POLICY RESEARCH WORKING PAPER 2631 Pension Reform in Hungary Hungary's pension reform package has been largely successful, significantly A Preliminary Assessment reducing imbalances in the pay-as-you-go system and the Roberto Rocha implicit pension debt while Dimitri Vittas introducing a mandatory, funded, privately managed pillar that seems to be operating fairly well despite initial problerns in the payment and registration systems and some regulatory weaknesses. Current shortcomings can be corrected by restoring the original 8 percent contribution rate to the second pillar and strengthening the regulatory and supervisory framework. The World Bank Europe and Central Asia Region Private and Financial Sectors Development Unit July 2001 I POLICY RESEARCH WORKING PAPER 2631 Summary findings Hungary is entering the fourth year of a multi-pillar for young workers, and most young workers effectively pension reform that has proved popular among workers opted for the new system. But the new system was despite initially lukewarm support from the government probably oversold as well, making older workers-who that succeeded the reforming government, and despite would be better off staying in the reformed PAYG the poor initial performance of capital markets because system-switch, too. of Russia's crisis in 1998. Roughly half the labor force The government has so far decided not to increase the joined the new system voluntarily. Most who switched contribution to the second pillar from 6 to 8 percent, as were younger than 40. originally planned, so efficiency gains in labor and capital Many people switched to the system because it offered markets may also be smaller than expected. more risk diversification. The pay-as-you-go (PAYG) Addressing projected deficits in the PAYG system may system, which had been severely damaged by repeated require further adjustments, such as delaying the manipulation of its parameters, clearly offered a low retirement age and shifting to indexed prices, reducing return on contributions. The new system is still net benefits to future generations. Reform has sharply predominantly PAYG. The first pillar accounts for more reduced the severe initial bias against future generations than two-thirds of the total contribution, but the new but hasn't eliminated it altogether. second pillar offers the chance of higher average returns The voluntary switching strategy achieves the same on contributions. outcome as a forced switch based on an arbitrary cutoff Most workers probably intuited the risk and returns age, while preventing legal problems and contributing to inherent in a pure PAYG system and mixed system, reduction of the implicit pension debt. But it leaves a few including the capital market risk in the second pillar and individuals worse off than if they'd chosen their best the political risk in the PAYG pillar. The new system option-a problem a well-designed public information offers better prospects of long-run risk-adjusted returns campaign can reduce. This paper-a product of the Private and Financial Sectors Development Unit, Europe and Central Asia Region-is part of a larger effort to disseminate ongoing research on pension reform in the Central European region. Copies of the paper are available free from the World Bank, 1818 H StreetNW, Washington, DC 20433. Please contact Lynn Gross, room H6- 148, telephone 202-473-7030, fax 202-522-0005, email address Igross@worldbank.org. Policy Research Working Papers are also posted on the Web at http://econ.worldbank.org. The authors may be contacted at rrocha@worldbank.org or dvittas@worldbank.org. July 2001. (29 pages) The Policy Research Working Paper Series disseminates the findings of work in progress to encourage the exchange of ideas about development issues. An objective of the series is to get the findings out quickly, even if the presentations are less than fully polished. The papers carry the names of the authors and should be cited accordingly. The findings, interpretations, and conclusions expressed in this paper are entirely those of the authors. They do not necessarily represent the view of the World Bank, its Executive Directors, or the countries they represent. Produced by the Policy Research Dissemination Center PENSION REFORM IN HUNGARY: A PRELIMINARY ASSESSMENT Roberto Rocha and Dimitri Vittas' ' Roberto Rocha is lead economist, the World Bank Regional Office for Central Europe in Budapest, and Dimitri Vittas is lead pension specialist, Research Department, World Bank. The authors are grateful to Patrick Wiese, for performing the actuarial simulations, and to Zoltan Vajda, Judit Spat, Csaba Feher, and Tibor Pamiczky for many fruitful discussions. TABLE OF CONTENTS 1. Introduction 2. The Situation of the Pension System Before the Reform The Rapid Maturation of the Pension System in the Post- War Period The Performance of the System in the 1990s The Future of the PAYG System in the Absence of Reform 3. The Hungarian Pension Reform General Description Switching Results Simulating the PA YG Reforms Simulating the Multipillar Reform with a Reduced Second Pillar Simulating the Multipillar Reform with the Original Second Pillar 4. The Structure and Performance of the Funded Pillars 5. Conclusions and Possible Lessons to Other Countries 1. Introduction In the summer of 1997, the Hungarian Parliament passed a proposal for a systemic reform to the pension system, involving substantive changes to the existing public pay-as-you-go (PAYG) system and the introduction of a new pension system. The new system comprises a smaller public PAYG system (the first pillar), and a new, funded and privately-managed system (the second pillar). A voluntary, funded and privately-managed pillar (the third pillar) already existed before the reform and continues to operate and grow. The new pension system started operating in January 1998, and became mandatory for all new workers entering the labor force after July of that year. Workers with accrued rights in the old PAYG pension system could choose to stay in the reformed PAYG system or switch to the new multi-pillar system. At the end of September 1999 (the deadline for switching to the new system), more than two million workers accounting for half of the labor force had decided to switch to the new pension system. Although some workers are expected to switch back to the reformed PAYG system until December 2002 (the extended deadline for switching back to the PAYG), the large number of switchers reveals the popularity of the reform among workers, particularly of workers under 40, which account for more than 80 percent of the total number of switchers. The Hungarian pension reform was the first systemic pension reform implemented in Central and Eastern Europe. Since then, a number of other countries have also implemented or are about to implement this type of reform, including Poland, Croatia, Latvia, Macedonia, and Kazakhstan. Multi- pillar systems are under preparation or consideration in several other countries of the region as well. The fact that Hungary pioneered this type of reform in the region, and that three years of implementation have passed, raises a natural interest in the Hungarian reform experience, and in the lessons that may be identified for other countries. The interest in the Hungarian reform experience may be enhanced by the fact that the center- right Government that succeeded the reformist center-left Government in mid- 1998-after six months of reform implementation-demonstrated just a lukewarm support for the reform. The lack of initial support was revealed in numerous official statements, claiming that the reform had not been well prepared and that it had an adverse impact on the public finances. One clear evidence of the lukewarm support was the new Government's decision to maintain the contribution to the second pillar at 6 percent, instead of increasing it gradually to 8 percent, as originally planned and prescribed in the legislation. This measure may have important implications for some particular cohorts and for market participants, and raises a number of issues which are relevant for other reforming countries. I This paper reviews the main components and objectives of the Hungarian pension reform, and makes a preliminary assessment of the first years of its implementation. The paper is structured as follows. The second section provides some background material, examining briefly the performance of the pension system before the reform, and showing long-run projections of the system in the absence of reform. The third section describes the overall reform package, examines the switching results, and provides a number of long-run actuarial simulations of the new multi-pillar system. The fourth section examines the structure and performance of the private pillars in the early stages of implementation. Finally, the fifth section provides some conclusions and identifies possible lessons for other countries. 