Abstract

Public pension systems are coming under increasing financial pressure around the world. As populations age, the number of pensioners is rising relative to the number of workers in many countries, resulting in increasing pension system deficits as spending on benefits rises relative to contribution revenue.

Public pension systems are coming under increasing financial pressure around the world. As populations age, the number of pensioners is rising relative to the number of workers in many countries, resulting in increasing pension system deficits as spending on benefits rises relative to contribution revenue.

Although Central America has more favorable demographics than many countries with low or even declining birth rates, its population will still age substantially in the coming decades. This, together with generous benefits relative to contributions and low retirement ages, makes many of Central America’s public pension systems unsustainable over the long run under their current parameters. If not addressed, these rising pension system deficits will crowd out private sector activity and may threaten macroeconomic stability.

Reforming public pension systems has been politically challenging both in Central America and the rest of the world. Pension reforms often face substantial political opposition, causing either delays in reforms—and therefore higher adjustment costs in the future—or reforms that are insufficient to put the system on a sound footing. Mobilizing broad political support for reforms can therefore be as important as the design of the reforms themselves.

Nonetheless, some countries in Central America have implemented significant reforms in recent years, ranging from the establishment of funded, defined-contribution systems in El Salvador and the Dominican Republic to the creation of multipillar systems in Costa Rica and Panama.

However, the region’s pension systems still face substantial challenges. Many of the region’s defined-benefit systems have large unfunded liabilities over the long run, while countries that switched to funded, defined-contribution systems face high transition costs and potentially large implicit liabilities due to minimum pension guarantees. Coverage also remains low, with less than 25 percent of the labor force contributing to the main pension system on a regular basis in most countries.

This chapter analyzes Central America’s pension systems, taking stock of recent changes and discussing options for future reform. To provide background for this analysis, the chapter first reviews the structure of the systems and assesses their sustainability. Given that almost all governments are currently considering pension reforms, the chapter also provides an overview of recent reform efforts. The chapter then highlights a number of options and directions for future reform, drawing on international experience with pension systems.

Basic Structures of Public Pension Systems

Central America has an increasingly wide range of public old age pension systems. Most of the region’s systems were started between the 1940s and 1970s as traditional pay-as-you-go, defined-benefit systems. In Guatemala, Honduras, and Nicaragua, the public pension systems still retain this basic structure. In contrast, El Salvador and the Dominican Republic both recently closed their traditional defined-benefit systems to new entrants, replacing them with defined-contribution systems of privately-invested individual accounts. In the middle of the spectrum are Costa Rica and Panama, both of which recently added second pillar defined-contribution systems while keeping their traditional defined-benefit systems as the first pillar of their pension systems. The main features of these various systems are summarized in Table 3.1 and described further below.

Table 3.1.

Main Characteristics of Mandatory Public Old Age Pension Systems

article image
Sources: Country authorities; and U.S. Social Security Administration, 2003, Social Security Programs Around the World: The Americas. Note:The table reflects current mandatory old age pension systems once any approved reforms have been fully phased in.

For old age, disability, and survivors insurance.

Once system is fully phased in in 2035. Replacement rates will be lower for high-income individuals.

Retirement age is 55 for miners, teachers, and the physically or mentally impaired. Workers with average earnings less than twice the minimum wage receive higher replacement rates (e.g., with 15 years of service, a worker earning below twice the minimum wage receives a 65 percent replacement rate, while a worker earning over twice the minimum wage receives a 57 percent replacement rate).

Once reforms are fully phased in in 2013. Applies only to monthly wages up to $500. Wages over this amount are taxed at a rate of 3.5 percent to help fund the scheme. In addition to its contribution, the government earmarks a variety of other taxes to help fund the system.

Contribution rate excludes administrative fees and 1.5 percent for a mandatory severance pay scheme.

Refers only to the contributory regime (see Appendix 3.1 for a description of the different regimes). Contribution rates will rise to these levels by 2008. Workers can retire at age 55 if they have accumulated enough assets to purchase an annuity equal to 150 percent of the minimum pension. Includes 1.6 percent for disability and survivor insurance and administrative fees.

Contributions include 1.3 percent of wages for disability and survivor insurance and 1.4 percent of wages for administrative fees. Retirement is permitted at any age if the insured has 30 years of contributions and the pension equals at least 60 percent of basic earnings or 160 percent of the current minimum pension.

Only applies on monthly wages over $500. On earnings over this amount, workers make an additional contribution of 3.5 percent to the defined-benefit scheme (the “solidarity contribution”).

Defined-Benefit Systems

The traditional defined-benefit systems in Central America are run by social security agencies that administer an array of benefits. Typically, these social security systems are divided into a long-term branch that provides old age, disability, and survivor pensions and a short-term branch that provides sickness, maternity, and work injury benefits. Old age pensions are the primary focus of this chapter, since they are the largest single social security expenditure item and the main source of long-run unfunded liabilities.

Old age pensions in Central America’s defined-benefit systems are determined as a percentage of pre-retirement income. At retirement, an individual’s average wage is calculated over some period of years (anywhere from three to 20 years, depending on the system). The individual’s pension is then set as a percentage of the individual’s average wage. This percentage, or replacement rate, generally rises with an individual’s years of contributions and varies from 50 to 100 percent (Table 3.1). In most countries, minimum and maximum pensions are also set in nominal terms. Once an individual’s pension is set at retirement, it is adjusted periodically through ad hoc, discretionary increases.

The defined-benefit systems are primarily financed by a payroll contribution that is split in some proportion between the employer and employee. These contributions are typically divided into portions that are earmarked for the long-term (old age, disability, and survivor pensions) and short-term (maternity, sickness, and work injury) branches of the social security system. The contributions for old age, disability, and survivor insurance vary by country from 3 to 13.5 percent of an employee’s wages.

In several countries, the central government also subsidizes the social security system directly. For example, in Guatemala the government pays 25 percent of benefits, while in Costa Rica, Honduras, and Panama, the government provides subsidies equal to 0.25 percent, 0.5 percent, and 1.04 percent of an employee’s covered wages, respectively. The Costa Rican and Panamanian governments also make a variety of additional transfers to their social security funds, including the proceeds of an earmarked alcohol tax (Panama) and profits of some state-owned enterprises (Costa Rica).

Most of the region’s defined-benefit systems have a weak link between contributions and benefits. In most cases, an individual’s pension is calculated as a percentage of his or her average wage over just the three to five years immediately before retirement. (Costa Rica, which calculates the average wage over 20 years, is the notable exception.) Such systems thus discourage individuals from earning and reporting high incomes during most of their career, since higher contributions outside of the years immediately before retirement do not affect one’s pension and are therefore a pure tax. Conversely, individuals have a strong incentive to earn and report very high amounts in the years immediately before retirement in order to boost their pensions. Short averaging periods are also inequitable in that they favor individuals with steep lifetime earnings profiles (typically higher-income individuals) over those who have flatter earnings profiles (typically lower-income individuals).

The defined-benefit systems operate largely on a pay-as-you-go basis. That is, current contributions primarily pay for current pension benefits, rather than being invested to fund current workers’ future pensions. However, all of the systems are partially funded in that they have built up at least some reserves.

Defined-Contribution Systems

In contrast, the Dominican Republic and El Salvador are in the process of switching to funded, defined-contribution systems. Both countries used to have traditional defined-benefit systems. However, El Salvador closed its defined-benefit system to new entrants in 1998, with a view to phasing it out. Workers under the age of 36 were automatically transferred into a new defined-contribution system of individual accounts. The vast majority of workers between ages 36 and 55 (for men) or 50 (for women) also voluntarily chose to switch to the new system. The Dominican Republic instituted a similar reform in 2003, with age 45 being the cut-off above which transfer to the new scheme was optional rather than mandatory.

Costa Rica and Panama have also introduced systems of mandatory individual accounts, but as supplements to, rather than replacements for, their defined-benefit systems. These countries thus have true multipillar systems. In Costa Rica, this system was adopted in 2000, while Panama’s system was approved in December 2005 and will be phased in over several years.

