5 Medium-Term Strategies for Long-Term Goals

Keimeir Kaizuka, and Anne Krueger
Published Date:
July 2006
  • ShareShare
Show Summary Details
Henry f. Aaron1

Japan faces four fiscal challenges. Its population is aging. Its economy is stagnant. Its budget is massively in deficit. And the organization of spending and taxing authority between the central and local governments is not well designed to encourage efficient provision of public services.

Among these four challenges, population aging receives the most attention. No one can deny that population aging poses serious challenges. Japan is the world’s most rapidly aging society, and population aging will force difficult trade-offs. Still, this problem is not unique to Japan. The cumulative demographic shock will be only slightly larger in Japan than it will be in several other nations where fertility rates are as low and life expectancy is nearly as high as Japan’s. To be sure, the speed of population aging in Japan is unique, but other nations’ fertility and life expectancy will soon get them to about the same place. The primary demographic distinction between Japan and other nations is that other nations now permit (or cannot prevent!) significant net immigration, and Japan does not yet allow much immigration.

Japan faces the future with two key advantages. First, it starts with a less generous pension system than those of most rich European nations and the OECD average.2 The Japanese government, with prudence and foresight, has already legislated significant reductions in pension obligations and has introduced provisions that automatically correct for certain future imbalances. These measures will eliminate much and, quite possibly, all of the accumulation of any additional shortfall in pensions.

Second, health care spending has been tightly controlled. Although Japan is well into the period of population aging, its health care spending as a share of gross domestic product remains lower than that of any other major industrial nation, except the United Kingdom. Powerful control mechanisms inhibit growth of health care spending. Critics allege that the organization of the Japanese health care system suffers from serious quality shortcomings arising from perverse incentives and insufficient funding. Still, such problems have not prevented Japan from achieving outstanding overall health indicators.

Even with these controls, population aging will cause both pension and health costs to rise sharply, and anticipated advances in medical technology will raise health care costs even more. As a result, further modifications in the financing and organization of both pensions and health care financing and delivery may prove necessary and desirable to help restore and maintain fiscal balance. But the menu of pension adjustments that would help control spending is well understood, and pension changes in Japan need not be nearly as large as those that will be required in many wealthy European nations.

I shall have more to say below on each of these issues, but they are not, in my view, Japan’s most difficult challenges. More serious is the condition of the Japanese economy as it enters the period during which pension and health costs are certain to rise. Prospects for sustained economic recovery remain cloudy. Virtually no one, to my knowledge, anticipates a return to economic growth at rates even remotely approximating the years before 1989. Nor should the failure to return to annual real growth of 6 percent or more be regarded as a failure. As a world leader in many industries, Japan will be able to grow in those fields only as fast as technology advances. But some Japanese sectors remain highly inefficient, in part because of continued regulation and public subsidy.

The Japanese budget deficit is currently the largest among developed OECD nations. While these deficits may be necessary, even desirable, given the sluggish state of the Japanese economy, dealing with them will become of urgent importance when and as the economy recovers. Furthermore, the distribution of spending authority and tax responsibility between the national and local governments stands as a major obstacle to efficient provision of public services and the effective control of public spending. Closing current deficits when economic conditions permit and reforming local public finance are essential precursors to dealing effectively with the fiscal challenges of population aging.


Pension policy has been studied and written about more than almost any other aspect of social policy. Most of the fundamental principles of pay-as-you-go (PAYG) finance are well understood. Still certain fallacies creep up, which may hinder formulation of sound policies. For example, some writers continue to bewail that some future age cohorts will receive pensions worth less than the taxes they have paid. I shall show that this result is unavoidable and also urge that Japan consider seriously introducing additional rules for changing pensions and taxes automatically in response to deviations of economic and demographic trends from assumptions used in drafting legislation.

Some basic principles

Under strict PAYG, current workers must pay a tax rate, t, equal to the ratio of the number of pensioners to the number of active workers, p, multiplied by the ratio of average pensions to average current earnings, r. Thus, if the pension fund is without reserves (hence, no interest income) and no reserves are being accumulated, t = p x r.

This simple relationship means that if p or r increases, t goes up. It also means that whenever per capita earnings grow faster than prices, t is lower in systems where the pension replaces a fixed proportion of retirees’ past earnings than it is in systems that replace the same proportion of active workers’ current earnings. The higher the rate of growth of earnings, the larger the difference. Put another way, the higher real wage growth is under systems that replace a constant fraction of retirees’ former wages, the lower is r. This relationship means that formula shifts along a continuum between price and wage indexing could be used as one way to maintain balance as other variables change.

Table 5.1.Ratio of pension wealth to average earnings
United Kingdom5.56.3
United States5.26.1
OECD average8.710.2
Source: OECD, 2005, Table 7.1, p. 67.

The variable, t, indexes the annual cost of pensions to workers, but it is not the best measure of the pension burden (or of the total fiscal burden) that a given cohort bears or imposes on the rest of the population. The pension burden, B, is given by the difference between the present discounted value of all pensions received by a cohort, Z, and the present value of all taxes that cohort has paid to support past pensions, T. That is, B = Z - T. If B is positive (negative), it means that a given cohort receives more (less) in benefits than it has paid in taxes. It also means that this difference needs to be made up by (accrues to) other cohorts.

The total fiscal burden that a cohort imposes on the rest of the population, B*, includes not only the pension burden, B, but also the difference between the cost of all other government services that cohort consumes (Z’) and all other taxes paid by the cohort (T′), B′.

Given the actual average lifetimes for all members of a given birth cohort, L, and the present value of all income earned by members of that cohort, Y, the average annual potential consumption by each cohort (ignoring inheritances and bequests) is (Y+B + B′)/L. 3 If longevity increases faster than worker productivity, members of a “self-sustaining” cohort (with B* = 0) must work an increasing fraction of their lives or experience a fall in annual consumption. This relationship underlies the automatic adjustment provision recently introduced to the Swedish pension system under which the replacement rate of each cohort will be reset based on estimated life expectancy (Kent Weaver, 2003/04).

Under PAYG pensions, early beneficiaries always receive more in pensions than they have paid for. These cohorts generate net pension burdens for successive generations. This situation can continue for many years, particularly if the pension system is periodically liberalized, as was the Japanese system in the mid-twentieth century. Thus, current retirees and older currently working age cohorts have B > 0. In contrast, younger actively working cohorts and those who will enter the labor force in the future will collectively receive pensions worth less than the taxes they will pay under current law. If benefits are cut further, as seems likely, the gap will be still larger.