2. The Situation of the Pension System Before the Reform2 The Performance of the System in the Post- War Period The Hungarian PAYG system matured rapidly in the post-war period, as reflected in the rapid increase in the system dependency ratio (the ratio of pensioners to workers), and the increase in the average replacement ratio (the ratio of the average pension to the average net wage). As shown in Figure 1, the system dependency ratio surpassed the old age dependency ratio already in the mid- 1 970s, due primarily to the low retirement age of women, who comprised an increasing proportion of the pensioner population, and whose life expectancy at retirement rose from 20 to 33 years in the last decades. The average replacement ratio also increased steadily in the post-war period (Figure 2), as a result not only of longer average contribution periods, but also of more generous benefits and more permissive eligibility rules. The increase in the number of pensioners and in the average benefit levels led to a steady increase in pension expenditures-from less than 5 percent of GDP in 1970s to more than 10 percent of GDP in 1990, requiring increasing contribution rates to balance the system. F1gw 1: (JdAgeadSysnI lyRaica, lS9- Figure 2: Ratio of Average Pension to Average Net , 1995Bcm°v Wage, 1950-1998 100 (in /) 83 1100 i 6D ~~~~~~~~~~~~~~80 - 60 40 : _ = = ~ _ 40 xl 20 0 15 9519)I519)17*1) 1)19511190 15 > 1951 1955 1960 1965 1970 1975 19W 1985 tggo 1990 m Lem 1950 1955 1960 1965 1970 1975 1980 1985 1990 1995 1998 2 By the early 1990s, total contribution rates to the PAYG amounted to around 34.5 percent of gross wages, including 30.5 percent for old age and survivors and around 4 percent for underage disability pensions. The Hungarian PAYG scheme arrived in the 1990s with difficulties to balance expenditures and revenues, despite charging one of the highest contribution rates in the world (Tables I and 2)3 During the 1990s the PAYG system was subject to further pressures, caused by a significant loss of revenues and a sharp increase in the system dependency ratio. As shown in Tables 1 and 2, the loss of revenues amounted to almost 3 percent of GDP before the reform, due primarily to an erosion of the tax base (the covered wage bill). Such base erosion was due not only to problems of ceilings and exemptions, but also to increasing evasion of the heavy payroll tax. The increase in the system dependency ratio amounted to 66 percent during the 1990s, most of which occurring in the first half of the decade (Figure 1). This dramatic increase in the dependency ratio was due to a reduction in labor force participation, increases in unemployment, and the maintenance of generous early retirement and disability schemes as a buffer against unemployment. The sharp increase in the number of pensioners implied strong pressures on expenditures and, combined with the revenue loss, would have resulted in very large PAYG deficits, in the absence of other correcting measures. The main correction that took place involved the manipulation of indexation parameters in the benefit formula, and which resulted in a significant drop of the average replacement ratio. A sharp real wage compression that took place in the mid-1990s (around 15 percent in 1995 and 1996) also contributed to the decline in real pensions, given the wage indexation of pensions. The final result of these measures was a drop in pension expenditures relative to GDP, and only modest PAYG deficits in the early and mid-1990s!4 Although these corrections prevented the emergence of large deficits in the PAYG system, they were perceived as arbitrary and unfair, diminishing the credibility of the PAYG in the eyes of the population. Furthermore, the scope for additional ad hoc corrections narrowed severely, making the system even more vulnerable to the demographic shocks projected for.the 215' century. As the consequences of a "do nothing" scenario were more widely understood, it became increasingly apparent that that the public pension system needed more fundamental reform. Before examining the long-run projections of the system, it must be noted that the PAYG deficits have increased somewhat after 1997 (the year when the reform was passed), but these deficits raise less concem, because they 2 This section draws on Palacios and Rocha (1998). 3The PAYG is defined so as to include contribution revenues, the pension expenditures of the Pension Insurance Fund, and the underage disability pensions of the Health Insurance Fund. It is arbitrarily assumed that the underage disability expenditures are covered by an equivalent amount of revenues. 3 are partly due to the creation of a second pillar, and because the actuarial imbalances of the system were already being addressed by the reform. Table 1: Revenues, Expenditures and Balance of the PAYG, 1991-99 (in % of GDP) 1991 1995 1996 1997 1998 1999 Contribution Revenues 11.0 8.9 8.4 8.3 8.2 7.8 Pension Fund 7.5 7.1 7.0 7.0 6.8 Health Fund (disability) 1.2 1.1 1.1 1.0 1.0 Pension Expenditures 10.5 9.1 8.5 8.3 8.7 8.8 Old age 6.8 6.3 6.2 6.4 6.4 Survivor 1.1 1.0 1.0 1.1 1.1 Disability 1.2 1.1 1.1 1.2 1.3 PAYG Balance 0.5 -0.2 -0.1 0.0 -0.5 -1.0 Revenue loss to second pillar 0.3 0.6 Pure PAYG Balance 0.5 -0.2 -0.1 0.0 -0.2 -0.4 Table 2: Base of Payroll Tax (% of GDP) and Contribution Rates (%), 1991-98 1991 1995 1996 1997 1998 1999 Covered Wage Bill/GDP 30.9 23.6 22.2 22.4 22.1 22.3 Total Contribution Rate 34.5 34.5 34.5 34.0 34.0 33.0 Pension Fund 30.5 30.5 30.5 30.0 31.0 30.0 Health Fund (notional) 4.0 4.0 4.0 4.0 3.0 3.0 The Future of the PA YG System in the Absence of Reform Along with the rest of Europe, Hungary will experience rapid population aging in the next few decades, a development that will submit the pension system to great pressures. The impact of these adverse demographic trends in the absence of reforms was assessed through the use of an actuarial model developed during the reform.5 The base year used for the actuarial projections is 1997-the year preceding the implementation of the reform. The main economic and demographic assumptions used in the actuarial projections are shown in Table 3. Table 3: Main Economic and Demographic Assumptions for Pension Simulations, 1997-2070 1998 1999 2000-2030 2000-2070 Economic Assumptions Real GDP Growth 5.0 4.1 3.0 2.7 Real Wage Growth 3.5 3.5 3.0 3.0 Inflation Rate 14.3 10.0 3.4 3.0 Unemployment Rate 9.0 8.0 7.0 7.0 Demographic Assumptions Population Growth -0.3 -0.3 -0.2 -0.3 Employment Growth 1.4 1.1 0.0 -0.3 Life Expectancy (Men) 65.1 65.0 71.8 76.4 Life Expectancy (women) 74.6 74.8 79.3 84.5 Note: Actual figures for 1998 and 1999. GDP growth is derived from wage growth. The labor share is assumed to remain constant. Life expectancy is years at birth, based on current mortality. 4Palacios and Rocha (1998) provide a more detailed analysis of the performance of pension revenues and expenditures during the 1990s. The actuarial model was developed by Patrick Wiese. 4 The economic assumptions include a decline in inflation rates to Western European levels, a constant labor share in GDP, and a moderate decline in the unemployment rate. The demographic assumptions are the same as the baseline scenario developed by Hablicsek (1995), and imply a declining population and a significant increase in the old age dependency ratio. The system dependency ratio also increases as a result, although this increase is somewhat moderated by the assumption that labor force participation rates converge gradually to the levels prevailing in Western Europe. This implies a moderate increase in labor force participation rates from the current levels, especially for women.6 Under these demographic and economic assumptions, and in the absence of reforms, the Hungarian PAYG system would generate growing deficits, as shown in Figure 3. The deficits would grow to around 2 percent of GDP at the end of the first decade of the century, and would converge to around 6.5 percent of GDP in 2070, the end of the projection period. The result is essentially due to the assumption of a declining rate of inflation and to adverse demographic trends. The decline in the rate of inflation implies increasing real average pensions and replacement ratios, because of the full backward wage indexation rule (prevailing before the reforms), and also because of smaller inflation- related losses in entry level pensions (the former benefit formula contained several indexation parameters that made entry level pensions very sensitive to inflation). Whereas the decline in inflation is the major cause of the early deficits, demographic factors dominate the results after the first decade. The projected fluctuations in the deficit closely mirror the old age and system dependency ratios, as shown in Figure 4. Both ratios increase between 1995 and 2017, followed by a ten year period of stability, followed by another increase. The old age dependency ratio nearly doubles to 65 percent at the end of the projection period, while the system lgume 3: Balmans of he ic Pension Sdeme inthe I Figure 4: Old Age and System Dependency Atsence of Iefoim, 1997-2070 Cm % of GW Ratios in the Absence of Reforms, 1997-2070 (in %) 0.0 -1.0 140- 120- -2.0 0 -3.0 ~~~~~~~~~~~~~~~~~80- -4.0 ~~~~~~~~~~~~~~~~~~60- 40- -5.0 ~~~~~~~~~~~~~~~~~~~~~~~~20 1997 2003 2009 2015 2021 2027 2033 2039 2045 2051 2057 2063 2069 199720032009 20152021 20272033203920452051205720632069 - Dependency ratio -Old age dependency ratio 6 Labor force participation for women in their 30s and 40s are assumed to increase from around 70 percent to around 75-83 percent. This assumption was adopted because labor force participation rates declined significantly during the early stages of the transition in the 1990s, from Hungary's historical levels and from the levels prevailing in most European countries. 5 dependency ratio grows to 120 percent. In the absence of reforms, balancing the pension system in 2070 would require increasing the contribution rate to more than 55 percent, or reducing the replacement ratio from around 60 percent to less than 3 5 percent. This would either cause even greater distortions in the labor market and/or great dissatisfaction and unrest among future pensioners. It is also clear that the absence of reforms would imply a massive burden on future generations, irrespective of whether the future imbalances were financed by higher contributions, lower replacement ratios, or general taxes. Recent calculations of generational accounts confirm that workers under 38 years of age would be net contributors to the system, and that the net tax burden on future generations would be particularly heavy.7 3. The Hungarian Pension Reform General Description of the Original Reform Package The reform package gave workers the choice to stay in a reformed PAYG or to switch to a new, mixed pension system until end-August 1999. Workers who initially opted for the new system will be able to return to the reformed PAYG until December 2002. After that date, workers will be permanently affiliated either with the reformed PAYG or with the new system. New entrants in the labor force after July 1998 have been automatically enrolled in the new mixed system. The reforms to the PAYG included the following main components: (i) a higher normal retirement age of 62 for both men and women (from 60 and 55, respectively); (ii) an increase in the number of years of service to be eligible for early