Abstract

ABSTRACT Most public old-age pension schemes around the world are based at le ast in part on the pay-as-you-go defined benefit (PAYGO DB) model. As these schemes have matured and the limitations of this approach have become more salient, pension experts have begun considering alternative models. The Notional Defined Contribution (NDC) model, which is also financed on a PAYGO basis, has emerged as one of the major new approaches. In the years ahead it may be combined with or possibly displace the funded defined contribution model as the major alternative to the PAYGO DB model. Drawing primarily on evidence from NDC schemes in 6 countries (Sweden, Italy, Poland, Latvia, the Kyrgyz Republic, and Mongolia), the goal of this paper is to describe the NDC model and to review its strengths and limitations relative to the major alternatives. A four pillar pension scheme is proposed to illustrate how a NDC pillar might be integrated into a multi-pillar scheme. One strength (relative to the PAYGO DB model) is that it makes a more explicit link between contributions and eventual pension benefits; however, the flipside of this strength is that it provides less adequate pension benefits to low-wage workers due to the lack of income redistribution. The fiscal burden of the transition is less than that associated with a shift to a funded defined contribution scheme, but NDC schemes lack many of the potential economic benefits associated with funded defined contribution schemes, such as contributing to economic growth. The NDC model may become common among the nations in the process of making the transition from centrally planned to market economies, among Western European nations, and among developing nations. It is less likely to be adopted in nations that currently have fully or partially privatized schemes in place. In the United States, the model will probably get relatively little attention in the debate over Social Security reform, at least for the foreseeable future. However, it is a model that would put workers with low wages and irregular work histories at less risk than the funded defined contribution alternative being actively considered in current debates about Social Security reform in the United States. 1 Prior to World War II national pension schemes paying modest benefits were in place in many industrial countries and a few developing na tions, particularly in Latin America. After the war, conditions were right for the expansion of existing schemes and the introduction of new schemes in nations that did not have them. During the interwar and war years the assets in several of the funded pension schemes were depleted, discrediting the funded pension model. This history, combined with the rapid population growth and rapid increases in wages, contributed to the increased popularity of the payas-you-go defined benefit (PAYGO DB) model during the post-war period. While many schemes emphasized the PAYGO DB model, there was much structural variation. Most did build up at least some reserves making them partially funded as opposed to pure PAYGO schemes. The defined benefit was sometimes means-tested, sometimes earnings-related, and sometimes a flat benefit (World Bank, 1994). In several developing countries in Africa and Asia, provident funds (publiclymanaged funded defined contribution schemes) were introduced While the trend in recent years has been away from the PAYGO DB model, the preponderance of public old-age pension schemes around the world today are based, at least in part, on this model Most nations that have relatively mature PAYGO DB schemes in place are facing current or projected problems financing these programs due to a combination of factors such as: program maturation, population aging, prior promises of overly generous benefits, changes in employment patterns (e.g., the trend toward early retirement), and in some cases fiscal problems associated with the transition to a market economy. The shift from a PAYGO DB scheme to a FDC scheme or to a multi-pillar scheme that includes a FDC component has come to be viewed by many public pension experts as a potential long-term solution to the projected financing problems most schemes face However, during the mid-1990s yet another model emerged that is based on the concept of notional accounts. Pension schemes (or pillars in multi-pillar schemes) based on the notional defined contribution (NDC) model have been introduced in Italy (1995), Latvia (1996), the Kyrgyz Republic (1997), Sweden (1999), Poland (1999), and Mongolia (2000. This paper seeks to describe and assess the strengths of this model relative to the PAYGO DB and FDC alternatives. It seeks to assess the utility of this new model as a potential pillar in a multi-pillar (alternatively referred to as a tier in a multi-tier) old-age security scheme. 