The great failure: Retirement pensions in Latin America


This blog was first published in Brookings Blog, on January 27, 2017.

Latin American countries created social insurance systems in the 1930s and 1940s with the promise of providing protection to all workers against various risks. These systems, the cornerstone of the region’s welfare states, were forged on the European model, with benefits financed through a contribution paid by firms and workers proportional to the latter’s wages. For this reason, they are at times labelled “contributory social insurance,” or CSI.

Over a half a century later, the promise remains unfulfilled. Although there are important variations across countries, considering the region as a whole, less than half of all workers are covered by CSI systems. Some countries have relatively high coverage rates: Uruguay 77 percent, Chile 71 percent and Brazil 64 percent. But these are the exception. For the majority, coverage rates are low: Mexico 40 percent, Dominican Republic 38 percent, Colombia 37 percent, Paraguay 23 percent, Peru 22 percent and Bolivia 17 percent. In Central America, with the exception of Costa Rica, coverage rates are 23 percent on average.

A basic design flaw explains why: In most countries only workers hired by firms that comply with the law are enrolled. These workers, known as formal workers, are mostly urban and employed by relatively larger firms, or directly by the government as public employees. The rest, known as informal workers, are employed in a variety of ways that leaves them without coverage. They can be self-employed or work in rural areas or in small firms in urban areas where it is difficult to enforce social insurance laws.

Aside from low, coverage is also unevenly distributed across income levels. Formal workers have on average higher wages than informal ones, often because they have more years of schooling, or as a result of the stronger bargaining power associated with working for larger firms or belonging to public sector unions. Thus, CSI fails to reach those who need it most.

The range of benefits offered by CSI systems varies across countries, but they are always bundled in a package that usually includes health, life, and disability insurance. In some cases, it also includes child allowances (Argentina); training programs (Colombia); or daycare services and housing (Mexico). In all countries, however, a key component of the package is pensions for retirement These pensions are motivated by the notion that workers on their own would fail to save sufficiently for their old age. Their consumption levels as retirees could then be substantially lower than when working. As a result, to “smooth consumption overtime,” governments force workers to put away a share of their wage for the future.


Retirement pensions started in Latin America under the so-called “pay-as-you-go” (PAYG) model, whereby the pensions of those already retired are paid with the contributions of those currently working. Once a worker qualified for a pension, she would get a regular payment from the common fund of contributions of active workers regardless of how many years she lived after retirement. As some would live more years than others, the risks of longevity were pooled.

Setting aside the coverage problem, initially PAYG systems worked well. Since there were few retired workers relative to those actively working, the contributions of those working were more than sufficient to cover the pensions of those retired. But a combination of causes eventually led to financial difficulties. First, in some cases, even at the start, from an actuarial point of view contribution rates were lower than those required to pay for benefits. Second, demographics changed so that overtime the proportion of people working relative to those retired gradually fell while life expectancy increased. Third, the region’s endemic macroeconomic crises limited the growth of formal employment. Finally, here and there, benefits were increased without matching contributions.

Gradually, PAYG systems began to generate deficits. Since retirement pensions took the form of legislated rights guaranteed by the state, those deficits had to be covered from the public purse. Today, many governments in Latin America use general revenues to cover the deficits of PAYG pension systems, de facto establishing a subsidy from all tax payers to the subset of workers who benefit from them. And since, as noted, those are workers with relatively higher wages, it follows that these systems introduce a regressive bias to public expenditures, a bias that can be quantitatively relevant. Brazil spends 4 percent of its GDP in subsidies to its PAYG system, Colombia 3.5 percent, Mexico 2.1 percent, El Salvador 2.0 percent, and Peru 1.7 percent. These figures can be put in perspective noting that countries in the region spend on average 0.5 percent of their GDP on conditional cash transfer programs targeted on the poor.