retirement without penalties to 40 years; (iii) increases in the penalties .for early retirement and in the rewards for late retirement; (iv) changes in the benefit formula designed to eliminate its explicit redistributive elements (i.e. a correction factor that penalized higher income workers); (v) a new tax regime; and (vi) a shift from backward net wage indexation to a "Swiss" indexation formula consisting of a combination of contemporaneous price and wage indexation (50 percent net wages, 50 percent consumer prices). The new legislation included detailed transition tables for the retirement age increase, the corresponding early retirement penalties, the minimum years of service for early retirement without and with penalties, and a new set of accrual rates which apply to gross rather than net wage history. The retirement age and the minimum years of service for early retirement started rising immediately but only reach their final state in the year 2009. The new legislation also included a transition period for the new 6 indexation formula, specifying the maintenance of full backward wage indexation for 1998 and 1999, followed by a mixed 70-30 percent (wage-prices) contemporaneous indexation formula for 2000, and finally a contemporaneous Swiss formula for 2001 and subsequent years. The new benefit formula and tax regime will become effective by 2013. Many of the changes reflected the Government's position that redistribution should be removed from the pension scheme. This was based on the Government's desire to tighten the link between contributions and benefits in order to improve compliance and the insurance characteristics in the system. An element of intra-generational redistribution was maintained, but in the form of a minimum top-up, means-tested pension benefit financed outside the pension system. As summarized in Table 4, workers deciding to stay in tlr reformed PAYG will pay a contribution rate of 30 percent of their gross wages, and will earn an accrual rate of 1.65 percent for each year of service.8 Workers who switch to the new system will have 22 percent of their gross wages channeled to the first (PAYG) pillar, earning an annual accrual rate of 1.22 percent for each year of service. The switching workers would earn the same 1.22 percent accrual rate for every year of service before the switching date. The switching workers would also contribute 8 percent of their gross wages to their second pillar accounts. This contribution rate structure would follow a two year transitional period- the contribution to the second pillar would be 6 percent in 1998, 7 percent in 1999 and 8 percent from 2000 onward. The overall contribution rate would remain at 30 percent and the contribution to the first pillar would be reduced accordingly. The first pillar of the new multipillar system applies the same rules as the reformed PAYG, including higher retirement age, minimum years of service, and indexation arrangements. However, the benefit formula was scaled down in proportion with the size of the contribution rates. Therefore, the annual accrual rate in the new first pillar was reduced to 1.22 percent, or roughly 74 percent of the 1.65 accrual rate which applies for those workers who remain in the reformed PAYG scheme. This corresponds to the ratio of the contribution rate to the PAYG paid by workers who switch to the multipillar scheme to the contribution rate paid by workers who do not switch, namely 22/30. For those who switch to the new scheme, the 1.22 accrual rate applies for both past and future years of participation in the system, implying that anyone who switches is voluntarily forfeiting approximately one quarter of their acquired rights in the process. Auerbach, Kotlikoff, and Leibfriz (1999) provide a general methodology of generational accounts and empirical results for several countries. Gal, Simonovits, and Tarcali (2000) construct generational accounts for Hungary. 8 Workers staying in the reformed PAYG can also obtain additional coverage from the voluntary, third pillar, which has been in existence since 1994. Vittas (1996) provides an early analysis of the Hungarian third pillar. 7 Table 4: Main Elements of the Reform Package "Old" PAYG "Reformed" PAYG New System First Pillar Second Pillar Contribution rate 30% 30%o 24-23-22% 6-7-8% Annual accrual rates Non-linear, equivalent Linear, 1.65% of Linear, 1.22% of n.a.. to 2% of net wage gross wage gross wage Pension Base Gradual increase to Gradual increase to Gradual increase to lifetime earnings lifetime earnings lifetime earnings Redistribution factor Yes No No Normal retirement age 60 (men); 55 (women) 62 (both) 62 (both) 62 (both) Early retirement rules Low penalty rates and Higher penalty rates Higher penalty rates minimum service and minimum service and minimum service years for early years for early years for early retirement retirement retirement Indexation of pensions Backward net wage Contemporaneous Contemporaneous indexation Swiss Swiss As discussed in greater detail in section 4, workers who switch to the new system and contribute for at least 15 years to the new second pillar are guaranteed a minimum second pillar benefit equal to 25 percent of the first pillar pension. The guarantee is modest for new workers, being equivalent to a minimum lifetime real return of only 0 percent p.a.. However, the guarantee is equivalent to a minimum lifetime real return of more than 4 percent p.a. for workers in their mid-forties, due to the shorter accumulation period. Therefore, the guarantee could be triggered for workers in their mid-forties who switch to the new scheme, although the amount required to meet this guarantee should not be significant. Changes in the Reform by the New Government The elections of May 1998 resulted in the departure of the Government which designed and implemented the pension reform, and which consisted of a coalition of the socialist party (a center-left party) and the party of free democrats (a center-right party). Another coalition Government formed by the party of young democrats (a center-right party) and the party of small shareholders (a right party) won the elections by a narrow margin. The new Government did not give emphasis to pension reform in its pre-election program, nor did it mention its intention to modify the ongoing reform during the campaign period, but has expressed less than full support to the reform during its tenure. One of the first measures announced by the new Government was the decision to maintain the contribution rate to the second pillar at 6 percent in 1999, instead of increasing it to 7 and 8 percent, as originally planned. Another important measure involved changes in the transition indexation rules negotiated during the reform preparation. More specifically, instead of maintaining the backward wage indexation in 1999 (which would have resulted in nominal pension increases of 18 percent), the Government announced an ad hoc increase of 14 percent. The announcement of these changes created uncertainty among workers and market participants, especially as they were not accompanied 8 by an announcement of the policies that would be followed in 2000 and in future years. Some politicians in the new coalition raised the level of uncertainty further, by announcing their intention to introduce more fundamental changes and even to roll back the reform entirely. Political factors may have motivated these initial sharp attacks on the reform. However, it also seems that some policy-makers became concerned with the transitional deficits caused by the loss of revenues to the second pillar. Technical discussions inside and outside the Government showed that the size of the transitional deficit was moderate and that this deficit was in any case neutral from the point of view of macroeconomic stability (as discussed below). During the second half of 1999 the general attitude towards the reform improved somewhat and the attacks subsided, but the Government still indicated that it would maintain the contribution to the second pillar at 6 percent until the end of its tenure (mid-2002). The Government also extended the option for workers to switch back to the PAYG, from December 2000 to December 2002. Actual Switching Outcomes The Government that passed the reform initially considered a switching strategy that involved forcing all workers below the age of 40 to switch to the new system and all workers above that age to stay in the reformed PAYG. However, it became increasingly apparent that a mandatory cut-off age could spark constitutional battles over accrued rights and prove too costly to implement. These problems led the Government to make the reform mandatory for new entrants (after July 1998) and voluntary for anyone with a contribution history in the old system. Moreover, the Government also decided to recognize accrued rights by making compensatory pension payments at the time of retirement, as in Argentina, and not by recognition bonds, as in the case of Chile, Peru, and Colombia. The rights earned under the old scheme are recognized by applying the accrual rates of the new first pillar. Since these accrual rates are lower than those implied by the old formula, switching workers voluntarily forego part of their accrued rights. The new system is still attractive to most younger workers, because the higher expected returns in the second pillar result in higher pensions in the new system under reasonable assumptions. Therefore, the valuation of past contributions in the context of a voluntary switch allowed the Government a certain measure of control over the speed of the transition and the size of early transition deficits, as well as a reduction in the implicit pension debt. As shown in Figure 5, if the contribution rate to the second pillar were set at 8 percent, as in the original reform package, and under conservative assumptions on returns and costs (returns of 2 percent above wage growth, annuity rate equal to wage growth, and operating costs and charges 9 amounting to 15 percent of contributions), workers below 36 years of age would tend to switch to the new system, whereas workers above that age would find attractive to stay in the reformed PAYG. The