Demographic projections suggest that the aging of the general population in many industrial countries will continue into the foreseeable future, and that in some countries the elderly will become, during the lifetime of today’s young, a large proportion of the total population. Given the existing structure of social support in many of these countries, the elderly are likely to command an increasing share of available resources. Many countries have recognized that the tax rates required to finance social security pensions for the retired will increase substantially, and several countries have begun to face up to the issues raised by this prospect.1

Abstract

Demographic projections suggest that the aging of the general population in many industrial countries will continue into the foreseeable future, and that in some countries the elderly will become, during the lifetime of today’s young, a large proportion of the total population. Given the existing structure of social support in many of these countries, the elderly are likely to command an increasing share of available resources. Many countries have recognized that the tax rates required to finance social security pensions for the retired will increase substantially, and several countries have begun to face up to the issues raised by this prospect.1

The Impact of Demographic Change on Social Security Financing

Demographic projections suggest that the aging of the general population in many industrial countries will continue into the foreseeable future, and that in some countries the elderly will become, during the lifetime of today’s young, a large proportion of the total population. Given the existing structure of social support in many of these countries, the elderly are likely to command an increasing share of available resources. Many countries have recognized that the tax rates required to finance social security pensions for the retired will increase substantially, and several countries have begun to face up to the issues raised by this prospect.1

The purpose of this paper is to examine the relationship between population aging and social security financing in the context of prospective developments in four countries—the Federal Republic of Germany, Japan, the United Kingdom, and the United States. In each of these countries the financing of social security has been widely discussed, and, to varying degrees and in different ways, programs have been, or are in the process of being, restructured so as to cushion against future financial pressures.

Section I of the paper describes projected demographic developments through 2025. In Section II these developments are incorporated into a simple model of a social security program. Although the model requires that actual programs be described only in terms of their essential characteristics, the model nevertheless embodies the principal trade-offs facing policymakers. Section III looks at how the model can be used to assess the implications of reform strategies that do not imply a radical departure from existing financing mechanisms. The section also discusses more radical reform that breaks with existing financing mechanisms. Section IV describes and analyzes recent policy developments in the four countries covered; Section V summarizes the discussion and draws some conclusions.

I. Demographic Trends Through 2025

The analysis that follows is, for the most part, based on a single demographic projection for each of the four countries. These projections share certain characteristics as regards future trends in fertility, mortality, and immigration.2

Demographic Assumptions

In all four countries the fertility rate (the average number of children per woman) has recently been below replacement level, which is around 2.1. The 1980 fertility rates were 1.42 for Germany, 1.75 for Japan, 1.78 for the United Kingdom, and 1.86 for the United States. In Japan and the United Kingdom, the fertility rate is projected to return to replacement level between 2010 and 2025. In the United States, the fertility rate is expected to increase but not quite reach replacement level. The fertility rate in Germany is expected to increase by as much as in the other three countries; however, because the 1980 fertility rate was relatively low, the rate of 1.65 projected for 2025 remains well below replacement level.

Mortality rates are expected to fall so that, between 1980 and 2025, male life expectancy increases by about 5 percent on average, whereas female life expectancy increases by about 3 percent. Improvements in life expectancy are greatest in the United Kingdom and the United States and smallest in Japan, although the range is narrow.

The projections for immigration differ widely because quite different patterns are expected in each country. For Germany, net emigration of about 1 million people before 1990 is anticipated; for the United Kingdom, steadily declining net emigration through 2015 is expected; for Japan, no immigration or emigration is anticipated; and for the United States, the steady net immigration of about 400,000 people a year is expected.

Population Structure

The assumptions about fertility, mortality, and immigration described above, combined with the 1980 population structures, suggest markedly different demographic developments for each of the four countries through 2025. These developments are summarized in Table 1. Total population is projected to fall steadily in Germany and will be 13 percent smaller by 2025 than in 1980. In Japan, total population is projected to increase by 10 percent between 1980 and 2000, then to increase a little further by 2010, and to be at that same level in 2025. Total population is projected to increase steadily in the United Kingdom, although the overall increase will be less than 5 percent. In the United States, total population is projected to grow fairly rapidly, with an overall increase of 33 percent.

Table 1.

Demographic Structure, 1980-2025

article image
Source: Heller, Hemming, and Kohnert (1986).

As a percentage of total population.

Table 1 shows how population aging is projected to occur in each country. Between 1980 and 2000, the increase in the proportion of the population aged 65 and over is minimal in the United Kingdom, modest in Germany and the United States, and much larger in Japan. As a result, by 2000 a considerable part of the difference in age structure observed among the four countries in 1980 will have been eliminated. After 2000, aging in Japan slows down, whereas the populations of Germany and the United States age more rapidly than previously. Aging also accelerates in the United Kingdom, but still remains relatively slow. By 2025, close to 21 percent of the population in these countries, on average, is 65 and over, compared with 17 percent in 2010, 15 percent in 2000, and 13 percent in 1980.

Figure 1 places the aging pattern for 1980-2025 in some historical perspective. It is clear that the population of Japan will continue to age faster than at any time in the recent past, at least up until 2020, The population of the United States will begin to age more rapidly from about 2005, after aging slowly between 1980 and 2005. But population aging in both Germany and the United Kingdom is expected to occur, except for a brief period in Germany (between 2000 and 2005), at a slower rate than at any time between 1950 and 1980.3

Figure 1.
Figure 1.

Percentage of Total Population Aged 65 or Over, Four Industrial Countries, 1950-2025

Citation: IMF Staff Papers 1987, 002; 10.5089/9781451972931.024.A003

Sources: OECD (1980); Heller, Hemming, and Kohnert (1986): and authors’ estimates.

In light of this historical perspective, why is concern about the implications of the aging of the populations of industrial countries emerging so widely and strongly in the 1980s? Everywhere except in Japan, aging is taking place at a slower rate than at any time since World War II. Clearly, a rapidly aging population poses more difficult problems of adjustment than a more slowly aging one, although rapid aging may be quite manageable when there are relatively few elderly to begin with. Once the elderly have become more numerous, however, even fairly slow aging can begin to look unmanageable. The current concern is not about rapid aging—although this is certainly of some immediate importance in Japan—so much as about the problems that may be created by half a century or more of continuous aging. One particular fear emerges from the way in which the financing of social security is affected by aging of the population.

II. Population Aging and Social Security Financing

This paper takes what Aaron (1986) refers to as a “nation-as-a-whole” view of social security financing. Programs are assumed to operate on a pay-as-you-go basis—as for the most part they do in the four countries covered—with all pensions being paid out of current revenue. If a social security financing problem arises, it is likely to take the form of conflict about how current resources are split between workers and pensioners. In particular, demographic developments imply that the financing of social security, as currently constituted, may require tax rates that are unacceptable to the working population.4

Methodology

The relationship between population aging and social security financing will be examined in the context of a simple model of a pay-as-you-go social security program. It is assumed that the average social security pension and the average social security tax payment are a fixed proportion of the same average earnings base, y. If there are R retirement pensioners and the average pension is p = βy, then total pension expenditure is Rβy, where β is the average replacement rate.5 This pension program is financed on a pay-as-you-go basis, so that a social security tax is levied on the working population at a rate α to meet concurrent pension expenditure. If there are W workers, then total social security tax revenue is Wαy, with the pay-as-you-go rule requiring that the tax rate be given by

α=βR/W=β/S,(1)

where S = W/R is the support ratio; that is, the number of workers per pensioner. With the replacement rate β fixed, the social security tax rate is inversely proportional to the support ratio. Thus, if projected demographic shifts can be transformed into changes in the support ratio, the implications of these shifts for social security tax rates follow directly.