Some critics bewail this shift as a sign of intergenerational inequity and seem to want to avoid forcing any cohort to have B < 0. This quest is futile. The pattern—B > 0 for early participants in PAYG pensions and B < 0 for some later cohorts—is inescapable over the long run unless the net return generated by a PAYG pension system indefinitely exceeds the appropriate discount rate, which growth theory indicates is impossible if saving and capital accumulation are approximately efficient. By definition, PAYG pension systems provide positive B to early beneficiaries—if B were not positive for early cohorts, the pension system would be a funded, not a PAYG system. The corollary of this net transfer to early cohorts is that net transfers from later cohorts are inescapable. The cumulative value of B to early cohorts has been called “legacy debt’.

Precisely when the cross-over from positive B to negative B occurs and whether the legacy debt is paid off or simply rolled over depends on political decisions regarding benefit levels (initial benefits and indexation after entitlement), ages of entitlement, the growth rate of earnings, growth of the labor force, and longevity. But the cross-over must occur. To bewail, as some do, the fact that future cohorts will suffer “intergenerational inequity” because they will pay taxes the present discounted value of which will exceed the pension benefits they receive is pointless. Once early cohorts received B > 0, later cohorts are doomed to receive B < 0. The only questions are: “Which ones?” and “How much?”

It is equally incorrect to argue that shifting from PAYG pensions to funded pensions is a way to avoid this intergenerational transfer. PAYG pensions enable early participants to consume more real goods and services than their incomes (excluding inheritances and bequests). These real goods and services must sooner or later come out of the consumption of later cohorts. Asking later cohorts to rely on funded pensions means that, during a transitional period, workers will have to forego more in consumption than would be necessary if PAYG pensions remained in effect. Changing from PAYG to a funded system shifts the generation for which B < 0. The transition can be fast or slow, depending on how sharply consumption is reduced for the transitional cohorts. Increases in saving and a shift to funded pensions may or may not be desirable for any number of reasons, but achieving such increases in saving through pensions does not eliminate the intergenerational transfers. These transfers are inescapable once early participants in PAYG pensions have received benefits greater than taxes paid on their behalf.

The rest of government

In evaluating the burden that a particular age cohort imposes on the rest of the population, it is illegitimate and highly misleading to focus on pensions alone. What counts is B*, not B alone. In considering whether current cohorts are paying their way, the balance between the taxes paid to support pensions and the value of the pensions that people will ultimately receive is only part of the equation. The fiscal balance between payments made to fund health care benefits and the value of health benefits people will ultimately receive is an additional part of the story. So also is the fiscal balance between taxes paid to support the rest of government and the benefits received.

Viewed through this three-part lens, older Japanese cohorts enjoyed positive pension transfers over their lifecycles. They are also enjoying positive transfers through the health care system, as per capita health costs, though low by international standards, have risen rapidly and are financed largely by current taxes that fall mostly on the economically active. Offsetting these gains, probably only in part, older cohorts received fewer public services of other kinds than they paid for. Budgets were regularly balanced, and much of public spending was devoted to restoration of public capital destroyed during World War II, the benefits of which will be spread over several generations. On balance, it seems likely that older cohorts were net beneficiaries, although I am unaware of any study on this subject.

The situation is sharply different for today’s younger working cohorts. They are paying pension taxes (earmarked taxes plus their share of general revenues) the present value of which is greater than that of the pensions they will eventually receive. They benefit from sacrifices of past cohorts made to restore public capital, and they are shifting to future cohorts much of the cost of current public spending.

The recent history of Japanese pension policy can be viewed through the simple arithmetic relations sketched above. For many years p was low because of high birth rates and relatively low life expectancy. As a result, t was low. Governments could, therefore, increase r and index current pensions to average earnings of active workers, a policy that extended to retirees the benefits of the extremely rapid growth of labor productivity. As fertility declined and longevity increased at historically unprecedented levels—as much as five years a decade during the mid to late twentieth century—t rose sharply.

Now, both p and r are rising, and, with them, t. The political analogue of the economist’s law of demand came into play. As the price of pensions rose, the quantity of pensions demanded by legislators—and, presumably, by the people who elect them—fell. The demand reduction took several forms—a reduced pension accrual rate, an increase in the age of initial eligibility, and a shift from wage- to price-indexing of benefits after their initial level was set. As the amount of PAYG pensions was reduced, elected officials responded by increasing the scope of defined contribution pensions. As noted, a shift to defined contribution pensions does not expunge the legacy debt created under the PAYG system, even if the added saving from a shift to defined contribution plans may be desirable on other grounds.

What is to be done?

Japan can respond to future increases in p in one or more of the only three ways available to any nation: lowering replacement rates (r), raising taxes (t), or taking steps to lower p by increasing the employment/ population ratio, whether by increasing female labor force participation or encouraging active workers to defer retirement and claim benefits at a later age. Encouraging workers to retire at a later age reduces pension burdens only if benefits are not actuarially adjusted to maintain the present value of benefits. Reductions in r would lower the relative economic status of the elderly, but may prove necessary for fiscal reasons. Japan has taken some steps to delay entitlement for pensions—that is to offset some of the increase in p from other sources.

As Japanese life expectancy increases from its already world-leading level, measures to lower the ratio of pensioners to active members of the labor force can help to avoid reductions in pensions relative to living standards for retirees or to minimize tax increases, which reduce net compensation of active workers. The largest opportunities for lowering this ratio are to boost female labor force participation and to extend working lives. If labor force participation rates remain where they were in the year 2000, the Japanese dependency burden (assuming that support costs for children are only 10 percent of those for adults) would increase more than 30 percent between 2005 and 2050. If labor force participation increases sharply, dependency rates would rise less than 10 percent (Burtless, 2004). 4

The increase in labor force participation behind so large a reduction in dependency costs is admittedly extreme. But such projections carry an important message. Policies that promote labor force participation by women, that encourage employers to retain older workers, and that make it financially attractive for workers to remain in the labor force are the most powerful direct instruments for countering the seemingly daunting fiscal consequences of population aging. Furthermore, they do so by reducing or eliminating the need either to raise taxes or to reduce annual pension amounts; in fact, increases in female labor force participation will increase pension amounts per elderly household.