higher replacement ratios of younger workers are essentially due to the effect of interest compounding over a longer number of years. This interest accumulation effect outweighs the reduction in accrued rights for workers under 36 years of age, but is not sufficient for workers above that age. Of course, these are rather conservative assumptions about pension fund returns. As shown in Figure 6, the rate of return-wage growth differential in the 1980s and 1990s was higher than 2 percent in a sample of countries with large funded systems, even in countries that imposed portfolio restrictions and/or followed very conservative portfolio strategies, such as Denmark and Switzerland. For the sake of illustration, assuming returns of 3 percent above wage growth, the equilibrium cut-off Figure 5: Replacem ent Ratios in the Old and New Systems for Each Cohort. Reforms 2% above wage growth (Gross Pension/ Gross W age, in% ) 765 b _ 5_ % 70 % 60% '=__________ 40% 22 24 26 28 30 32 34 36 38 40 42 44 46 48 50 52 lMM onoPillar M ulti 8% l _ _ _ Multi 8% (in 2003) B Multi 6% Figure 6: Real W age G rowth and Real Returns on Pension Funds (1984- 1996) 8 - -- -- --- - - - - - - - - - - - - -- - - - - - - - - - - - - - - - - - - - 6 - --- ---- -- - - ----- -- -- - - - -- - - -- - - - 4 - - - 02--- - LLL: lLLr -2-Ireland UKChile Netherlands Denrnark Switzerland,- -2 | Real Return of Pension Funds *Real W age Growth 3Difference Source: EFRP (1996) and OECD (1998) 10 age would increase to 38 years of age, as shown in Figure 7. If returns were assumed to be as high as those shown in Figure 6, the cut-off age would exceed 40 years of age. Figure 7: Replacement Ratios in the Old and New Systems for Each Cohort. Returns 3% above Wage Growth (Gross Pension / Gross Wage, in %) 80 60 i 40 22 24 26 28 30 32 34 36 38 40 42 44 46 48 50 52 54 56 58 60 62 - Reformed PAYG -Multipillar with 8% |. | _ _ _Multipillar with 8% in 2002 . Multipillar with 6% The actual switching outcome was in line with other switching experiences and largely met initial expectations.10 As shown in Figures 8 and 9, the number of switchers was significant in the first few months of implementation, kept increasing throughout 1998 and 1999, and accelerated in the last two months before the final deadline for switching (September 1999), reaching approximately 2 million workers or nearly half of the labor force. At that point, more than 80 percent of switchers consisted of workers below 40 years of age, and more than 80 percent of workers in their 20s and early 30s had switched to the new system. The increase in the number of switchers after that date reflects primarily the new entrants to the labor force, which are mostly young workers. By December 2000, roughly 90 percent of workers in their 20s and early 30s were enrolled in the new system. There may be a number of explanations of why workers continued switching, despite the uncertainty caused by the absence of political support to the reform by the new Government. For one, the possibility to switch back to the reformed PAYG by the end of 2002 may have eliminated the perception of risk associated with early switching. Also, some workers may have speculated that the 9 A positive difference between the return on capital and wage growth is also a condition of dynamic efficiency (see, e,g, Barro and Sala-i-Martin). The returns on pension fund assets shown in Figure 6 cannot be used as direct evidence of dynamic efficiency because they contain a risk premium on equity. However, see Feldstein (1995) for a discussion of the dynamic efficiency condition in the presence of risk for the US case, and Kotcherlakota (1996) for a discussion of the equity risk premium. 10 See Disney and Whitehouse (1992) for an analysis of the opt-out experience in the case of the UK, Rofman (1996) for a description of the Argentine experience, and Palacios and Whitehouse (1999) for a comparison of switching outcomes in a sample of reforming countries. 11 originally envisaged contribution rate of 8 percent would sooner or later be restored. Indeed, if the contribution rate were increased to 8 percent in 2003 by the same Government or a new Government (after the 2002 elections), the equilibrium cut-off age would be only slightly reduced, as shown in Figures 5 and 7. Finally, workers switching to the new system may have probably assigned a great weight to the political risk associated with the PAYG scheme. The perception of significant political risk associated with the public PAYG is justified, in view of the manipulation of the parameters of the Hungarian PAYG benefit formula in the late 1980s and early 1990s (section 2). The decision of the new Government to abandon the indexation formula, negotiated and agreed for the period of transition, and established in the 1997 law, provides another example of how the parameters of a public PAYG system may be easily changed by subsequent legal amendments. Figure 8: Number of Switchers to the New System (in absolute numbers and in % of labor force) 2,500,000 0.6 2,OD0,000 0E5 - 1,500,000 i 1,000,000 0.2 500,000 i.1 98. 1. 98. 11. 98. 111.98. IV. 99. 1. 99. 11. 99. 111.99. IV. 00. 1. 00. II. 00.111. 001V. -u. r% of Labor Force Number of Switchers7 12 Figure 9: Actual Switching Outcome, December 2000 (as a % of the labor force in each cohort) 1000 _ ~O 600 40.0 20.0 0.0 15-24 25-29 30-34 35-39 40-44 45-49 50-74 Cohort It is noteworthy, however, that the failure to restore the original 8 percent contribution in the next few years could lead to a significant reduction in the equilibrium cut-off age. As shown in Figures 5 and 7, maintaining the contribution to the second pillar at 6 percent indefinitely would reduce the cut-off age to around 28-33 years of age, depending on the assumption for pension fund returns in the long-run. It is difficult to assess whether and when the contribution rate to the second pillar will be increased to 8 percent. However, it is clear that there are a few cohorts that might be better-off switching back to the reformed PAYG, if the original contribution rate is not restored. Simulating the PA YG Reforms It is useful to present the simulations of the reform in two stages. First, the various measures designed to improve the balance of the PAYG are examined and contrasted with the no reform scenario. Second, the direct fiscal impact of the introduction of the second pillar is examined, in combination with the PAYG reforms. The macroeconomic assumptions used are the same as those in Table 3, with the exception of a slightly higher growth rate of the labor force in the scenarios which include an increase in the retirement age (which also implies slightly higher rates of GDP growth, given the assumption of- a constant labor share). This section highlights the impact of the two major reform measures, namely, the increase in the retirement age and the shift toward mixed indexation. The changes in the benefit formula and the tax treatment have an impact on particular workers but little or no impact on the aggregate balance of the PAYG (Palacios and Rocha (1998)). As shown in Figure 10, the new retirement age rules reduces significantly the projected deficits. This is due not only to the increase in the statutory normal retirement 13 age over time, but also to the increase in the penalties for early retirement and in the minimum years of service for early retirement. While it is difficult to predict retirement behavior in the face of the new penalties or the average number of contribution years, reasonable assumptions suggest an increase in the effective retirement age for men and women of roughly 2 and 5 years respectively. The longer working period raises pensions and replacement ratios, given the accrual rates, but the higher pension is received for fewer years and some individuals continue to contribute to the scheme. The net effect is an average annual reduction in future deficits of about 1.5-2 percent of GDP. While the retirement age increase has an important impact, it is only when the new indexation method is added to the reform package that the PAYG scheme moves into an extended period of surplus. Figure 10 shows how these surpluses peak in 2013, which is the year when the baby boom cohorts begin to retire. Later, deficits reemerge when a second demographic shock hits the PAYG scheme around Figure 10: Balances of the Public Pension Scheme After PAYG Reforms, 1997-2070 (in % of GDP) 2 0 2i 9 2009 20'1!-..242±2 2033 Z0 _ 51 2057 2063 2065 -2 -3 -4-- - - - - - - - - - - - - - -5 -6 -u-- No reform - - - Retirement Age Increase - Retirement Age Increase + Swiss Indexation 2035. With an increase in life expectancy of two years per decade assumed, the retirement age increase to 62 is not sufficient to offset the demographic developments and to maintain a constant retirement duration in the long run. These two major parametric reforms generate an average annual improvement in the PAYG balance of more than 4 percent of GDP during the projection period. The implicit pension debt, measured according to an accrued benefit obligation (ABO) definition", is reduced from 309 percent of GDP in the no reform scenario to 241 percent of GDP with the PAYG reforms, a reduction of more than 20 percent. 11 The PAYG is assumed to be terminated in 1997 and the future pension obligations (accrued as of December 1997) are discounted at the rate of wage growth. 