Equation (1) forms the basis of the first set of social security tax rate projections—which may be thought of as the “1980 status quo”— reported in this paper. These projections take as their starting point social security pension expenditure in i980 and the pay-as-you-go tax rate needed to finance it, which in turn reflects the demographic structure, age-specific labor force participation rates, and unemployment rates. The social security programs included are the old-age component of the social insurance program in Germany, the old-age component of the full range of public programs in Japan, the flat-rate and earnings-related state pensions in the United Kingdom, and the old-age and survivors’ component (OASI) of the social security program in the United States.6 From published information on these programs, it is reasonably straightforward to calculate the 1980 values of α.7 The variables R and W reflect the age structure of the population, pension age, and the labor market factors noted above. Then, from equation (1), β is derived as a residual. The 1980 values of α, β, and S = W/R are shown in Table 2.8

Table 2.

Social Security Tax Rates, Replacement Rates, and Support Ratios, 1980

article image
Source: Authors’ estimates.

In percent.

In percent. Derived by using equation (1) of the text; that is, β = α S.

Workers per pensioner.

With β fixed, a projection of α requires a projection of the support ratio S. This projection in turn is derived from the demographic projections, under the assumptions that pension age and age-specific labor force participation rates remain at their 1980 values and that unemployment reverts to 1980 rates (or below) by 2000.9 The resultant support ratios are shown in Table 3. The fall in support ratios in each country by 2025 mirrors the increase in the proportion of the total population aged 65 or over shown in Table 1, although the decline differs among countries because pension age and labor force participation rates (and hence retirement age) also differ.10 The most dramatic decline occurs in Japan, where the number of workers per pensioner drops by more than half, and the most modest decline occurs in the United Kingdom, where the support ratio remains between about 2.0 and 3.0 for the entire projection period. This is the range ultimately reached in Japan and the United States, whereas in Germany the support ratio falls below 2.0 between 2000 and 2010 and then falls significantly thereafter.

Table 3.

Support Ratios, 1980-2025

(Workers per pensioner)

article image
Source: Authors’ estimates.

In moving from projections of the support ratio to projections of social security tax rates, one must take into account an additional factor not reflected in equation (1). In each of the four countries discussed in this paper, pensions are related to some measure of a workers’ previous earnings and to the number of working years spent in the program. Such programs have a maturation period during which average replacement rates increase as people approach retirement, having spent an increasing share of their working life in the program. Well-established programs, such as that in Germany, are fully mature. The program in the United States is newer but is nearly mature because of its relatively short qualifying period (ten years). Like the U.S. program, Japan’s major programs are newer, but the Japanese programs are still immature because of their much longer qualifying period (35 years in the case of the Employees’ Pension Plan). In the United Kingdom, the earnings-related program only came into effect in 1978 and will not fully mature until halfway through the twenty-first century. Because of the immaturity of the programs in Japan and the United Kingdom, the average replacement rate β for these countries is allowed to grow steadily between 1980 and 2025. Thus, if the subscript zero indicates 1980 and the subscript t indicates any future year (so that t =45 indicates 2025), then

αt = β0(1+b)t/St,

where b is the rate at which the replacement rate is assumed to grow. Given that β0= α0S0,

α1=α0(1+b)tS0/St,(2)

with b = 0 for Germany and the United States and b > 0 for Japan and the United Kingdom. The growth rates in the latter two countries have been derived from the maturity adjustment implicit in official projections.11 These growth rates are a little less than 0.6 percent a year in each case, and they imply that the average replacement rate will increase from 35.9 percent in 1980 to 46.4 percent by 2025 in Japan, and from 28.7 percent in 1980 to 37.5 percent by 2025 in the United Kingdom.

Results

Projected social security tax rates implied by the changes in support ratios shown in Table 3 are reported in Table 4. In all four countries, the “1980 status quo” projections suggest that social security tax rates will be markedly higher in 2025 than they were in 1980, with much larger increases after 2000 than before. Overall, the tax rate less than doubles in Germany, the United Kingdom, and the United States, whereas it more than triples in Japan. From being the lowest in 1980, the tax rate in Japan increases to nearly 20 percent by 2025, a figure that is exceeded only in Germany, where the tax rate rises to over 25 percent by 2025. In the United Kingdom and the United States, the social security tax rate increases to about 17 percent.

Table 4.

Social Security Tax Rates, 1980-2025: Pensions Fixed Relative to Average Gross Earnings

(In percent)

article image
Source: Authors’ estimates.Note: Rates are derived by using equation (2); that is, αt = α0(l + b)t S0/St with b = 0.00572 for Japan, b = 0.00595 for the United Kingdom, and b = 0 for Germany and the United States.

As with demographic developments, it is interesting to compare prospective changes with past ones. Although a time series of pay-as-you-go tax rates is not readily available, the Organization for Economic Cooperation and Development (OECD (1985)) reports the ratio of social security pension expenditure to gross domestic product (GDP) for the period 1960-80. The projections in Table 4 can be converted to a similar base, which has been done in Figure 2.12 This chart provides little indication that the ratio of social security pension expenditure to GDP will, in general, grow faster after 1980 than before; indeed, in each country future growth is markedly slower over much of the period between 1980 and 2025 than growth before 1980.

Figure 2.
Figure 2.

Ratio of Social Security Pension Expenditure to GDP, Four Industrial Countries, 1960-2025

Citation: IMF Staff Papers 1987, 002; 10.5089/9781451972931.024.A003

Sources: OECD (1985) and authors’ estimates.Note: The 1960-80 data relate to the OECD definition of social expenditure on pensions, which is somewhat broader than the definition of social security pensions used in this paper. The OECD definition includes all benefits paid to the elderly, the disabled, and survivors, as well as government employee pensions.

Pension Increases

The above projections assume that the average pension is fixed relative to average gross earnings. Thus the average pension starts as a specified proportion of average gross earnings and is then increased after retirement in line with the growth in average gross earnings. Only in Germany was such a provision for pension increases after retirement built into legislation in 1980, although pension increases were capped in 1979, 1980, and 1981, with the result that pensions declined in value relative to average gross earnings. In Japan, legislation provides only for post-retirement increases in line with the cost of living, although pensions have been periodically increased to reflect real income growth in the economy as a whole. The United Kingdom and the United States appear to be more firmly committed to, at the most, increasing pensions to reflect price movements. In the United Kingdom, however, pension increases have been more generous in the past, reflecting in part periods with less restrictive legislation and in part government discretion. In the United States, pension increases have also been more generous in the past, although this was due in part to an error in the pension formula, now corrected, that provided for double indexation (relative to price changes) between 1972 and 1977 (see Munnell (1977) for more details).

