaluation framework proposed by Mitchell and Zeldes (1996), we pose four questions:

Pension systems are useful in pooling a range of risks from income loss due to disability, to economic insecurity due to living longer than anticipated. Chile's new system provides some earnings insurance due to the linkage between contributions and eventual benefits in the AFP system. It also affords a degree of protection against poverty with its first-pillar minimum benefit. In addition, the system requires mandated participation in privately provided disability insurance, though little has been written evaluating that segment of the system. So far there is relatively little known about how much income security the AFP accumulations will generate in retirement, since the system is as yet quite young. There is even less known about how well the system will perform in terms of its payouts at the end of the contributor's worklife. From other sources, mainly the United States, there is reason to suspect that annuity markets do not work particularly well (Mitchell, Poterba and Warshawsky, in preparation). Given the potential for adverse selection in individual annuity pools, this deserves more study in Chile as well.

Above we mentioned some of the issues pertinent to the minimum pension program While a larger "cushion" or safety net will improve the redistributive effects of the system, it also induces evasion, discourages saving, and probably reduces labor supply in the formal sector. It remains to be seen how big a problem these will prove to be, but evidence on evasion and noncontribution may be cause for concern.

As yet there is little direct evidence on this point. It does appear that the formal labor sector has grown in Chile, as compared to the old system, in part because payroll taxes are lower than previously. Nonpension private saving, however, appears to have declined as pension saving in Chile has risen (Holzmann 1996), which suggests that there may be some substitution going on between retirement saving vehicles. In addition, during the system's first decade of life, AFPs earned large positive real rates of return(approximately 12 percent real per year(which probably discouraged nonpension saving. By contrast, over the last couple of years, AFP average real returns became negative (real (2.5 percent in 1995), a disappointing result that may discourage pension participation and may drive private (nonpension) savings rates up again.

There is little convincing econometric evidence that the Chilean system increased overall national saving (Holzmann 1996). This is partly because nonpension private saving fell to offset rising pension saving, as just noted. Also, the government dissaved over most of the last decade, using its budget surplus to pay current retirees, mitigating part of the pain of the transition cost. There appears to be little strong net new national saving.

On the other hand, many argue that Chile is in better economic condition today because it enacted a wide-ranging set of financial, regulatory, and insurance market reforms to support the afp system. As a whole, these reforms helped enhance the appeal of Chilean capital markets. In this sense, the economic reform package as a whole contributed to economic growth and national saving, but it is virtually impossible to prove that the pension reforms per se raised national saving rates.

Looking Ahead

Several other Latin American nations have adopted variants on the Chilean model. For instance, Peru recently implemented a private pension account system.

One difference was that, at least initially, Peruvian workers had the option of participating in the new private retirement accounts or staying in the old public system (which was not intended to be phased out). This meant that there would be long-term duplication of administrative costs, and it opened up the possibility of "gaming" the minimum benefit via optional contributions to each system through time. The continuation of the old plan in Peru also meant that high payroll tax rates continued to be projected into the future, rather than winding down, as anticipated under the Chilean plan. To its credit, the Peruvian government is revamping the plan as these issues emerge. However, system participants were not given as clear a message from the outset as in the Chilean case, so the Peruvian reform got off to a slow start.

Argentina's 1994 social security reform plan was in many ways less ambitious than Chile's, with its intention of maintaining a relatively large public pension benefit while offering workers a one-time option to move part of their payroll taxes to a capitalized plan similar to Chile's AFPs. Now in its growth phase, the Argentine system is experiencing high administrative costs, but pension accumulations are beginning to mount up, and early evidence suggests that the majority of the eligible workforce is participating in the private capitalization accounts.

In Uruguay, where the size of the implicit old-system social security debt was measured at twice the national gdp, reformers are working to establish private defined-contribution pensions as well, each with their own variant on the two-pillar scheme (Mitchell 1996; Kane and Palacios 1996). Here, too, a large public pension role is envisioned, as large if not larger than Chile's. The Mexican government is also taking brave steps to revitalize its old-age security system, recognizing the crisis it faces with deteriorating benefits and payroll taxes of almost 25 percent of earnings under its existing program. Like Chile, Mexico intends to establish individually managed private pension accounts. Unlike Chile, it has decided to cut administrative costs by using the national tax agency to collect the payroll taxes.

There are signs that the Chilean social security reform model has become an export to developed countries as well. In the United States, for instance, the Social Security Advisory Council and numerous other research and policy groups recently examined how the Chilean system might be relevant. ₅ The U.S. National Academy of Social Insurance along with the Pension Research Council also have major studies under way on social security reform, including ones on the potential consequences of privatization in large nations with well-developed capital markets. Australia, Sweden, and other developed nations also face population aging in tandem with insolvent public old-age support programs, and here, too, social security reform is on the policy agenda.

As the Chilean model becomes more widely understood and debated, it is important to understand both how it works and where it requires strengthening. In the end, a social insurance program(including a pension system(cannot eliminate all risk. The best it can do is to optimally and transparently spread some of the risks affecting its participants, so as to have as few negative behavioral effects as possible (Mitchell and Fields 1994). A healthy national pension system pools individual risk, including the risk of any one person outliving his retirement accumulations. Chile's system does spread some longevity risk by requiring that at least a portion of all retirement accumulations be annuitized.

