Abstract
El Salvador: Selected Issues
The Salvadoran Pension System: in Search of Sustainability1
El Salvador’s pension system has been marked by continual imbalances and insufficient action to address them. The move to a defined contributions system from the late-1990s proved no panacea as transition costs surged while coverage stayed low. The recently-submitted proposal for a mixed system would return most of the private segment to a public PAYG pillar, helping fiscal cash flow and reducing measured public debt and deficits (for purely accounting reasons). But the draft eschews parametric reforms and thus would not improve the system’s sustainability, while introducing new risks. A viable reform strategy should include: (i) deep parametric reforms and (ii) a credible commitment to “fiscalize” remaining pension benefits via creating adequate fiscal space in non-pension accounts.
A. Background
1. El Salvador’s pension system emerged gradually as a heavily subsidized and narrow PAYG scheme (Lazo et al. (2010)). For a long time, the pension contribution rate was only 3.5 percent in the private sector (of which the government paid 0.5 percentage point) and 5–12 percent in the public sector (depending on sub-sector and period). Just 15 years of contributions were sufficient for a pension, while average replacement rates ranged from 80 to 100 percent.2 As a result, in the private sector contributions of only 6 months’ worth of wages were sufficient to earn a right to a life-time pension, with expected post-pension life span of around a quarter-century, implying severe actuarial imbalances and large implicit subsidies. A study carried out in 1994 estimated that without reforms the public system would have run deficits from 1997 and that its reserves would have been depleted by 2009. At the same time, the system was estimated to have covered less than a quarter of the economically active population.
2. In the late 1990s, El Salvador embarked on a transition to the defined contributions (DC) system. The PAYG system was to be phased out with a law enacted in 1996 (effective from 1998), which obliged older population (women over 50 and men over 55 as of 1998) to stay in the PAYG system, while the young (under the age of 36 in 1998) were mandated to switch to a new system based on individual accounts. The remaining individuals were given an option to choose between the PAYG benefits (linked to a wage earned over the last 10 years prior to retirement) and the new system. In the event, some 90 percent opted to be in the new system.3 It was believed that the new system would help bolster formal labor force participation and hence coverage.
3. The defined contributions segment that was introduced in 1998 has been managed by private pension funds. Initially, five pension funds were set up, but eventually only two funds remained.4 The funds are tasked with allocating assets and awarding pension benefits, and have the flexibility to contract services (contributions collection, information processing, etc.) with the exception of direct management of the investment portfolio. Commissions represent the primary source of income, and maximum commissions were initially set at 3.5 percent of wages (over a quarter of the contributions flow), but these were cut several times and have been at 2.2 percent of wages since 2011. About one-half of the commission represents expenses on insurance policies to guarantee financing of disability and survivorship benefits. Supervision is conducted by the Pensions Superintendency.
4. The 1996–98 reform had also adopted limited parametric adjustments to improve fiscal sustainability. The contribution rate was gradually increased to 13 percent for the private sector and 14 percent for the public sector. Required years of contributions were raised to 25 years. The qualifying period for a pension calculation in the PAYG segment was changed from the last 3 years of salary to 10 years. As a result, the replacement rate in the PAYG segment was reduced from 100 to 86 percent for public sector employees and from 80 to 78 percent in the private sector. Overall, the 1996-98 reform was estimated to have improved the long-term pension sustainability, roughly halving its unfunded liabilities. Still, the PAYG segment remained very generous by international standards, with the replacement rates being high, and with retirement ages and required contribution periods being low comparatively.
5. The reform implied nontrivial fiscal “transition costs.” Most of pension-related social contributions were re-directed away from public revenue to private individual accounts, while residual PAYG pension entitlements remained to be financed by the budget. This caused fiscal pension deficits of around 2 percent of GDP. Additional (initially fairly modest) transition cost obligations arose from the need to credit past contributions made within the PAYG system for those who opted to transfer to the new system, and a decision to guarantee a certain minimum pension to those who complied with the main qualifying requirements but whose balances in individual accounts were insufficient to generate a pension. Still, the transition costs, while not insignificant, initially appeared to be “bounded,” and were expected to fall over time, converging to zero by 2030 or so.
6. The authorities opted to absorb the transition costs via reserve drawdown and borrowing rather than deficit-reducing measures. The funding of PAYG deficits initially relied on tapping “technical reserves” from prior years. But these were fully drawn in 2001, well before expectations, while domestic budget funding options became more limited.5 In 2001-05, the key funding source was borrowing in international markets, with some $1.2 billion estimated to have been issued for pension-related financing over 5 years. But this source was problematic due to its high interest cost (8–9% annually in mid-2000s). In 2006, it was decided to resort to private pension fund financing of public pensions. The pension funds had ample liquidity and appeared to be a stable source at a more acceptable cost. To secure this financing on favorable terms, the pension funds were mandated to purchase new government pension bonds (CIP bonds), at an interest rate (LIBOR plus 75 basis points) that initially seemed not far from market levels and was around 6 percent in 2006.6 The rate however was dragged down to 1–1½ percent by the 2008-09 global crisis, staying low through 2016.