In terms of a “1980 status quo,” it would appear that fixing pensions relative to earnings is more reasonable in Germany and Japan than in the United Kingdom and the United States. Table 5 illustrates the potential effects that breaking the link between pensions and earnings can have on social security tax rates. If pensions are fixed at their real 1980 levels, other than for increases to reflect the maturing process, then in all countries the social security tax rate will be lower in 2025 than in 1980. Because the programs in each country relate starting pension levels to previous earnings, however, pensions can only be held constant in real terms after retirement, in which case the future tax rate will reflect the demographic and labor market characteristics of the retired population. There is no straightforward way of capturing this type of provision in the model used in this paper. Official U.K. projections, however, suggest that, in the long term, maintaining the real value of pensions after retirement significantly reduces future social security tax rates compared with linking pensions to earnings. In terms of the projections reported in Table 4 for the United Kingdom, if pensions were increased after retirement by 1 percentage point less than the growth rate of earnings (with earnings growing at 8 percent a year in nominal terms), the 2025 social security tax rate would be about 13.5 percent rather than the reported 17.2 percent.13

Table 5.

Social Security Tax Rates, 1980-2025: Pensions Fixed at 1980 Real Levels

(In percent)

article image
Source: Authors’ estimates.Note: Rates are derived from the equation αt = α0[(1 + b)/(1 + g)]tS0/St, where g is the assumed rate of growth of real earnings. This rate of growth has been equated with the rate of growth of productivity. Thus g = 0.02 for Germany, g = 0.03 for Japan, g = 0.014 for the United Kingdom, and g = 0.016 for the United States.

The significant effect on social security tax rates of alternative provisions relating to pension increases after retirement gives rise to the question of whether long-run projections based on pension increases that are linked to average earnings—especially if the legislated provision diverges from actual changes in average real earnings—provide a true guide to future tax rates. The answer depends on whether a deviation from what has happened in the past can be sustained. This paper assumes that if changes in average real earnings are not reflected in pension levels, divergences and mismatches in living standards will result— between workers and pensioners and between older and younger pensioners—that are probably not sustainable in the long run.

Sensitivity to Principal Assumptions

The preceding analysis is based on a single, probably optimistic, demographic projection that corresponds to the baseline projection reported in Heller, Hemming, and Kohnert (1986). That study also made use of a less optimistic “greater-aging” demographic projection, which we have also used here. Under these “pessimistic” demographic assumptions, fertility rates are expected to remain at their 1980 levels except in the United States, where a small decline is assumed. Life expectancy is assumed to increase by a little less than twice as much as in the baseline case; that is, average male life expectancy increases by 9 percent, average female life expectancy by 6 percent. It is also assumed that the number of immigrants to the United States is reduced by 25 percent in the greater-aging case. These revised assumptions yield both lower projections of total population in each country and lower support ratios; the consequences of these lower projections for social security tax rates are indicated in Table 6. By 2025, tax rates will be significantly higher in Germany, Japan, and the United States; in the United Kingdom, the difference between the baseline and greater-aging population projections will be relatively small.

Table 6.

Support Ratios and Social Security Tax Rates, 1980-2025: Pessimistic Demographic Assumptions with Pensions Fixed Relative to Average Cross Earnings

article image
Source: Authors estimates.Note: For these “pessimistic” assumptions, see Section II (under “Sensitivity to Principal Assumptions”).

Workers per pensioner.

In percent. See Table 4 for derivation.

In addition to the demography, the economic environment may turn out somewhat less favorably than allowed for in the projections. Lower productivity growth than expected would have no impact on the projections, however, since pensions are assumed to increase in line with average earnings. Only if the link between pensions and earnings is severed will higher or lower average productivity growth affect the projections. The assumed unemployment rate, in contrast, has a direct effect on the projections because a higher unemployment rate reduces the size of the working population.

III. Social Security Reform

The foregoing analysis is fairly mechanical and makes no attempt to assess whether the social security tax rates generated by the projection exercises are sustainable, from either an economic or a political point of view. The economic arguments—which revolve around the supply-side effects of high tax rates—cannot be resolved. In any event, political considerations are likely to dominate economic considerations, and the consensus emerging among today’s politicians is that future generations of workers will find the currently projected social security tax rates unacceptable. Thus social security reform is again a topical subject, with the focus of attention on how to accommodate the financial pressures associated with population aging.

In the existing pay-as-you-go programs, the range of broad policy options is limited. In terms of the model used above, the trade-off is between social security tax rates and pension levels (as reflected in the average replacement rate) and pension age. Before exploring this tradeoff further, however, we will outline an alternative approach to pension reform in the face of demographic pressure. A pay-as-you-go pension program can be viewed as an implicit intergenerational contract, under whose terms each working generation supports the retired—and expects to be supported in turn by future working generations. Musgrave (1981) has examined how contracts with different terms divide demographic and economic risks between working and retired generations and has assessed the effect of amending the terms of the existing contract in various ways.

Changing the Pension Contract

With respect to demographic change, under pay-as-you-go arrangements the cost of adjustment falls on the working population.14 The rules determining pensions tend to be fixed, and the tax rate is adjusted to guarantee that income matches expenditure. With pensions fixed relative to average gross earnings, population aging results in continuously increasing tax rates, with a corresponding decrease in the relative net incomes of workers. According to Musgrave (1981), this type of contract “contains a built-in potential for collapse” (p. 103), Continuously increasing tax rates without any compensating reduction in relative pension levels may induce a future generation of workers to renege on the contract. An alternative intergenerational contract, where the cost of aging is shared more evenly by working and retired generations, could reduce the likelihood of such an outcome.15

One way to achieve fairer cost sharing is to fix the average pension not as a proportion of average gross earnings, but as a proportion of average net earnings. If the social security tax is the only tax, then average net earnings are given by n = (1 - α*)y, where α* is the social security tax rate. The average pension is then p = β*n = β*(l - α*)y, where β* is the average replacement rate defined relative to net earnings, and total pensions expenditure is Rβ*(1 - α*)y. Social security tax revenue is again Wα*y, with pay-as-you-go financing now requiring that

α*/(1α*)=β*/S.(3)

The social security tax rate α* is inversely related to the support ratio, and average net earnings and the average pension change in the same proportion, reflecting the change in the retention rate (l-α*), as the support ratio worsens. This is the sense in which cost sharing is equitable.16

Taking into account increasing maturity, the future social security tax rate can be written (using the notation employed in Section II) as follows:

αi*/(1αi*)=β0*(1+b)t/St=[α0/(1α0)](1+b)tS0/St,

since β0*=α0S0/(1α0). It then follows from equation (2) that αi*/(1αi*)=αt/(1α0),17 and therefore that

αi*=αt/(1α0+αt),(4)

which is the equation used to derive the social security tax rates reported in Table 7.