3 THE STRUCTURE OF NDC SCHEMES The NDC model has some characteristics associated with PAYGO DB models and some associated with FDC schemes. The NDC model (also referred to as the PAYGO DC model) can be viewed as a variant of the PAYGO DB model with a number of provisions designed to assure a much closer link between contributions and benefits Contributions to NDC accounts are notional, not capitalized. If they were capitalized, appreciation from year to year would be based on trends in capital markets, but this is not the case with NDC accounts. The indexing procedure used varies from one country to another, but in all cases it is linked to the change in wages, the wage sum (the total wage base subject to the payroll tax, a measure based on trends in both wage levels and the number of workers contributing), or to GDP growth. In Sweden and Mongolia, the indexing is based on the change in wage levels A distinctive component of NDC schemes is the way in which the starting pension benefit is calculated when the notional assets are annuitized. While there are differences from one country to another, they all incorporate some mechanism to adjust for changes in life expectancy that take place over time. In Sweden, for example, the formula is based on age-specific unisex life expectancies at the time when the worker's cohort reaches age 65 . The procedure used is very similar in the other countries with NDC schemes, although there is some variation with respect to how often estimates of life expectancy are updated ). Retirement pensions need to be indexed in some way to keep up with changes in the standard of living (wage increases) or at least inflation (price increases); however, the Kyrgyz Republic has not yet passed legislation to do this (P. Castel, personal communication, July 5, 2002). In Sweden the formula takes into consideration both inflation and changes in the rate of economic growth. If the rate of economic growth is above 1.6%, the annual adjustment exceeds the rate of inflation; if the rate of economic growth is below 1.6%, it falls below the rate of inflation In some countries, such as Sweden and Poland, the employer and the employee contribute equal shares to the payroll tax; but in Italy, Latvia, the Kyrgyz Republic, and Mongolia, the employer share is larger. In most countries, some of the payroll tax is credited to the NDC account and some is not. In several of these countries (see below) most of the contributions not credited to the NDC accounts are diverted to individual FDC accounts. In Mongolia, that part of the payroll tax not credited to the NDC accounts is used to pay administrative costs as well as to fund disability benefits, survivor benefits, and the minimum pension . In Sweden, the payroll tax is 18.5% --evenly split between the employer and the employee. Of this amount, 16% is credited to the worker's NDC account . In Poland, the total payroll tax is much higher (32.52%), but the amount credited to the NDC account is actually lower than in Sweden (12.22%) . In Italy, the total payroll tax is 32.8% with 23.91% contributed by the employer and 8.89% by the employee with the full amount credited to the notional account . In Latvia, the payroll tax is 23.58% for the employer and 9% for the employee with 20% credited to the NDC account . In the Kyrgyz Republic, the payroll tax is 29% with the employer contributing 24% and the employee 5% (P. Castel, personal communication, July 5, 2002). In Mongolia, the employer contribution is 13.5% and the employee contribution is 5.5% with 15 percentage points being credited to the NDC account . In each country that has introduced an NDC scheme, there are provisio ns for making the transition from the old PAYGO defined benefit scheme to the new scheme. When an NDC scheme replaces a PAYGO DB scheme, some way must be found to compensate those who had made contributions for many years to the PAYGO DB scheme; these transition procedures provide a mechanism to do so. In most countries, the transition will be quite gradual with pension benefits for many years being based on a complex weighting of entitlements earned in connection with both schemes; that is, pension benefits are derived in part on the basis of the old scheme and in part based on the new scheme as a function of wage or contribution history, year of birth, and the number of years of contribution to the respective schemes at the time of retirement. Since the first of these NDC schemes was introduced in 1995, it will be several decades before all of those retiring will derive all of their pension benefits on the basis of these new schemes. In Latvia, there was a translation of acquired pension rights under the prior scheme into a notional credit which was then added to the person's NDC account . In Mongolia, workers born before 1960 will remain with the prior scheme 7 . In the Kyrgyz Republic, pensions will be based on a weighted sum with one component based on the five best years prior to 1996 and a second based on the notional capital accumulated in the NDC account since 1996. In the Kyrgyz Republic, there is also a third component, the base pension (see below). The transition process in Poland, Italy and Sweden is more complicated; in all three countries, workers are broken into three categories based on year of birth. In Poland, workers in a specified age range (that excludes the youngest and the oldest workers) have the option of participating in the new scheme; those born after 1968 must participate in the new NDC scheme, those born between 1949 and 1968 have the option of being covered by the new NDC scheme alone or in combination with a FDC scheme, and those born before 1949 must remain with the scheme in place before the NDC scheme was introduced (Chlon-Dominczak, 2002). In Italy, only those who entered the work force