Many policies exist to encourage labor force participation. Some are inexpensive and depend heavily on social attitudes and rules of thumb. As a perhaps eccentric example, the proportion of women to win positions with U.S. symphony orchestras increased greatly when auditions began to be performed behind a curtain that concealed the sex of the applicant. Similar devices for removing or reducing the role of the worker’s sex in hiring, promotion, and wage setting could change the demand-side incentives governing women’s decisions about the acquisition of human capital and about entering and remaining in the labor force. Other policies are more expensive and obvious, including subsidized day care for children available only if the mother is employed more than a certain number of hours, as is commonplace in Sweden, the country with the highest labor force participation of women. Japanese female labor force participation among women aged 30-50 is 20 percentage points lower than it is in Sweden and 10 percentage points lower than that in the United States (International Labor Organization, LABORSTA INTERNET, <>).

Japan already has among the highest rates of labor force participation among older workers and by far the highest participation rate among people over age 65 (see Table 5.2). Still labor force participation among older men declined sharply between 1970 and 2000—by 14 percentage points for men aged 60-64 and by 21 percentage points for men age 65 or older. Although mandatory retirement rules have been relaxed, current practices of large Japanese companies still do not encourage labor market participation at ages at which most workers remain capable and energetic. Although Japanese labor force participation among cohorts over age 55 remains comparatively high, Japan has a considerable reservoir of potential workers (see Table 5.2), whose increased labor force participation could alleviate the prospective increase in Japan’s dependency burden (Burtless, 2004). U.S. research indicates that retirement decisions depend only weakly on modest changes in pension amounts, but are much more sensitive to the age at which pensions are first payable and to loss of pension wealth if retirement is deferred (sometimes called “pension force”). Japanese public pension laws already generously increase pension amounts if workers delay initial claims, thereby maintaining pension wealth if retirement is deferred.

Table 5.2.Labor force participation rates by older people, selected nations, 1950-2000
New Zealand45.925.15.7
United Kingdom51.436.74.4
United States56.041.49.9

Pensions: changing the defaults

The Japanese government adopted major and extremely controversial changes in its pension system in 2004. That legislation sharply increased earmarked taxes.5 It significantly reduced replacement rates on a phased basis. The annual adjustments in currently payable benefits were shifted from wage- to price-indexing, less a flat percent intended to help offset the cost-increasing effects of rising life expectancy. It gradually increased the age at which benefits will be payable.

The 2004 reforms also included automatic adjustment provisions of a limited sort. In particular, the formula used for updating earnings received in the many years before a worker’s average career earnings are computed will be lowered based on the evolution of two demographic factors. The wage index used for adjusting past earnings is to be lowered based on the anticipated decline in the total number of people enrolled in public pension plans. A second adjustment will offset the anticipated increase in program costs resulting from increased longevity. This change will deprive those currently receiving benefits of some or all cost-of-living adjustments for many years. The combined reduction in benefits is estimated to be 0.9 percentage points for each year these adjustments are in effect. These two changes were set to terminate in 2023 and 2025, respectively (Takayama, 2005). These same two factors are to be applied to reduce the price index used for updating currently payable benefits. Since nominal benefits are not to be cut and prices are currently falling, the adjustment to currently payable benefits is not yet operative.

These changes were estimated to restore long-term balance between revenues and benefits. Indeed, given official assumptions, the present value of taxes exceeds the present value of benefits over the long run. Even with these changes, however, the share of total program costs coming from general revenue transfers will have to increase from one-third to about one-half of outlays by 2009. Eventually, the increased wage taxes are projected to be more than sufficient to cover the reduced pension liabilities.6

Whether these changes will turn out to be adequate to maintain long-term balance between revenues and outlays is far from clear. Long-term projections are notoriously sensitive to modest changes in key assumptions. Professor Takayama has calculated the sensitivity of balance to the discount rate. But future balance depends on other economic and demographic factors as well.

Forecasters who are today trying to anticipate economic conditions in 2025 are not materially better equipped to do their job than were forecasters in 1985, who predicted conditions in 2005 based on the impressive economic growth of the 1970s and 1980s. Should birth rates rebound, as official projections assume, and if robust economic growth resumes, large surpluses may eventuate. It might then be possible to lower taxes or raise benefits. But fertility may remain below rates assumed in official projections and productivity growth may remain sluggish. In that event, further increases in earmarked taxes, additional reductions in benefits, or increases in the age of eligibility may be necessary. Policy makers would have to decide how much of each.

Because the controversy surrounding the 2004 legislation was so intense and the potential resentments of younger workers that they are being treated unfairly are so serious, I believe that serious consideration should be given to extending the rules under which changes in pension taxes and benefits are adjusted automatically when prospective pension outlays and revenues become unbalanced for any reason. Furthermore, it is unclear why the legacy debt should be paid by taxes levied in proportion to current wages. I shall take up both issues in turn.

Automatic adjustments

The 2004 reforms limited automatic adjustments to indexation of past wages based on reductions of enrollment in the pension plan and formula adjustments based on anticipated increases in longevity. But imbalances can result from other events, such as deviation of the real discount rate from anticipated levels. The principle of automatic adjustment could be extended so imbalances (or possibly only deficits; surpluses are much less problematic than deficits) from any source would be corrected automatically by a default formula that would take effect unless the Diet acted to modify it.

In broad outline, the government or an independent organization would project long-term pension fund balance, with account taken for general revenue transfers. Long-term balance would include some allowance for reserve accumulation and maintenance. Reserve policy could range from pay-as-you-go (only a contingency reserve) to partial funding (building a reserve to some multiple of annual benefit payments) to full funding. If the financial situation changed, provisions to restore balance would be triggered automatically. Surpluses and deficits should probably be treated asymmetrically. The Diet would, of course, retain the authority to suspend or modify the default adjustments. Legislators could establish as a default any desired mix of changes in earmarked taxes, pension accrual rates, or age of initial entitlement. It would be possible to set as a default that all of any imbalance would be made up entirely by reduction in replacement rates, delay in initial entitlement, increases in earmarked taxes or by some combination of these changes. The adjustments triggered by projected deficits might well differ from those triggered by projected surpluses.