The shortfalls of PAYG pensions motivated important design changes. Starting in Chile in the 1980s, and then in Mexico, Peru, El Salvador, Colombia, Argentina, and Bolivia in the 1990s, countries turned to systems where contributions would be deposited directly in workers’ individual accounts (as opposed to a common fund in the PAYG system); and where pensions would be proportional to the amounts accumulated by each worker. The new system, usually known as “defined contribution,” would avoid cross-subsidies from one set of workers to another. A novel feature was that deposits to workers’ accounts would be managed by private firms, known as pension fund administrators. These firms, on one hand, would actively compete among themselves for workers’ accounts; on the other, would invest workers’ funds to maximize expected returns. At the time of retirement, the accumulated funds would be used to purchase the worker an annuity; longevity risks, rather than being absorbed from a no-longer existent common fund, would be shifted to private insurance companies, protecting workers.

The new system was expected to have important benefits. First, it would eliminate the regressive features of the PAYG system. Since benefits would simply depend on accumulated contributions, there would be no need for government subsidies. Second, it would stimulate workers’ voluntary savings since their additional savings would be deposited directly in their personal accounts and reflected in the size of their pensions. Third, it would give greater depth to financial intermediation, augmenting the supply of long-term savings that, through the pension fund administrators, could be channeled to long-maturity investments.

Time has shown that some of these benefits have failed to materialize. Certainly, eliminating the regressive system of cross-subsidies was beneficial. On the other hand, the system has done little to stimulate voluntary savings; few workers have channeled additional resources to their accounts. Further, the market for workers’ individual accounts has been far from competitive. On the demand side, workers as consumers of financial products for retirement had difficulty comparing the various combinations of fees and investment options offered by pension fund administrators, particularly when the “product” that workers were buying (or rather, were being forced to buy) would be delivered many years from today. On the supply side, there were few private firms competing, partly because the presence of economies of scale in the administration of funds naturally led to a monopolistic market structure. The result was that returns to workers, net of commissions, were lower than would have been under more competitive conditions.


Critically, however, the change from PAYG to defined contribution pensions left unchanged the system’s original design flaw. Pensions would still be financed from a contribution paid by workers and firms proportional to workers’ wages; or, put differently, pensions would still be associated with workers’ formal status. Thus, despite the change, informal workers continued without coverage.

Experience has shown that labor markets in Latin America are more complex than what was envisaged at the time when these changes were carried out. In particular, many workers have periods of formal and informal employment over their working lives. A formal worker might voluntarily leave his job to try his luck with his own business and be self-employed; or, he may be fired and fail to find another formal job and instead work informally for some time. There are, of course, a large number of possibilities. Depending on the country, on any given year between 15 and 20 percent of workers change employment status from formal to informal, or vice versa. Thus, it is more accurate to refer to “when workers are formally employed” rather than to “formal workers,” and similarly for informal ones. On average, workers with more education, and thus with higher wages, have longer spells of formal employment; and only a small subset of workers—public sector ones being a relevant example—are employed formally throughout their working lives.

Formal-informal transits represent a major challenge for contributory systems since workers save for their pension only when formally employed; the so-called contribution density, the share of the time that workers contribute to their pension relative to the time that they work, will be less than one. Clearly, the lower the contribution density, the lower is the pension relative to the worker’s wage (the so-called replacement rate). This makes it harder to reach the consumption smoothing objective. A good example is Chile, the country with the oldest defined contribution system: replacement rates are in the order of 35 percent (so that the pension is 35 percent of the worker’s wage), which compares unfavorably with replacement rates of about 70 percent in member countries of the Organization for Economic Cooperation and Development.

But there is another more troublesome implication of formal-informal transits. Pension systems in the region, of either variety, usually require workers to contribute a minimum number of years to qualify for even the minimum pension. This requirement varies from country to country: 20 years in the case of Peru, 23.5 years in Colombia, and 25 years in El Salvador and Mexico, for example. When formal-informal transits are large, or there are long spells of informal employment, workers will not accumulate the required years of contributions to qualify for a pension.