Table 7.

Social Security Tax Rates, 1980-2025; Pensions Fixed Relative to Average Net Earnings

(In percent)

article image
Source: Authors’ estimates.Note: Rates are derived by using equation (4); that is, αi*=αt/(1α0+αt).

A comparison of the tax rates in Table 7 and the rates reported in Table 4 reveals that the reduction in social security tax rates implied by the switch in pension contracts taking effect in 1980 is between 2 and 3 percentage points by 2025 in Germany and Japan, and about 1 percentage point by 2025 in the United Kingdom and the United States. If the definition of net earnings were to take into account not only the social security tax but all direct taxes on income—a procedure that can be justified if it is believed that pensioners should share the costs of all government activities, and not just the costs of pension provision—the 2025 tax rate would be a little lower; that is, about 1 percentage point in Japan, about ¾ of 1 percentage point in Germany, and about ¾ of 1 percentage point elsewhere.18 But the effect of the fairer cost-sharing formula remains modest and suggests that, although such a switch may be desirable in itself, it may not shift enough of the burden of population aging to the retired so as to guarantee the survival of the pension contract. More direct intervention is then required.

Pension Levels and Pension Age

In the context of a pay-as-you-go social security system, any attempt to contain the growth of pension expenditure so as to limit the increase in tax rates must focus on pension levels or pension age.

Pension Levels

Pension levels are a function of replacement rates, and the focus of attention will now shift from average to marginal rates (that is, the proportion of earnings, appropriately defined, that are replaced by a pension at the date of retirement) and to the provision for pension increases after retirement. Among the ways in which marginal replacement rates could be changed, two would directly affect social security tax rates: the number of pensionable years of employment could be limited, or accrual rates could be reduced. Alternatively, the pension base could be changed: for example, qualifying earnings could be reduced through the use of floors and ceilings, or, where past earnings were taken into account in determining a pension, intervening inflation and real earnings growth could be either partially compensated for or ignored. The above changes need not be applied uniformly. Rather, replacement rates could be made a decreasing function of earnings, with a view to targeting social security expenditure more directly at pensioners who need income support.

As noted above, altering pension levels by limiting pension increases after retirement has potentially large financial impact. We have argued, however, that pension levels are unlikely, in the long run, to move vastly out of line with earnings; nor is it desirable that they do so. One argument against linking pensions to earnings is that the real incomes of pensioners will fall when real earnings fall. But it seems more reasonable to argue that pensioners should not be afforded more inflation protection than workers themselves can secure. If pensions are to be linked to earnings, however, there is a compelling case for linking them to some measure of net earnings. In response to the financial pressures created by demographic change, there is a strong case for having the fortunes of the retired population linked directly to those of the working population that is paying its pensions. Linking pensions to net earnings, on retirement and thereafter, achieves this objective.

Pension Age

Pension age is different from retirement age in that there is early or late retirement, a difference that is reflected in the participation rates of those in the labor force close to pension age. Pension age and other provisions of the pension program necessarily affect participation rates, as do a range of other factors that could themselves be influenced by policy so as to increase the size of the work force (for example, the availability of child-care facilities). We will focus, however, on the effects of changes in pension age.

Increasing pension age has a direct effect on the support ratio— because it increases the number of workers and reduces the number of pensioners—and on social security tax rates. In this paper the effects of two alternative pension-age scenarios are considered. In the first scenario, pension age is increased by one year in 2000 and by one year in 2010, with adjustments to labor force participation rates close to pension age made to reflect the implied disincentive for a worker to retire.19 The impact of this change is shown in Table 8. In the second scenario, pension age is assumed to increase by one year every five years, starting in 1995 and ending in 2015. The impact of this change is shown in Table 9.

Table 8.

Support Ratios and Social Security Tax Rates, 1980-2025: Phased Increase in Pension Age of Two Years with Pensions Fixed Relative to Average Gross Earnings

article image
Source: Authors estimates.Note: Pension age is assumed to increase by one year in 2000 and by one year in 2010.

Workers per pensioner.

In percent. See Table 4 for derivation.

Table 9.

Social Security Tax Rates, 1980-2025: Phased Increase in Pension Age of Five Years with Pensions Fixed Relative to Average Gross Earnings

article image
Source: Author’s estimates.Note: Pension age is assumed to increase by one year every five years, starting in 1995 and ending in 2015.

Workers per pensioner.

In percent. See Table 4 for derivation.

To compare these alternative pension-age scenarios with the 1980 status quo position shown in Table 4 and with the effect of a revised pension contract as shown in Table 7, Figure 3 combines these four separate projections for each of the four countries. As outlined above, the revised pension contract has limited financial effect and, with the exception of Japan, a two-year increase in pension age is of greater benefit. By 2025, social security tax rates would be about 3 percentage points lower in Germany than in the 1980 status quo, about 2 percentage points lower in the United Kingdom, between 1 and 2 percentage points lower in the United States, and about 1 percentage point lower in Japan. A five-year increase in pension age implies significantly lower social security tax rates in all countries, again with the exception of Japan, where the tax rate is reduced by less than 3 percentage points. The reduction is between 3 and 4 percentage points in the United Kingdom and the United States, and nearly 7 percentage points in Germany.

Figure 3.
Figure 3.

Impact of Alternative Policy Measures on Social Security Tax Rates, Four Industrial Countries, 1980-2025

Citation: IMF Staff Papers 1987, 002; 10.5089/9781451972931.024.A003

The Funding Option

The preceding analysis has been set in the framework of a pay-as-you-go social security program, which is the norm throughout the developed world. The alternative—actuarial funding—is common in the private sector and is also used in the public sector (in public enterprises and local government) but has been little used as a means of social security financing. There are three main reasons why pay-as-you-go programs have proved so popular. First, in a funded system, where pensions are paid out of accumulated savings, it takes many years to build up a claim to a reasonable pension. In contrast, under a pay-as-you-go system, a full pension can be paid immediately, with a transitional gain to retirees and some older workers. Second, because pensions are paid out of workers’ earnings, they can be increased to reflect changes in prices or earnings without claiming an increasing share of these earnings. It is difficult— although not impossible—for funded programs to match this provision. Third, as Holzmann (1986) indicates, a pay-as-you-go system appeals to voters, in that a promise can be made to increase future pensions that does not entail any immediate financial costs. Indeed, the cost of such provisions can be delayed many years.

We have shown above, however, that adverse demographic developments can turn against pay-as-you-go programs; moreover, the impact of such demographic change is reinforced by the sort of voter-oriented promises mentioned above. One question that then arises is whether funding could be part of the solution to the problems caused by these adverse developments. By “funding” we mean two things: first, taxes (or, more appropriately, contributions) are determined according to the benefits being accrued; second, pension claims are backed by an accumulated reserve. According to these criteria, neither the Japanese nor the U.S. social security program can be regarded as funded, despite the existence of “funds,” In both countries the fund, which is invested in government securities, is used to smooth social security tax rates. These rates are not determined on an accrual basis, and the social security program will ultimately revert to a full pay-as-you-go basis.