14 The PAYG reform measures shown in Figure 10 provide the starting point for the analysis of the introduction of the second pillar. To this point, the package results in a significant improvement in the finances of the PAYG during the next decade, followed by a gradual erosion of the surpluses at the end of the following decade. The pension system would record deficits again at the end of the projection period, and the elimination of these deficits would require further reforms to the PAYG, such as a further increase in retirement age (say, to 65 years, as in most OECD countries) or the adoption of price indexation. Nevertheless, the reform measures achieve a significant reduction in future deficits and the implicit pension debt. In fact, the reforms produce some small but significant surpluses . Despite the presence of these surpluses, the Government never gave serious consideration to a reform package limited to improving the PAYG scheme, as it saw four serious shortcomings in such a solution: First, the accumulation of surpluses in the PAYG would provide for an easy opportunity to reverse the reforms through politically-motivated benefit increases. Second, it would have created a new role for the public pension fund as an asset manager. There was little reason either from historical Hungarian or international experience to believe that such an arrangement would lead to efficient investment allocation or good corporate governance.12 Third, this solution was unlikely to contribute to the type of capital market development which a multipillar package was capable of generating. Finally, the promise of higher returns in the private scheme, even after taking into account higher administrative costs, helped offset the benefit reductions in the PAYG scheme and simultaneously diversified the workers' risk in the long-run. This positive aspect of the overall package was instrumental in generating support, especially among younger voters. In view of these and other perceived advantages, the new system was designed to divert the savings generated by the reform to privately-managed pension funds referred to as the second pillar. Simulating the Multipillar Reform with a 6 Percent Contribution to the Second Pillar The simulations of the full multi-pillar package were performed assuming initially that the number of switchers will stay roughly the same as of September 1999, and that the contribution rate to the second pillar will be maintained at 6 percent indefinitely. Figure 11 confirms the obvious fact that allowing switching workers to divert part of their contributions to the second pillar causes an immediate revenue loss to the PAYG. The revenue losses increase rapidly to around 0.8 percent of GDP in the first 4-5 years of the reform, and then keep increasing at a more gradual pace to reach 1.4 percent of the GDP in the third decade, when most of the active population will be enrolled in the new system. The PAYG deficit would tend to increase at the same pace, but the PAYG reforms described above more than offset the revenue loss, allowing a reduction in the deficit. The system is actually 12 Iglesias and Palacios (1999) show that publicly-managed pension funds have performed very poorly. 15 2 1.5 0.5 -0.5 -1.5 1997 2003 2009 2015 2021 2027 2033 2039 2045 2051 2057 2063 2069 -PAYG Reforms Before Second Pillar -Multipillar Reforms with 6% Contribution projected to generate small surpluses at the end of the decade, even considering the revenue losses, but turns into deficits again in the second decade, with the retirement of the baby boom generations. The PAYG deficit peaks around 2022 but then improves, as the first significant cohorts to receive first pillar benefits in the new system retire (i.e. those in their mid-thirties in 1998). The difference between the PAYG balances with and without the opt out to the second pillar increases until about that time and narrows thereafter. By 2040 the deficit in the multi-pillar system is actually smaller than the PAYG deficit had the second pillar not been introduced. These results are driven by two factors. First, the replacement ratio of the first pillar of the new system is about three fourths of the replacement ratio in the reformed PAYG. Thus, the temporary imbalances between replacement ratios and contributions created by the opt out starts to taper off after the cohorts in the mixed system begin to retire. Second, the reform involves a reduction of about one fourth of the accrued rights of workers who opt for the new system. As a result, the valuation for the years of contribution under the old system are lower than what would have been generated by the old benefit formula. To determine the first-order impact of the reform on national savings the public and private pension savings need to be combined, as shown in Figure 12. The PAYG balances with the second pillar opt out are reproduced in Figure 12, together with the net private contributions to the second pillar (gross contributions plus interests minus redemptions), and the sum of the two balances. Total pension savings peak at around 2011, decline thereafter, following the decline in the PAYG balances, but increase again after 2020, when the first significant numbers of workers begin to retire in the new system. The contribution of the pension system to savings decline in the third decade due to demographic aging. The private scheme starts maturing, with increasing redemptions and smaller net positive contributions, while the public scheme shifts into deficits. The net result is a decline in total pension savings to around I percent of GDP. In order to eliminate the public deficits and increase total savings, the authorities would have to increase further the retirement age and/or shift to price indexation. "S Figure 12: Public, Private and Total Pension Balanees with 6% Contribution to 2nd Pillar (in % of GDP) 23 0.53 1_ .1. 1997 2003 2009 2015 2021 2027 2033 2039 2045 2051 2057 2063 2069 Public Balance (Multipillar) N- N et Private Balance | - Total Pension System Balance The final impact of the pension reform on national savings will depend on the reaction of voluntary private savings to the individual measures of the reform, and on the reaction of the Government to the path of the PAYG deficits and surpluses. The reaction of voluntary private savings to the reform is difficult to estimate numerically, although it is possible to identify some of the major changes that might occur. The increase in the retirement age could induce some decline in private savings, whereas the change in indexation would imply a decrease in expected retirement income, inducing some increase in private savings.13 The expectations of higher returns on the contribution to the second pillar (relative to the PAYG) could have a positive or negative effect on voluntary savings, depending on the relative sizes of the income and substitution effects. Therefore, net impact of all these factors on voluntary private savings is in principle ambiguous, and would in any case be dampened by the existence of liquidity constraints. The reaction of the Government to the path of future PAYG deficits will be primarily influenced by Hungary's objective of joining the EU by the year 2005. Meeting the objective of EU membership will require an effort to reduce the general Government deficit from around 4-4.5 percent of GDP (the levels that prevailed in the late 1990s) to levels below the Maastrich ceiling of 3 percent of GDP. The expected reduction of the General Government deficit to levels below 3 percent of GDP suggests that the transition will be primarily tax-financed and that savings effects might be stronger. However, additional increases in national savings arising from offsetting reductions in the Government's deficit are likely to be moderate, as the General Government deficit is already close to the Maastrich ceiling, the projected PAYG deficits are moderate, and the actuarial projection even predict a period of PAYG surpluses. 13 These are the responses predicted by the overlapping generations model. See, e.g., Kotlikoff (1989). 17 2 1.5 0.5 -1 -1.5 1997 2003 2009 2015 2021 2027 2033 2039 2045 2051 2057 2063 2069 | ~~~PAYG Reforms - M ultipillar Reforms with 6% Contribution -MultipillarReforms with 8% Contribution On the whole, there is no reason to believe that the direct impact of the reform on national savings would be substantially stronger than indicated in Figure 12. Of course, to the extent that the reform has a positive impact on Hungary's growti performance, there could be additional indirect effects on savings, resulting from endogenous interactions between savings and growth, and these effects are not reflected in Figure 12.14 The perception that the Hungarian transition would be primarily "tax-financed", due to the PAYG reforms and the overall reduction in the General Government deficit, led many observers to state that the reform would benefit excessively future generations, to the detriment of current generations, who would be forced to "pay twice". However, whereas it is true that the reform package has reduced significantly the burden on future generations, it has not been sufficient to fully restore intergenerational balance, given the severe initial bias against these generations. Calculations of generational accounts for Hungary indicate that the pension reform has reduced the net burden on future generations by roughly three fourths, but that future generations are expected to remain net contributors to the system.15 The reduction of the initial intergenerational imbalance has been achieved by increasing slightly the burden on current workers, as older workers and pensioners have not been significantly affected by the reform and have remained net beneficiaries of the system. The calculations of generational accounts for Hungary also show that current workers would have incurred even greater losses if the second pillar had not been introduced. What the calculations do not capture, however, is the possible positive impact of the reform on Hungary's growth performance and on the welfare of current generations. 14 The interactions between pension reform, growth, and the welfare of different generations are examined in Kotlikoff (1995) and Corsetti and Schmidt-Hebel (1995). Loayaza, Schmidt-Hebel, and Serven (2000) provide a recent survey of the empirical literature on savings. '5 Gal, Simonovits, and Tarcali (2000). 