As with defined-benefit schemes, the defined-contribution schemes in Costa Rica, the Dominican Republic, and El Salvador are financed by payroll contributions divided between the employer and employee. However, instead of going to pay current benefits, the payroll contributions are invested in individual accounts with private pension funds, known as pension fund administrators (Administradoras de Fondos de Pensiones–AFPs). Upon retirement, individuals choose either to make programmed withdrawals from the account or use the account to purchase a private annuity (or some combination of the two).2 Both the Salvadoran and Dominican systems guarantee that the resulting pensions will be at least at some minimum level. In the Dominican Republic, the minimum pension is equal to the minimum wage and funded by 0.4 percent of the payroll contribution. In El Salvador, the minimum pension is determined each year in the annual budget (in 2006, it was $114 per month, or 72 percent of the minimum wage) and paid by the central government. A portion of an individual’s payroll contributions pays for the AFPs’ administrative expenses. In the Salvadoran and Dominican systems, a portion of the contribution is also used to purchase private life and disability insurance, since there is no first-pillar social security system that provides these benefits. In all countries, the AFPs are regulated by a superintendent of pensions.

In Panama, the defined-contribution scheme will apply only to workers with monthly wages over $500. These workers will be affiliated with the first-pillar defined-benefit scheme on the first $500 of their monthly wages, together with workers earning less than $500 (approximately the average wage). On earnings over this amount, workers will contribute 3.5 percent to the first pillar (as a “solidarity contribution”), with the remainder of their contributions (10 percent of wages) going into individual investment accounts that will be managed by the social security agency. Individuals will be able to withdraw the amount in these accounts gradually upon retirement. In the very long run, only the individual investment accounts, which are expected to be sustainable, will remain as the $500 per month limit shrinks in real terms, assuming the government sticks to its intention to keep the $500 limit fixed in nominal terms (see Appendix 3.1 for more details on this recent reform).

Comparison of Key Parameters

Coverage

Most Central American pension systems cover a relatively small portion of workers. Only about 15 to 20 percent of the labor force contributes to the main pension system in El Salvador, the Dominican Republic, Guatemala, Honduras, and Nicaragua (Figure 3.1). These low numbers partly reflect the large informal sectors in these countries, as well as the fact that public sector workers are covered by separate, special regimes in some countries. The notable exceptions to low pension coverage are Costa Rica and Panama, which have achieved coverage rates of approximately 50 percent due to their larger formal labor markets.

Figure 3.1.
Figure 3.1.

Pension Coverage: Number of Contributors as a Share of the Labor Force, 2003

(In percent)

Sources: Country authorities; World Bank, World Development Report database; and IMF staff estimates.

Because coverage applies mainly to the formal sector (where incomes are higher than for the population as a whole), pension beneficiaries tend to be relatively high-income individuals. Consequently, government subsidies to pension systems (such as those mentioned in the previous subsection) tend to be quite regressive. For example, one study finds that, in Guatemala, 80 percent of net pension subsidies (government transfers to pension systems to fill gaps between contributions and benefits) accrue to the richest 20 percent of the population, while the poorest 20 percent receive less than 2 percent of net pension subsidies (Lindert, Skoufias, and Shapiro, 2006).

Contribution Rates

Contribution rates in Central America are somewhat low by international standards, but have been rising recently in some cases. Contribution rates in the region for old age, disability, and survivor insurance range from 3 percent of earnings in Honduras to 13.5 percent in Panama,3 a range that is somewhat lower than the Latin American average (excluding Central America) of 15 percent and well below the average of 19 percent for member countries of the Organization for Economic Cooperation and Development (OECD) (Figure 3.2). However, the trend is toward higher rates—Costa Rica, the Dominican Republic, and El Salvador all effectively raised their total contribution rates in the process of introducing defined-contribution systems; Nicaragua raised its contribution rate from 5.25 to 10 percent in 2000; and Panama recently adopted reforms that will raise its contribution rate from 9.5 percent in 2006 to 13.5 percent by 2013.

Figure 3.2.
Figure 3.2.

Contribution Rates for Mandatory Old Age, Disability, and Survivor Pensions

(In percent)

Source: U.S. Social Security Administration, 2003, Social Security Programs Throughout the World.Note: OECD = Organization for Economic Cooperation and Development.1 Rates gradually rising to this level by 2035.2 Rates gradually rising to this level by 2008.3 Rates when recent reforms are fully phased in in 2013. Contributions on monthly wages over $500 also have a defined-contribution element.4 Simple average.

Some pension systems also place a ceiling on earnings that are subject to contributions. The application of this ceiling varies across countries. For example, in Honduras it is relatively low at only about twice average earnings, while in Nicaragua it is largely nonbinding at nearly 10 times average earnings.4 In Costa Rica, there is no ceiling at all. Since ceilings are placed on maximum pensions, the absence of a ceiling on earnings effectively taxes the earnings of very high-income workers to subsidize the rest of the system.

Retirement Ages

Like contribution rates, retirement ages in Central America are somewhat low by international standards. Standard retirement ages range from 55 to 65, with 60 being the most common (Table 3.1). Costa Rica, El Salvador, Honduras, and Panama have lower retirement ages for women than for men, despite women’s longer life expectancy. Average retirement ages for the region are similar to the averages for the rest of Latin America, but well below the OECD average (Figure 3.3). This is true despite the fact that average life expectancy at age 60 in Central America is basically the same as in OECD countries for men and only somewhat lower for women (larger differences in life expectancy at birth are mainly due to differences in infant and child mortality).5 However, the minimum contribution requirements—15 to 39 years, depending on the country (Table 3.1)—mean that some workers have to work beyond the standard retirement age before becoming eligible for old age pensions, especially given the large informal sector that results in many workers contributing only during a portion of their working career.

Figure 3.3.
Figure 3.3.

Standard Retirement Age and Life Expectancy at Age 60

(In years)

Sources: U.S. Social Security Administration, 2003, Social Security Programs Throughout the World; United Nations, 1999, Population Aging; and country documents.Note: Simple averages for each group. Life expectancy is expressed in terms of age 60 plus the expected remaining years of life, given that an individual has reached age 60. OECD = Organization for Economic Cooperation and Development.

Benefits

Average pensions in the region are comparable to international averages, despite the relatively low contribution rates and retirement ages. In the defined-benefit systems, average old age pensions range from 38 to 86 percent of average earnings (Table 3.2), levels that are broadly similar to the average of 57 percent across OECD countries.

Table 3.2.

Defined-Benefit Systems: Average and Minimum Old Age Pensions

(Percentage of economy-wide average earnings)

article image
Sources: Country authorities; Organization for Economic Cooperation and Development (2005); and IMF staff estimates.Note: Data are for 2003, except for Honduras (2002). Economy-wide average earnings are calculated as total annual covered earnings divided by the number of active contributors. OECD = Organization for Economic Cooperation and Development.

However, there is substantial variation in average pensions across countries. Panama’s pensions are the high-end outlier at 86 percent of average earnings, a ratio that may be due in part to Panama’s high minimum pension (41 percent of average earnings).

There is also substantial variation across age groups within countries, with average pensions typically falling with age. For example, in 2001, the typical 65 year old in Costa Rica had a pension equal to 67 percent of the economy-wide average wage, while the typical 78 year old had a pension that was only 34 percent of the economy-wide average wage. These differences across age groups reflect some combination of pensions growing more slowly than wages once they are set at retirement and recent retirees having more years of contributions and therefore higher replacement rates. These factors also explain why average pensions in most countries are only 40 to 50 percent of average earnings, despite statutory replacement rates of 50 to 100 percent (Table 3.1). More years of contributions for more recent retirees may imply, however, that average pensions will rise over time. In any event, the mix of low contribution rates, low retirement ages, and relatively high benefits is feasible in most of Central America’s defined-benefit systems only as a result of the currently favorable demographics. Over the longer run, this combination is unsustainable.

Sustainability

Defined-Benefit Systems

Most of the region’s defined-benefit systems have run moderate cash flow surpluses and built up reserves in recent years (Table 3.3). These surpluses are due to the region’s relatively young population (Table 3.4), which results in a high ratio of contributors to pensioners. The one exception is Panama, which has already begun to run cash flow deficits due to its more generous benefits.

Table 3.3.

Defined-Benefit Systems: Balances and Reserves, 2004

(In percent of GDP)

article image
Sources: Country authorities; and IMF staff estimates.Note: For old age, disability, and survivor insurance.

Estimated out-turn for 2005. Reserves are total assets in investment accounts for the entire social security system times the portion of earnings on these assets that are allocated to the old age, disability, and survivor’s branch.

Table 3.4.

Projected Population Aging

(In percent)

article image
Source: U.S. Census Bureau international database: http://www.census.gov/ipc/www/idbnew.html.