Pay-as-you-go and funding are two different ways of establishing a claim to future resources. The former does so through an inter-generational contract the terms of which dictate that each working generation will transfer part of its income to the retired, whereas the latter permits individuals to transfer resources from their working years to their retirement years, A switch from pay-as-you-go to funding does not affect the resources transferred to the retired unless, of course, funding is viewed as implying a benefit structure radically different from that associated with a pay-as-you-go program.20 The financing method used, however, may affect the ease with which a specific pension claim can be met, since the method of financing can affect capital accumulation and, as a result, the resources available to meet future pension payments.

If the government were to fund social security by building up a reserve of government securities, the result would be the effective equivalent of a pay-as-you-go program, and aggregate saving would be unaffected by the choice between the two financing mechanisms. Only if there were perceived to be a difference between pay-as-you-go and this form of funding would aggregate saving be affected. Alternatively, the government could accumulate private sector assets, although the extent to which this action contributes to an increase in the capital stock would depend on the reactions of private investors and the corporate sector to the resultant change in the structure of asset demand and supply.21 Clearly, the effects of alternative social security financing methods on aggregate saving and capital accumulation cannot be determined a priori. Moreover, where aspects of these relationships have been subjected to empirical investigation, few clear and uncontroversial conclusions have been produced,22 Thus, it cannot be concluded that there are strong grounds for substituting funding for pay-as-you-go because of the beneficial effect of the former on the capital stock; even if there were, other fiscal and monetary policy instruments probably provide the government with more direct means of influencing both public and private saving. Furthermore, any shift from pay-as-you-go to funding would involve transitional losses, since at least one generation has to pay for the pensions of a previous generation (under pay-as-you-go) while providing for its own pensions (under funding).

Although funding offers the private sector several advantages—the most important are tax concessions and a measure of financial security for program members—its signaling function is of clear value to both the private and public sectors. The accrual basis requires that taxes (or contribution rates) reflect the liabilities that are being incurred. This is not a feature of the pay-as-you-go system, although such a link could be established without resorting to funding; rather than building up a fund, government could levy higher taxes than under pay-as-you-go and reduce current borrowing. The result would be that, if compensating deficits did not build up on other programs, increases in future tax rates would be correspondingly contained. This principle underlies the notional funding practice common in the public sector in the United Kingdom (see Kay (1983)). It is also partially reflected in the current trust fund arrangement in the United States, in that between 1987 and about 2025 workers will be paying more than is necessary under pay-as-you-go while, for a number of years, future generations will be paying less. In the interests of equity between generations, however, the taxes paid and pensions received by each generation should ideally bear a fixed relationship to that generation’s earnings (Aaron (1985)).23 In effect, and in contrast to current arrangements in the United States and elsewhere, the rate of return successive generations receive under the social security program would be brought into line.

IV. Social Security Policy

The preceding analysis makes it clear that pay-as-you-go social security programs, as currently constituted, will come under severe pressure from the forces of demographic change. Many countries have recognized this, including the four examined in this paper. But countries have also recognized that pay-as-you-go financing has limited potential for flexibility in response to demographic change. As reflected in the model used in this paper—the issue of the appropriate basis for pension increases aside—the social security tax rate, the replacement rate, and pension age are the only direct policy instruments. Expenditure control requires either lower replacement rates, a higher pension age, or some combination of the two. If social security expenditure does not fully adjust, then tax rates must inevitably increase.

Recent Policy Responses

This trade-off is exemplified in the policy responses that have occurred in Germany. Japan, the United Kingdom, and the United States.

Germany

The early 1970s was a period of rapid expansion of the social security system in Germany that was marked by extensions in coverage to groups outside the work force and by the introduction of generous early retirement provisions that brought pension age and effective retirement age down to 63 (and later to 60 for unemployed men). During the period social security expenditure as a percentage of GDP increased rapidly, reflecting in part this expansion, and in part demographic developments and real wage growth, with real growth in wages affecting both starting pension levels and pension increases that were linked to increases in average earnings (see Figure 2).

As we pointed out earlier, by the beginning of the 1980s Germany was relying on ad hoc post-retirement pension increases in place of the specific indexation mechanism previously used—increasing pensions in line with average gross earnings (averaged over the previous three years)—to control the growth of social security pension expenditure in response to the pressures of an aging population, increasing unemployment, and a falling retirement age (itself largely a response to increasing unemployment). The capping of pension increases during 1979-81 led to a drop in pensions (not only relative to average earnings, but also in real terms; see Cumming (1983)) and to a decline in social security expenditure as a percentage of GDP (see Figure 2). After 1981, pensions were again increased in line with average gross earnings, but this practice was soon abandoned with the onset of recession, and ad hoc procedures (in particular, delayed increases) were reinstated. At the end of 1984, however, these procedures were replaced by a formal arrangement when the authorities decided, in a supplementary law to the 1985 budget, that, in principle, pensions would be adjusted in line with average net earnings.24

Japan

The social security system in Japan underwent rapid expansion throughout the 1970s as coverage grew, pension levels improved as the programs matured, and the population aged. Social security expenditure as a percentage of GDP increased correspondingly (see Figure 2), The prospect’ of still larger expenditure increases after 1980 made social security reform a central issue in Japan.

The 1985 Pensions Reform Act represents a comprehensive plan to restructure the Japanese social security program. There are two principal elements to the reform. The first is the integration of the two largest public programs, the National Pension Plan and the Employees’ Pension Plan. Because the first of these pays flat-rate pensions and the second pays earnings-related pensions, the integrated program will pay a combination of the two. There will also be some extensions of coverage, an aspect of the plan that brings about a significant rationalization of the existing system. The second element of the reform attempts to tackle the financial difficulties created by rapid population aging.

As pointed out earlier, the Japanese social security program is still immature. The integrated system, however, will cap earnings-related benefits at their current levels in real terms for single-earner couples and reduce them for single people or two-earner couples, effectively eliminating any further maturation of the program. The average new pensioner has now been in the program about 32 years, compared with the maximum of 40 years. The long-run average replacement rate has therefore been reduced by about 20 percent. Significant increases in contributions are also proposed, some to take effect immediately and some in the future. Although the growth in benefits will be moderated, by 2025 social security tax rates will still be more than twice as high as they are now. Thus the government has also indicated that the pension age may have to be increased at some time in the future and that the basis for pension increases may have to be changed.

The United Kingdom

Social security expenditure as a percentage of GDP grew slowly but steadily in the United Kingdom between 1960 and 1980 and is expected to continue, although at a slightly slower rate, through 2000 (see Figure 2). Thereafter, as demographic pressures mount and the costs of the earnings-related component of the social security program—introduced in 1978—build up, the ratio of social security expenditure to GDP will increase more rapidly, although not as fast as in Japan and Germany.

After only a few years of operation the new earnings-related program was being criticized—in particular, the detailed workings of the scheme, its long-run costs, and the structure of pension provision as a whole that it implied (see, for example, Hemming and Kay (1982)). The government, sharing misgivings about the scheme (its costs in particular), developed a reform strategy.