The following example illustrates in general terms how such calculations might be performed. The first step would be computation of the present value of earmarked taxes and of projected pension costs under stated economic assumptions (for example, real wages and yields on reserves), demographic assumptions (for example, birth rates, immigration, and longevity), and behavioral assumptions (for example, labor force participation and marriage patterns), and reserve accumulation policies. If changes in economic, demographic, or behavioral assumptions create imbalances, some combination of changes in earmarked taxes, replacement rates, or age of initial eligibility would be automatically triggered. The character of these “default” adjustments would strongly influence the course of actual adjustments because changing them would require a political majority on behalf of one of a large number of possible alternatives, each of which would appeal to different groups. Accordingly, the selection of the default policies is and should be intensely political.

For the sake of putting a specific option forward for discussion, I suggest that the default consist in equal parts of tax rate and benefit changes. The tax change would be an increase (or decrease) in the earmarked tax rate by an equal number of percentage points at each date. To the extent that the imbalance results from changes in longevity, the benefit changes would consist entirely of an increase (or decrease) in the age of eligibility for pensions, phased in gradually after a delay of, say, five years. For imbalances from all other causes, the benefit changes would consist of changes in replacement rates—again, phased in gradually after a delay. For example if an increase in longevity created a projected deficit with a present value of ¥2X trillion, taxes would be increased enough to raise revenues by ¥X trillion, and age of eligibility for pensions would be increased enough to reduce pensions with a present value of ¥X trillion.7

I believe that some form of automatic adjustment provision, more comprehensive than those in the 2004 legislation, would be highly desirable, given the sharp and relatively sudden changes in economic and demographic prospects Japan has confronted. If these automatic adjustments were made relatively often—possibly more frequently than the five-year intervals between current official pension studies in Japan—each adjustment would be small. Furthermore, the automatic character of the adjustments would reduce the amount of “political capital” that governments would have to invest in maintaining a key foundation of economic security, leaving political energies for tackling other matters.

Legacy debt

Because deceased pensioners, current retirees, and older working cohorts paid taxes in the past worth less than the pensions they have received or will receive, an unfunded liability or legacy debt exists, even though future cohorts are now slated to pay for all of their benefits. This situation prevails in the United States today. All of the projected U.S. social security deficit measured over the next 75 years is accounted for by benefits that were paid in the past or that will be paid to current retirees, the present value of which exceeds the present value of taxes paid on behalf of those beneficiaries. U.S. workers now are paying more than enough taxes to buy the pensions they are slated to receive, given prevailing economic, demographic, and behavioral assumptions. According to the report of the U.S. social security actuaries (Board of Trustees, 2005, p. 60, Table IV.B7), the unfunded liability of all benefits for workers, past, present, and future, is $11.1 trillion. But the unfunded liability for past and current participants is $12.0 trillion. The difference, $0.9 trillion, represents the present value of taxes in excess of benefits that future participants will pay. Younger U.S. cohorts are paying fully for the benefits they will receive under current law.

According to a recent study by Professor Noriyuki Takayama (2005), a similar situation prevails in Japan. He estimates that under the 2004 reforms, liabilities generated by past contributions exceed those contributions by ¥570 trillion, but that liabilities generated by future contributions will fall short of those contributions by ¥230 trillion based on specific demographic, economic, and behavioral assumptions. Both estimates exclude government general revenue transfers.

For future cohorts, this legacy debt is functionally similar to the stock of public debt. In both cases, the debt is a sunk cost resulting from decisions by past governments from which many then-active taxpayers derived no direct benefit.8 As with the public debt, one can pay it off, carry it forward by paying just interest, or defer interest payments, borrow and increase the debt. In any case, it is hard to identify an economic justification for distributing the burden of amortizing this debt differently from payments on the national debt—which is to say, by general revenues. It may well be desirable to earmark a portion of the most widely shared taxes—on income or value added—but it is hard to understand why the burden of paying it off should be shared in proportion to future earnings, while none is distributed in proportion to capital income or consumption.9

Health care

Despite rapid population aging, health care spending as a share of gross domestic product remains lower in Japan than it is in all other major developed nations, excepting Great Britain. Though low, the share of health care spending in GDP rose faster during the 1990s in Japan than it did in all other developed nations, save the United States and Switzerland. The principal reason why the health care share in GDP rose in Japan was sluggish growth of GDP, which averaged only 0.1 percent a year from 1990 through 2001.10 The contrast between low health care spending and outstanding health status indicators in Japan is particularly striking, especially to someone from the United States, where per capita health care spending vastly exceeds that in any other nation, but such health indicators as life expectancy and infant mortality lag those of many other nations. Japan leads the world in longevity, has among the lowest infant mortality rates, and spends remarkably little on health care services.

Various authors have speculated on why health care spending in Japan has remained so low (for example, Fukawa and Izumida, 2004; Fukawa, 2002; Ikegami and Creighton Campbell, 2004,). Fukawa and Izumida (2004) list six factors: review of appropriateness of care before payment is rendered; the relatively paucity of health care providers; low prices paid to health care inputs, including physicians; rigorous price controls through fee schedules, especially for high-technology services; the single-payer system; and the comparatively good health of the elderly.

It is difficult to evaluate the relative importance of these factors. But the combination of low fees for high-technology services, utilization review, and the comparatively low remuneration of physicians, combined with the fact that physicians dispense drugs, as well as prescribe them, are likely to be powerful constraints.11 Pay of Japanese physicians relative to average earnings is comparable to that in the United Kingdom and more than 40 percent lower than that in the United States. Japanese surgery rates are about one-third of those in the United States. U.S. physicians who prescribe surgery increase their incomes, but those who prescribe medication do not. U.S. physicians may be denied payment if they have not secured approval for surgery from managed care plans, but such approval is seldom withheld. In Japan, physicians who prescribe surgery face possible review and are paid according to a strict fee schedule, while those who prescribe medications will be reimbursed more than their cost. Incentives matter. Japan spent nearly 29.5 percent of its health budget on drugs in 1993, compared to 11.3 percent in the United States and 17.1 percent in Germany (Fukawa, 2002, p. 11).