In the future, the systems’ demographics will deteriorate significantly. Over the coming decades, the ratio of contributors to pensioners will fall dramatically in Central America as the population ages. In particular, if contributors and pensioners remain constant fractions of the age 20–59 and 60+ populations, respectively, then the number of contributors supporting each pensioner would fall by about threefold in most countries (Table 3.4). Moreover, the share of elderly that are pensioners is likely to rise as pension systems mature. For example, in Guatemala old age pensioners currently constitute only about 5 percent of the population age 65+, whereas contributors constitute 18 percent of the working age population. As current contributors age, it is likely that the pensioner share of the elderly population will rise closer to 18 percent. While the low coverage of the labor force also leaves room for expanding the share of the working age population that contributes, this would probably only offset the rising pensioner share of the population, at best, under current policies.

The effect of a worsening ratio of contributors to pensioners can be seen by looking at the basic arithmetic of pay-as-you-go systems. In a pure pay-as-you-go system, sustainable pension parameters must satisfy the following condition: r= cX, where r = the average gross replacement rate (the average pension benefit divided by the average wage), c = the contribution rate, and X = the ratio of contributors to pensioners.6 From this expression, one can see that, if X falls on average from 7.8 to 2.8 as projected, then either contribution rates would have to be raised threefold or average pensions would have to be cut threefold to maintain the current balance. While the pension systems’ current surpluses and reserves give them some cushion against such a demographic shock, significant changes to the parameters of the systems—either higher contribution rates, lower benefits, or higher retirement ages to improve the ratio of contributors to pensioners—will nonetheless be required eventually, especially considering that population aging will likely strain the survivor (mainly elderly widows and widowers) and disability portions of social security systems as well.

Another way to see the systems’ unsustainability is to examine their internal rates of return. The internal rate of return is the discount rate that equates the present discounted value of an individual’s expected contributions and pension benefits. In other words, the internal rate of return is the rate of return that an individual effectively receives on his or her contributions. In a pure, steady-state, pay-as-you-go system, the sustainable real internal rate of return is the real growth rate of the wage bill (perhaps 4 percent over the long run, assuming labor force growth of 1.5 percent and real wage growth per worker of 2.5 percent).7,8 In a fully-funded system, the sustainable internal rate of return is equal to the real return on investments (perhaps 3 to 6 percent, depending on how much risk the system is willing to bear).

The parameters of Central America’s current defined-benefit systems entail real internal rates of return that are unsustainable. Real rates of return for stylized cases are shown in Table 3.5. The rates range from 5½ to 13½ percent, with the recently-reformed Panamanian system being the outlier on the low end. In general, these internal rates of return are quite high and will only grow over time as longevity improves. Such rates of return are clearly unsustainable, indicating that current contribution rates are insufficient to fund the promised replacement rates and retirement ages over the long run.

Table 3.5.

Defined-Benefit Systems: Real Internal Rates of Return on Contributions

(In percent)

article image
Source: IMF staff estimates.Note: For old age benefits once all approved reforms are fully phased in. Based on a stylized individual who contributes for 30 years, spread out evenly over the individual’s working career before retirement at the standard retirement age. Assumes that the old age benefit share of contributions is proportional to the old age benefit share of expenditure and that pensions-in-payment grow with inflation. Mortality rates are based on Nicaraguan data. Results do not differ significantly if Costa Rican mortality rates or an alternative number of years of contributions are used.

To fully assess the sustainability of pension systems, however, proper actuarial analyses are required. Such analyses project future cash flows based on detailed data on the profile of existing contributors and pensioners, together with assumptions regarding the future path of economic and demographic variables. Unfortunately, such thorough actuarial analyses are not readily available for all countries. Moreover, the results of actuarial analyses can be sensitive to assumptions regarding long-run growth rates, interest rates, and increases in longevity, as well as the time horizon chosen, which makes direct comparisons across analyses for different countries difficult.

Nonetheless, available actuarial analyses confirm that many of Central America’s defined-benefit systems are unsustainable. For example, recent analyses indicate that Nicaragua’s old age, disability, and survivor pension system will begin running cash flow deficits around 2017–18. Panama’s defined-benefit system is already in deficit, although recent reforms are expected to gradually phase out the system (and therefore its deficit) over the very long run. Costa Rica’s system was projected to go into deficit by 2018, but the 2005 reforms extended this date to 2040 according to the government’s actuarial analyses (Superintendencia de Pensiones, 2005).9

Defined-Contribution Systems

The region’s funded, defined-contribution systems are generally sustainable by construction, except to the degree that low pensions trigger government guarantees. Two countries have defined contribution systems as a first pillar—the Dominican Republic and El Salvador. In theory, these systems should be self-funding, with each beneficiary simply receiving whatever retirement benefits can be purchased by his or her accumulated savings. However, if this level of benefits is very low, the government will be required to subsidize pensions to bring them to the guaranteed minimums that exist in both countries.

The magnitude of future government liabilities arising from minimum pension guarantees will depend on the rates of return that contributors earn on their savings and their years of contributions. Table 3.6 shows the replacement rates (pension benefits as a share of pre-retirement earnings) that retirees would be able to achieve in the Dominican and Salvadoran systems in certain stylized cases, depending on the rate of return that individuals earn on their investments and their years of contributions. The estimates indicate that real rates of return of less than 5 percent in the Dominican Republic and 4 percent in El Salvador are likely to lead to low replacement rates (10 to 30 percent) that will likely trigger significant fiscal costs to fund minimum pension guarantees. The size of fiscal liabilities generated by these first-pillar, defined-contribution systems will thus depend crucially on investment performance.

Table 3.6.

Replacement Rates under Different Rates of Return in First-Pillar, Defined-Contribution Systems

(In percent)

article image
Source: IMF staff estimates.Note: For a male beginning work at age 20 and retiring at age 60; estimates assume that contributions are spread out evenly over the individual’s working career, that current mortality rates will apply by the time the individual retires, and that the annuity mark-up over actuarial value is 15 percent.

The Dominican Republic and El Salvador also face significant legacy costs from their old defined-benefit systems. When these countries changed their systems, most workers switched to the new defined-contribution systems and stopped contributing to the old systems. However, the old defined-benefit systems retained their obligations to pay pension benefits to current retirees and older workers who opted to remain in the old system. Moreover, in 2003 a government decree in El Salvador guaranteed that individuals who opted to move to the new defined-contribution system (those between ages 36 and 50/55 at the time of the reform in 1998) would receive a pension at least equal to what they would have received under the old system (despite the fact that these individuals’ contribution rate in the new system is 1 percentage point lower than for those who opted to stay in the old system). The vast majority of pensioners in this cohort will need a significant top-up at retirement in order to meet this guarantee, since the returns on their individual accounts will almost certainly be less than the very high implicit internal rates of return that they would have received under the old system. This guarantee will thus significantly increase the legacy costs of old system.

These old systems will run substantial deficits during the transition due to these legacy costs. For example, El Salvador’s old system currently requires transfers from the central government of nearly 2 percent of GDP, a figure that is projected to rise to 2.7 percent of GDP by 2018, before gradually declining to near zero by 2040. In the Dominican Republic, transfers from the central government to the legacy system were 0.8 percent of GDP in 2005.

Administrative Costs

Central America’s defined-benefit systems vary somewhat with regard to their administrative costs. Costs range from 3 to 11 percent of total contributions (Table 3.7). Overall, administrative costs are significantly higher than in some developed countries. For example, administrative costs in the United States are only 0.5 percent of contributions. Such differences may reflect economies of scale and other country-specific factors. However, they may also indicate an opportunity to lower administrative costs in some Central American systems.

Table 3.7.

Administrative Costs in Defined-Benefit Systems

article image
Sources: Country authorities; and IMF staff estimates.Note: For old age, disability, and survivor portion of social security schemes in 2004, unless otherwise noted.

For 2003.

For 2001.

For complete social security system.

Administrative costs in the privately-administered, defined-contribution systems are somewhat higher than in the publicly-administered, defined-benefit systems. In Costa Rica, roughly half of AFPs charge the maximum allowable for administrative fees—4 percent of contributions and 8 percent of returns—although fees on contributions go as low as 2.5 percent for some AFPs. In the Dominican Republic, administrative fees are set at 0.6 percent of an individual’s wages (0.5 percent for AFPs and 0.1 percent for the pension supervisor), plus a certain percentage of returns over benchmark yields. In El Salvador, the maximum allowable fee is 1.4 percent of wages. These fee levels reduce effective contributions (and therefore balances at retirement) by roughly 13, 8, and 11 percent in Costa Rica, the Dominican Republic, and El Salvador, respectively.10

Investment Policies

Available data indicate that both defined-benefit and defined-contribution plans invest the vast majority of their assets in domestic government securities and bank deposits. For example, 70 percent of the assets in Costa Rica’s defined-benefit plan were invested in government bonds at end-2003. The defined-contribution plans (AFPs) have similar portfolios—for example, government and central bank securities account for 89 and 83 percent of investments in the defined-contribution systems in Costa Rica and El Salvador, respectively. In some cases, such portfolios may be due in part to regulatory limits on other types of investments. In the Dominican Republic, investments are primarily with banks, since AFPs are not allowed to invest directly in government paper. However, banks in turn invest a significant portion of the AFPs’ deposits in government securities.11 The real rate of return on AFPs’ investments has varied from 1 to 15 percent in recent years because of macroeconomic volatility (especially in the Dominican Republic due to the recent banking crisis).