1X Simulating the Multipillar Reform with a 8 percent Contribution to the Second Pillar The reluctance of the Government to raise the contribution rate to 8 percent, allegedly because of the larger transitional deficit, raises the question of what would be the impact of this measure on the balances of the system and on the economy. This analysis is summarized in Figures 13 and 14, where it is assumed that the contribution to the second pillar is raised to 8 percent in 2003. The number of switchers is assumed to be the same as in Figures 11 and 12. Figure 13 reveals the obvious fact that the increase in the second pillar contribution from 6 to 8 percent would imply an additional loss of revenues to the PAYG and slightly larger PAYG deficits. The difference between the two lines would be 0.3-0.4 percent of GDP in the next decades. However, these larger deficits would be offset by larger private surpluses, leading essentially to the same overall pension balance, as shown in Figure 14 (by comparison with the overall balance in Figure 12). Therefore, the increase in contribution rates to the second pillar would not produce any adverse impact on key macroeconomic variables such as inflation and the current account. Figure 14: Public, Private and Total Pension Balaces with 8% Contribution to 2nd Pillar (in % of GDP) 4 3 2 0 1997 2003 2009 2015 2021 2027 2033 2039 2045 2051 2057 2063 2069 Public B alance (M ultipillar) Net Private B alance - -_- Total Pension System B alance The difference between the two scenarios depicted in Figures 11-14 is clearly artificial, as it assumes that the number of switchers is the same. If a large number of workers switches back to the PAYG, the scenario with a 6 percent contribution could actually prove worse than the 8 percent contribution scenario at the end of the projection period, as it would imply a smaller reduction of accrued rights, a larger implicit pension debt, and possibly larger PAYG deficits in the long-run. The failure to restore the original contribution rates could also imply some efficiency losses. A smaller second pillar implies a weaker link between contributions and benefits with possible adverse 19 effects in the labor market. A smaller second pillar also implies more limited capital market effects, due to the slower growth and smaller size of pension funds. It is very difficult to determine the quantitative impact of these two effects on economic performance in the long-run with any degree of accuracy, but it is bound to be negative.'6 Simulations of Alternative (Counterfactual) Reform Scenarios As mentioned above, during the first stages of the reformn the new system was criticized for various reasons, including the generation of a transitional deficit. Some of the critics in Hungary claimed that a system of notional defined contribution (NDC) accounts would have generated the same positive results without generating the deficits. To examine whether this criticism is valid, this section provides counterfactual simulations of an NDC type of reform, and compares them with the pure defined benefit (DB) reform and the Multi-pillar reformn examined above. Figure 15 shows projections of the Hungarian pension system with all the major paratnetric reforms that have effectively been adopted (e.g. retirement age increase and Swiss indexation), and the counterfactual adoption of an NDC scheme as of January 1998. Therefore, workers retiring after that date would receive a DB benefit based on their accrued rights under the old DB formula, and an NDC benefit based on their notional balances accumulated after that date. Two counterfactual NDC schemes are simulated, the first with a notional interest rate equal to GDP growth, and the second with an interest equal to GDP growth plus one percent. The schemes convert their notional balances into annuities assuming the same respective notional interest rates in the payment period. The two lines are compared with the pure DB reform and Multi-pillar reform shown in Figure 11. As shown in Figure 15, the NDC scheme paying an interest rate equal to GDP growth would produce smaller deficits than the Multi-pillar and the reformed DB systems at the end of the projection period, while the NDC scheme paying GDP growth plus one percent ends up with larger deficits. However, the NDC system paying just GDP growth would still fare worse than a pure DB reform in the first 40 years. This is because the NDC scheme would "transform" the employment growth that happens in the early and middle stages of the transition into higher pension benefits, something that the DB scheme does not do. In the very long-run, the NDC scheme paying GDP growth generates a better balance because it pays an interest rate lower than wage growth (GDP grows at a lower rate due to decline in employment). The multi-pillar reform produce a higher deficit than an NDC paying GDP growth, but again this is due to the partial diversion of contributions to the 16 See Levine and Zervos (1996) for an empirical analysis of capital market development and economic growth, and Holzman (1996) for an analysis of this effect in the Chilean case. 20 Figure 15: NDC Reforms X DB and Multipillar Reforms 2 I . 0 -I __I -2 - - - - - - - - - - 1997 2003 2009 2015 2021 2027 2033 2039 2045 2051 2057 2063 2069 DB Reforms -M M u Iti Pillar | N NDC with interest= gdp growth l - - - NDC with interest = gdp growth+l% I second pillar. The overall pension balances produced by the multi-pillar reform are better than those produced by the NDC schemes. The computation of the implicit pension debt under these alternative reform scenarios provides another interesting tool for comparisons across scenarios. As shown in Table 5, the implicit pension debt (IPD) falls in all the reform scenarios, but is lowest in the multi-pillar scenario, due to the voluntary reduction of accrued rights by switchers. The IPD of a pure DB reform is initially lower than that generated by an NDC paying GDP growth, because of initially lower replacement ratios. In the long-run, the NDC paying a modest interest rate would produce a lower IPD, but again this is because the replacement ratios would drop under an NDC scheme paying GDP growth rates. A pure DB reform would also be capable of producing lower IPDs if accrual rates and replacement ratios were reduced. Table 5: Implicit Pension Debt under Different Reform Scenarios (in % of GDP) Scenario Base Year 1997 2010 2020 2030 No Reform 309 340 363 378 Pure DB Reform 241 226 246 254 Multi-Pillar Reform 232 208 216 214 NDC with i= gdp growth 248 233 245 232 NDC with i= gdp growth + 1% 257 255 281 288 The reforming Government decided not to adopt a pure NDC reform for the same reasons why it decided not to adopt a pure DB reform (e.g. excluding a second pillar). The initial surpluses produced by a pure NDC reform could be depleted by politically-motivated increases in pension benefits, by increases in the central budget deficit, or by political interference in the management of the assets leading to negative returns. Moreover, an NDC scheme would still be operated by the public sector on a PAYG basis and would not have the risk-diversification properties of a mixed 91 system. Finally, an NDC scheme would not contribute to the development of capital markets that Hungary and other transitional countries need in the current stage of their transformation 17 4. The Structure and Performance of the Funded Pillars Evolution and Structure Before the 1997 reform, Hungary had already a voluntary private pension scheme (the third pillar) which started operating in 1994. The concession of generous tax incentives led membership to increase to I million affiliates by 1999, the equivalent of 25 percent of the labor force. The average contribution has been about 5 percent of the average wage, and total assets reached 1.5 percent of GDP in 1999. Affiliates in the voluntary system tend to be middle-to-high income workers above 35 years of age and employed in larger enterprises. The industry was initially very fragmented, including a large number of very small and poorly managed funds, but a process of mergers and liquidations reduced the number of active voluntary funds from 270 in 1994 to 160 in 1999. The pension funds in the mandatory (second) pillar were constructed as mutual associations, as in the third pillar. The system started operating in January 1998 with only 38 licensed funds, as a result of stricter licensing criteria, and the number of funds had already fallen to 25 by end- 1999, as a result of mergers. Concentration in the second pillar is much higher than in the voluntary pillar-by mid-1999, the five largest funds accounted for 78 percent of all members and 73 percent of total assets, and only 8 funds had less than 5,000 members. Participation in the mandatory pillar increased rapidly, reaching 2 million workers in September 1999, or the equivalent of 50 percent of the labor force, and will continue expanding in the future due to new entrants in the labor force. The total assets of second pillar institutions rose at a rapid pace, reaching 0.9 percent of GDP in 1999, and are expected to exceed the assets of the third pillar in less than three years, due to the much larger membership (already twice as large as the voluntary pillar) and the higher contribution rate. The Regulatory Framework The regulatory framework is generally sound and similar for both pillars, although it is stricter for the mandatory pension funds'8. For example, voluntary pension funds can be established without 17 See Disney (1999) for a critical assessment of NDC schemes. However, most of the objections to NDC schemes raised above would not apply in the cases where the NDC scheme is smaller and just part of a broader and truly mixed pension system, such as in the case of Latvia, Poland, and Sweden. See, e.g., Lindeman, D., M. Rutkowsky, and 0. Sluchinky (2000) for an updated account of recent reforms which include a first pillar built on NDC basis. 