The Baltic states inherited a PAYG pension system from the Soviet Union (Box 2.1). Economic developments of the early 1990s had a tremendous impact on the pension systems in ail transition economies, including the Baltic countries. Economic depression accompanied by growing unemployment (open and hidden) undercut the pension system’s financing while adding large numbers of new beneficiaries.1 At the same time, high inflation rates that were not immediately fully matched by nominal pension increases contributed to a sharp decline in the real value of pensions and flattening of benefits across different groups of beneficiaries.2

PAYG in Early Transition Years

The Baltic states inherited a PAYG pension system from the Soviet Union (Box 2.1). Economic developments of the early 1990s had a tremendous impact on the pension systems in ail transition economies, including the Baltic countries. Economic depression accompanied by growing unemployment (open and hidden) undercut the pension system’s financing while adding large numbers of new beneficiaries.1 At the same time, high inflation rates that were not immediately fully matched by nominal pension increases contributed to a sharp decline in the real value of pensions and flattening of benefits across different groups of beneficiaries.2

The pension systems were also hampered by problems of tax compliance. Decentralization of economic management and privatization of state enterprises undercut, at least initially, the state’s ability to enforce tax collection, while low real wages, high unemployment, inflation, and uncertain political and economic prospects provided additional incentives to both employers and employees to avoid paying payroll taxes. In all three countries, pension system dependency ratios (that is, the number of pensioners divided by the total number of declared employed) were much higher than the demographic dependency ratio (the share of persons over 60 years divided by the number of people between 15 and 59 years), reflecting extensive early retirement and a sizable share of informal sector activity.3 Partly as a result, the tax authorities managed to collect only about 70–75 percent of what theoretically could have been collected with full compliance.

In addition, the pension systems became, over time, the most important de facto (if not de jure) sources of welfare benefits to the growing number of unemployed and poor. To counter declining economic activity and rising unemployment, most transition economies, including the Baltics, relaxed or bent pension rules to provide benefits to displaced older workers (Table 2.2).4 This use of the pension system to deal with rising unemployment arose from serious shortcomings in alternative social protection schemes,5 including unemployment insurance, as well as the lack of financial and real assets owned by the unemployed and retired in the Baltics.6

Table 2.2.

The Baltic States: Impact of Economic Crisis on the Number of Pensioners

article image
Source: National statistical agencies and staff estimates.

Includes both full- and part-time employment.

Given these pressures, it is not surprising that pension spending was high and rising in the early years of transition. In Latvia, for example, pension spending rose by 4 percentage points of GDP over the period 1991–95, increasing to over 10 percent of GDP In the two other Baltic countries, pension expenditures were more modest at 6–7 percent of GDP (Tables 2.32.5). This difference reflected both a somewhat worse demographic profile for Latvia and, by 1995, a higher average replacement rate (Tables 2.62.8). In all three countries, financial balance was largely maintained through a combination of high payroll taxes, relatively low replacement rates, and increased transfers from the central government. Looking forward, however, it was clear that demographic pressures would make it increasingly difficult to maintain this balance without substantial reforms.

Table 2.3.

Estonia: Financial Performance of the Social Insurance Fund

article image
Sources: Social Insurance Fund and Ministry of Finance; and IMF staff estimates and projections.

Largely interest Income from Social Insurance Fund’s bank deposits.

End-year stock. In 1998, including EEK 180 million of deposits in closed banks.

Includes net lending.

Table 2.4.

Latvia: Financial Performance of the Social Fund

article image
Sources: Social Fund and Ministry of Finance: and IMF staff estimates and projections.
Table 2.5.

Lithuania: Financial Performance of the SoDra

article image
Sources: Social Insurance Fund (SoDra) and Ministry of Finance: and IMF staff estimates and projections.
Table 2.6.

Estonia: Pension Indicators

article image
Sources: Social Insurance Fund and Ministry of Finance: and IMF staff estimates.

Average pension as a percentage of gross average wage.

Table 2.7.

Latvia: Pension Indicators

article image
Sources: Social Insurance Fund and Ministry of Finance; and IMF staff estimates.