Special Regimes

Many Central American countries have special pension systems for civil servants and uniformed personnel. The basic structure of these systems is similar to that of the regular defined-benefit systems, but typically with even more generous benefits (lower retirement ages, higher replacement rates, etc.) relative to contributions. The importance of these special systems varies across countries. For example, in Honduras the unfunded liabilities of these special systems are much larger than those of the regular system. In contrast, the special pension system in El Salvador for government employees is being phased out as part of the transition to the new private defined-contribution system. Special regimes are not the main focus of this chapter, in part due to a lack of detailed actuarial analysis and publicly available information. However, the fundamental issues facing these systems, particularly the unsustainability of benefits, are in many cases similar to—or more severe than—those facing the regular pension systems.

Several countries also have special noncontributory regimes for low-income groups. Noncontributory, means-tested pensions are provided to the elderly poor and other vulnerable groups in Costa Rica (financed by a 5 percent employer payroll tax and 20 percent of sales tax revenue), the Dominican Republic (financed by the government), and Nicaragua (financed by the regular pension system).

Recent Reform Developments

The need for pension reform has been recognized by all governments in Central America. While reform efforts cover a wide spectrum, they are uneven across countries. As pointed out above, some countries (the Dominican Republic and El Salvador) are in the process of privatizing their systems—that is, switching to a fully-funded, defined-contribution system of privately-managed individual accounts. Others (Costa Rica and Panama) have recently chosen to adopt multipillar systems. These countries are now considering or undertaking second-generation reforms to address ongoing unsustainable initial pillars (in multipillar systems) or high transition costs and potentially large contingent liabilities (in privatized systems).

Some countries have started to address unsustainability through parametric reform (reform of pension system parameters such as contribution rates and benefit formulas). A number of countries (Costa Rica, the Dominican Republic, and Panama) have started to raise retirement ages, increase contribution rates, tighten eligibility requirements, or change a combination of parameters. Appendix 3.1 provides a country-by-country overview of such recent reform efforts. Despite these reforms, however, further reforms are necessary.

The experience of reform efforts in Central America also demonstrates how difficult it is to move pension reform forward and generate or sustain sufficient public support. Fractured or polarized political landscapes and elections in Nicaragua (2006) and Guatemala (2007), for example, have delayed reforms. The experience of Panama and El Salvador, in turn, demonstrates how public opposition can lead to weaker reforms. In Panama, a strong parametric reform to address the system’s actuarial imbalance was ultimately withdrawn as a result of public opposition, despite passage by the assembly in mid-2005. To address public concerns, the government engaged in a five-month consultative process with civil society, which ultimately resulted in passage of a revised but somewhat weakened reform in December 2005. In El Salvador, authorities tried to ensure popular support of the privatized system by offering costly guarantees to those who chose to switch to the new privatized system.

Pension Reform and Development of Local Capital Markets

At present, financial systems in Central America are largely bank-based, and capital markets remain underdeveloped. In particular, equity and corporate bond markets are limited to a handful of companies (except in Panama). Instead, public debt markets account for the lion’s share of market-based financing in Central America, and much of this is in foreign currency (Panama and El Salvador officially adopted the U.S. dollar as legal tender). Furthermore, local markets in domestic currency tend to be short term.

Some argue that privatizing parts of the pension system (as in Costa Rica, El Salvador, and the Dominican Republic) can have the positive side effect of fostering capital market development, especially through the creation of pension funds and an increase in institutional investors.12 In particular, such proponents suggest that decentralizing and increasing the number of managers of pension assets can promote development of “institutional capital” and competitive pricing of financial assets.13

Pension reforms that reduce fiscal liabilities could also contribute to the development of capital markets in Central America by reducing macroeconomic vulnerabilities and increasing national savings. The lack of developed capital markets in the region reflects in part persistent and relatively high levels of fiscal deficit and debt. In the absence of reform, rising pension system deficits may thus further undermine capital market development. Conversely, pension reforms that reduce unfunded liabilities (e.g., by raising retirement ages and contribution rates and reducing replacement rates) could promote capital markets by reducing fiscal deficits and associated macroeconomic vulnerabilities. Such reforms may also increase national savings, which in turn could be associated with positive externalities such as the deepening of financial markets, increased liquidity, and lengthened maturities. The establishment of benchmark yield curves, in turn, provides the foundation for the development of corporate debt markets.

To date, however, privately-managed pension pillars have had limited success in promoting capital markets in Central America or Latin America overall (World Bank, 2006), perhaps in part because reforms have not been sufficiently aimed at reducing unfunded pension liabilities, there has been a lack of complementary reforms to security laws and regulations, and pension funds have been very tightly regulated (Roldos, 2004). The decentralization of pension asset management may also have come at the cost of reducing administrative economies of scale.

Moreover, requiring pension funds to invest in underdeveloped domestic markets can lead to distortions and potential asset price bubbles. It also heavily exposes pension systems to country-specific risk. Indeed, the experience from Latin America demonstrates that most pension funds are not well diversified and are highly exposed to sovereign risk (World Bank, 2006). However, rapidly relaxing restrictions on foreign investment could lead to capital outflows that could undermine macroeconomic management. Such relaxations should thus be managed gradually.

Reform Options

Pension reform in Central America needs to be tailored to each country’s specific circumstances and preferences. However, some common features of reform approaches can be identified, given the commonality of issues across many of the region’s pension systems. In particular, policymakers may wish to consider the suggestions that follow in terms of reform.

Parametric Reform of Defined-Benefit Systems to Make Them Sustainable

As noted earlier, many of the region’s defined-benefit systems are unsustainable due to the aging of populations and to benefit formulas that are very generous relative to contributions. Some combination of higher contribution rates, higher retirement ages, and lower replacement rates is therefore needed to balance contributions and benefits over the long run. Such adjustments should be the first priority of pension reform.

How such adjustments are allocated among contribution rates, retirement ages, and replacement rates will vary according to country-specific circumstances and preferences. In countries where contribution rates are relatively low (e.g., Honduras and Guatemala), contribution rates could be increased somewhat. In many countries, however, the scope for raising contribution rates may be limited, given the adverse consequences that higher rates may have on development of the formal sector.14 In contrast, there is scope in many countries to scale back replacement rates (which reach up to 100 percent) to more moderate levels and to raise retirement ages to reflect increased longevity. Retirement ages for men and women could also be equalized in some countries.

Parametric reforms will likely need to be phased in over time to allow individuals to adjust their retirement planning to the new rules and make such reforms politically palatable. However, if reforms are delayed, unfunded liabilities will continue to accumulate, necessitating more dramatic and abrupt reforms in the future. Policymakers should thus take early action to allow a smooth and gradual adjustment process.

Parametric reforms could either be done on their own or as part of a shift to a defined-contribution system. Shifts to defined-contribution systems in Central America typically result in implicit benefit cuts, since rates of return on investments are lower than internal rates of return on existing defined-benefit systems. Shifts to defined-contribution plans can therefore be one way to introduce parametric reforms. This approach may be appealing to policymakers, since the change in systems can make implicit benefit cuts less transparent and therefore more politically acceptable. For example, many Salvadorans appeared not to realize the implicit benefit cuts involved in such a shift, since most voluntarily chose to shift from the generous defined-benefit system to the less generous defined-contribution system. However, such “stealth” parametric reforms entail risks—in El Salvador, those who chose to switch eventually realized that they would be getting a lower pension. They then lobbied the government to issue the 2003 decree guaranteeing that those who switched would get at least as large a pension as they would have gotten under the old defined-benefit system, expanding the transition costs and undoing much of the initial reform.