What happens to these workers? Their savings are returned to them in one lump-sum payment at the time of retirement, to do with them as they please. They will not have access to an annuity, in case they were in a defined contribution system, or to a life-long payment from the common fund, in case they were in a PAYG system; the risks of longevity will be borne by them. Further, the objective to smooth consumption between their life as workers and their lives as retirees—the raison d’etre of pensions—will be partly defeated. This situation, unfortunately, is the rule rather than the exception. In Colombia, it is estimated that only 25 percent of workers contributing to their pension will actually qualify for one. In Mexico, less than one-third will. For a majority of those contributing, the promise of a pension will be unfulfilled, surely a major social and political issue in the future.

It is worth highlighting that this problem is present whether the pension system is of the PAYG or defined contributions variety. In the end, it is a reflection of the failure of the underlying assumption made when contributory systems started in the Region: that eventually all workers would be formal, and would contribute to their pension during most, if not all, of their working lives. Pension systems —of any variety—will hardly be able to smooth consumption overtime if workers are only forced to save for their pension during, say, half of their working lives.

Importantly, the problem is not a result of low contribution rates. Peruvian workers contribute 13 percent of their salary to their retirement pension; a rate higher than that of Canada or the United States. But this does not improve workers’ pattern of formal-informal transits, and it is estimated that less than half of those contributing will qualify for a pension upon reaching retirement. In fact, raising the contribution rate might worsen the problem. Consider this: Peruvian workers, if employed formally, are forced to save 13 percent of their salary in an account that they cannot pledge as collateral for any other loan, nor use in case of an emergency; and, when they reach 65 years of age, are more likely than not to have their savings returned to them in one lump-sum payment (with interest net of commissions). From this point of view, it is not surprising that workers will perceive that a part of their contribution, rather than being a future benefit for them, is really a tax associated with their formal status.


Lack of pension coverage, particularly among low income workers, represents a major social problem. In response, starting in Brazil in the 1990s but then spreading to practically all of Latin America, countries have introduced pensions to the elderly even if they never contributed to the pension system when they were workers, or even if they never participated in the labor force. Because these pensions are financed from general government revenues and not from a tax on wages, they are usually labelled as “non-contributory pensions,” although of course this is a misnomer, as all tax payers contribute to general revenues (at times they are also called “social pensions,” another misnomer, as if contributory pensions were not social). Although again there is variation across countries, these pensions are given to people usually after 65 or 70 years of age. The amounts paid are the same to all who receive them, although the rules to qualify vary: in some cases, subject to a means test, in others subject to the beneficiary not having access to a contributory pension, and in other cases universal. This variation is reflected in their fiscal costs, which can range from 0.1 percent of GDP in Peru, to 1.0 in Brazil, and 1.2 percent in Bolivia (with a regional average of 0.5 percent).

Non-contributory pensions serve a critical role in reducing old-age poverty, and from this perspective they are very welcome. They also contribute to extend the coverage of retirement pensions. But it is hard to argue that they help to increase the coverage of contributory systems. In fact, the opposite is more likely. Depending on amounts and conditionality, workers will eventually ask: “If when I reach old age I can obtain a pension without being forced to save at all while working, why should I participate in the contributory system when it is not even assured that I will qualify for a pension?” And because the non-contributory pension is more significant for low wage workers than for others (as a share of their income), this question will be more relevant to them than to those with higher wages.

Moreover, recall that formally employed workers have to pay for a bundled package of benefits, not only for retirement pensions. If “non-contributory health programs” are also offered, as is the case in seven countries in Latin America, the relevance of the question posed above is increased. It is useful to note that, on average, governments in the region spend 1.7 percent of their GDP on non-contributory programs (retirement pensions, health, and others). From a social perspective, this is understandable and indeed desirable. But from an economic perspective the result is that informal employment is implicitly subsidized (while formal employment is taxed to the extent that workers undervalue the benefits of participating in the contributory system).  This combination of taxes and subsidies is particularly worrisome for two reasons: First, because there are large productivity differences between the formal and informal sectors; and second, because stagnant productivity growth is the main reason why the region has grown slowly in the last decades (ignoring the 2003-2008 commodity-driven boom). 