7. In the event, the transition costs quickly escalated, reflecting a vicious circle between falling returns and lack of policymakers’ resolve to adopt tangible measures. With the returns on pension fund investments falling perceptibly in the first few years of operations,7 the authorities yielded to political pressure to equalize pension benefits of those who had opted for the DC system to those enjoyed by PAYG pensioners. This resulted in ad-hoc decisions (in 2003 and 2006) granting top-ups to various population cohorts at the expense of the budget. Additional steps, such as significant periodic increases in the guaranteed minimum pension, also increased the pension system’s future fiscal burden. In 2008-09, declines in the LIBOR and correspondingly the rates of return on pension bonds due to the global financial crisis added to these costs and extended them over time. Unlike the original transition costs, these new unfunded liabilities reflected not so much the generous defined benefits of the PAYG system, but longer-term costs from projected declines in replacement rates in the DC system raising the probability of recourse to the minimum pension guarantee in the future.
Unfunded liability and Cost of Top-Ups to Pension Benefits
(Millions of USD, NPV terms)
Citation: IMF Staff Country Reports 2016, 209; 10.5089/9781498342346.002.A004
Source: Salvadoran authorities and Fund Staff estimates and projections.Projected Fiscal Burden of Servicing Pension Debt
(Percent of budget envelope)
Citation: IMF Staff Country Reports 2016, 209; 10.5089/9781498342346.002.A004
Sources: Salvadoran authorities and Fund staff esimates and projections.8. The evolving system is now approaching several critical junctures. First, at the urging of the Constitutional Court, in early 2016 parliament adopted a decision to substantially raise interest rates on pension bonds, which would increase the fiscal burden of pensions going forward. Second, the upper (45 percent) limit on mandatory purchases of key pension bonds (“CIP-A”) is expected to be reached in 2016. This limit was raised in the past, and the government had committed not to increase it further. Should it keep this commitment, it can no longer rely on captive financing from the pension funds. Third, it is estimated that, starting in 2016, total contributions would become smaller than the needed issuance of pension bonds. Thus, financing the transition costs “within the pension system” would be unrealistic even if the government chose to continue to rely on captive financing. Finally, the large younger cohort that was not entitled to a choice of the system and received no benefits from the top-ups to date would begin retiring in 2017. As their replacement rates are projected to drop relative to those of previous cohorts, there would be political pressures to top up their pensions, creating a major precedent if the line on the top-ups cannot be drawn.
B. Diagnostic of the Current (Pre-reform) Situation
9. El Salvador’s pension system problems of fiscal, social, and actuarial sustainability require a thorough analysis. A broad stock-taking of the pension system is warranted to go beyond specific pension system issues and also focus on linkages with El Salvador’s economic and institutional features: including the fiscal position, the social protection system, and the labor market. A cross-country perspective could additionally help measure the severity of any policy gaps and viability of options to address them, as many other countries are grappling with imbalances of a similar nature.
Fiscal Sustainability
10. On the surface, El Salvador’s implicit public pension debt – a summary indicator of pension system’s impact on fiscal sustainability – does not look high in a cross-country comparison. The budget obligations for defined benefit pensions and the projected triggering of the minimum pension guarantee are estimated at around 100 percent of GDP in NPV terms.8 This estimate is not straightforward to put in a cross-country perspective given the differences in horizons, discount rates, the nature of the systems, and other variables. The World Bank (2004) study that aimed to arrive at somewhat comparable estimates put those in the range from 30 to 500 percent of GDP, with El Salvador being at the low end at the time (some 50 percent of GDP at the 3% discount rate in 1999/2000), to be comparable to the 3% discount rate that El Salvador currently uses. Since then, El Salvador’s implicit debt has roughly doubled as a percent of GDP. While there is no comprehensive updated calculation, a comparison with EU countries, Colombia, and Mexico indicates that for El Salvador the assessed liability still remains at the low end of the cross-country range.9
Implicit Public Debt in 1999/2000
(Percent of GDP, NPV discounted at 2% and 4%)
Citation: IMF Staff Country Reports 2016, 209; 10.5089/9781498342346.002.A004
Source: World Bank (2004)11. Upon closer scrutiny, however, the cross-country comparisons of implicit debt understate El Salvador’s fiscal sustainability challenges. The indicator as commonly calculated (see World Bank (2004)) included only spending obligations, excluding a favorable impact from “earmarked revenues,” such as pension contributions. At the same time, the second pillar-based systems such as those of El Salvador are treated differently, in that they exclude spending obligations, presumably because they are automatically offset by matching revenues. This makes the above El Salvador’s NPV calculations biased downward compared with countries that have greater reliance on PAYG systems. The latter also generate significant revenue earmarked for pension spending (both from specific contributions and general taxation), but this revenue is not allowed to reduce unfunded spending obligations in any proportion.10 Thus, El Salvador’s actuarial calculations would not be comparable to those of countries with PAYG systems. And even compared to funded systems only, this measure for El Salvador would tend to yield, by construction, a lower unfunded liability, as most other such systems (e.g., Chile, Mexico, etc.) have more extensive PAYG elements than El Salvador.