It should be stressed that shifting to a defined-contribution system reduces a pension system’s unfunded liabilities only to the degree that it reduces expected benefits, increases contributions, or increases the return on the system’s assets. Reforms that simply move from a defined-benefit to a defined-contribution system without altering these underlying parameters do little to solve the sustainability problem. Indeed, privatizations that attempt to shift from an unfunded, defined-benefit system to a funded, defined-contribution system without addressing the underlying imbalance between contributions and benefits in the old system can actually worsen the overall system’s finances by transforming implicit unfunded liabilities (which typically have no legal guarantee and therefore can be altered) into explicit government liabilities (that have a legal guarantee and therefore cannot be altered without triggering a default), as evidence suggests happened with Argentina’s pension reform (Cuevas and others, 2006).

For those countries that shifted to a defined-contribution system, further parametric reform of their old defined-benefit systems would help reduce transition costs. As noted earlier, El Salvador and the Dominican Republic have shifted to defined-contribution systems but still face substantial transition costs because of payments of pensions to those who remain in the old systems. These old systems still offer relatively high replacement rates and low retirement ages; gradually adjusting these parameters to more internationally comparable levels would substantially reduce transition costs.

Tightening the Link between Contributions and Benefits to Reduce Labor Market Distortions

In most of the region’s defined-benefit systems, benefits are only weakly linked to contributions. As a result, contributions in many systems are largely a pure tax, with all of the associated labor market distortions.

One way to tighten the link between contributions and benefits in defined-benefit systems would be to increase the number of years over which an individual’s average wage is calculated. As noted earlier, in most systems, an individual’s pension is calculated as a percentage of his or her average salary only over the three to five years immediately before retirement. To reduce the resulting distortions and inequities, the region’s systems could move to the common international practice of basing pensions on average earnings over 25 to 40 years, with missing years counted as zeroes and wages from past years typically converted into current dollars using an index of the average economy-wide wage. Costa Rica recently adopted such a reform, gradually expanding the wage-averaging period from four to 20 years.

Another way to tighten the link between contributions and benefits would be to move to a defined-contribution system; however, this may come at the cost of increased financial market risk for participants.15 As noted above, defined-benefit systems often have a poor link between contributions and benefits either because benefit formulas are poorly designed (e.g., with the short wage-averaging periods discussed above) or poorly understood due to their complexity. In contrast, defined-contribution systems have a clear link between contributions and benefits—individuals can easily see that each dollar contributed adds to their account. However, the volatility of investment returns in defined-contribution systems creates significant volatility in the replacement rates that any individual or generation receives. For this reason, countries may prefer to maintain a defined-benefit system for their first-pillar, universal public pension system, but improve the tightness and transparency of the system’s link between contributions and benefits.

One system that tries to create a tight and transparent link between contributions and benefits while insuring pensioners against financial market volatility is the notional defined-contribution (NDC) system. NDC systems mimic defined-contribution systems in that they credit contributions in individual accounts that earn a rate of return. Upon retirement, the amount in the account is paid out through some combination of annuitization, programmed withdrawals, or lump sum distributions. NDC systems are thus technically defined-contribution systems and have a very clear link between contributions and benefits. However, NDC systems are like defined-benefit systems and unlike most defined-contribution systems in that the return that the individual accounts earn is not determined by investments in financial markets, but is set and guaranteed by the government, thereby reducing the volatility of replacement rates. Typically, the rate of return is tied in some way to the government’s ability to pay benefits. For example, the return may be set equal to the growth rate of the economy or the wage bill covered by the system. NDC systems are typically unfunded in that current benefits are paid out of current contributions, but such systems could also build up reserves so that they are funded.

NDC systems can be phased in through a variety of mechanisms. One of the simplest is to calculate pensions as a weighted average of what benefits would be under current law and what they would be if the individual had been in an NDC system all along, with the weights varying by birth cohort. For example, to phase in an NDC system over 10 years, those turning 65 next year would receive a pension equal to 10 percent of the NDC pension and 90 percent of the current law pension; those turning 65 in the year after that would receive a pension equal to 20 percent of the NDC pension and 80 percent of the current law pension, and so forth until the NDC pension is fully phased in. Such a transition mechanism is being used in Sweden.16

Strengthening Social Protection

Indexing pensions-in-payment to consumer prices (after replacement rates have first been reduced to sustainable levels) would strengthen pensions’ income protection function. Currently, most of the region’s defined-benefit systems adjust pensions-in-payment annually largely on an ad hoc basis, which has several drawbacks:

  • The discretionary nature of ad hoc increases makes systems vulnerable to political pressures;

  • The erratic nature of ad hoc adjustments raises pensioners’ uncertainty regarding the future path of their pensions in real terms, reducing the system’s insurance value; and

  • Ad hoc adjustments create inequities, since pensioners with similar contribution histories can receive different rates of return on their contributions due to differences in ad hoc adjustments that may occur during their retirement years.

These problems could be avoided by automatically indexing pensions to consumer prices (as in most OECD countries), thereby fixing pensions in real terms and removing discretion for any additional upward adjustment.17 In countries where pensions-in-payment would have otherwise been increased in real terms, such indexation would reduce pension costs. However, indexation will increase costs in countries that rely on nonindexation to reduce the real value of pensions that are set at unsustainably high levels at retirement. In these cases, it may be necessary to first reduce replacement rates to sustainable levels before indexing pensions-in-payment to consumer prices. To safeguard the government’s fiscal flexibility in times of financial stress, it may also be useful to include “safety valves,” such as setting a limit (e.g., 5 percent) on the increase that can occur automatically in any given year, with adjustments for inflation over this amount dependent on government approval based on an assessment of the system’s finances.

It may also be possible to strengthen pension systems’ social protection objectives by reducing the systems’ redistributive elements and using the proceeds to finance well-targeted transfers to the poor outside of the systems themselves. Most public pension systems have some redistributive element, whether it be via minimum pensions or providing relatively higher internal rates of return to pensioners with fewer years of contributions. However, recipients of minimum pensions may not necessarily be poor, due to nonpension income, and pensioners with fewer years of contributions may still have high pensions due to high earnings and short wage-averaging periods. Moreover, the vast majority of the poor are likely to fall outside of the pension system altogether, since they are not heavily involved in formal sector employment. Consequently, pension systems’ redistributive mechanisms may result in substantial leakage to the nonpoor, and subsidization of pensions from general revenue is likely to be poorly targeted and even regressive. Governments may thus wish to consider whether poverty (old age and otherwise) could be more effectively combated by reducing some of the current redistributive elements in pension systems, scaling back general pension subsidies, and using the proceeds to fund better-targeted transfers to the poor outside of the contributory pension system. Costa Rica, the Dominican Republic, and Nicaragua have begun reforms along these lines by establishing noncontributory, means-tested pensions.

Expanding coverage would also provide better protection against poverty among the elderly and improve short-term cash flows. As noted earlier in this chapter, most Central American pension systems cover less than 20 percent of the labor force. Expanding coverage through better enforcement of mandatory contributions will therefore be critical to ensuring that pension systems provide adequate protection against poverty among the elderly. Expanding the contribution base will also improve short-term cash flows. However, long-term imbalances will grow as coverage expands, since more individuals will enter systems that promise unsustainably high internal rates of return. Parametric reform is thus critical to place defined-benefit systems on a sustainable footing before expanded coverage magnifies the long-run imbalance between contributions and benefits.

Reducing the Contingent Liabilities of Defined-Contribution Systems

As noted earlier in this chapter, the defined-contribution systems in the Dominican Republic and El Salvador both guarantee minimum pension levels. As a result, these governments may incur significant liabilities to fund these minimum pensions, especially if real returns on private accounts are modest. These contingent liabilities could be reduced by measures such as raising the minimum age at which one can retire with a minimum pension to 65 or by modestly increasing contribution rates.

Improving Investment Performance

Investment performance could be improved by gradually facilitating more international diversification. Most Central American pension plans (both defined-benefit and defined-contribution) invest almost all of their assets domestically, mainly in government securities and bank deposits. As a result, these plans are highly exposed to risks specific to Central America. These risks could be reduced by increasing investment in well-diversified, professionally-managed, and well-regulated portfolios of international assets. Diversification into international assets should be done gradually, however, in order to minimize any macroeconomic disruptions that may be caused by a shift in capital flows.

Governments should also avoid directing pension plans (whether defined-benefit or defined-contribution) to lend to social and politically-favored projects. While such projects may be worthy, these should be funded from the budget rather than through pension funds in order to ensure that the costs of such projects are transparent and that they compete with other priorities for public funding. Governments may also wish to preclude public defined-benefit plans from investing their reserves in domestic assets other than government securities or bank deposits, since investments in domestic equities or real estate run the risk of becoming politicized. Public pension institutions are also not generally well placed to actively manage and operate commercial enterprises and real estate.