Rapid growth in Japanese health care spending as a share of GDP seems certain to persist. Young age cohorts, who are relatively low consumers of health care, are shrinking and elderly cohorts, who are the high consumers of health care, are growing. Health care spending on the elderly, relative to the non-elderly, is higher than in other developed nations.12 If one uses a crude index of the ratio of health care spending by those over age 65 to those under age 65, the increase in the proportion of the population over age 65 will raise health care spending by about 30 percent from 2000 through 2030, or 2.3 percent of GDP.13 This procedure may overstate the impact of population aging on health care spending. If per capita spending is a function of expected years until death, population aging has a much smaller effect on per capita health care spending than if outlays depend on years since birth.14

One aspect of the Japanese health care system may overstate health care spending on the elderly. Average lengths of stay in Japanese hospitals are longer than in most other nations and several multiples of lengths of stay in U.S. acute care hospitals. An important explanation is the fact that the elderly in the United States who require long-term care are resident in nursing facilities of various kinds, not in hospitals. Those who require little more than help with activities of daily living who are not cared for at home are normally resident in so-called “custodial” facilities; those who require physical or other forms of therapy reside in so-called skilled- or intermediate-care nursing homes. In Japan, far more elderly people live with their children than in the United States. But those who do not are to be found in far greater proportions than in other developed nations in hospitals. Hospitalization of those who do not require the full range of services offered by a hospital almost certainly results in inferior care, as the amenities and ambience of hospitals are poorly suited to the wants and needs of people without acute conditions. For this reason, observers have urged the Japanese to provide more adequate social services and even residential accommodation for the elderly. Care outside hospitals would almost certainly be a less costly way of caring for those elderly who are not acutely ill but are currently resident in hospitals. If care in nursing homes and the community were well designed, it would also raise the quality of treatment of the frail elderly. But total public spending on the frail elderly would almost certainly increase over time because comfortable nursing home accommodations and supportive community services would enable young relatives to shift the burden of caring for the elderly to others.

The impact of technological change on total health care spending (to say nothing of the effect on the value of medical services) is likely to be far larger. Scientific advance promises a steadily lengthening menu of beneficial but costly interventions that no wealthy nation will elect to do without. Indeed a continuation of the gap between annual growth of health care spending and annual growth of GDP were to persist, Japan would look forward to increases in health care spending much larger than any realistic projection of pension burdens (see Table 5.3).

Table 5.3Health care spending as a percentage of GDP, alternative assumptions regarding gap between growth of health care spending and GDP
YearAnnual gap between growth of health care spending and GDP
[Actual 1990-2000]

Based on criticisms that both Japanese and foreign observers have leveled at the Japanese health care system, some considerable increase in health care spending seems warranted. Shortcomings include the following:

  • Hospitals on the average are small. Such units cannot at reasonable cost maintain a diverse range of services and house staff necessary for high-quality care.

  • Current fee schedules discourage services that advanced hospitals provide, including surgery and intensive care (Fukawa, 2002, p. 9). In addition, hospitals have been used not only for acute care, but also for long-term chronic care of people who require services which hospitals are poorly equipped to provide.

  • The fee schedule has been used to control the use of various procedures. Using fee schedules to control cost is a recipe for inefficiency, because the value of procedures is not uniform across patients. Low fees discourage both low- and high-priority uses of a particular technology. Perhaps out of awareness of this inefficiency, Japan has begun to experiment with bundled payments for episodes of illness (Ikegami and Creighton Campbell, 2004, pp. 29-30). This method of expenditure control leaves to providers the decision on what services to provide.

  • The increasing fraction of health care devoted to the elderly has raised problems of fiscal imbalance among prefectures and municipalities. A heavy concentration of elderly enrollees in a community, for example, can create undue fiscal burdens. The solution is to pool across the nation cost variations related to age or to other factors over which individuals have little or no control. One way would be to fund care for the elderly separately. Another would be to provide equalizing grants or direct subsidies.

  • In order to hold down public costs, consideration is being given to permitting health care providers to charge patients fees in addition to the payment made under the official fee schedule. Such a modification would doubtless attract new providers, but it would also lead to differentiation of care based on ability to pay out-of-pocket for it.

  • A variety of measures is under consideration to encourage hospitals and providers to be more solicitous and informative in their dealings with patients.

  • As noted above, improvements in care of the elderly outside hospitals is overdue.

  • These and other changes are more likely to raise than to lower the growth of health care spending. What is apparent beyond dispute, however, is that the key to preventing health care spending from becoming an increasing financial burden is not to be found within the health care system, but must result from a restoration of sustained economic growth. The Japanese health care system already has pushed administrative controls to, and perhaps beyond, the point at which service quality is compromised and access to modern medical technology is curbed.15 Population aging means that an increasing share of the population will be in age brackets that consume large amounts of medical care.16 The advance of medical technology means that the medical profession will increasingly be able to provide services to extend the duration of vigorous and healthy life. The people of no modern nation will—or, in my view, should—accept the denial of such services.

New policies

For 16 years, the Japanese economy has experienced repeated recessions and little growth. Analysts at home and from abroad have struggled to identify the causes of this stagnation. Some have stressed the difficulty of adjusting asset values that became inflated during the bubble economy. Others point to an unwillingness to deregulate or to cease subsidies to unproductive economic sectors. Still others blame large investments in comparatively valueless public works. Finally others emphasize errors of timing of macroeconomic policy, with stimulus started too late and ended too soon.17

Meanwhile, as the stagnation of the 1990s extended into the new millennium, the challenge of population aging loomed larger and moved closer. Until recently, official demographic projections repeatedly and inexplicably embodied the assumption that the drop in birth rates would be only temporary. Every two or three years, new projections were released in an effort to catch up with the reality of declining birth rates. Only recently have official projections incorporated the assumption in population projections that the current completed birth rate per woman of about 1.35 would persist until 2050.18 Even now, however, the possibility of further declines has not been built into official projections.19

That future population aging will pose fiscal challenges is beyond doubt. The current question concerns what policy makers can do now to help prepare for those challenges. Most of the answers to this question are not unique to Japan, but apply in varying measure to other developed nations facing similar long-term fiscal challenges.


The first step in preparing for future fiscal challenges is to restore Japanese economic growth. This statement is so obvious that it is banal; the hard question is what policies would accomplish that objective. Officials and analysts throughout Japan and around the world have been absorbed by this question for a decade and one half.20 The menu of ways to restore growth is familiar. Economic deregulation, which has been underway for more than a decade, should continue and be extended. The government should reduce the level and improve the selection of public works investments, many of which are now judged to have produced few economic benefits and generated sizeable environmental disruption. In this connection, the proposed privatization of Japan Post holds the potential of closing off a protected source of capital for public corporations and ending the diversion of vast sums into low-return investments. This potential will be realized only if the independence from government control of managers of the privatized Japan Post is genuine. The continued reduction in the number and value of non-performing loans would encourage banks to seek to increase lending.