Other risks are not fully pooled by Chile's system. For example, a worker experiencing a sudden drop in earnings because of skill obsolescence or injury will see this pay cut reflected in his defined-contribution pension contributions (and eventual retirement benefits). The existence of the first-pillar minimum public pension payment mitigates this risk somewhat, but also gives rise to moral hazard as spelled out above. The fact that Chile's system is almost exclusively invested in Chilean assets also makes it vulnerable to a different type of risk, namely country-specific macroeconomic risk. More needs to be learned about why greater diversification does not occur; as other countries import and adapt their own variant of the model, it is likely that more diversification will make sense.

Finally, designers of the Chilean system and pension experts everywhere must acknowledge the fact that there is economic risk worldwide that no pension system can protect against. It is imperative for reformers to acknowledge that risk of this sort is inherent in any pension system, so as to give workers and retirees a better sense of how their future income streams might vary.

Of course, economists know that there is no such thing as a risk-free promise, but others sometimes forget this fact and expect their pension system to deliver complete security when indeed it cannot in good faith be offered. Only with a realistic vision of what a retirement system can provide will it be possible to design a better performing plan for old-age income.

Cortazar, Raul. "The New Chilean Pension System: Lessons After 15 Years." Unpublished paper, Santiago, Chile, June 1995.
Diamond, Peter. "Privatization of Social Security: Lessons from Chile." Revista de Analysis Economico, June 9 (1) 1994: 21-33.
Diamond, Peter, and Salvador Valdes-Prieto. "Social Security Reforms," in Eds. B. Bosworth, R. Dornbusch, and R. Laban. The Chilean Economy: Policy Lessons and Challenges, Washington, D.C.: Brookings Institution, 1993.
Edwards, Sebastian. "The Chilean Pension Reform: A Pioneering Program." Unpublished paper presented at a NBER conference on "Privatizing Social Security," Cambridge, Mass., August 1994.
Holzmann, Robert. "Pension Reform, Financial Market Development, and Economic Growth: Preliminary Evidence from Chile." Working Paper, International Monetary Fund, August 1996.
Hubbard, R. Glenn, J. Skinner, and S. Zeldes. "Expanding the Life Cycle Model: Precautionary Saving and Public Policy." American Economic Review, May 1994. 8(2): 174-179.
James, E., and D. Vittas. "Mandatory Savings Schemes." In Securing Employer Provided Pensions: An International Perspective. Eds. Z. Bodie, O.S. Mitchell and J. Turner. Pension Research Council, Phila.: Univ. of Pennsylvania, 1996.
Kane, Cheikh, and Robert Palacios. "The Implicit Pension Debt." Finance and Development. 33(2). June 1996: 36-38.
Marcel, M. and A. Arenas, Social Security Reform in Chile. Inter-American Development Bank, Occasional Papers No. 5, 1992.
Mitchell, Olivia S. "Administrative Costs in Public and Private Pension Systems." NBER Working Paper No. 5734, August 1996.
Mitchell, Olivia S. "Social Security Reform in Uruguay: An Economic Assessment." Pension Research Council Working Paper. The Wharton School, University of Pennsylvania. 1996.
Mitchell, Olivia S. and Gary Fields. "Designing Pensions for Developing Countries." Report to the ESP Division of the World Bank. Pension Research Council Working Paper, The Wharton School, University of Pennsylvania, September 1995.
Mitchell, Olivia S., James Poterba, and Mark Warshawsky. "The Market for Annuities." In preparation.
Mitchell, Olivia S., and Steven Zeldes. "A Framework for Analyzing Social Security Privatization." American Economic Review, May 1996.
Myers, Robert J. "Chile's Social Security Reform After Ten Years," Benefits Quarterly, Third Quarter, 1992: 41-55.
Pinera, Jose. "Principios y Fundamentos del Sistema Privado de Pensiones en Chile." In Analisis del Sistema Privado de Pensiones en Chile. Santiago, Chile: Congreso Iberoamericano, Asociacion de Administradoras de Fondos de Pensiones, 1991.
Quinn, Joseph, and Olivia S. Mitchell. "Prospects for Social Security Reform." American Prospect, May-June 1996, (26): 76-81.
Schmidtt-Hebbel, Klaus. "Pension Reform Transitions from State Pay-As-You-Go to Fully Funded Systems." Unpublished Paper. World Bank, June 1993.
World Bank. Averting the Old-Age Crisis: Policies to Protect the Old and Promote Growth. Policy Research Report. Policy Research Department, The World Bank: June 1994.
Zurita, S. "Minimum Pension Insurance in the Chilean Pension System." Revista de Analisis Economico, June 9(1) 1994: 105-126.

Footnotes

Note 1: See, for instance, Diamond 1994; Diamond and Valdes-Prieto 1994; Edwards 1996; James and Vittas 1996; Marcel and Arenas 1992; Myers 1992; Pinera 1991; World Bank 1994. Back.

Note 2: Edwards (1996) p. 11. Additional voluntary contributions are permitted to the AFP retirement funds as well. In addition to the 10 percent pension payroll tax, another 3.5 percent of payroll must be paid to the AFP for administrative costs as well as mandatory survivor and disability insurance. Back.

Note 3: Early retirement is possible from age 55, as long as the worker has at least 10 years of contributions, his pension accumulation is at least 110 percent of the government minimum pension, and his pension would amount to at least half of his average wages over the last decade. Back.

Note 4: For U.S. research on behavior along these lines see Hubbard et al. (1994). Back.

Note 5: For further information, the final report of the Technical Panel on Trends in Income and Retirement Saving may be accessed via the internet at: http://www.ssa.gov. An alternative site is lexis.pop.upenn. edu/aging_html. Back.