Average pension as a percentage of gross average wage

Table 2.8.

Lithuania: Pension indicators

article image
Sources: Social Insurance Fund and Ministry of Finance; and IMF staff estimates.

Average pension as a percentage of gross average wage.

The Pension Systems at Independence

Upon regaining independence, all three Baltic states inherited from the Soviet Union a standard, publicly managed. PAYG pension system. Under this system, administered by a stale agency and financed by mandatory payroll taxes, those who worked paid for the benefits of current pensioners. Pensions were granted for several reasons; old-age, disability, and survivors’ pensioners were the three largest groups of benefit recipients. The pension system provided multiple retirement and benefit rules; while the standard retirement age was set at 60 for men and 55 for women, employees of heavy industry and mining, teachers, police, and other categories of employees could retire earlier. The initial pension benefit was typically based on years of service and earnings during the most recent years preceding retirement. However, years of service and earnings were adjusted to account for service in difficult conditions, and years of service included credit for noncontributory periods, such as child care, university studies, or military service. Benefit rules favored certain occupational groups, and some sectors offered supplementary benefits to their former employees. These occupation-related supplements contributed to large differences in benefits among pensioners with similar education, work tenure, and preretirement incomes. In general, the system was perceived as redistributive, with notional rates of return much higher for workers with relatively low labour income and even negative for high wage earners. (This argument is often challenged as lower educated wage earners start contributing social taxes much earlier in their careers and contribute for longer periods.)

While pension benefits were relatively generous, it is important to note that pensioners in the Soviet Union and other central and eastern European (CEE) transition countries relied almost exclusively on the state pensions for income during retirement. Under the Soviet Union’s economic system, individuals were not expected, or even allowed, to accumulate meaningful financial assets during working years (see Table 2.1). While households could save for purchases of major “big-ticket” items, such as appliances, cars, or, to a limited degree, housing, saving to derive capital income was not permitted.

Table 2.1.

The Baltic States: Household Wealth Held in Bank Deposits

article image
Source: International Financial Statistics.

Foreign current/deposits of the private sector.

Until the late 1980s, the PAYG systems in the Soviet Union and most other CEE countries were financially stable, generating financial surpluses dial were used to finance other parts of the government. High tax compliance was virtually assured because almost all employees worked for state-controlled institutions. Also, the relatively favorable demographic situation and high employment levels supported the financial stability of the system. In particular, in the three Baltic states, the elderly (55 years and older) comprised only 22.7 percent of the population in 1990 while, as a result of the baby boom during the 15 years following 1945. 41 percent of the population was of working age (between 25–54 years old).

Reforming the PAYG System

The stabilization of their economies in the mid-1990s allowed the Baltics to begin to focus on longer-term issues, in particular the sustainability of their existing pension systems. The large number of “baby boomers” born in the 1950s was expected to provide the PAYG system with significant financial surpluses during the first decade of the 21st century. If the existing parameters of the pension plan were left unchanged, however, the retirement of these baby boomers would lead to unsustainable deficits in the PAYG system.7 The demographic situation in the Baltics was further complicated by a sharp drop in birthrates in the 1990s, with total fertility per woman declining from over 2 in the late 1980s to 1.2 in Latvia, 1.3 in Estonia, and 1.5 in Lithuania in 1999 8 Finally, life expectancy at retirement is projected to continue increasing gradually, as the Baltic health systems modernize and the standard of living increases.9 As a result, for the Baltic states as a whole, the share of the 25–54 year olds in total population is projected to first increase from 41.6 percent in 2000 to 43.1 percent in 2010 and then decline to 38.5 percent in 2040 (Figures 2.12.4. During the same period, the share of the population 55 years and older will increase dramatically, from 25.3 percent in 2000 to 38.3 percent in 2040.

Figure 2.1.
Figure 2.1.

Population Distribution by age in 1990 and Projected for 2040

Source: Country authorities and IMF staff estimates.
Figure 2.2
Figure 2.2

Estonia: Demographic Projections, 1990–2040

Source: Country authorities; and IMF staff calculations.
Figure 2.3
Figure 2.3

Latvia: Demographic Projections, 1990–2040

Source: Country authorities; and IMF staff calculations.
Figure 2.4.
Figure 2.4.