But none of these measures will suffice without vigorous consumer demand that contributes directly to current economic activity and eventually calls forth investment by companies to meet that demand. The Bank of Japan has done a great deal to pump up the money supply and to drive down nominal interest rates. Even at the very low lending rates of recent years, however, real rates remain significantly positive because of domestic deflation. Although the timing of changes in fiscal policy has been criticized, few can criticize the amount of current fiscal stimulus as too small. Indeed, observers of the Japanese economy are torn between the clear current need to sustain economic stimulus and the risks associated with a rapidly growing official debt resulting from large budget deficits. The 2005 Japanese budget is running large deficits of 6-7 percent of GDP, despite stated efforts by successive governments to cut those deficits. The result is rapidly increasing government debt (with gross debt approaching 160 percent of GDP and net debt approaching 100 percent of GDP).

Whether gross or net debt better indexes the severity of Japan’s fiscal imbalance remains in dispute. Conceptually, net debt is the correct measure, but the national government faces a number of serious fiscal risks: that debt service costs could rise; that government assets could turn out to be over valued; and that the national government might be forced to spend considerable sums to honor a portion of its contingent liabilities for local government debt.21 In these circumstances, gross debt may come closer to measuring the actual “net” government obligations.

These eventualities are interrelated. Should they materialize simultaneously, the fiscal situation facing the national government could quickly become critical. These risks are well recognized by foreign and domestic observers (OECD, 2004). After restoring vigorous economic growth, closing this fiscal gap is perhaps the most important single step that Japanese policy makers can take to cope with the fiscal problem of population aging.

Two steps may now be possible that would encourage current consumption; one would promise future reduction of budget deficits. To offset the subsidy to delay consumption that results from falling prices, consideration should be given to enacting now a graduated increase in the value-added tax (VAT), beginning perhaps a year hence and continuing in periodic steps until the rate has been raised enough to reduce the budget deficit by a target amount. For example, the cyclically adjusted primary deficit is currently about 5 percent of GDP. The VAT, now 5 percent, generates revenue equal to approximately 2 percent of GDP. Thus a policy of raising the VAT by 2 percentage points in each of the next three years would encourage consumption over that period, relative to consumption later, and promise to lower the primary deficit by roughly one-half. I present this example for illustrative purposes only—different rate changes in terms of size and timing may be preferable—in order to give specificity to the general approach.22 Were the current budget not so deeply in deficit, it would seem sensible to cut the VAT rate temporarily, followed by steeper or more sustained increases in the VAT rate. The objective, in any event, is to put consumption in general “on sale” because higher prices are promised in the future.

A second way to encourage consumption would be to assure the middle-aged and elderly Japanese that pension benefits for those over a certain age will be maintained at current levels for a specified period. These age cohorts have seen promised benefits substantially lowered. They would be imprudent not to fear further cuts. Such uncertainty is bound to dampen consumption. Ending this uncertainty would have a positive effect, although the size of the effect is bound to be speculative.

Beyond Trinity

For many years, analysts have pointed to an imbalance between the predominance of taxation, which is carried out by the central government, and of expenditures, which are made predominantly at the local level. The central government severely circumscribes the way local governments can spend targeted grants and even limits uses of block grants. Nonetheless, this separation of taxing authority and spending responsibility has been criticized, correctly in my view, as a recipe for fiscal irresponsibility. Local officials spend funds they have not had to raise and lack incentives for frugality and careful management. Pork barrel politics has led to large investments in low-value public works that some have likened to pyramid building.

The Trinity reform, launched in 2002, sought to ameliorate this problem by: (1) cutting earmarked grants to local governments; (2) cutting the local allocation tax, and (3) transferring tax authority to local governments. A sizeable reduction in grants to local governments has already been implemented.

The reform is surely a step in the right direction, but several commentators have observed that it did not go far enough.23 The imbalance between taxing and spending authority persists. While it would be possible to close the overall budget deficit by simply raising taxes or lowering spending—and both steps will surely be part of any program to restore fiscal balance—it will be difficult to sustain balance or to improve the efficiency of public expenditures unless the structure of relations between the national government and local governments is changed more extensively even than is envisaged in the Trinity reform. Trinity has lowered transfers from the center and shifted some tax responsibility to sub-national governments.

Shifting authority to sub-national governments to set rates and bases of more of the taxes that generate the revenues they spend will help. But large differences in fiscal capacity are a reality. So is the concentration in certain areas of high-cost populations, such as the elderly. These facts mean that transfers from the central government to prefectures and municipalities will not, and should not, end, as they are necessary to enable the poorer or more heavily burdened communities to maintain adequate public services. But they do not mean that the national government should continue to underwrite debt service costs of sub-national governments through block grants. And they do not mean that local authorities should be shielded from the consequences of expenditures on poorly selected public works or other inefficient outlays.

Two additional tests will indicate whether fiscal authority has actually been devolved to sub-national governments. The first test is whether grants from the central government to sub-national authorities are capped or effectively limited. The second is whether sub-national governments are required to borrow marginal funds without central government guarantee in the open market from private bond buyers who are free of any requirement to buy such bonds and those same governments are required to pay debt service from locally generated funds. Bond purchases by the Fiscal Investment and Loan Program (FILP), or similar entities, would presumably continue, but in strictly limited amounts.

Requiring local governments to sell bonds without central government guarantee to unconstrained buyers would expose prefectures and municipalities to a form of fiscal discipline that is difficult to duplicate through even the sternest warnings from higher level administrators. Forcing localities to sell bonds beyond some specified amount without implicit national government guarantees in unsheltered markets and requiring them to finance debt service from locally generated funds would dramatically intensify the need for efficiency and parsimony in the management of public spending. Bond ratings would quickly and ruthlessly identify governments that were making bad investments or living beyond their fiscal capacity. Debt finance for public investments may sometimes be warranted, but private bond-buyers who are investing their own funds in bonds that are not guaranteed will scrutinize the quality of such debt more rigorously than will government organizations, such as FILP.