A variety of circumstances characterize the status quo of pensions in the region. In Chile, the country that pioneered the defined contribution model and where the first cohorts of beneficiaries are reaching retirement age, there is disappointment because replacement rates are below expectations. This has led some voices to propose a return to PAYG pensions. A similar proposal is under discussion in El Salvador. These proposals are unfortunate since, as argued, they will not fix the underlying problem. In Brazil, reform of contributory pensions is at the heart of the unavoidable fiscal correction currently underway. In Argentina, a few years ago workers’ deposits in their individual savings accounts were nationalized as part of a return to the PAYG system. In Peru, last year legislation was passed eliminating the obligation to buy an annuity even for workers who do qualify for a contributory pension (de facto, turning pension fund administrators into saving fund administrators); workers are now allowed to take 95 percent of the funds accumulated in their individual account immediately at the time of retirement. In Colombia, the PAYG system co-exists with the defined contribution one, and workers arbitrage between the two. In parallel, many countries are saddled with the actuarial deficits of their PAYG systems and will have to deal with the burden of subsidizing them from the government’s budget for many years to come, regressive as these expenditures may be.

In parallel, demographic transition in Latin America implies longer life expectancies and the fading of the “demographic bonus” (so that the elder population, say 70 years of age or more, will grow faster than the rest). Larger cohorts of workers will be reaching retirement, and a majority will do so without a pension, because they worked informally all their lives and never contributed to one; or because even if they did contribute, they failed to accumulate the minimum years of formality required to qualify for one. In this context, one can visualize further expansion of non-contributory pensions, putting further pressure on already stressed fiscal budgets, on one hand; and negatively affecting the incentives to participate in contributory systems, on the other. At the same time, there may be increasing dissatisfaction, perhaps even a feeling that an important promise is being unfulfilled, with difficultly to foresee political implications.


There are usually no simple explanations for complex problems and retirement pensions are not the exception. But two elements are central. First, there has been a persistent underestimation of the challenge posed by informal employment. The expectation was that it would gradually fade away, as a result of faster growth or of increased education. In hindsight, this has not been the case. Informality is not a transitory nuisance; it is a structural characteristic of many countries, with deep and complex roots. Nevertheless, economists and policymakers have yet to fully reflect this fact in the technical design of contributory pension systems; the core building block—forcing workers to save through a tax on wages—persists. Economists have long debated the merits of PAYG versus defined contribution systems. Economists have also debated how to best regulate pension fund administrators, or how to adjust the retirement age, or how to deepen the market for annuities, or how to stimulate voluntary savings with nudges and the like. And yet, important as all these debates are, they are clearly second order compared to the challenge that informal employment represents. Informality is the elephant in the room that needs to be faced directly and placed at the center of the analysis.

Second, Latin America’s political systems have difficulty dealing with the long-lived nature of pensions. Projecting a deficit in a PAYG system 10 or 20 years hence is a weak incentive to induce a timely response. Similarly, a projection that over half of workers contributing to the pension system will not qualify for one is a weak incentive to act, grave as the implication of that projection is. A political calculus of the inter-temporal distribution of costs and benefits is, more often than not, unfavorable. The thinking may be that pensions are a future problem that a future government will have to deal with. If so, this thinking is flawed: The future has been with us for some years already. But the region has not developed the legal framework to induce presidents and congresses to put this problem high in the public policy agenda; often the urgent has dominated the important. In the end, the institutions necessary to effectively manage a process with long lags between benefits and costs have been absent.

Share this