12. In line with implicit debt, the projected pension deficits do not seem to be large in the baseline, but this hinges on an unlikely assumption that the current rules under the defined contribution system could be fully maintained. The baseline projection shows that those deficits (which essentially reflect public spending on transition costs) would peak at above 3 percent of GDP during 2020–25, and subside gradually over the next few decades. Such a path would again put El Salvador at the lower end among regional peers, including compared to similar pension systems, in terms of the level of the pension-related public spending. The somewhat lower deficits largely reflect the dominance of the self-financed defined-contributions pillar in El Salvador relative to other countries. But it is unlikely that the defined contribution system could be maintained in the current form given the political pressures to keep higher replacement rates and the cash flow problems that could emerge under unchanged rules. One question would be whether El Salvador could avoid further top-ups to pension benefits, which would mark a major break with the practice of the past decade or so.
El Salvador Pension Deficit
(Percent of GDP)
Citation: IMF Staff Country Reports 2016, 209; 10.5089/9781498342346.002.A004
Source: Salvadoran authorities.Pension System Public Spending of LA Countries with a Significant Second Pillar Component
Citation: IMF Staff Country Reports 2016, 209; 10.5089/9781498342346.002.A004
Sources: IDB, Salvadoran authorities, and IMF staff estimates and projections.13. The impact of the likely slippages relative to the current rules could be large. In a hypothetical scenario whereby full top-ups (the complete matching of defined benefits) would continue, public pension spending would be driven by demographic factors, and, other things equal, could exceed 7 percent of GDP by 2050. Additional costs could arise from the very low coverage (less than a quarter of the population), which may not be sustainable. Increasing the coverage rate would entail correspondingly larger public spending. To be sure, there could be various cost-effective ways of increasing coverage via means-tested expansion of benefits to the poor.
El Salvador: Projected Pension Expenditure, 2013-2050
(Percent of GDP)
Citation: IMF Staff Country Reports 2016, 209; 10.5089/9781498342346.002.A004
Source: Fund staff estimates.14. Even assuming that El Salvador’s fiscal pressures from the pension system could be contained within the “baseline” projection, the liquidity pressures could raise more serious problems than those faced by most peer countries. First, as a fully-dollarized economy El Salvador tends to have particularly stringent fiscal funding constraints and hence smaller room for maneuver. Second, in this context, the incremental pressure on funding costs in the next few years would be larger than the baseline deficit path would suggest. This is because the room for financing offered by private pension funds since 2006 has been largely exhausted. Assuming that the pension funds can no longer finance any public spending, the increased funding needs could amount to 3–3½ percent of GDP.
Social Sustainability
15. Pension system sustainability assessment needs to go beyond its fiscal gap impact. The full adoption of the defined contribution system would eventually (if only technically) “solve” the issue of fiscal sustainability over the long term, but the new pressure point would be the adequacy of pension benefits, including: (i) adequacy of pension levels; (ii) adequacy of coverage, and (iii) distributional adequacy. As explained below, all of these aspects raise serious sustainability concerns, though it is difficult to develop precise metrics for this assessment.
16. Falling pension levels. In the baseline scenario projection developed by the authorities, replacement rates are projected to roughly halve over the next generation—from almost 70 to around 40 percent on average. They would fall further for some categories of the population: for example, high-income female workers would see replacement rates of below 25 percent. In addition, should financial returns continue to remain close to the low levels of the past few years, the replacement rates would fall further for all categories of pensioners. A substantial proportion of future pensioners would see replacement rates in the range of 30–40 percent or below, making them among the lowest in Latin America. Apart from the low levels, the relatively rapid reduction in the replacement rates over time could also impact sustainability. Finally, about half of contributors will likely not earn a right to a pension but would instead receive a lump-sum benefit, which would be clearly insufficient to provide for an adequate pension.
Estimated Replacement Rates in the Pension System
Citation: IMF Staff Country Reports 2016, 209; 10.5089/9781498342346.002.A004
Sources: Salvadoran authorities and Fund staff estimates and projections.17. Low coverage. Only 24 percent of workers in the labor force are paying pension contributions. At the same time, less than 11 percent of those over 65 years of age receive any pension benefits. Based on comparable numbers, these are some of the lowest rates of coverage in Latin America. The defined contributions segment has been supplemented with other pillars (see Table 1 in the Appendix). The zero pillar was introduced in 2009, whereby a small noncontributory monthly pension of US$50 (basic universal pension) was allocated for low-income individuals in the poorest municipalities. While the program can expand coverage in a desirable way and at a reasonable cost, its scale has so far been very small. The third, voluntary pillar remains negligible. Finally, based on the experience of other countries, expecting significant and rapid increases in coverage would be unrealistic, not least because incorporating the shadow economy into the formal sector is at best a long, drawn-out process requiring a major effort.