22 any minimum capital or membership, even when they intend to offer annuities. In the mandatory pillar, there is a minimum size requirement of 2,000 members for a pension fund and 25,000 members for those that offer annuities. In addition, annuity providing funds must create a capital reserve of HUF 100 million. The law stipulates that pension funds are separate legal entities and their assets must be segregated from those of sponsoring employers or other founders. The funds are in principle governed by their members through their boards, and the law contains detailed rules on fund governance that aim to preserve the independence of boards and management. In practice, however, most fund members do not exercise their voting rights or participate in general assemblies, making the pension funds resemble more mutual funds, which are effectively controlled by their sponsors. The possibility of fund members to exit poorly performing funds (switching across funds is allowed with few restrictions) is expected to exert more discipline on fund managers than formal voting rights and board rules. Pension funds are required to employ the services of certified asset managers, accountants, auditors, actuaries, and custodians. A minimum capital requirement on external asset managers is imposed, amounting to HUF 100 million for small pension funds and HUF 500 million for pension funds with over HUF 2 billion in assets. Pension funds must use a single custodian institution, which is responsible for the safekeeping of assets and for ensuring compliance with asset allocation policies and prescribed investment limits. To ensure asset diversification and limit the risk exposure of pension funds, investments are subject to quantitative limits by asset class as well as individual investments. No individual investment may exceed 10 percent of the assets of the fund (fund limit) or 10 percent of the equity of the issuer (issuer limit). The ceiling on equity holdings is 30 percent of the total asset portfolio, and the ceiling on foreign assets was initially set at zero, but is scheduled to increase to 30 percent of the total portfolio by 2003. Quantitative investment limits on Hungarian pension funds- appear reasonable, and are unlikely to be binding and to affect marginal investment decisions, at least in their early years of operation. Investment restrictions are expected to be relaxed gradually, as the capital market develops and the country approaches the date of membership in the European Union. Although the regulatory framework is generally sound, there are also some important weaknesses in both regulation and enforcement that have not yet been corrected. Market valuation of assets is required only once a quarter, while the crediting of declared interest is effected only once a Is Policy issues in pension fund regulation and supervision are discussed in Davis (1995), Demarco and Rofman (1998), 23 year. This allows significant room for the creation of hidden reserves and the manipulation of reported investment returns. Pension funds are required to disclose their terms and conditions to new members and to send annual statements to all their members. They must also indicate clearly their returns and fees. However, the lack of more precise rules and guidelines for computing and disclosing costs and returns still creates room for different practices across funds and problems for comparing fund performance. Two types of guarantees are offered in the second pillar. First, the second pillar pension benefit cannot be lower than 25 percent of the pension benefit from the first public pillar. This guarantee is backed by a central guarantee fund to which all pension funds contribute. Second, pension funds must make up any shortfall in individual returns, if investment performance falls below the return of a portfolio of long-term government bonds by more than 15 percent. The minimum return is backed by a minimum reserve equal to 0.5 percent of total member assets. Returns exceeding 40 percent of the benchmark are placed in reserves. These guarantees built support to the reform but present some complications. The first guarantee implies a lifetime real return guarantee of only 0 percent p.a. for young workers, but a lifetime guarantee of about 4 percent p.a. for workers in their 40s. The probability that this guarantee will be called for these older workers is not negligible, especially if the contribution is kept at 6 percent indefinitely. The second guarantee was expected to complement the first and improve capital protection, but contains some flaws in design. For one, using a bond index as a benchmark for the second guarantee seems to have distorted pension fund investments in favor of bonds. More importantly, the excess reserves are imposed at the level of the pension fund, not the asset managei9. The two guarantees together imply access to a central guarantee fund without putting the capital of the asset manager at risk first, raising the issue of moral hazard in the behavior of asset managers. Performance Unfortunately, good data on the operating fees and costs and the investment returns of the private pension funds are not readily available. Operating fees have absorbed between 5.5 and 7.5 percent of contributions in voluntary funds and between 7.5 and 11 percent in mandatory funds. These rates are lower than those reported for Latin American pension funds, where operating fees frequently amounted to more than 25 percent of contributions in the first years of operation. Three reasons may account for this difference. First, in Hungary, there are additional charges for asset management and external administration, and these seem to add between 50 and 100 basis points for most funds. Second, sponsors of pension funds in Hungary may have absorbed a higher proportion of operating Queisser (1998), Rocha et al (1999) and Vittas (1997, and 1998). 24 costs. Employer-sponsored funds, in particular, are reported to subsidize the operations of pension funds by providing rent-free premises and by not charging for the time of staff who are involved in the administration of pension funds. Third, Hungarian pension funds seem to spend less on marketing and on commissions to selling agents than their Latin American counterparts. Data on investment returns are even less satisfactory. Although most pension funds publish some data, the numbers do not seem to include unrealized capital gains or losses, nor accrued but not yet received dividends and interest. To shed some light on the performance of pension funds, investment returns were simulated by using average quarterly asset allocations of pension funds for 1998 and 1999, and the total market returns on bank deposits, bonds, and equities. Assuming a quarterly investment horizon, gross investment returns in 1998 amounted to 17 percent for the mandatory funds and to 14 percent for the voluntary funds against an inflation rate for the year of 14 percent. For the first three quarters of 1999, annualized investment returns were respectively 13.4 and 12.8 percent for the mandatory and voluntary funds, against an inflation rate of 10 percent. Allowing for asset management and custodial fees, it is likely that real returns on individual accounts have been positive in real terms, though not much above zero. The returns of pension funds in 1998 and 1999 were low in real terms, primarily as a result of the financial crises of recent years, particularly the collapse of Russian markets. These crisis led to a sharp increase in short-term interest rates, capital outflows, and a sharp decline in the return on equity and bonds during 1998. Equity prices recovered in 1999, but pension fund managers decided to maintain a low share of equity in their portfolios (less than 10 percent in the case of mandatory funds), in view of the overall volatility in world capital markets, and did not benefit from the recovery in equity prices. Real returns are expected to increase to higher levels in future years, in line with the expected increase in the share of domestic and foreign equities in portfolios. 5. Conclusions and Possible Lessons to Other Countries Hungary has entered the fourth year of implementation of a multi-pillar pension reform that has proved popular among workers, despite the initial lukewarm support from the Government that succeeded the reforming Government, and also despite the poor initial performance of capital markets, due to the Russia crisis of 1998. Approximately half of the labor force has joined the new system voluntarily. Most of the switchers are workers below 40 years of age, and the young workers who switched account for more than 80 percent of the labor force in their age group. 9 Rocha et al (1999) and Vittas (1998) examine in more detail the problems in designing pension guarantees. 25 The decision to switch was motivated by a number of factors, the most important being the better risk diversification properties of the new system. The credibility of the PAYG system had been severely damaged by repeated manipulation of its parameters and the PAYG offered clearly a low return on contributions. The new system is still predominantly a PAYG system, as the first pillar accounts for more than two thirds of the total contribution, but contains a new second pillar which offers the prospects of higher average returns on contributions. Most workers probably understood intuitively the risk and return characteristics of a pure PAYG system and a mixed system, including the capital market risk in the second pillar and the political risk in the PAYG. The new system offers better prospects of long-run risk-adjusted returns for young workers, and most young workers effectively opted for the new system. However, some overselling of the new system probably occurred as well, making older workers switch, although they would be better-off staying in the reformed PAYG. The Government has decided so far not to increase the contribution to the second pillar from 6 to 8 percent as originally planned, alleging that such an increase would increase the size of the transitional deficit and possibly produce macroeconomic imbalances. However, the increase in the contribution to the second pillar would not have any adverse impact on the economy, as the increase in the PAYG deficit would amount to only 0.3-0.4 percent of GDP, and would in any case be offset by larger private savings. This decision has also violated the objectives and the internal consistency of the original reform package, in which contribution rates and accrual rates were jointly calculated so as to produce an equilibrium cut-off age around 35-40 years of age. Clearly, many workers have switched counting on a higher contribution rate, and if the 8 percent original contribution is not restored in the next few years, some of these workers will be worse-off. The failure to restore the original contribution rates may also produce smaller efficiency gains in labor and capital markets than originally anticipated. The PAYG is still projected to produce deficits in the long-run, despite the PAYG reforms. Addressing these projected deficits will require further adjustments in the future, such as further increases in the retirement age and a shift to price indexation. These future adjustments are bound to result in some reduction in net benefits for future generations. Independent calculations of generational accounts for Hungary suggest a similar outcome-the reform package has reduced sharply the severe initial bias against future generations, but has not eliminated the bias entirely. Future generations remain net contributors to the pension system, even after the reform. Therefore, the criticism frequently voiced in Hungary-that the reform had benefited excessively future generations to the detriment of current generations-seems unjustified. 