The key test of such bond market reforms and limits on grants would be how the central government responds when the first sub-national government failed to meet its debt service obligations. The most effective way to show local officials the need for careful management would be to allow a fiscally troubled municipality to default on its obligations. So severe a shock may not be economically desirable or politically feasible in the Japanese context. But if municipalities that fail to honor their debt obligations can be forced to bear a significant price, such as a temporary loss of administrative autonomy or a mandatory increase in local tax rates, the message would not be lost on others.


Population aging, which is the source of much of the worry about Japan’s future fiscal problems, has two striking characteristics. First, not much can be done about it, other than by dramatically changing immigration policy, which raises issues that far transcend fiscal policy. Second, rather than being bad news, increased longevity is the culmination of human efforts spanning centuries to extend the duration and quality of life.24 Longevity is not a calamity, but a cause for celebration. Projections of increasing health care spending likewise reflect the twin blessings of increased longevity and the development of new medical diagnostic and therapeutic technologies with enormous potential for improving human welfare. No sensible person would regard a cessation or major slow-down in the advance of medical technology—the cause of most of the projected growth of health care spending—as a desirable way to ameliorate the fiscal challenge of rising health costs.

There are only two humane and practical ways to lessen the real fiscal shock of an increasing dependent population. One way is to facilitate increased labor force participation—by women and by older people of both sexes. The other is to spur rapid increases in productivity. Growthpromoting policies include deregulation, cessation of subsidies to inefficient industries, continued emphasis on investments in human capital, and the termination of physical investments in projects of little economic value. When the Japanese economy resumes economic growth, it will become necessary to raise taxes and cut spending to close budget deficits. Left unattended, fiscal imbalances as large as those that now exist could easily become unmanageable.


The views expressed here are my own and do not necessarily represent those of the trustees, officers, or staff of The Brookings Institution. I thank my colleague, Barry Bosworth, for constructive criticisms and helpful suggestions, and Professor Noriyuki Takayama for correcting my misunderstandings of the operation of the recent pension legislation.

A recent OECD paper reports on replacement rates for earners at various positions on the earnings distribution. Because pensions rise with income, the total pension burden depends more on replacement rates for above average earners than on replacement rates at the average. For retirees at 1.5 times average earnings, gross replacement rates are as follows: France, 51 percent; Germany, 46 percent; Italy, 79 percent; Japan, 44 percent; the Netherlands, 68 percent; Sweden, 65 percent; OECD average, 52 percent. Relative replacement rates for workers with lower or higher earnings differ from those of average earners, but Japan is lower than all of the above nations and the OECD average for higher earners and lower than all but Germany for lower earners (OECD, 2005, p. 49).

In computing T, it is essential to include taxes imposed on the cohort to finance general revenue transfers to the pension plan. The value of Z in Japan, measured at retirement age, like that of its annual pension replacement rates, is higher than that of the United States, but lower than those of major European nations, other than the United Kingdom (see Table 5.1).

Specifically, Burtless computed the increase in the dependency burden if participation rates of people between ages 20 and 24 and over age 60 return to the highest rate observed between 1950 and 2000 and that rates for five-year cohorts of people aged 24-44 go to the higher of the rates currently found in the United States or France.

Concern has been expressed that higher payroll taxes will lead to the loss of as many as 1 million jobs because employers will be loathe to bear the added taxes associated with hiring.

Professor Takayama is concerned about the possibility that overfunding may unduly restrain aggregate demand, that increased wage taxes may cause employers to lower demand for labor, and that lowered take-home pay may induce workers, especially older workers, to curtail labor supply. I believe that these concerns may be misplaced for several reasons. The first, which Professor Takayama notes, is that the assumptions other than interest rates may turn out to be overly optimistic. The currently projected surpluses lie in the future—for now, general revenue transfers remain necessary and are projected to grow for several years. In the end, the anticipated surpluses may or may not be realized. Second, legislatures rarely find it difficult to dispose of surpluses when and if they emerge; eliminating deficits is far more problematic. Third, neither economic theory nor research (admittedly based on U.S., not Japanese, data) supports the contention that higher payroll taxes will cause employers to substitute capital or other factors for labor. Rather, increased wage taxes, whether levied on workers or paid by employers, are thought to be borne by workers in the form of lowered cash wages or other fringe benefits. And although reduced take-home pay, resulting from increased wage taxes, may affect labor supply, the direction of the effect is theoretically indeterminate. Finally, what counts for aggregate demand is the total impact of the public finances, not the partial effect of wage taxes.

The approach sketched here differs from that in Sweden, where all adjustments to funding imbalances take the form of changes in benefits. In practice, the Japanese government (as well as the U.S. government) has reacted to imbalances in the past by a combination of benefit reductions and tax increases.

To distinguish legacy debt from imbalances generated in the future, it is clear that some assumption must be made regarding allowable benefit changes for current retirees and those nearing retirement.

Some U.S. observers have proposed earmarking revenues from the estate tax to help pay for pension benefits.

In 2000, Japanese health care spending totaled 7.6 percent of GDP, compared to 9.3 percent in France, 10.6 percent in Germany, 8.2 percent in Italy, 8.6 percent in the Netherlands, and 8.4 percent in Sweden. Annual growth in health care spending was comparatively rapid, 1.7 percent annually from 1990 to 2001 (Huber and Orosz, 2003).

For example, reimbursement for MRIs in Japan is approximately one-fourth of the average U.S. cost. Despite low remuneration, Japan has more scanners than all of Europe and more on a per capita basis than the United States.

The ratio of health care spending on those age 65 and over to those younger than 65 ranges from 2.8 in Germany, 3.0 in France, 4.0 in the United Kingdom, 4.1 in Australia, 4.4 in the United States, 4.8 in Canada, and 5.3 in Japan. Despite this high ratio, Japan spends so little on health care in general that the spending on the elderly adjusted for the proportion of the population that is elderly is lower in Japan than in all of these countries except the United Kingdom (Reinhardt, 2000).

This projection uses the ratio cited in note 12 above and ILO projections of population growth. It ignores the increase in health care spending arising from the fact that the average age within each of these two broad age ranges will also be increasing and does not take into account any impact of technological change on the age pattern of health care spending.

For a careful examination of this issue in the context of projected increases of U.S. government spending on pensions and health care, see Shoven (2004). Shoven summarizes several studies and concludes that health care spending is better correlated with “years until death” than with “years since birth.”