Estimated Coverage in Latin America and the Caribbean, 2011
(Percentage of population above 65)
Citation: IMF Staff Country Reports 2016, 209; 10.5089/9781498342346.002.A004
Sources: Salvadoran authorities, Fund staff estimates and projections, and Interamerican Development Bank.Percentage of Contributors versus Employed
Citation: IMF Staff Country Reports 2016, 209; 10.5089/9781498342346.002.A004
Sources: Salvadoran authorities, Fund staff estimates and projections, and Interamerican Development Bank.18. High inequality. The benefits disproportionally accrue to high-income pensioners in all key segments of the system. Thus, 5 percent of individuals receive 20 percent of pensions in the new defined contributions segment, while 20 percent of high-income individuals receive 50 percent of benefits in the PAYG segment that is being phased out. This inequality was a product of the pension system’s initial design of overly generous benefits for the few. Since the system’s participation was dominated by the relatively well-off parts of the population, the benefit generosity was de-facto regressive. This effect was reinforced by follow-up policy adjustments that were significantly aimed at protecting existing benefits. The shift to a defined contributions-based system did not correct any distributional disparities as it does not envisage re-distribution from the rich to the poor. In addition, poorer people are more likely to be among those that do not accumulate 25 years of contributions that are necessary to earn a pension.11 The noncontributory (first) pillar would mitigate those differences, but it remains extremely small in El Salvador. As a result, the distribution of pension system’s implicit subsidies remains strikingly skewed to the richer segment of the population.
Pension System Subsidization, El Salvador, 2011
(Proportion of public spending per income quintiles, in percent)
Citation: IMF Staff Country Reports 2016, 209; 10.5089/9781498342346.002.A004
Sources: Salvadoran authorities, Fund staff estimates and projections, and Interamerican Development Bank..Actuarial Sustainability
19. An integrated actuarial perspective is also important for gauging sustainability of a contribution-based system. While a macro-actuarial analysis was already used in the assessment of unfunded fiscal liabilities, a micro-based actuarial evaluation may be essential to assess incentives and labor market issues. For example, micro-level distortions could affect substantially assumed macroeconomic scenarios. Micro-actuarial sustainability is key to boosting participation in the system, which is particularly important given the high share of informality in El Salvador. In this context, the following stylized facts stand out.
20. Large disparity between contributions and benefits. A statutory contribution rate of 13 percent would translate into an effective contribution rate of 10.8 percent (net of commissions). Assuming zero return on contributions, this rate would be sufficient to accumulate only 2.7 years’ worth of salary payments.12 However, conditional life expectancy at the time of retirement in El Salvador is on average 25 years.13 At the current (defined benefit) replacement rate of around 70 percent, there would be a large micro-actuarial imbalance, whereby the amount of expected contributions is about 6½ times lower than the expected benefits they would need to finance. Alternatively, under the same assumptions, a fully-balanced actuarial system (both at micro and macro levels) would imply, in a steady state, a replacement rate of only around 11 percent, which would most likely be unsustainable. The above back-of-the-envelope calculation would be broadly consistent with the fact that the total amount of pension contributions is currently approximately equal to total pension benefits, since the number of contributors is about 4-5 times the number of current beneficiaries.14 But this ratio is projected to worsen sharply over the next few decades with population aging, as the micro-actuarial imbalance could give rise to macro imbalances.
21. Delays in parametric reforms are a key reason for the observed actuarial imbalances. The veil of self-financing of the defined contributions system appears to have eased the urgency of parametric adjustments, as retirement ages, years of contribution, and contribution rates have persisted unaltered since 1998 at levels substantially below those of other countries in the region. Retirement ages have not been changed for a much longer period of time and – along with Venezuela and Bolivia – El Salvador has been one of the most visible outliers in this regard.
Minimum Retirement Age to Receive Full Pension in Latin America, Female
Citation: IMF Staff Country Reports 2016, 209; 10.5089/9781498342346.002.A004
Sources: Salvadoran authorities and Fund staff estimates and projections.Minimum Retirement Age for a Full Pension in Latin America, Male
Citation: IMF Staff Country Reports 2016, 209; 10.5089/9781498342346.002.A004
Sources: Salvadoran authorities and Fund staff estimates and projections.22. Sub-par financial returns. In principle, robust financial returns could help sustainability of a defined contributions pension system, other things equal. In the case of El Salvador, however, there are several factors that have hampered those returns. First, the low-yielding government CIP bonds (both “A” and “B” series, with the limit of 45 percent applying only to the “A” series) required to be purchased accounted for about 60 percent of the pension fund portfolios in late-2015, carrying an interest rate of around 1.3 percent at the time, compared to that of over 6 percent for the remainder of the portfolio.15 Second, aside from the mandatory purchases, there remain restrictions on asset allocations within pension fund portfolio, particularly limiting the amount that they could invest abroad. Third, and more broadly, the Salvadoran economy has been growing at a relatively slow pace over the past few decades, so it would be unrealistic to expect that very high rates of return could be sustained. As a result of the above reasons, the profitability of Salvadoran pension funds has been among the lowest in the region and trending down.
Profitability of Pension Funds, 2012-15
(Real, in percent)
Citation: IMF Staff Country Reports 2016, 209; 10.5089/9781498342346.002.A004
Source: International Association of Supervisory Bodies of Pension Funds (AIOS).23. Weak incentives to contribute. The low returns dent contributors’ willingness to participate in the pension system. While contribution payments are mandatory for employers and employees (excluding the self-employed), tax and contribution evasion is high in El Salvador. In this regard, surveys show a considerable contributor preference to pay for health benefits as opposed to pension benefits. The unwillingness to honor pension contributions is broad-based, but is particularly high among the poorer workers. One reason for this could be that contributors may feel that the pension system would not offer adequate benefits in the future. In any case, this contributor preference is also visible in the data on actual collections, whereby health-related contributions have been reported to be somewhat higher than the pension contributions.