Lithuania: Demographic Projections, 1990–2040

Source: Country authorities; and IMF staff calculations.

The push to reform the Baltic pension systems was given increased impetus by a number of other specific factors. First, while unemployment rates had begun to decline, a large number of potential workers lacked the incentive to give up early retirement or disability pensions as they could both work (often in the informal sector) and draw benefits. Second, since pensions depended only on years of service and not on the level of contributions, the pension systems provided little incentive for employees to contribute into the system or pressure their employers to comply with social tax rules.10 Third, high payroll taxes not only discouraged tax compliance and encouraged the growth of the shadow economy but also tended to increase labour costs.11 Fourth, the share of pension expenditure in total government outlays was relatively high, and changing fiscal priorities—in particular growing demand for infrastructure investment, education and medical services, and the need to finance preparations for EU accession—made the search for budgetary savings in the pension area unavoidable. Finally, the push to limit the relative size of pension expenditures was aided by a growing relative affluence of pensioners compared with other social groups, in particular families with young children.12

Starting in the mid-1990s, all three Baltic countries began reforms primarily addressed at scaling down the PAYG system’s expenditures and establishing a closer link between lifetime contributions and benefits. New formulas to link retirement benefits to past contributions were imposed, schedules for gradually raising the retirement age were put in place, indexation rules were modified to help ensure sustainability, and rules governing eligibility for benefits were gradually tightened. The most comprehensive, and earliest, reform was carried out in Latvia, while Estonia and Lithuania initially addressed pension reform in a more piecemeal fashion. However, at least some of the initial benefits of reform in Latvia were undone by subsequent ad hoc increases in pension benefits.

In each case, these reforms have been carried out within a broader framework envisaging the eventual introduction of a three-tier system. The first tier would be a modified PAYG system, with stronger links to contributions and a minimum pension to protect the lifetime poor. The second tier would be a fully funded system of privately managed savings accounts financed by a portion of the payroll tax. The third tier would be voluntary, privately managed pensions, organized primarily through employers (see Section III and Box 3.1).

Strengthening the Link Between Contributions and Benefits

Each of the Baltic countries has moved to reduce the redistributive element of their pension systems by linking individual pension benefits to lifetime contributions. The introduction of such a link was aimed at improving tax compliance, including by providing incentives for participation in the formal labour market and enhancing the transparency of the pension system. Reforms in Latvia 13 made this link most pronounced, while the pensions systems in Estonia and Lithuania continued to include significant elements of redistribution.

In 1995, Latvia became the first CEE country to implement the reform of its PAYG pension system based on the introduction of the “notional defined contribution system”.14 The new PAYG system was designed to mimic a contribution-based pension that would be offered in the private sector by an efficient insurance company. The system provides individual accounts to all workers paying the social tax, with those contributions earmarked for the pension system credited to these accounts. In fact, however, social tax payments are used to pay benefits to current retirees, with the individual accounts serving as a record-keeping mechanism. The “capital” in each account earns a rate of return equal to the growth of the sum of wages on which contributions are collected (the contribution wage base), and participants receive annual statements of their contributions and accumulated “savings.” At retirement, the pension paid is equal to the total capital in the person’s account, divided by the projected life expectancy of the beneficiary on retirement.15

Some element of redistribution was maintained in Latvia through the introduction of a minimum guaranteed pension. All those who reach the age of 65 (and those who reach 60 and have 10 years of service) are eligible for this pension, with benefits set at the level of the social assistance pension. Since April 1, 1998, this pension has provided about 30 lats (US$50) per month.

Hidden subsidies in the form of pension credits for noncontributory periods were also removed. All such subsidies were made explicit, as contributions to the pension fund for time spent in higher education, in military service, or at home taking care of children, are made through transfers from the state budget, with contributions based on the minimum wage. The cost of these transfers to the state budget is estimated at roughly 0.3 percent of GDP per year.