Two commentators suggest that balance billing is not needed in order to attract new providers because “wealthy people can and do go abroad for [costly medical care services] if they wish” (Ikegami and Creighton Campbell, 2004, p. 33). For citizens of a country as rich and technologically sophisticated as Japan to have to seek care abroad seems oddly incongruent.

An IMF study (2004, p. 55) reports that increased medical and other care for the elderly will increase by 4 percentage points of GDP over the next 20 years.

For a review of the frustrating economic history of the 1990s, see Lincoln (2005).

Even now, official projections assume that birth rates will begin to climb after 2050. The reasons for such an assumption are difficult to fathom. There is also the possibility that birth rates will continue to fall: the actual birth rate in 2003 was only 1.29 (IMF, 2004, p. 55).

According to the National Institute of Population and Social Security Research, birth rates will continue falling until they reach 1.1 and will not rebound.

For a careful review of the causes of economic stagnation, see Kuttner and Posen (2001), who review the evidence on whether the failure of the Japanese economy to recover was traceable to the ineffectiveness of traditional macroeconomic tools—monetary expansion and fiscal stimulus—or the failure to apply them with sufficient vigor. They conclude that the tools work, but were not used aggressively enough. See also the comments by Fischer (2001, pp. 161-73).

See Lincoln (2005), who argues in favor of emphasizing net debt, and the OECD (2004, p. 64), which is agnostic on which is the better indicator.

The OECD (2004, p. 86) makes the same recommendation.

For a scathing criticism of policies of the Koizumi administration toward local governments, see DeWit and Yamazaki (2004). For a more tempered criticism, see Doi (2004). In a different article, Yamazaki and DeWit (2003) write: “There can be little doubt that a well-managed and full-scale [decentralization] reform would profoundly reshape the state and help revive Japan’s once vaunted economic competitiveness. This is because it is an essential condition for cutting through the country’s entrenched porkbarrel politics, which emphasizes wasteful and often environmentally damaging public works. Japan’s highly centralized fiscal system was very adept at spreading around the fruits of the country’s previously high economic growth. But over time, political, bureaucratic and business interests came to parasitize much of the enormous flow of funds from the centre to the subnational governments, leaving the fiscal system largely unable to adapt to the new needs of a rapidly changing economy and society.” <>.

For a perhaps extreme example of the view that population aging threatens disaster, see Kotlikoff and Burns (2004). They write, for example (pp. xi-xii): “Let your mind wander toward the future…You see a country [the United States] whose collective population is older than that in Florida today. You see a country where walkers outnumber strollers…You see a government in desperate trouble. It’s raising taxes sky high…It’s also printing tons of money to ‘meet’ its bills…You see major tax evasion, high and rising rates of inflation, a growing underground economy, a rapidly depreciating currency, and more people exiting than entering the country. You see political instability, unemployment, labor strikes, high and rising crime rates, record-high interest rates. You see financial markets in ruin. In short, you see America plunging headlong toward third world status.”


    Board of Trustees, 2005, The 2005 Annual Report of the Board of Trustees of the Federal Old-Age and Survivors Insurance and Disability Insurance Trust Funds, Washington: Board of Trustees.

    Burtless, G.,2004, “Pension Policy in Developed Countries: Assessment of Alternative Reforms in Response to Population Aging,”paper presented to the Ninth Annual Seminar of the National Institute of Population and Social Security Research, Tokyo, December16.

    DeWit, A., and Yamazaki, Y.,2004, “Koizumi’s Flawed Political Economy of Decentralization,”October14, <>

    Doi, T.,2004, “A Missing Link in Decentralization Reform in Japan: Trinity Reform Package,”Policy Research Institute, Ministry of Finance, Discussion Paper No. 04A-08, April.

    Fischer, S.,2001, “Comments and Discussion on “The Great Recession: Lessons for Macroeconomic Policy from Japan,’”Brookings Papers on Economic Activity, Vol. 2, pp. 161-73.

    Fukawa, T.,2002, Public Health Insurance in Japan,Washington: World Bank Institute.

    Fukawa, T., and Izumida, N.,2004, “Japanese Healthcare Expenditures in a Comparative Context,”Japanese Journal of Social Security Policy, Vol. 3, No. 2,December, pp. 51-61.

    Huber, M., and Orosz, E.,2003, “Health Expenditure Trends in OECD Countries, 1990-2001,”Health Care Financing Review, Vol. 25, No. 1, pp. 1-22.

    Ikegami, N., and Creighton Campbell, J.,2004, “Japan’s Health Care System: Containing Costs and Attempting Refom,”Health Affairs, Vol. 23, No. 3, May/June, pp. 26-36.

    IMF, 2004, Japan: Selected Issues, Washington: IMF, July13.

    Kent Weaver, R.,2003/04, “Design and Implementation Issues in Swedish Individual Pension Accounts,”Social Security Bulletin, Vol. 65, No. 4, pp. 38-56.

    Kotlikoff, L. J., and Burns, S.,2004, The Coming Generational Storm, Cambridge, Mass.: MIT Press.

    Kuttner, K. N., and Posner, A. S.,2001, “The Great Recession: Lessons for Macroeconomic Policy from Japan,”Brookings Papers on Economic Activity, Vol. 2, pp. 93-160.

    Lincoln, E. J.,2005, “Japan: Long-Term Budget Challenges,”paper presented to the Conference on Long-Term Budget Challenges and Fiscal Stability in the G-7, Washington, June2-4.

    OECD, 2004, “2004 Economic Review—Japan,”Paris: OECD, November18, Box 3.1.

    OECD, 2005, Pensions at a Glance: Public Policies Across OECD Countries, Paris: OECD.

    Reinhardt, U.,2000, “Health Care for the Aging Baby Boom: Lessons from Abroad,”Journal of Economic Perspectives, Vol. 14, No. 2, Spring, pp. 71-83.

    Shoven, J.,2004, “The Impact of Major Improvements in Life Expectancy on the Financing of Social Security, Medicare and Medicaid,” in H. J.Aaron and W. B.Schwartz,eds., The Impact of Molecular Biology on Medicine and Society, Washington: Brookings Institution.

    Takayama, N.,2005, “The Balance Sheet of Social Security Pensions in Japan,”Discussion Paper, February15.

    Yamazaki, Y., and DeWit, A.,2003, “Fiscal Cuts or Common Sense? Fiscal Decentralization in Japan,”June17,, <>

    Other Resources Citing This Publication