The government’s initial proposal to counter cost increases was to phase out earnings-related pensions and to make occupational pension coverage mandatory. In response to criticism of this proposal, however, the government decided to retain the earnings-related pension scheme but to eliminate some of its more costly and less essential provisions.25 These include switching from the highest-earning 20 years to a lifetime basis for pension determination, changing the replacement rate from 25 percent to 20 percent, shifting a larger part of the responsibility for pension increases for employees who are not full members of the scheme (by virtue of the quality of their private pension) to the private sector, and reducing widows’ pensions. The pension changes are to be phased in beginning in 1988, In the long run (by about 2033 in official projections), and without reform, earnings-related pensions would have accounted for about half the total costs of social security pension provision; the reforms described above will reduce these costs by about half. Rather than making occupational pension schemes mandatory, the government is now going to provide incentives to encourage the growth of this sector.

The United States

The future growth of social security expenditure relative to GDP in the United States, with an unreformed social security program, would also be modest (see Figure 2). But social security reform in the United States has proceeded faster than in any of the other countries analyzed in this paper; the program was substantially reformed in 1983, with the result that future expenditure growth will be slower than under the 1980 status quo. The major changes can be divided into those of a short-term nature and those of a long-term nature. For the short term, it was decided to increase the social security tax rate, to delay annual benefit increases by six months, to bring new federal employees and all employees of nonprofit organizations into the program, and to tax half of the benefits received by high-income individuals. For the long term, an increase in pension age—from 65 to 67—is to be phased in between 2000 and 2027, while early retirement benefits are to be lowered, and delayed retirement credits increased, so as to promote later retirement.26

Assessment

Section III described the alternative approaches that might be taken to social security reform within a pay-as-you-go framework; it is clear that the four countries examined in this paper have followed different courses. In Germany, the basis for pension increases has been switched from gross to net earnings; in Japan, replacement rates are to be held constant or reduced, and it looks as though social security tax rates will be deliberately allowed to increase markedly, at least at present; in the United Kingdom, replacement rates will rise less than initially envisaged; and in the United States future increases in social security tax rates are to be limited by an increase in the retirement age.

Table 10 attempts to capture the financial impact of the reforms described above. The estimates shown relate, for the most part, to the principal features of reform packages as they are best reflected in the projection model described earlier. Thus, in the case of Germany, the 2025 social security tax rate—with recent reforms—of 23.1 percent allows for pension increases related to changes in earnings net of social security taxes. If all direct taxes on income are taken into account, the corresponding estimate would be 22.4 percent.27 In Japan the replacement rate has been frozen at its 1985 level. For the United Kingdom, the full impact of the 1985 reforms on social security expenditure is reflected in the reported estimate; this estimate, not surprisingly, coincides with the official estimate adjusted to the same basis. Recall that the reported estimate assumes that pension increases are linked to average earnings; if pensions are increased in line with prices, the 2025 U.K. social security tax rate is close to that prevailing in 1980. In the United States, the reported tax rate takes into account a two-year increase in pension age (up to 2025 rather than 2027). Official projections (United States (1985)) indicate that the pay-as-you-go rate for OAS1 in 2025 will be about 12.5 percent rather than the 15.1 percent reported in Table 10. This difference reflects several factors: the assumed link between pension increases and average earnings as opposed to cost-of-living increases; the impact of the 1983 amendments other than the phased increase in pension age; and differences in modeling (in particular, the fuller analysis of behavioral responses to the social security amendments contained in the official projections).

Table 10.

Social Security Tax Rates, 1980 and 2025, Under 1980 Status Quo and with Recent Reforms

(In percent)

article image
Source: Authors’ estimates. For assumptions underlying the estimates, see Section IV (under “Assessment”).

For the present, the current arrangements in the United States look stable, as do those provisions in the United Kingdom that reflect the 1985 reforms. The arrangements in Japan are also probably sustainable, although with the relatively low pension age there is scope to limit future increases in tax rates by increasing pension age, which is under consideration. The German situation is the most intriguing. Even with pensions linked to net rather than gross earnings, the social security tax rate (or at least the part that pays for pensions) is set to increase to well over 20 percent by 2025. This increase may well combine with other social security taxes and the income tax to produce a total personal tax burden that would be judged unsustainable.28 If this possibility is soon recognized, then recent developments in the other three countries are certainly instructive, since those countries acknowledge what is widely regarded as a crucial requirement of pension reform: the need to announce future changes well in advance.

Major pension reforms cannot be introduced in haste. Many older workers and pensioners make irreversible labor market and savings decisions on the basis of expectations set up by prevailing social security arrangements. To violate these expectations may undermine the contractual basis of social security pension programs and diminish the widespread support that such programs have attracted in the past. But social security pension schemes have weaknesses, one clear example of which is the way they are affected by demographic change. Reform in this area, and in others, is manifestly desirable. Change must be gradual, however, to give people time to adjust. Because demographic pressures will not really begin to be felt until after the turn of the century, there is time for the necessary reforms to be phased in.

V. Summary and Conclusions

Expenditure on social security pensions is expected to grow rapidly between 1987 and the end of the first quarter of the twenty-first century, principally because of population aging. Demographic trends imply a substantial increase in the number of retired people as the baby-boom generation—born between 1945 and 1960—reaches retirement age. This generation is currently in the work force and is sharing the cost of a relatively modest pension burden. For the next 30 years or so, the net effect of having this large birth cohort move through the population age structure will be to contain the growth of social security tax rates. Thereafter, social security taxes will begin to escalate and in some countries may reach a level that, it is widely felt, would be unsupportable.

In this paper we have not attempted to explore the full range of equity and efficiency issues associated with demographic change and social security financing. Although these are important issues, they would take us into the areas of the distributional characteristics of alternative programs, the interaction of the social security system with tax or transfer systems, and the appropriate role of the public and private sectors in pension provision. Our purpose has been more modest: to examine the effect that some modifications to existing social security programs would have on future social security tax rates.

We have focused on four industrial countries in which policy developments are of particular interest (the Federal Republic of Germany, Japan, the United Kingdom, and the United States). Our analysis of each is for the most part based on a single, probably optimistic, demographic projection. Although, by historical standards, populations will not age particularly quickly between 1980 and 2025, the share of the elderly in the population of these countries will rise, on average, from 12.5 to 20.5 percent. It is assumed that social security is financed on a pay-as-you-go basis, with pensions initially fixed relative to average gross earnings. Demographic developments then govern the projections. Under such an assumption, social security tax rates will increase significantly in each country, but more so in Germany and Japan than in the United Kingdom and the United States.