Workers Not Disposed to Contribute any Amount to Pension or Health Funds, per Income Deciles
Citation: IMF Staff Country Reports 2016, 209; 10.5089/9781498342346.002.A004
Sources: Salvadoran authorities, Fund staff estimates and projections, and Interamerican Development Bank.Overall (Sustainability) Assessment
24. The current pension system is unsustainable overall, reflecting a complex—and partly offsetting—interaction between fiscal, social, and actuarial sustainability aspects. In the baseline scenario (e.g., assuming the current defined contribution rules can be maintained indefinitely), fiscal sustainability is a short-to medium term problem, which gets “resolved” in the long term at the expense of social sustainability. However, the large size of basic actuarial imbalances indicates that, in the long term, it would be impossible to maintain both fiscal and social sustainability without parametric adjustments. In particular, either there would be a need for large permanent fiscal subsidies for pensions (which are not currently envisioned and anyway likely unaffordable due to El Salvador’s overall fiscal position) or replacement rates would have to drop to unsustainable levels for the system to be financially viable.16 Overall, there is little evidence that the pension system would be capable of fulfilling its core function: a reliable and efficient provision of decent retirement income to a significant part of the population.
25. In addition to the fundamental actuarial disequilibria, institutional shortcomings in dealing with transition costs create an unfavorable dynamic and increase uncertainty. First, it remains unclear how the transition costs could be ultimately resolved. For example, while the general budget has the legal obligation of servicing pension bonds, the incremental fiscal gap and its implications have not been internalized in the broader fiscal adjustment strategy. The authorities’ draft fiscal responsibility law and medium term fiscal framework documents treat pension and non-pension accounts separately, with the key non-pension indicators and targets set largely independently of the pension system developments and needs. Second, there is neither a framework for evaluation, nor a process that could impact future top-ups or other adjustments in the pension system due to fiscal sustainability concerns. Partly for this reason, these top-ups have proliferated and their cost has been snowballing. Third, the short-term solution to the fiscal cash flow problem, borrowing from pension funds at low rates to finance current pension payments in the publicly-managed system, has created a vicious (political economy) cycle. The borrowing depresses future pension benefits in a defined contribution system, in turn putting additional pressure on the government to prop up those benefits at the expense of the budget. The resulting adverse dynamics and uncertainty suggest that the system is not evolving in the right direction, underscoring the rationale for reform.
C. The “Mixed System” Proposal
Draft Amendments to Pension Law17
26. In February 2016 the government submitted legislation to parliament proposing to address pension system problems by halting the transition to a defined contributions model. Instead it suggested adopting a “mixed” system on a permanent basis. The main features would be the following:
Transfer of assets to the public sector. Pension contributions on all (portions of) salaries that are below a threshold (2 minimum (non-agricultural) wages, over one-half of pension contributions flow) would be reallocated to the public sector. Assets corresponding to the past contributions below the same threshold would be moved to the public sector.
Flat public pension benefits. Pensioners in the newly formed public pillar would earn a minimum “contributory” pension (currently $207 per month, or 82 percent of the minimum wage, adjustable annually based on inflation and subject to availability of fiscal resources), regardless of the precise amount contributed as long as contributions are below the two minimum wages threshold. The reform would rescind a major top-up to pension benefits adopted in 2006.
Downsizing of the defined-contributions pillar. Contributions above the same threshold (involving about 20 percent of contributors) would earn pension benefits based on returns that the pension funds can generate. Pension funds would continue to administer the pensions system, but their commission would be cut from 2.2 percent to 1.9 percent in the remainder of the private system and to 1 percent in case they were to be contracted to manage the public segment. There would be some relaxation of restrictions for pension funds to diversify their investment, although their effective ability to do so would depend on implementation.
Absence of parametric reforms. No (saving-generating) parametric reforms are envisioned and no major steps to generate additional resources would be considered, with the exception of tightening contributions collections. In effect, the contribution rate for public sector workers has been proposed to be reduced from 14 to 13 percent, which would align it with that of the private sector.
Other selected features. A special entity, National Institute of Pensions, would be created to manage the new public sector pillar. Workers who do not comply with the minimum requirement of 25 years of contributions would receive reimbursement not as a lump sum, but in installments. In this case, they would continue to enjoy free access to the public health system. Regular actuarial reports assessing the long-term liabilities of the pension system are envisioned every two years (which is in line with the best practices).
Assessment
27. The reform would move the pension system closer to a PAYG type, which entails both advantages and disadvantages. On the one hand, the new system would offer more predictable replacement rates for the low-income workers, which would initially range from 41 to 82 percent, and would be higher for lower income pensioners. On the negative side, the reform would forgo some of the labor market (closer link between pension benefits and contributions) and fiscal (automatic financing of pension benefits) advantages, although as explained above those advantages were not well-realized in practice. On the whole, it is difficult to make a strong ex-ante argument in favor of one particular system: the result would largely depend on the institutional capacity to implement the system well. It will also depend on the actuarial calculations, which admittedly have a margin of error and are sensitive to particular scenarios.