in 1996 and after must participate in the NDC scheme and those with 18 years or more of coverage under the old scheme are required to remain with that scheme. This leaves a group with fewer than 18 years of coverage under the old scheme as of 1996 who had the option to switch to the new scheme. For those who did switch, their pension will be based on a weighted combination of pension credits linked to each of the two schemes (Hamann, 1997). In Sweden, those born after 1954 will have a pension based entirely on the new scheme. Those born before 1937 will have a pension based entirely on the old system. And those born between 1938 and 1953 will receive a pension based on a weighted average of contributions to the old and the new schemes NDC schemes are designed to reward those who remain in the labor force for more years and to penalize those who retire early. In Italy, for example, the projected replacement rate (62%) is basically the same under the new NDC scheme (the Dini reforms) as under the prior scheme (the Amato reforms) at age 62 (with 37 years of contribution); however, the replacement rate will increase for those who remain at work until age 65 (with 40 years of contribution) (74% under the Dini reforms vs. 66% under the Amato reforms). Similarly, for those who retire early (at age 57 with 35 years of 9 contribution) the replacement rate is lower under the new scheme (50% vs. 59%) . While the NDC schemes are designed to keep a balance between contributions and pension benefits paid out, it cannot be assumed that there will be a balance under all conceivable demographic and economic scenarios. For this reason, some schemes, such as those in Sweden and Poland, build in provisions for a reserve or buffer fund. Sweden is able to draw on pre-existing funded accounts that were in place for several decades prior to the introduction of the NDC system. In Poland, part of the payroll tax is set aside in a special reserve fund for just such contingencies . In Sweden, there is also a provision to abandon the standard mechanism of indexing of notional assets if the implicit pension debt moves above a specified level . There are other components of the pension schemes in nations with NDC accounts that are not actually part of the NDC tier itself, but are very relevant to the impact of the overall scheme. Although the NDC model is non-redistributional, redistribution can be accommodated within the scheme, by transfer from general government revenues. For instance, most countries have a minimum pension. These pensions, generally paid for out of general government revenues, are designed to ensure that low-wage workers who have contributed for a specified number of years will be guaranteed at least a basic minimum pension benefit. They are not considered social assistance. The minimum pensions are typically means tested, but only for other sources of pension income. The minimum pension is not part of the NDC, but many pension policy experts argue that it is needed due to the lack of any income redistribution in the NDC component itself. Depending on how generous the minimum pension is, the net 10 results when combined with the NDC component can be more redistributive, less redistributive, or about as redistributive as the DB scheme being replaced. The Kyrgyz Republic and Italy are partial exceptions; neither have minimum pensions per se. In the Kyrgyz Republic, there is a base pension, a flat amount added to the earning-related pension for each year of service-between 5 and 20 years for women and between 5 and 25 for men ). This pension was set in 1997 to the minimum wage (about 29% of the average wage). The plan is not to increase it until it declines to 12% of the average wage. The pension is available at age 60 for men and 55 for women, but these ages will gradually increase to 63 for men and 58 for women by 2006 (P. Castel, personal communication, July 5, 2002. In Italy, under the Dini reforms (the NDC scheme) there is no minimum pension, but there is a means-tested social assistance pension for those age 65 and older ). All other countries with NDC schemes have minimum pensions. The minimum pension is typically available to workers (or, in the case of Sweden, residents) when they reach a specified eligibility age. Many countries require payroll tax contributions for a specified number of years. The eligibility age ranges from a low of 55 (for women in the Kyrgyz Republic) to a high of 65 in Sweden (for both genders). In some countries, the eligibility age is the same for both genders (e.g., Latvia at age 60, Sweden at age 65), but in others it differs by gender (e.g., 60 for women and 65 for men in Poland, 55 for women and 60 for men in Mongolia and the Kyrgyz Republic). In some countries, such as Sweden, the minimum pension is quite generous, but in others it tends to be very modest Another way most NDC schemes accommodate redistribution is by providing notional credit for certain categories of people who are out of the paid labor force or not subject to payroll taxes for certain reasons. Typically, the government makes contributions out of general revenues to the fund used to pay pensions and at the same time a corresponding amount of notional credit is recorded in the NDC account for the covered person. The contingencies covered vary from one country