The Estonian authorities have also taken steps recently to link future pensions to contributions, albeit less closely than under the Latvian model. Beginning January 1, 1999, notional accounts were also introduced in Estonia. The new pension benefit formula consists of three components: a National Pension Rale (NPR) determined by parliament each fiscal year: a length of service component; and a contributions-related component, based on the worker’s income subject to contribution relative to the average income subject to contributions.

The national pension is granted to individuals not eligible for old-age, work-incapacity, or survivors’ pension, and provides minimum-income support for the elderly. Responsibility for these pensions has been shifted to the general budget, strengthening the insurance principles underlying the pension system. To qualify for the national pension, the person must reach retirement age and have resided in Estonia for at least 5 years prior to claiming the pension. The amount of the national pension is set as a percentage of the NPR and is determined by parliament every fiscal year; currently it is set at the equivalent of roughly US$47 per month.

The importance of length of service as well as the existence of a national pension rate in the benefit formula flatten the benefit structure and weaken the link between contributions and benefits in Estonia. It has been calculated, for instance, that even if the service-related component were completely phased out, an individual contributing 40 percent more than the average individual each year for 40 years would receive a pension only 13 percent higher in present value than that of the average person.

Under special circumstances, the government will make contributions to the pension fund at the rate applicable to the monthly minimum wage on behalf of qualifying individuals, including nonworking parents and the registered unemployed. Prior to the recent reforms these contributions awarded workers in these categories a full year of service and a significantly higher benefit. The new pension formula—which gives credit according to contributions paid—will correct this defect. The overall budget costs of the disability benefits, national pensions, and contributions for special categories of participants are estimated to be quite small.

In Lithuania, the reform effort also established a partial link between pension benefits and contributions. Under the 1994 reforms, pension benefits were calculated as a function of years of service (basic pension) and earnings (supplementary pension). This design was meant to ensure sufficient redistribution of benefits, while still introducing a relationship between lifetime earnings, contributions, and retirement benefits. Benefits were to be modest, providing at most a 40 percent replacement rate. While all citizens, including farmers, sole proprietors, and the self-employed, are covered by the basic pension, only persons employed under labour contracts are covered by the supplementary pension. To receive a full basic pension (set at 110 percent of minimum subsistence level) workers had to have at least 30 years of service, with workers with 15 to 29 years of service receiving proportionally lower benefits, and those with less than 15 years not eligible for the basic pension.

The supplementary pension implicitly links benefits to contributions in Lithuania. Benefits depend on a worker’s average lifetime earnings relative to average wage. Under this system, a record is kept of a worker’s wages, expressed relative to the average national wage for each year of employment, with a maximum allowable multiple of average wage of 5. At retirement, a worker’s 25 highest observation are averaged to compute the average lifetime earnings relative to the average wage. Benefits are computed by multiplying the worker’s average lifetime relative earnings by the average wage in the period prior to retirement, with credit given at the rate of 0.5 percent for each year of service.

Raising the Retirement Age

In all three countries, gradual increases in retirement age play a key role in the strategy of ensuring long-term sustainability of the PAYG pension system. Changes were as follows:

  • In Latvia, the 1995 pension legislation foresaw a gradual increase in retirement age for women from 55 to 60—the retirement age for men—by 2004, and it was anticipated that the retirement age would, in the longer term, increase to the age of 65 for both sexes. (Retirement age has subsequently been increased more rapidly. See Box 2.2.) In any case, as pension benefits were to be actuarially fair, so that expected lifetime benefits were unaffected by an individual’ retirement age, the system provided incentives to work beyond these ages.16’ In addition, special early retirement rights for particular occupations were phased out.17 The minimum pension guarantee did not apply to anyone taking a pension before age 60. Those retiring early were entitled only to an actuarially fair pension; it was expected that most would find this pension to be so low that they would continue to work, greatly reducing the number of early retirees.

  • In Estonia, starting on January 1, 1994, the normal retirement ages for men (60 years) and women (55 years) were revised each year by three months for men and six months for women, so that a retirement age of 63 years would be reached for men in 2006 and for women in 2010. The New Social Tax Act introduced the possibility of early retirement; an individual can retire up to three years before the statutory retirement age with a reduced pension. The adjustment factor is 0.4 percent per month in advance of normal retirement age, which is meant to be actuarially fair on average.