Controlling Administrative Costs

The efficiency of both the region’s public defined-benefit systems and the private defined-contribution systems could be enhanced by controlling and reducing administrative costs, especially in cases where these costs are particularly high (e.g., Honduras). For the public defined-benefit systems, this may mean re-engineering pension institutions to streamline employment costs and enhance use of information technology. For private defined-contribution systems, governments may wish to investigate whether the prescribed level of AFP fees could be lowered and whether regulations adequately promote competition on the basis of administrative costs. Lower AFP fees could perhaps be facilitated by restricting (as El Salvador recently did) how often workers can change AFPs and by limiting marketing costs.

Better Integrating Special Regimes and the Regular Systems

Several countries could achieve significant efficiency and equity improvements by better integrating special regimes (e.g., for civil servants) with the regular system. For example, in Honduras an array of special systems for different types of government employees remain. In these cases, integrating these special systems into the regular pension system could provide more uniformity of treatment, facilitate mobility across sectors (currently hindered because moving from one sector’s plan to another can reduce benefits), and cut administrative costs through economies of scale.

Repealing Excessively Generous Tax Treatments for Pensions

The fiscal cost of pension systems could be reduced by repealing the excessively generous tax treatment that pensions sometimes receive. The commonly-recommended international practice is to tax income placed into pension plans once (but only once), either by making pension contributions nondeductible or by taxing withdrawals from pension accounts.18 However, in some Central American countries (e.g., El Salvador), pension contributions are tax deductible when made, earn interest and dividends tax-free, and are tax-free upon withdrawal. As a result, the original earnings escape the income tax net altogether, giving the pension preferential treatment relative to other forms of income. Repealing this preferential treatment (i.e., by taxing pension withdrawals in countries where contributions are tax deductible) would reduce the tax preference given to the pension system and increase fiscal revenue. It would also likely be progressive, since participants in pension plans tend to be relatively high-income individuals and income tax systems tend to be progressive.

Improving Transparency

Better publication of pension systems’ financial statements and actuarial reviews would improve transparency. Most actuarial reviews in the region are either unpublished or could be strengthened in their analysis. Publishing more detailed data on the finances and actuarial imbalances of pension systems would increase public awareness of the systems’ difficulties, which should both build support for reforms and better inform public debate on possible solutions.19

Collecting and publishing information on the distributional incidence of government subsidies to pension systems may also prompt reform, since existing evidence indicates that participants in public pension systems in Latin America tend to be relatively high-income individuals, implying that general government subsidies to these systems are regressive (Lindert, Skoufias, and Shapiro, 2006). Such subsidies exhaust public resources that could be used for more pro-poor spending in such areas as primary education or health care.

Conclusions

Despite the implementation of reforms in some countries and reform attempts in others, further reforms are necessary. The pension systems in Central America remain either unsustainable or face high transition costs and potentially large contingent liabilities.

While all of the defined-benefit systems are currently running moderate cash flow surpluses (except Panama) and have built up reserves, the systems are in long-run imbalance. In the future, spending will grow more rapidly than revenue as the ratio of working age persons to the elderly falls from an average of about eight today to less than three in 2050. Internal-rate-of-return calculations corroborate the unsustainability of the systems, given that most have real rates of return close to 10 percent, which is substantially higher than the real growth rate of the wage bill and the growth performance of the economies over recent decades.

Pension reforms in Central America need to be tailored to specific circumstances—there is no onesize-fits-all solution. However, some important recommendations are applicable across many countries. First and foremost, any reform should focus on sustainability. Depending on the country’s circumstances, this might call for an increase in contribution rates or retirement ages, a reduction in benefits, or some combination of reforms. Reforms of defined-benefit systems should also aim to strengthen the link between contributions and benefits by lengthening wage-averaging periods. Links could also be improved by moving more toward notional defined-contribution or regular defined-contribution systems. However, the latter may expose participants to more financial market risk. Privatizing pension systems and switching to defined-contribution systems without addressing the systems’ underlying shortcomings can also have adverse effects by converting implicit unfunded government liabilities (which typically have no legal guarantee and therefore can be altered) into explicit ones.

Countries that have moved to a defined-contribution system still need to implement further parametric reforms of their old defined-benefit system to reduce transition costs. In those countries, there is substantial scope for such reforms, given that the old systems offer relatively high replacement rates and low retirement ages.

Recent experience also highlights the difficulties involved with mobilizing public support for pension reform. In some countries this has caused reforms to be abandoned while in others it has led to a substantial weakening of the parametric components of the reforms. Therefore, to increase the likelihood of success, policymakers might want to:

  • Provide stakeholders with multiple options. This approach was successfully followed in countries such as Barbados, where the authorities held broad consultations with stakeholders before adopting an extensive parametric reform. The consultations stressed that some combination of parametric reform was unavoidable, but that the distribution of adjustments across contribution rates, retirement ages, and benefits could be weighted in many different ways. Such broad consultations have also recently facilitated parametric reform in Costa Rica and Panama.

  • Publish actuarial reviews and other information that highlight the magnitude of the problem, the costs of inaction, and the relative trade-offs between different options.

  • Study and publish the distributional effects of the current system and proposed reforms. Such studies may prompt reform, since they are likely to show that the beneficiaries of subsidized pension schemes are mainly higher-income individuals. Similarly, governments may wish to express the opportunity costs of pension subsidies in concrete measures, such as the loss of x number of medicines or primary school textbooks. The redistribution of income from future to current generations implicit in most pay-as-you-go systems could also be quantified and highlighted to foster reforms that reduce the generosity of benefits for current pensioners.

As in other countries of the world, the real choice in Central America is not between reform or no reform, but between gradual and moderate reforms now or sudden and drastic reforms later. The later reform measures are adopted, the harsher they will have to be to address imbalances that will continue to accumulate under the current systems. Moreover, political resistance to pension reform may increase as pensioners become an increasing share of voters. Undertaking pension reform soon should therefore be a key priority for policymakers.

Appendix 3.1. Country-by-Country Descriptions of Recent Reforms

Costa Rica

Costa Rica started the process of reforming its old age pension system in 1990, with a view to enhancing financial sustainability. In 1990, retirement ages for the defined-benefit pension system under Costa Rica’s social security administration (Caja Costarricense de Seguro Social– CCSS) were raised from 57 years to 61 years and 11 months for men, and from 55 years to 59 years and 11 months for women. In 1992, a general pension law was enacted whereby 19 different pension regimes were unified and consolidated under a single defined-benefit system, which was charged to the central government’s budget, and incentives were given for affiliates to transfer to the CCSS regime.20 Finally, the Worker Protection Act of 2000 transformed the national pension system into a four-pillar system by adding a mandatory defined-contribution pillar based on individual accounts and a voluntary defined-contribution pillar to the existing defined-benefit first pillar and the noncontributory, means-tested poverty alleviation pillar. The 2000 reform also transferred to the first pillar 15 percent of the profits of some public enterprises, with the amounts to be negotiated each year.

In April 2005, after a broad and extended consultation process, changes were introduced to the defined-benefit pillar that have improved its financial sustainability. The reform raised workers’ contributions by 0.5 percentage points every five years between 2010 and 2035, which will add 3 percentage points to the current 2.5 percent contribution rate. Replacement rates were also subjected to a progressive scale that effectively increased the rate for lower-income brackets and reduced it for higher-income brackets; as a result, average replacement rates were reduced by about 1.5 percentage points.21 In addition, benefits are now based on average real earnings over the last 20 years of contributions, compared to average nominal earnings over the last four years of contributions before the reform. This reduces the scope for distortions, abuses, and inequity. The changes will be phased in gradually: for those aged 55 and over by the time of the reform, the old regime applies; for individuals between ages 44 and 55, the incidence of the changes varies inversely with age; for people under 44 years old, the reform applies in full. Finally, the minimum number of monthly contributions for those retiring at 65 was raised from 240 to 300. Overall, the reform is estimated to extend the financial sustainability of the regime by three decades.22

However, the reform was insufficient, and important challenges remain. In particular, the coverage of the pension system, though high relative to the region, is still unsatisfactory; evasion (lack of insurance or underreporting of salaries) and payment arrears are widespread; and administrative efficiency in terms of collection, investment performance, and benefit allocation could be improved. The special pension regimes in the government budget include privileges relative to the social security first pillar and are not subject to the same parameters as those introduced in the 2005 reform. And though the reform gave substantial breathing space to the social security administration’s defined-benefit regime, further parametric revisions will be required to extend the life of the system beyond 2054. Finally, the design and structure of the mandatory and voluntary defined-contribution pillars may need rethinking, especially regarding the high administrative costs.