to another, but most include credit to a parent (typically the mother) who takes time off from work to care for a young child. The amount of time allowed varies from country to country, as does the level of compensation provided. In Poland, the government is also negotiating a program of "bridging pensions" with the labor unions, which would allow workers in physically demanding occupations such as mining to retire early and still receive benefits before they reach legal retirement age Italy has a rather generous program which offers benefits for childcare until the child reaches age six, as well as benefits for care of the elderly . Latvia offers a credit for a maximum of 1.5 years per child (for up to two children) . Latvia also contributes for sickness and disability leave as well for military service. For covered noncontributory periods, transfers are made from general 12 government revenues to the pension fund at the minimum wage level and a corresponding amount is credited to the person's NDC account . In Sweden, there are contributions for those enrolled in higher education and those in the military as well as those who are eligible for disability or unemployment benefits. In addition, there is coverage for up to four years of parental leave per child with credit being based on the most favorable of several options: 75% of the average earnings for all covered persons; the individual's own earnings the year before the child's birth; or a fixed amount indexed to the covered wage per capita . Poland contributes benefits to personal accounts for unemployment, childcare, and care for the disabled The government contribution to the pension system and to the parent's NDC account is at the level of the minimum wage Some countries with NDC schemes also have or plan to have a FDC scheme as well. In Sweden, in addition to the 16% of the payroll tax credited to its NDC system (and used to pay benefits to current pension recipients), another 2.5% is allocated to funded individual retirement savings accounts. In Poland the split is more even, with 12.22% credited to the NDC component and 7.3% to the funded individual accounts . In Italy and the Kyrgyz Republic, there are no mandatory funded individual accounts. In Mongolia, there are plans to add a funded component in 2005 starting at 3% of the payroll tax and then gradually increasing to 7.5% by 2020 (Bender 13 & MacArthur, 2000). While Italy does not have mandatory funded accounts, it does have tax incentives for voluntary individual accounts With such schemes, part of the worker's pension is subject to the risk associated with trends in wage levels and part is subject to the risks of fluctuations in stock and bond markets. For a table providing a comprehensive overview of the major features of the NDC schemes in Italy, the Kyrgyz Republic, Latvia, Mongolia, Poland, and Sweden, see Appendix 1. THE PROS AND CONS OF NDC SCHEMES When assessing the strengths and limitations of the NDC approach, ideology and political philosophy do matter; what is a strength to one analyst is often a limitation to another. For example, many conservative analysts will see the lack of redistribution as a strength of the NDC model, whereas many liberal analysts will view this attribute as a weakness. For this reason, an effort will be made to qualify many of the arguments presented in this section. Since the se NDC schemes are all new, there is currently little hard evidence as to how the y work in actual practice. Some of the problems that will eventually emerge have not had the time do so, and some of the problems that have already emerged will turn out to be minor and temporary. It is also important to keep in 14 mind that some of the purported strengths (and limitations) are relative to the FDC approach while others are relative to the traditional PAYGO DB approach. One strength of the NDC model relative to the PAYGO DB model is that it will, at least in the long run, help keep pension benefits in balance with available payroll tax revenues. Policy makers in Sweden, for example, believe that it will be possible to keep the payroll tax at the current 18.5% level. It may turn out that it is not possible to do so, but this is the goal and expectation of current Swedish pension policy experts. While the size of the specified payroll tax varies from one country to another, in most cases the expectation is that it will not increase in the decades ahead. Italy is an exception. Italian pension policy experts recognize that in order to maintain the current level of benefits, a 5% increase in taxes per generation will be required (although this is an improvement from the old PAYGO DB system, under which the required increase would have been 9% per generation) The NDC model goes a long way toward dealing with possible labor force and demographic changes and the potential funding consequences of such changes. All NDC schemes are designed to deal with changes in the rate of economic growth and related macroeconomic phenomena, particularly changes in wage levels. All NDC schemes have some provisions designed to deal with population aging (including adjustments in the formula for computing the annuity based on notional assets at retirement), and some have provisions designed to deal with fluctuations in the size of the labor force. For example, the Latvian and Polish schemes (but not the Swedish or Mongolian sche mes) automatically adjust benefits when there are changes (including reductions) in the number of workers paying into the system. 