  • In Lithuania, legislation was passed in 1994 to raise the retirement age from 55 to 60 at the rate of four months annually for women and from 60 to 62.5 at the rate of two months annually for men, with the ultimate retirement age attained by the year 2009. Subsequently, under a World Bank-supported Structural Adjustment Loan (SAL) program the government accelerated these increases at the rate of six months yearly for both men and women, and increased the retirement age for women further, to 62.18 It was expected that the retirement age would have to increase to 65 by the year 2024 for men and women to ensure financial stability of the PAYG system.

Indexation Provisions

In the Baltics, where wages are low relative to industrial countries, but are expected to grow rapidly, an important issue is the extent to which retirees will share in productivity gains. To reduce the first pillar deficit, and allow an increase in the size of the second pillar over time, a government may wish to adopt at the outset a pension indexation formula that is below that of the wage growth rate. While such an indexing mechanism would have important benefits for the financial health of the PAYG scheme, it would imply a steady decline in the replacement rate for retirees relative to prevailing wage rates, although the replacement rate relative to the retirees’ last wage could continue to grow in real terms, assuming that real wages grow economy-wide. If pension coverage falls too much relative to prevailing wages, such indexation schemes could generate political pressure for additional ad hoc benefit increases.

In Latvia, affordability of the PAYG was reinforced through changes in its indexing provisions. During the accumulation period, contributions in the “notional” account are indexed to the inflow of resources to the pension system, so that liabilities do not grow more rapidly than revenues. Upon retirement, pensions are indexed to the Consumer Price Index (CPI) until 2002, and after that to a mix of wages and prices. This will improve the finances of the Pension Fund, and avoid rising liabilities during an economic downturn, but implies a declining replacement rate, relative to the prevailing wage, over time. Pensions are also adjusted for life expectancy, automatically adjusting to demographic changes.

Estonia has not yet developed a schedule for indexing pensions for inflation. Instead, pensions have been increased periodically, often before elections; there has been a political understanding that pensions would be maintained at about 40 percent of the average wage. The lack of explicit procedures is a serious potential weakness in the pension system, which is expected to be addressed under new legislation in 2000. Three possibilities are being considered: indexation to CPI, to the growth rate of the wage fund, or to a weighted average of the two. The last proposal has considerable merit in that it would reduce the liabilities to current pensioners with respect to full indexation to the wage fund but would still allow pensioners to partly benefit from the expected increases in real wages.

Latvia: The 1999 Pension Amendments

In Latvia, problems with the transition rules following the 1995 pension reform contributed to higher-than-expected pension spending. In particular, the fact that pension capital for new retirees was based only on the most recent year or two of contributions provided a strong incentive for working pensioners to have their benefits recalculated based on most recent earnings while some employees near retirement age even borrowed money to make large contributions. In these ways, a few gained entitlement to pensions up to six times the average. Parliament responded by imposing a ceiling on all pension benefits of about three times the average wage. In addition, amendments introduced in 1999 prohibited pension recalculations, although this was not made retroactive.

Several other changes were introduced, restoring some of the redistribution of the previous system, and chipping away at the sustainability that had been built into the pension reform. In particular, those who had contributed during the first years after the transition would have their capital valued at least at the average wage, even if their own wage was lower; and 80 percent of the guaranteed minimum pension was extended to all women retiring before age 60, regardless of the pension they should have received based on the benefit formula. The latter contributed to a more rapid-than-expected increase in the number of retirees in Latvia, and the provision is currently under review.

The fall in tax revenues associated with the Russia crisis forced the Latvian government to reconsider the level of pension spending and develop proposals to restore the original intent of the reform—bringing the pension system into balance. In August 1999, the Latvia parliament passed a series of amendments to the pension law. The retirement age for men was to be gradually raised from 60 to 62 by January 1, 2002; and for women from 57.5 to 59 on January I, 2000, and by 6 months each year until it reached 62 in 2006. The amendment cancelled the right to retire early (at age 55) available to women, and all working pensioners with monthly pensions above 60 lats were to lose benefits during 2000–05. The amendments also limited the eligibility for the minimum pension to those who contribute for more than 30 years. Other proposals included limiting cost-of-living indexation to once per year and returning to backward-looking indexation, and annulling special pension increases to pensioners who reached the age of 80.