The new administration of President Oscar Arias has pledged to deepen pension system reform, in particular with respect to the noncontributory pillar, whose coverage and benefits it intends to improve.

Dominican Republic

The Dominican Republic embarked on an ambitious reform of its pension system (old age, disability, and survivor insurance) in August 2001. The reform is scheduled to be phased in over a 10-year period and involves moving from a pay-as-you-go, defined-benefit system to privately-managed individual accounts. To integrate different groups within the same legal and institutional framework, three regimes were envisaged:

  • A contributory regime designed for both private and public employees, under which participation is mandatory, except for workers above the age of 45 who could receive benefits under the old system;

  • A subsidized regime designed to ensure that the elderly poor receive a minimum pension financed out of general revenue; and

  • A subsidized-contributory regime designed to cover self-employed workers with earnings above the minimum wage, whose contributions would be supplemented by government subsidies that would vary inversely with a worker’s income.

The reform of the pension system is being implemented gradually, though delays have arisen owing largely to the economic crisis of 2003–04. The key steps involving each of the regimes are detailed below.

The contributory regime was formally created in 2003. It requires employers and employees to contribute a combined 7 percent of the employee’s earnings (increasing to 10 percent by 2008) to the employee’s choice of one of eight pension funds. By end-2005, pension fund assets reached 2½ percent of GDP. The superintendency of pensions is evaluating measures to diversify investment opportunities (including real estate) and reduce the administrative costs of pension fund management companies.

The subsidized regime was launched in August 2004, although social assistance to the elderly is to be expanded gradually. The first stage envisioned providing minimum pensions to 150,000 poor households by the end of 2005.

The financial viability of some elements of the pension reform is being reconsidered in light of the fiscal constraints resulting from the banking crisis. The government is considering reducing the generosity and slowing down the expansion of the social assistance components of the reform, recognizing the need to bring public debt down from the current 50 percent of GDP to more comfortable levels.

El Salvador

The pay-as-you-go, defined-benefit system was replaced in 1998 by a system of privately-managed individual accounts. Participation in the new system was mandatory for workers under 36 years old and new entrants to the labor force, while women 50 and over and men 55 and over had to remain in the old system. Persons in the 36 to 50/55 age group had the option to remain in the old system or move to the new system. About 85 percent of these participants chose to move to the new system, as this was the default option. Recognition bonds, calculated at the time of transfer, were issued for each worker who switched. These bonds mature at the time of the participants’ retirement.

The move to the new system created substantial transition costs because most of the contributors shifted to the private pension funds, while pension payments for older workers under the old system continue to require public funding.

The transition costs were significantly exacerbated by a decree in 2003 that offered a government guarantee to those who chose to switch to the new system. The decree guaranteed that those who chose to switch would receive a pension under the new system that is at least as large as the pension they would have received under the old system. This guarantee will require the government to top-up pensions at retirement for the vast majority of individuals in this cohort, since the returns on their individual accounts will almost certainly be less than the very high internal rates of return that they would have received under the old system. Consequently, this decree significantly increased the transition costs, which are projected to total 2 to 3 percent of GDP per year for the next several decades.

The Salvadorian government is exploring options to reduce the transition costs, including parametric reform of the old system that would reduce these costs while preserving the defined-contribution system with individual pension accounts.

In the second half of 2006, the obligations of the old pay-as-you-go system were transferred to a newly created fiduciary fund. Given the legal set-up of the fund, this change reduced the size of the measured nonfinancial public sector deficit. However, this (essentially accounting) change did not significantly affect the fiscal burden.

Guatemala

The Guatemalan authorities are in the process of devising a reform program to put the public pension system on a more sustainable path, but have not yet announced a specific plan. The authorities intend to address issues raised in the latest Financial Sector Assessment Program (FSAP) report, which suggests that, in the short term, Guatemala should:

  • Conduct actuarial studies of the pension schemes for the private sector and civil servants and develop an actuarial capacity within the government and the social security administration;

  • Complement actuarial modeling with comprehensive studies on possible reform options to address deficiencies;

  • Develop principles for external audits and actuarial valuations of complementary pension funds and use moral suasion to encourage plan sponsors to follow them; and

  • Review the fiscal framework of retirement products offered by financial institutions.

Furthermore, in the medium term, the FSAP suggests that Guatemala should:

  • Prepare a reform agenda for the public pension schemes for the private sector and civil servants;

  • Strengthen the overall governance and accountability framework of the public pension schemes; and

  • Develop a proper regulatory and supervisory framework for complementary pension funds.

However, the prospect for social security reform during the tenure of the current administration is likely to diminish as the country enters the next presidential election campaign, scheduled for September 2007.

Honduras

In 2005, the Honduran government drafted a pension reform plan whose main features include parametric reforms of the existing defined-benefit first pillar; the gradual switching of government employees to the universal first pillar and the phasing-out of their current special regimes; and the introduction of a defined-contribution second pillar. The reform includes several positive elements that aim to correct some of the problems with the current system. However, the reform does not appear to fully put the pension system on a sustainable footing.

Since a new government was elected in November 2005, prospects for the previous administration’s plan are unclear, since the new administration may develop a reform plan of its own.

Nicaragua

Nicaragua operates a pay-as-you-go, defined-benefit system. The system was originally put in place in 1955, the current law was enacted in 1982, and many of the parameters of the current system have been established by decree. While the system offers very generous benefits relative to contributions, it has historically remained in balance due to a young and growing population and the erosion of real pensions through inflation. Indeed, the system’s cash flow is not projected to deteriorate significantly over the next decade. However, as the macroeconomic situation stabilizes and the population ages, the system will eventually require fundamental structural reform to be financially viable.

The authorities of the former government intended to prepare a reform strategy that would have put the public pension system on a sound financial footing, without relying on government subsidies. A previous reform in April 2000 aimed at transiting to a funded pension system, but this plan was dropped when a reassessment indicated that the transition costs would be prohibitively high. The national assembly approved a new social security law in May 2005 that increased benefits without raising contributions and weakened the system’s governance structure. The president vetoed the law but was overridden by the assembly. Nonetheless, implementation of this law was deferred until after the November 2006 elections, and the next steps will be considered by the new government.

Panama

In Panama, the need for comprehensive pension reform became apparent by the early 2000s. During the 1990s, several assessments of Panama’s pension system indicated a fundamental imbalance between contributions and benefits. By the early 2000s, financial reserves, which covered only a fraction of existing pension rights, started to decline and were projected to be exhausted by 2013.

Forging a political consensus on reform proved difficult. A national dialogue called by then-President Mireya Moscoso in 2001 failed to reach agreement on changes, as employers pressed for raising contribution rates and tightening eligibility and workers defended existing benefits. In 2004, the new government of President Martin Torrijos developed a parametric reform designed to tackle the system’s actuarial imbalance, and the reform was passed by the assembly in June 2005. However, in response to strong popular opposition, the bill was not implemented, and President Torrijos launched a new national dialogue with civil society.

Following a five-month consultation process, the assembly approved a revised reform system in December 2005. It established a dual pension system in which individual pension accounts will coexist with a reformed defined-benefit scheme (after a transition period, the dual pension system will become fully effective in 2008). The reformed defined-benefit scheme will be similar to the partially-funded scheme in place before the reform, but with gradual parametric changes. Starting January 1, 2008, new members earning monthly wages of $500 or less will contribute solely to the defined-benefit system; workers earning more than $500 will contribute to the defined-benefit system on the portion of their salary up to $500, and to mandatory individual accounts managed by the social security administration on the portion exceeding $500, along with a “solidarity” contribution of 3.5 percent of their salary in excess of $500.

The new system will be phased in gradually. Under the transitional arrangements, new workers and workers age 35 or younger on January 1, 2006 will have the option to switch to the dual system by January 2008. Workers older than 35 will continue to contribute to the defined-benefit system based on their entire salary and will not have the option of contributing to individual accounts. It is estimated that about 70 percent of current and prospective affiliates are earning less than $500 a month and would therefore qualify solely for the defined-benefit system.

Increased financing for the defined-benefit system will come from the following main sources:

  • Contribution rates (employee and employer combined) will be raised from 9.5 percent of wages to 11 percent in 2008, 12.5 percent in 2011, and 13.5 percent in 2013.

  • The solidarity contribution (3.5 percent) levied on wages in excess of $500 will partially offset the loss of contributions channeled to individual savings accounts.