26 A more ambitious reform package would have involved larger initial surpluses (and smaller deficits in the steady-state) and a significant reduction in the high contribution rates, with stronger positive effects on savings and on the labor market. However, the erosion in the tax base that occurred in the 1990s ruled out any ambitious plan to reduce contribution rates, and the additional PAYG reforms that could have allowed larger initial surpluses and/or reductions in contribution rates (e.g., a direct move toward price indexation) were ruled out for political reasons. Despite these shortcomings, any preliminary assessment of the Hungarian pension reform would need to conclude that the reform has been successful, especially considering the severe constraints imposed by initial conditions (e.g. large fiscal deficits, high contribution rates, high tax rates, very adverse demographic trends). The reform has reduced significantly the imbalances of the PAYG system and the implicit pension debt, while also introducing a mandatory, funded, and privately-managed pillar that seems to be operating fairly well, despite the initial problems in the system of payments and registration, and some weaknesses in the regulatory framework. Moreover, the current shortcomings can be corrected over the next few years, by restoring the original 8 percent contribution rate to the second pillar and strengthening the regulatory framework. The Hungarian reform also suggests that a voluntary switching strategy achieves essentially the same switching outcome as a forced switch based on an arbitrary cutoff age, while avoiding legal problems and contributing to the reduction of the implicit pension debt. The disadvantage of this strategy is that it leaves a few individuals worse-off relative to their best option. A well-designed public information campaign may minimize the occurrence of these cases. The implementation of the new second pillar met initial difficulties, as the information, registration, and payments systems had not been yet fully developed. These initial difficulties caused an initial discomfort and, although these technical problems were eventually solved, they show that more attention to practical and technical aspects of implementation is required from policy-makers. Finally, the Hungarian reform would benefit from a stronger regulatory and supervisory framework, including several aspects of asset valuation and disclosure. Although no major problems have been reported, and these areas of the regulatory framework are expected to be improved in the near future, a stronger regulatory framework could have been introduced from the start of the program without major technical difficulties. 27 REFERENCES Auerbach, A., L. Kotlikoff, and W. Leibfritz (1999). Generational Accounts Around the World. NBER-The University of Chicago Press. Barro, R. and X. Sala-i-Martin (1995), Economic Growth, McGraw-Hill. Corsetti, C. and K. Schmidt-Hebel (1995). "Pension Reform and Growth". World Bank Policy Research Working Paper No. 1471. Washington DC. Davis, E. P. (1995), Pension Funds, Retirement-Income Security, and Capital Markets - An International Perspective, Clarendon Press, Oxford. Demarco, G.; R. Rofman, and E. Whitehouse "Supervising Mandatory Funded Pension Systems: Issues and challenges", Unpublished Manuscript, The World Bank, January 1998. Disney, R. (1999), "Notional Accounts as a Pension Reform Strategy: An Evaluation". Social Protection Discussion Paper No. 9928, December. The World Bank. Washington D.C. Disney, R. and E. Whitehouse (1992), "The Personal Pensions Stampede", The Institute for Fiscal Studies, London, UK. European Federation of Retirement Provision (1996), European Pension Funds: Their Impact on European Capital Markets and Competition. EFRP, London. Feldstein, M. (1995), "The Missing Piece in Policy Analysis: Social Security Reform", National Bureau of Economic Research Working Papers, No. 5413. Feldstein, M. (1995), "Would Privatizing Social Security Raise Economic Welfare?", National Bureaus of Economic Research Working Papers, No. 5381. Gal, R., A. Simonovits, and G. Tarcali (2001), "Generational Accounting and the Hungarian Pension Reform", unpublished manuscript, TARKI, Budapest, Hungary. Holzmann, R. (1996), "Pension Reform, Financial Market Development and Economic Growth - Preliminary Evidence for Chile. IMF Working Paper, WP 96/90, IMF Staff Papers 1997. Iglesias, A., and R. Palacios (2000), "Managing Public Pension Reserves: Evidence from the Intemational Experience". Social Protection Discussion Paper No. 0003, January. The World Bank. Washington D.C. Kotcherlakota, N. (1996), "The Equity Premium: It's Still a Puzzle", The Journal of Economic Literature, Vol. XXXIV, No 1, March. Kotlikoff, L. (1989), What Determines Savings?, The MIT Press, Cambridge, Massachusetts. Kotlikoff, L. (1995), "Privatization of Social Security: How it Works and Why it Matters", National Bureau of Economic Research Working Papers, No 5330. Levine, R. and S. Zervos (1996), "Stock Market Development and Economic Growth", The World Bank Economic Review, Vol 10, No. 2, May. 28 Lindeman, D., M. Rutkowski, and 0. Sluchynsky (2000), "The Evolution of Pension Systems in Eastern Europe and Central Asia: Opportunities, Constraints, Dilemmas, and Emerging Best Practices", unpublished manuscript, The World Bank, Washington D.C. Loayza, N., K. Schmidt-Hebel, and L. Serven (2000); "Savings in Developing Countries: An Overview", The World Bank Economic Review, Volume 14, No.3 (September). Organization for Economic Cooperation and Development (1998), Maintaining Prosperity in an Ageing Society. OECD. Paris. Palacios, R. and R. Rocha (1998), "The Hungarian Pension System in Transition". Social Protection Discussion Paper No. 9805, April. The World Bank. Washington D.C. Also in Bokros L. and J.J. Dethier, eds. (1998): Public Finance Reform during the Transition, The Experience of Hungary. The World Bank. Washington D.C. Palacios, R. and E. Whitehouse (1998), "The Role of Choice in the Transition to a Funded Pension System". Social Protection Discussion Paper No. 9812; September. The World Bank, Washington D.C. Queisser, Monika (1998), "Regulation and Supervision of Pension Funds: Principles and Practices", OECD Development Centre, International Social Security Review, Vol. 51, April-June, p. 1-21. Rocha, R., J. Gutierrez, and R. Hinz (1999), "Improving the Regulation and Supervision of Pension Funds: Are There Lessons from the Banking Sector?", Working Paper No. 9929, The World Bank, December. Forthcoming in Holzman R. and J. Stiglitz, eds. (2000). New Ideas About Social Security. The World Bank. Washington DC. Vittas, D. (1997), "Investment Rules and State Guarantees for Mandatory Private Pension Funds", Unpublished Manuscript. The World Bank, Washington DC. Vittas, D. (1996), "Private Pension Funds in Hungary: Early Performance and Regulatory Issues", Policy Research Working Paper 1638. The World Bank, Washington DC. Vittas, D. (1998), "Regulatory Controversies of Private Pension Funds". Unpublished Manuscript. The World Bank, Washington DC. 29 Policy Research Working Paper Series Contact Title Author Date for paper WPS2615 Trade, Growth, and Poverty David Dollar June 2001 E. Khine Aart Kraay 37471 WPS2616 Reforming Land and Real Estate Ahmed Galal June 2001 D. Dietrich Markets Omar Razzaz 34995 WPS2617 Shanghai Rising in a Globalizing Shahid Yusuf June 2001 S. Yusuf World Weiping Wu 82339 WPS2618 Globalization and the Challenge Shahid Yusuf June 2001 S. Yusuf for Developing Countries 82339 WPS2619 Do Banks Provision for Bad Loans Michele Cavallo June 2001 E. Mekhova in Good Times? Empirical Evidence Giovanni Majnoni 85986 and Policy Implications WPS2620 Who Owns the Media? Simeon Djankov June 2001 R. Sugui Caralee McLiesh 37951 Tatiana Nenova Andrei Shleifer WPS2621 Does Indonesia Have a "Low-Pay" Deon Filmer June 2001 H. Sladovich Civil Service? David L. Lindauer 37698 WPS2622 Community Programs and Women's David Coady June 2001 L. Wang Participation: The Chinese Experience Xinyi Dai 37596 Limin Wang WPS 2623 Trade Liberalization in China's Elena lanchovichina June 2001 L. Tabada Accession to the World Trade Will Martin 36896 Organization WPS2624 Are Incentives Everything? Payment Varun Gauri June 2001 H. Sladovich Mechanisms for Health Care Providers 37698 in Developing Countries WPS2625 Australia's Experience with Local Garry Pursell June 2001 L. Tabada Content Programs in the Auto Industry: 36896 Lessons for India and Other Developing Countries WPS2626 Mandatory Severance Pay: Its Donna MacIsaac June 2001 H. Sladovich Coverage and Effects in Peru Martfn Rama 37698 WPS2627 With the Help of One's Neighbors: Harold Alderman June 2001 P. Sader Externalities in the Production Jesko Hentschel 33902 of Nutrition in Peru Ricardo Sabates Policy Research Working Paper Series Contact Title Author Date for paper WPS2628 Monopoly Power and Distribution in Hanan G. Jacoby June 2001 P. Kokila Fragmented Markets: The Case of Rinku Murgai 33716 Groundwater Saeed Ur Rehman WPS2629 Bridging the Digital Divide: How George R. G. Clarke July 2001 P. Sintim-Aboagye Enterprise Ownership and Foreign 37644 Competition Affect Internet Access in Eastern Europe and Central Asia WPS2630 Parallel Imports of Pharmaceutical Mattias Ganslandt July 2001 L. Tabada Products in the European Union 36896