These amendments met with strong resistance from a number of opposition parties that forced a national referendum on the bill. In response, the government submitted to parliament and won approval for a softer package of amendments. Specifically, the speed at which the retirement age was to reach 62 for both men and women was reduced. Starting January 1, 2000, men could retire at 60.5 and starting July 1, the retirement age for women will increase to 58. The retirement age for both men and women will increase by six months annually until it reaches 62. At the same time, the revised amendments allowed men and women with at least 30 years of service to retire at the age of 60 (men) or two years before the official retirement age (women) at 80 percent of the full pension. The revisions also scaled down the penalty to working pensioners: such pensioners with monthly pensions above 60 lats will receive 60 lats rather than zero. The revisions also returned to the original policy of guaranteeing a minimum pension of 30 lats to all pensioners, but left unchanged the original proposals regarding indexation provisions and the elimination of increases to those over 80. While the final amendments were weaker than those originally proposed, they appear to have had a positive impact on the finances of the Pension Fund thus far in 2000.

In Lithuania, pensions are indexed to a weighted average of two components. The first component is the minimum living level which is adjusted by the government on an ad hoc basis. The second component is based on the average wage in the economy.

Tightened Eligibility for Disability Pensions

In response to the rapid growth in the number of disability pensioners, the Baltic countries also moved to tighten eligibility for such pensions. In Estonia, modifications to the disability benefits were implemented in 1999, linking the work-incapacity pension to the length of service and the establishment of minimum contribution period. Prior to this change, invalidity pensions were flat rate benefits determined by the degree of invalidity. The work-incapacity pension will be paid only to permanently work-incapable individuals, and will be paid only during the working age years. When the individual reaches the retirement age, the benefits will be switched to an old-age pension.19

Experience with Reforms to Date

The experience with pension reform in the Baltics has been mixed. On the one hand, increases in retirement age and tightening of eligibility for benefits strengthened the finances of the pension funds in all three countries. In Latvia, in particular, social fund deficits of nearly 2 percent of GDP in 1993–94 were converted to surpluses averaging about ½ percent of GDP during 1995–97 (Table 2.4). On the other hand, pension expenditures as a share of GDP were significantly higher in all three countries in 1999 than in 1995. Some reforms, in particular in Estonia, are still quite new, and their full effects have yet to be felt.

Gains from reform have also been partially undone by ad hoc benefit increases in Latvia and Estonia. In Latvia, with an election looming in the fall of 1998, the government used a portion of the fiscal surplus to increase pension indexation beyond that provided for in the law,20 and pensioners received, in addition, a one-time increase averaging 15 percent. As a result, the social fund was in deficit in 1998–99, reserves were fully run down, and the pension fund had to borrow from the central government (Box 2.2). In Estonia, in late 1998, the parliament facing elections in early 1999 voted to increase average pensions by over 20 percent in real terms in 1999, contributing together with a recession to a deficit in the social fund of about 1 percent of GDP. These experiences underline the need for a rules-based approach to increases in pension benefits.

In Lithuania, the authorities have let benefits increase according to their formula. While several ad hoc adjustments to the minimum subsistence level (to which the basic pension is linked) led to a significant rise in pension spending, financial balance was generally maintained through economic growth. However, since the formula does not allow for a fall in pensions, with the fall in wages after the Russian crisis, the social insurance fund has more recently incurred deficits.

Taken together, while reform of the PAYG pillar has moved in the right direction, more needs to be done. Retirement ages should be increased further, and eventually reach industrial country levels. Benefit indexation should be set with due regard to its impact on PAYG finances and be taken out of the political arena. Finally, further strengthening of the link between contributions and benefits may lead to enhanced labour market efficiency and greater tax compliance.