28. The implementation of the proposal would not significantly improve the fiscal position. Actuarial deficits post-reform would be initially lower than in the pre-reform scenario, but significantly higher in the outer years. The impact on NPV of unfunded liability would be ambiguous: there would be some savings from rescinding one major benefit top-up of 2006, but also some additional costs - the net impact would depend on a detailed micro-simulation of the effects of reform. The authorities’ calculation presented at the time of the draft law submission showed a modest decline in the unfunded liability from 99 to 90 percent of GDP. However, some of the parameters of this calculation need to be updated, and it is impossible to rule out a scenario whereby the NPV of the unfunded liability could increase in the post-reform relative to pre-reform scenario. In addition, the calculation of unfunded liability somewhat unrealistically assumes that there would be no increases in the minimum pension, which would be would be particularly costly in this post-reform scenario.
Actuarial Deficit of the Pension System of El Salvador
(Millions of US dollars)
Citation: IMF Staff Country Reports 2016, 209; 10.5089/9781498342346.002.A004
Source: Salvadoran Authorities29. The reform would have substantial accounting benefits of lower measured public debt and deficits. By bringing assets and contributions within the government sector, measured fiscal deficits could initially decline by about 1½ percent of GDP, reflecting not only contribution revenues but also lower net interest payment on pension bonds. At the same time, “headline” public debt would decline by about 8 percent of GDP. As per the actuarial analysis, these accounting improvements would however not imply an improvement in the underlying fiscal position: while the deficits would initially be lower they would grow at a faster rate.
30. The proposal entails moderate fiscal cash-flow benefits. The extent of those benefits would largely depend on the amount of (partly temporary) financing that could be generated from transferred assets. From a flow perspective, the liquidity benefits for the budget from shifting to the mixed system would not be large: the budget would get most pension contributions revenues, but it would have reduced financing from pension funds. Still, there could be a moderate increase in cash flow benefits in the event that the government would continue to rely on pension fund resources to finance its expenditures. Its capacity to do so is however unclear given that the authorities have emphasized the need for greater diversification of pension fund assets in the mixed system, with the purchases of pension bonds not proposed being mandatory anymore.
31. The reform could have adverse implications for pension fund operations and the broader investment climate. Taking into account reduced commissions, the business of the pension funds could decline considerably. In addition, there could be substantial uncertainty over the dynamism of the higher-income segment of the pension system that would provide supplementary pensions. Such segments are typically more volatile, and some instability could be expected due to the uncertainty over new rules, including whether the pension funds would have enough autonomy to manage their assets. There could also be increased withdrawal from pension funds by those participants who fulfilled requirements for lump-sum reimbursement. More broadly, the signal of transferring assets to the public sector could be interpreted by parts of the investor community and the public as confiscation, with repercussions for the investment climate.
32. The proposal would have an ambiguous effect on alleviating social sustainability problems. The draft law does not include steps to broaden coverage among the poor, for lack of resources to enhance the noncontributory pillar. With respect to income distribution, the overall progressivity of the reform package is difficult to assess in the absence of detailed data: it includes both progressive (a robust flat pension for the lower income contributors) and regressive (a reduction in calculated lump-sum benefits, which generally accrue to the poor)18 elements.
33. Overall, the reform would have a rather small (and ambiguous) impact on sustainability, but may entail sizable implementation risks. The latter could comprise a possibility of political, legal, institutional, and financial disturbances and uncertainty involving pension fund assets and liabilities as well as pension rights. In addition, there would be substantial technical implementation risks due to limited government capacity to manage the transfer and activity of pension accounts. Finally, it is possible that, consistently with the practice of the last few years, the proposal would be augmented with elements that could increase its fiscal cost either in the process of parliamentary approval or subsequent implementation.
D. Recommended Reform Strategy
34. A more ambitious and comprehensive strategy is needed to deal with the root causes of the pension sustainability problem. Such a strategy could be sufficiently flexible to maximize consistency with elements of the authorities’ current proposal, or any proposal that could emerge as part of a societal consensus. In this context, the share of private versus public sector ownership, or defined contributions versus defined benefit schemes, would likely be secondary – in principle, either set-up fo the system could be consistent with sustainability requirements. There are however technical pension system parameters that need to be targeted for sustainability in light of the initial situation and cross-country experience.
35. Parametric reforms are the most significant policy tool for tackling the actuarial imbalances. The disparity between contributions and benefits could be bridged either by increasing contributions or raising the length of working life relative to the retirement period. Parametric reforms are the only direct tools for regaining this balance. Indirect tools such as tax breaks to incentivize greater coverage would be much less effective, and could have significant collateral damage, such as a potential loss of tax revenues.
36. Unlike most other steps, parametric adjustments to retirement ages would tend to have beneficial, simultaneous effect on both fiscal and social sustainability aspects in El Salvador. Increases in retirement ages (and required years of contributions), other things being equal, would generate fiscal savings in a defined benefit system and also help raise replacement rates in a defined contributions system. Such increases in retirement ages, and aligning those for men and women, are long overdue in El Salvador given its outlier status on retirement ages with most of its regional peers. By contrast, the case for increasing contribution rates is much less clear cut due to potential disincentives to participate in the labor market, and the fact that in El Salvador the contribution rates are not low compared to most countries in the region.