15 What happens when the provisions for keeping the scheme in financial balance do not work, that is, when pension benefits start to exceed contributions? While NDC schemes are designed to make automatic adjustments in pension benefit levels based on demographic changes and fluctuations in the economy, this does not mean it will always be possible to keep the scheme in balance based on these provisions alone. For just such contingencies, Sweden has created a reserve fund that includes assets that could be liquidated in such a contingency. In addition, Sweden has a special "brake" that can be activated if the imbalance gets above a specified level: at that point the formula for adjusting pension benefits each year (a formula that does include a provision for lower benefit increases in the event the economy does not perform well) can be temporarily While a case can be made that the NDC model goes a long way toward creating a balance between payroll tax revenues and pension benefits, it does not by itself offer a solution for those countries that face a serious short-term imbalance Several of these countries have found ways to deal with their short-term pension financing problems via the set of reforms that included the introduction of the NDC scheme, reforms that add up to some combination of tax increases and benefit cuts According to many, a second strength of the NDC model is that it is more transparent than the PAYGO DB alternative. However, there is no consensus on what exactly transparency means, and some analysts argue that the NDC model is in fact less transparent in important ways than the PAYGO DB system. Those who say that NDC 17 accounts are more transparent point out that the worker can at any time check to see how much is in his or her account and knows that the amount is a function of past payroll tax contributions that are being indexed based on a clearly specified formula that will seem fair to most workers. Such transparency is enhanced when, as in Latvia, the government issues a recalculated table of age-specific average life expectancies on a periodic basis that can be used to adjust projections of post retirement annuity (pension) benefits This transparency may increase political support for the program, particularly among more affluent workers, as covered workers can expect to get out in direct proportion to what they put in. In comparison with a PAYGO DB scheme, an NDC scheme has the potential to foster a greater sense of ownership. One reason for this is that a worker covered by an NDC scheme periodically gets a statement from a government agency stating the explicit value of the credit (or notional assets) in the notional account and how much that value has changed in recent months. For similar reasons, it is more likely that the NDC scheme will be viewed as a savings plan than as just another tax Those who argue that the NDC system is less transparent say that, although it is clear how much is credited to a worker's account at any time, the rate of return is less clear than under a PAYGO DB system. The benefit formula under a PAYGO DB system is fairly straightforward, while it is much more difficult for workers to estimate their pension based on an NDC formula, since the amount paid out is dependent on such factors as changes in life expectancy and the basis (average wage, wage sum, or GDP) on which accounts are indexed. Pensions so indexed do tend to be less 18 volatile than those based on equity, which are subject to stock market fluctuations. Although NDC systems might be less transparent in terms of expected benefits in comparison to PAYGO DB systems, they are still more transparent than FDC systems. The system is arguably also more transparent in terms of redistribution . As discussed above, although the NDC system itself is nonredistributional, redistribution can be (and in most cases is) incorporated through such means as childcare credits and minimum pensions. The funds for these no longer come from payroll taxes, but from the government's general revenue . This results in a clear separation between benefits based on contributions and redistribution programs. Redistribution is no longer obscured by complex benefit formulas, as it may be in a PAYGO DB system. There are also analysts who argue that NDC systems are a less politically transparent way of reforming the pension system. Because many of the benefits of an NDC system could be achieved through parametric reforms, some analysts see NDC schemes as a matter of tactical packaging, allowing policymakers to enact what would be otherwise unpopular reforms Many analysts contend that a third strength of the NDC system is that it is likely to lead to an increase in the average age of retirement as it removes the economic disincentives to continued labor force participation implicit in many PAYGO DB schemes (1) the notional assets already in the account will grow for an additional year, In addition, it will not take long for workers to realize that if they retire early, pension benefits will be low and that for each year retirement is delayed benefits will increase substantially