A feature of pay-as-you-go financing, as reflected in these projections, is that the burden of adjustment to adverse demographic developments falls on the working population, whose tax rates adjust to pay for the benefits that the pension programs define. It is this feature that has led some observers to conclude that the implicit pension contract that exists between successive generations contains the seeds of its own destruction. One generation of workers will ultimately walk away from its obligation to provide the preceding generation with pensions. How can such an eventuality be avoided? This paper explores some possible solutions. One involves letting pensions automatically adjust as the population structure changes, in a way that apportions the adjustment costs equally between the working and retired populations. Fixing pensions relative to net earnings achieves this; however, the impact of such a switch on future tax rates is limited. Another possibility is to adjust retirement age, and in each of the four countries a two-year or, more significant, a five-year increase in pension age has a marked effect on tax rates. Within the framework developed in this paper, if expenditure growth is not controlled in one of these ways, then the replacement rate has to be reduced directly. The alternative is to let tax rates rise.

Having presented the basic model and results, we should emphasize that the exercise we have described is wholly mechanical. It does not point to the need for expenditure control, nor does it provide a guide to appropriate policy responses, although it can indicate how effective certain policies might prove. Presenting policy choices quite starkly—as the model does—and then trying to summarize actual policy responses so that they can be analyzed in terms of the model plainly does not do full justice to the reforms undertaken in the countries examined. Even though the model is not all-embracing, it does still embody the trade-offs facing the policymakers charged with controlling the growth of social security expenditure, and therefore the model provides a guide to the implications of choosing between broad alternative strategies. In this context, two concluding comments should be made.

By linking pensions to earnings and by equating increases in earnings and productivity, the possibility that economic growth will provide the means to support the retired has been ruled out. To the extent that pensions are not indexed to earnings but are instead designed to preserve pensioners’ real income levels, economic growth can provide the means to support current pension arrangements. But this possibility does not make reform any less urgent; indeed, it compounds the problems. Our concern in this paper has been the costs (or risks) associated with demographic change. Because existing programs do not have any built-in capability to adjust in response to these costs (and risks), reform is needed. As we emphasized earlier, whatever adjustment is planned, preparations have to be made well in advance so that, if major structural changes are involved, people are given the time to make appropriate behavioral adjustments. Taking account of the costs (and risks) associated with future economic developments reinforces the need for advance preparation.

A final comment should be offered on the pay-as-you-go financing mechanism itself. We again emphasize that the choice between pay-as-you-go financing and funding depends on the respective effects of these arrangements on the real economy, as well as on aspects of monetary and fiscal management. A switch to funding may affect the ability of the economy to support a particular level of pension expenditure. Given the social security benefit structure, however, the switch cannot affect the required level of expenditure and, as a result, the resources transferred to the elderly. But this does not preclude the possibility that introducing an element of funding—either because it will contribute to capital accumulation or, more realistically, because it will endow a pay-as-you-go scheme with greater intergenerational fairness—will help accommodate demographic and economic developments in a way that secures the future of existing social security programs.

REFERENCES

  • Aaron, Henry J., “Social Security: Problems of Maturity,” paper presented at the Fourth Congress of the International Institute of Public Finance, Madrid, August 1985 (Saarbrucken, Germany: forthcoming in congress proceedings).

    • Search Google Scholar
    • Export Citation
  • Aaron, Henry J., “When Is a Burden Not a Burden? The Elderly in America,” The Brookings Review (Washington), Vol. 4 (Summer 1986), pp. 1724.

    • Search Google Scholar
    • Export Citation
  • Aldrich, Jonathan, “The Earnings Replacement of Old-Age Benefits in 12 Countries, 1969-80,” Social Security Bulletin(Washington), Vol. 45 (November 1982), pp. 311.

    • Search Google Scholar
    • Export Citation
  • Cumming, Christine, “Social Security in Germany and the United Kingdom,” Federal Reserve Bank of New York Quarterly Review (New York), Vol. 8 (Spring 1983), pp. 1325.

    • Search Google Scholar
    • Export Citation
  • Germany, Federal Republic of, Haushaltsbegleitgesetz, Sections 79 AVG and 1271 RVD (Bonn, December 22, 1984).

  • Heller, Peter S., Richard Hemming, and Peter W. Kohnert, Aging and Social Expenditure in the Major Industrial Countries, 1980-2025, Occasional Paper 47(Washington: International Monetary Fund, September 1986).

    • Search Google Scholar
    • Export Citation
  • Hemming, Richard, and John A. Kay, “The Costs of the State Earnings-Related Pension Scheme,” Economic Journal(London), Vol. 92 (June 1982), pp. 30019.

    • Search Google Scholar
    • Export Citation
  • Holzmann, Robert, “Pension Reform: Sharing the Burden,” OECD Observer(Paris), No. 138 (01 1986), pp. 310.

  • Kay, John A., “The Finance of Public Sector Pension Schemes,” Panel Paper 20, in Bank of England Panel of Academic Consultants, The Economics of Pension Arrangements(London, March 1983).

    • Search Google Scholar
    • Export Citation
  • Munnell, Alicia H., The Future of Social Security, Brookings Studies in Social Economics (Washington: The Brookings Institution, 1977).

  • Musgrave, Richard A., “A Reappraisal of Social Security Financing,” in SocialSecurity Financing, ed. by Felicity Skidmore(Cambridge, Massachusetts: MIT Press, 1981).

    • Search Google Scholar
    • Export Citation
  • Organization for Economic Cooperation and Development, OECD Demographic Trends, 1950-1990 (Paris, 1980).

  • Organization for Economic Cooperation and Development Social Expenditure, 1960-1990: Problems of Growth and Control (Paris, 1985).

  • Social Security Amendments of 1983 P.L. 98-21 (Washington, April 1983).

  • Takayama, Noriyuki, “Japan,” The World Crisis in Social Security, ed. by Jean-Jacques Rosa (San Francisco: Institute for Contemporary Studies, 1982; Paris: Bonnel, 1982).

    • Search Google Scholar
    • Export Citation
  • Thompson, Lawrence H., “The Social Security Reform Debate,” Journal of Economic Literature(Nashville, Tennessee), Vol. 21 (December 1983), pp. 142567.

    • Search Google Scholar
    • Export Citation
  • United Kingdom, National Insurance Fund: Long-Term Financial Estimates, Report by the Government Actuary on the First Quinquennial Review Under Section 137 of the Social Security Act 1975(London: H.M. Stationery Office, 1982).

    • Search Google Scholar
    • Export Citation
  • United Kingdom, Reform of Social Security: Programme for Action Cmnd. 9691(London: H.M. Stationery Office, 1985).

  • United States, Department of Health and Human Services, Social Security Administration, Board of Trustees of the OASDI Trust Funds, 1985 Annual Report of the Old-Age, Survivors Insurance and Disability Insurance Trust Funds (Washington: Government Printing Office, March 1985).

    • Search Google Scholar
    • Export Citation
*

Mr. Halter, who holds degrees from Stanford and Oxford Universities, works for McKinsey and Company, Inc. This paper was written while he was an intern in the Fiscal Affairs Department of the Fund.

Mr. Hemming, Senior Economist in the Fiscal Affairs Department when this paper was written, is now Senior Economist in the Asian Department of the Fund. He holds degrees from Sussex and Stirling Universities.