37. Despite their ultimate effectiveness, the parametric reforms would still have a relatively slow impact and thus need to be introduced soon. In light of accumulated delays in adjusting parameters and the established practice of grandfathering pension beneficiaries, the effects of new measures would be relatively back-loaded. Thus, realistically, the reforms could only partially reduce the unfunded liability that results from “transition costs,” but can play a crucial role in generating long-term savings, including supporting the viability of the defined contribution system in the future. 19 Also, those savings could be used to create additional fiscal space for the much needed increases in the coverage of the system, particular in expanding the coverage of the basic universal pension. In any case, given that it takes time for the reforms to have (significant) budgetary effect, they need to be introduced soon.
El Salvador: Example of Effects of Selected Pension Measures on Reducing Unfunded Pension Liability 1/ 2/
(In percent of GDP, NPV terms)
The effect of each measure is partial and could be differengt in a package.
The calculation does not take into effect the beneficial effects of these reforms on raising replacement rates in the new system, which are important for improving “social sustainability.”
El Salvador: Example of Effects of Selected Pension Measures on Reducing Unfunded Pension Liability 1/ 2/
(In percent of GDP, NPV terms)
| Changing the method of calculating benefits | 19 |
| Increasing minimum contributions period to 30 years | 11 |
| Raising retirement age by 5 years | 9 |
| Raising contribution rates | 4 |
The effect of each measure is partial and could be differengt in a package.
The calculation does not take into effect the beneficial effects of these reforms on raising replacement rates in the new system, which are important for improving “social sustainability.”
El Salvador: Example of Effects of Selected Pension Measures on Reducing Unfunded Pension Liability 1/ 2/
(In percent of GDP, NPV terms)
| Changing the method of calculating benefits | 19 |
| Increasing minimum contributions period to 30 years | 11 |
| Raising retirement age by 5 years | 9 |
| Raising contribution rates | 4 |
The effect of each measure is partial and could be differengt in a package.
The calculation does not take into effect the beneficial effects of these reforms on raising replacement rates in the new system, which are important for improving “social sustainability.”
38. To buy time until the parametric reforms make an impact, it would be essential to maintain confidence in the viability of the system and the rules of the game. With medium-term transition costs being largely exogenous over the next few years, there is little alternative to creating room in non-pension fiscal accounts to address these, which, over the next decade, would likely approach 2-3 percent of GDP annually (although this would depend on whether the current pension reform proposal goes ahead). In the meantime, policymakers should strive to find balance between realistic, affordable, and socially acceptable replacement rates in the system and design incentives for greater contributions compliance and formal labor market participation. Finally, some moderate re-distribution relative to the current system may be warranted given large effective subsidization of the rich. For example, a progressive taxation of the highly unequal pension benefits would both create fiscal space (by raising revenues) and address the distributional concerns.
Appendix I. El Salvador’s Pension System
El Salvador Pension System Structure
El Salvador Pension System Structure
| Sources of funds | Qualifying conditions | Benefits (2013) | Coverage Administrative | |
|---|---|---|---|---|
| 0 pillar | Budget, program cost is 0.07% of GDP | 70+ years of age, reside in municipalities with extreme poverty and not receiving any type of pension benefit | $50 per month, 25% of minimum wage | 8% of eligible population, less than 1% of total population Administered by municipalities (with means-testing and a comprehensive social assistance package) |
| 1 pillar (being phased out) | Initially recognition bonds, since 2006 forced borrowing from pension funds at LIBOR+0.75%, which is nearing its limit in 2016 | Born before 1962, 25 full years of work; at least age 55 for women and 60 for men | Benefits a proportion of last 10 years of wages; Replacement rate of around 70% | Less than 2% of total population A special institution (FOP) is in charge of financing from pension funds. ISSS, and INPEP manage administrative issues |
| 2 pillar (core pension system) | Social contributions (13% private sector and 14% public sector) and returns on pension fund assets (accumulated contributions and recognition bonds) | Born after 1962, 25 full years of contributions; at least 55 for women and 60 for men; early retirement possible if accumulated contributions sufficient to finance a pension 60% above a minimum pension | Pension equal to an annuity stream from assets. Minimum pension of $207 month. Replacement rates currently 65-70%, falling | Around 10% of eligible beneficiaries; 25% of contributors; less than 1% of total population currently Two private pension funds compete for contributors and administer individual accounts. Restrictions on asset allocation important (floor for investing in government bonds; ceiling for investing abroad) |
| 3 pillar | Voluntary contributions | Mostly work as addons to pillar 2 (same conditions) | Tax benefits | Negligible Pension funds |
El Salvador Pension System Structure
| Sources of funds | Qualifying conditions | Benefits (2013) | Coverage Administrative | |
|---|---|---|---|---|
| 0 pillar | Budget, program cost is 0.07% of GDP | 70+ years of age, reside in municipalities with extreme poverty and not receiving any type of pension benefit | $50 per month, 25% of minimum wage | 8% of eligible population, less than 1% of total population Administered by municipalities (with means-testing and a comprehensive social assistance package) |
| 1 pillar (being phased out) | Initially recognition bonds, since 2006 forced borrowing from pension funds at LIBOR+0.75%, which is nearing its limit in 2016 | Born before 1962, 25 full years of work; at least age 55 for women and 60 for men | Benefits a proportion of last 10 years of wages; Replacement rate of around 70% | Less than 2% of total population A special institution (FOP) is in charge of financing from pension funds. ISSS, and INPEP manage administrative issues |
| 2 pillar (core pension system) | Social contributions (13% private sector and 14% public sector) and returns on pension fund assets (accumulated contributions and recognition bonds) | Born after 1962, 25 full years of contributions; at least 55 for women and 60 for men; early retirement possible if accumulated contributions sufficient to finance a pension 60% above a minimum pension | Pension equal to an annuity stream from assets. Minimum pension of $207 month. Replacement rates currently 65-70%, falling | Around 10% of eligible beneficiaries; 25% of contributors; less than 1% of total population currently Two private pension funds compete for contributors and administer individual accounts. Restrictions on asset allocation important (floor for investing in government bonds; ceiling for investing abroad) |
| 3 pillar | Voluntary contributions | Mostly work as addons to pillar 2 (same conditions) | Tax benefits | Negligible Pension funds |
References
Acuña R., 2005, “Pension Reform in El Salvador,” Social Protection Discussion Paper No. 0507, World Bank.