  • The central government will provide substantial transfers to the defined-benefit scheme. In the long run, only the defined-contribution component, which is expected to be sustainable, will remain as the $500 per month limit shrinks in real terms.

Eligibility requirements will also be tightened. The minimum contribution period necessary to receive a full pension will be raised from 180 months at end-2007 to 216 months in 2008 and to 240 months in 2011. Arrangements will be introduced to mitigate the impact of the step-wise changes in eligibility requirements for pension benefits of workers within two years of the retirement age and having at least 180 monthly contributions. These workers will be entitled to a reduced pension. However, the retirement age will be unchanged unless mandatory annual actuarial reviews conclude that raising it is necessary to preserve the sustainability of the defined-benefit scheme.

The reform is expected to reduce the actuarial imbalance of the system by markedly increasing contribution rates, tightening eligibility requirements, and introducing safeguards that call for a revision of the main parameters if periodic actuarial assessments point to resurging imbalances. In this regard, increases in the retirement age may be required at some point in the future. Overall, the reform is an important step in the right direction, as it limits the contingent fiscal burden of the previous pension scheme and tightens the link between contributions and benefits.

References

  • Acuña Ulate, J.A., 2006, “Reformas recientes al régimen de invalidez, vejez y muerte,” in Los sistemas de pensiones: Presente y futuro, ed. by G. López and C.E. Umaña (San José, Costa Rica: Academia de Centroamérica).

    • Search Google Scholar
    • Export Citation
  • Cuevas, A., M. González, D. Lombardo, and A. LópezMarmolejo, 2006, “Pension Privatization and Country Risk” (unpublished; Washington: International Monetary Fund).

    • Search Google Scholar
    • Export Citation
  • Holzmann, R., 1997, “Pension Reform, Financial Market Development, and Economic Growth: Preliminary Evidence From Chile,” IMF Staff Papers, International Monetary Fund, Vol. 44, No. 2, pp. 14978.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Holzmann, R., and E. Palmer, eds., 2006, Pension Reform: Issues and Prospects for Nonfinancial Defined-Contribution Schemes (Washington: World Bank).

    • Crossref
    • Search Google Scholar
    • Export Citation
  • International Monetary Fund, 2001, Government Finance Statistics Manual 2001 (Washington: International Monetary Fund).

  • Lindert, K., E. Skoufias, and J. Shapiro, 2006, “Redistributing Income to the Poor and the Rich: Public Transfers in Latin America and the Caribbean” (unpublished; Washington: World Bank).

    • Search Google Scholar
    • Export Citation
  • Organization for Economic Cooperation and Development, 2005, Pensions at a Glance (Paris: OECD).

  • Roldos, J., 2004, “Pension Reform, Investment Restrictions and Capital Markets,” IMF Policy Discussion Paper No. 04/04 (Washington: International Monetary Fund).

    • Search Google Scholar
    • Export Citation
  • Settergren, O. and` B. Mikula, 2006, “The Rate of Return of Pay-As-You-Go Pensions Systems: A More Exact Consumption-Loan Model of Interest,” in Pension Reform: Issues and Prospects for Nonfinancial Defined-Contribution Schemes, ed. by R. Holzmann and E. Palmer (Washington: World Bank).

    • Search Google Scholar
    • Export Citation
  • Shoven, J., and Slavov, S. 2006, “Political Risk Versus Market Risk in Social Security,” NBER Working Paper No. 12135 (Cambridge, Massachusetts: National Bureau of Economic Research).

    • Search Google Scholar
    • Export Citation
  • Superintendencia de Pensiones, 2005, “Informe al Comite de Vigilancia del Regimen de Invalidez, Vejez, y Muerte” (San José, Costa Rica: Superintendencia de Pensiones).

    • Search Google Scholar
    • Export Citation
  • Walker, E., and F. Lefort, 2002, “Pension Reform and Capital Markets: Are There Any (Hard) Links?Social Protection Discussion Paper No. 0201 (Washington: World Bank).

    • Search Google Scholar
    • Export Citation
  • World Bank, 2006, “Pension Reform and the Development of Pension Systems: An Evaluation of World Bank Assistance,” Operations Evaluations Department Sector Study (Washington: World Bank).

    • Search Google Scholar
    • Export Citation
1

The authors would like to express their appreciation to Cleary Haines for her excellent research support.

2

It remains to be seen whether an adequate annuity market will actually be available when individuals in the privatized systems begin to retire in large numbers. Individual annuity markets have had difficulties developing in some countries due to adverse selection problems.

3

Once recent reforms are fully phased in by 2013.

4

Unless otherwise noted, “average earnings” in this chapter refers to the average of earnings on which social security contributions are due.

5

While commonly used, it is unclear whether life expectancy is the proper basis for setting retirement ages. If the rationale for mandatory public pension systems is to provide for elderly persons who cannot work (and to preclude the moral hazard arising from the provision of such benefits), then the appropriate metric for determining the retirement age may be morbidity (the degree to which poor health at a given age prevents work) rather than life expectancy. However, cross-country morbidity data are less readily available.

6

This expression can be derived as follows: total contributions =cCw (where C = the number of contributors and w = the average wage) and total benefits = pB (where p = the average pension and B = the number of beneficiaries). Setting total benefits equal to total contributions in each period (as in a pure pay-as-you-go system) yields pB = cCw. Rearranging yields (p/w) = c (C/B). Defining r (the replacement rate) = p/w and X (the ratio of contributors to pensioners) = C/B yields r = cX. Note that this equation ignores administrative costs.

7

While this result is a close approximation for many pay-as-yougo systems, it strictly applies only for pure pay-as-you-go systems (no reserve fund) that are in a steady-state (the difference between the average age of contributors and pensioners is constant). See Settergren and Mikula (2006).

8

In the very long run, 4 percent wage bill growth is likely to be optimistic, since labor force growth in many developed countries is approaching zero and real wage growth per worker is closer to 2 percent.

9

See Appendix 3.1 for more information on these recent reforms.

10

Assuming a 30-year contribution history, 4 percent annual wage growth, and 8 percent nominal returns. The estimate for the Dominican Republic assumes an average charge on returns of 2 percent.

11

Allowing AFPs to invest directly in government securities may improve their returns, since capital market imperfections and intermediation costs may prevent banks from fully passing on the return on government securities to their depositors.

13

In the Dominican Republic, domestic capital market development has been curtailed by the portfolio restrictions imposed on pension funds.

14

If individuals receive a market rate of return on their contributions, then contributions are in theory savings rather than a pure tax. However, in most cases, weak and nontransparent links between contributions and benefits, together with high discount rates (i.e., myopia), result in contributions having or being perceived as having a high taxation element, with all of the associated labor market distortions.

15

Defined-benefit systems also subject participants to some risk, including political risk and the risk that permanent economic shocks will eventually force an adjustment of benefits (Shoven and Slavov, 2006).

16

Fully fledged NDC systems are also being introduced in Italy, Latvia, and Poland. For more on NDC systems, see Holzmann and Palmer (2006).

17

In some systems, pensions are indexed to wages rather than consumer prices, which requires lower initial pensions at retirement (assuming positive real wage growth). However, indexation to consumer prices is generally viewed to be preferable, as it is unclear why pensioners would want rising real consumption profiles in retirement. Moreover, even if pensioners prefer such profiles, they could still achieve them under price indexation by saving a portion of their initial pensions.

18

These are equivalent in present value if cash flows are discounted by the rate of return earned on contributions and the income tax rate is the same at the time of contributions and withdrawals.

19

Governments could also enhance fiscal transparency by adopting the accounting methodologies in the Government Finance Statistics Manual (IMF, 2001).

20

However, the National Teachers Union, the Finance Ministry, and the judiciary maintained separate regimes. For a summary of these reforms, see Acuña Ulate (2006).

21

For instance, with 360 monthly contributions, the replacement rate for individuals earning less than twice the minimum wage was raised from 60.9 percent before the reform to 62.5 percent, while the replacement rate for individuals earning over eight times the minimum wage was reduced from 60.9 to 53 percent (Acuña Ulate, 2006).

22

Thus, while the regime’s reserves were projected to be depleted by 2024–28 before the reform, they are now estimated to last until 2054 (Acuña Ulate, 2006).

Cited By

  • View in gallery

    Pension Coverage: Number of Contributors as a Share of the Labor Force, 2003

    (In percent)

  • View in gallery

    Contribution Rates for Mandatory Old Age, Disability, and Survivor Pensions

    (In percent)

  • View in gallery

    Standard Retirement Age and Life Expectancy at Age 60

    (In years)