An earlier version of this paper was prepared for the Joint Japanese/OECD Conference on Health and Pension Policies in the Context of Demographic Evolution and Economic Constraint, Tokyo, November 25-28, 1985.

1

Throughout this paper, the term “social security” is used to refer only to a pension program, as is the practice in the United States. In many other countries, the term is used to describe a wider range of programs.

2

These projections have been chosen because of their shared characteristics. They correspond to the baseline demographic projection described in Heller, Hemming, and Kohnert (1986), where a fuller description of the underlying assumptions can be found. These assumptions are broadly similar to those used to produce recent official mid-range population projections in the countries concerned.

3

These statements are based on the rate of change of the proportion of the population aged 65 and over, represented by the slopes of the curves in Chart 1.

4

Note, however, that the social security financing issue can be viewed in other ways. For example, it maybe that a relatively large age cohort retires after having saved and added more to productive capacity than smaller cohorts, and so has at least partially financed its apparently unacceptable claim to current resources (see Aaron (1986)).

5

The average replacement rate, which is the ratio of the average pension to average earnings at the same point in time, should be distinguished from the more widely referred to marginal replacement rate, which is the ratio of the full pension to average earnings (or an average of previous earnings) on retirement. From now on, the term “replacement rate” will refer to the average rate.

6

A brief description of these programs is contained in Appendix II to Heller, Hemming, and Kohnert (1986).

7

Social security tax rates are levied on both employees and employers; a refers to the sum of the employees’ and employers’ rates. For Germany and the United Kingdom, where the social security programs are run on a purely pay-as-you-go basis, these rates are the actual tax rates associated with the programs concerned. The U.S. program is backed by a trust fund, but in 1980 the actual tax rate was close to the pay-as-you-go rate. In Japan, the multiplicity of programs with widely ranging characteristics, and the fact that social security is partially funded, make estimation of the pay-as-you-go tax rate difficult. The assessed rate is well below the actual rate, which in 1980 was 10.6 percent for men in the largest program, the Employees’ Pension Plan. In all cases, the pay-as-you-go tax rates have been calculated net of any government subsidy to the social security program, and such subsidy is assumed to remain unchanged in percentage terms over the projection period.

8

When one takes into account differences in coverage, the different bases used to compute taxes and pensions, and government subsidies, the estimated average replacement rates reported in Table 2 are not obviously inconsistent with the marginal replacement rates reported in Aldrich (1982). We do not have sufficient information to make a more careful comparison of the two sets of replacement rates.

9

The assumed unemployment rates from 2000 onward are 3.0 percent in Germany, 2.0 percent in Japan, 6.0 percent in the United Kingdom, and 5.5 percent in the United States.

10

Pension ages are 63 for Germany, 60 for Japan, 65 for the United States, and 65 for men and 60 for women in the United Kingdom. In Japan different schemes have different pension ages. For example, the Employees’ Pension Plan has pension ages of 60 for men and 55 for women, whereas the National Pension Plan has pension ages of 65 for both men and women.

11

For Japan, the official projection relates to only one program, the Employees’ Pension Plan, as reported in Takayama (1982). For the United Kingdom, the official projection is taken from United Kingdom (1982).

12

It has been assumed that real GDP growth reflects changes in employment, as assumed in the support ratio projections, and the following annual rates of growth of average productivity: 2.0 percent for Germany, 3.0 percent for Japan, 1.4 percent for the United States, and 1.6 percent for the United Kingdom (see Heller, Hemming, and Kohnert (1986) for further details and discussion). Note that the OECD definition of pension expenditure is broader than that used in this paper. The two series of expenditure to GDP ratios have been integrated by using the 1980 ratio implied by the social security tax rates reported in Table 2 as a base.

13

Based on the estimates in Table 14 of United Kingdom (1982).

14

Although Musgrave’s analysis is in terms of risk bearing (that is, who pays if demographic developments are less favorable than expected?), this paper focuses on the burden of costs (that is, who pays for given specific demographic developments?). The difference between these two approaches relates only to whether a particular change is anticipated or unanticipated.

15

A similar problem arises when pensions after retirement are increased in line with prices rather than earnings. Adverse demographic developments do not affect real pension levels; therefore, the full cost of adjustment is borne by workers.

16

As well as the costs (risks) associated with demographic developments, those associated with economic developments—for example, an anticipated or unanticipated slowdown in productivity growth—are also of interest. If pensions are fixed relative to either gross or net earnings, then workers and pensioners share economic costs and risks. If pensions are fixed in real terms, however, such costs and risks are borne by workers.

17

For Japan and the United Kingdom it is assumed that the replacement rate defined in terms of average net earnings, β*, increases at the same rate as the replacement rate defined in terms of gross earnings, β; that is, b. Since b reflects the increase in the average length of the working life spent in a program, this assumption seems appropriate.

18

These estimates are based on the assumption that the overall rate of income tax increases only as a result of increasing social security taxes. But other public expenditure programs—in particular health care—are affected by demographic developments, with the result that the financing of nonpension expenditure will also require higher tax rates (see Heller, Hemming, and Kohnert (1986)).

19

These adjustments are fairly arbitrary, involving no more than an extension of the age range over which participation rates immediately prior to the original pension age are applied.

20

For example, under funding it may seem more natural to define the contribution level and to pay pensions on the basis of the fund’s financial performance, rather than to define the benefits as at present. Moreover, in the absence of indexed securities, provision for pension increases after retirement may be removed or at least limited under funding.

21

A question may also arise about the justification for heavy public investment in the private sector, which would certainly rest uneasily with widely held misgivings about public ownership.

22

For a review of this evidence, see Thompson (1983) and the references cited therein.

23

Such an outcome would only emerge under pay-as-you-go if the aggregate earnings of each successive generation were to increase at a constant rate.

24

Based on information reported in Germany (1984). No further details relating to this change are available.

25

For a description of these reforms, see United Kingdom (1985).

26

These reforms are described in Social Security Amendments of 1983 (April 1983).

27

If account were taken of the impact of demographic change on other social programs, the reduction in future tax rates would be larger; see Heller, Hemming, and Kohnert (1986, p. 35).

28

Heller, Hemming, and Kohnert (1986) reported that payroll tax rates in Germany will have to rise by about 15 percentage points between 1980 and 2025 to pay for projected increases in government social expenditure, and that payroll taxes will reach about 43 percent. This figure is nearly double the corresponding social security tax rate. Note also that in Japan payroll tax rates are projected to increase by over 20 percentage points (with social security tax rates projected to increase by about 7 percentage points). In the United Kingdom payroll tax rates are projected to increase by about 6 percentage points, whereas in the United States the projected increase is only 3 percentage points.

IMF Staff papers: Volume 34 No. 3
Author: International Monetary Fund. Research Dept.
  • View in gallery

    Percentage of Total Population Aged 65 or Over, Four Industrial Countries, 1950-2025

  • View in gallery

    Ratio of Social Security Pension Expenditure to GDP, Four Industrial Countries, 1960-2025

  • View in gallery

    Impact of Alternative Policy Measures on Social Security Tax Rates, Four Industrial Countries, 1980-2025