Lazo, J. F., and others, 2010, “Reforma al sistema de pensiones: cobertura, brechas de género y poder adquisitivo,” Universidad Centroamericana “José Simeón Cañas.”
Stiftung Marktwirtschaft, 2015, “Honorable States? EU Sustainability Ranking 2015,” http://www.stiftung-marktwirtschaft.com/wirtschaft/themen/generationenbilanz.html
World Bank, 2004, “Implicit Pension Debt: Issues, Measurement and Scope in International Perspective,” Social Protection Discussion Paper Series No. 0403 (Robert Holzmann, Robert Palacios and Asta Zviniene).
Prepared by Bogdan Lissovolik
These replacement rates reflected higher average number of years of contributions than 15, but in any case replacement rates could not fall below 55 percent with the minimum requirements fulfilled.
In late 1990s, there was optimism over capacity to generate financial returns in El Salvador and globally, but it proved unfounded. In addition, the individuals making a choice may have implicitly assumed that the government would move to limit their losses on the downside, an assumption that proved at least partly correct.
The quick reduction in the number of pension funds and lack of new players subsequently raised concerns about barriers to entry (see Acuña (2005)).
The full dollarization regime adopted in 2001 likely constrained the government’s funding options.
This rate was however somewhat below market related levels considering the pension bond’s 25-year maturity and graduated back-loaded principal repayment schedule.
Nominal returns were partly influenced by the adoption of the full dollarization regime in 2001, which helped to substantially reduce interest rates.
In El Salvador, only defined contributions pensions have clear funding sources, while other pension liabilities have to be financed in an ad-hoc way, and would ultimately require a fiscal effort (either to pay the benefits immediately or service the debt burden over time). The estimates given herein exclude additional liabilities for military pensions, which however would have a limited impact (about 0.1 percent of GDP annually).
El Salvador’s NPV of around 100 percent of GDP is below Colombia’s and Mexico’s recent figures of 130 percent of GDP. In Emerging Europe, Stiftung Marktwirtschaft (2015) estimated implicit debt in the range of 100 to 500 percent of GDP.
In any case, sustainability would depend on the overall fiscal position and not only the profile of pension spending.
While those who cannot fulfill the years of contributions requirements would receive a lump-sum amount corresponding to their savings, this benefit would be clearly falling short from the social assistance perspective. Also, unlike those who earn a pension, the lump-sum beneficiaries would not enjoy a minimum pension guarantee, or free access to the public health care system.
The assumption of zero return on contributions may not seem realistic, but may not be far off in light of El Salvador’s fully dollarized economy with low inflation and low growth, as well as the specific distortions that depress its financial returns (see below).
This is an average of 29 for women at age 55 and 21 for men at age 60.
The calculation is sensitive to inclusion or exclusion of various groups of beneficiaries (e.g., disability and survivor pensioners).
The different risk profile of assets would not explain the difference in returns, in part because a significant chunk of the remainder of the portfolio is composed of (high-yielding) Eurobonds, whose risk profile is not higher than that of the pension bonds.
In this regard, the actuarial perspective indicates that a fall in average replacement rates could be yet deeper than the “baseline” scenario would indicate. This may reflect the role of some particular assumptions behind the baseline scenario that is modeled by the authorities, such as that of a “closed population” after 2045. As a result, the aging population pressures may be underestimated in the baseline.
This discussion is based on the draft law submitted to Congress in February 2016. Substantial changes to the draft were reported to be considered, but their status was unclear as of late-May and is not discussed here.
It has however been reported that this regressive element of the package could be reconsidered in parliament.
In El Salvador, it has been estimated that relatively ambitious parametric reforms would absorb only about 30 percent of the transition costs in NPV terms. At the same time, this package would allow to raise replacement rates in the defined contributions system by 10-12 percentage points.
