Japan: Selected Issues
Author:
International Monetary Fund
Search for other papers by International Monetary Fund in
Current site
Google Scholar
Close

A sustained decline in fertility rates underlies a rapid aging and decline of Japan's population. This will have profound social and economic implications. The paper illustrates the difficult situation facing Japanese fiscal policy in the years ahead. The findings of this paper indicate that there may be a role for foreign exchange interventions in providing stimulus at the current conjuncture. Deposit insurance reform is a central element in the government strategy to strengthen the Japanese banking system. The unemployment-deflation puzzle in Japan has been explained.

Abstract

A sustained decline in fertility rates underlies a rapid aging and decline of Japan's population. This will have profound social and economic implications. The paper illustrates the difficult situation facing Japanese fiscal policy in the years ahead. The findings of this paper indicate that there may be a role for foreign exchange interventions in providing stimulus at the current conjuncture. Deposit insurance reform is a central element in the government strategy to strengthen the Japanese banking system. The unemployment-deflation puzzle in Japan has been explained.

I. Population Aging and Its Macroeconomic Implications

by Hamid Faruqee

A. Introduction

1. Population aging will figure as a prominent feature of the Japanese economic landscape over the next century. According to demographic projections, Japan will have one of the the highest ratios of elderly dependents among the major industrial countries by the end of this year, despite having the lowest share just a decade ago. This dramatic shift is expected to continue well into the new century and will likely have profound social and economic implications.

2. At the policy level, the implications of an aging population have several dimensions in Japan. The social security system—consisting mainly of health care and pension benefits—will inevitably face rising costs associated with the increasing share of elderly dependents.1 In the absence of further policy reforms, the central government could be saddled with growing unfunded liabilities associated with these entitlement programs. In particular, the pension system, which is partially funded, will face a severe shortfall in meeting future pension obligations at prevailing contribution rates.

3. This chapter develops a general equilibrium framework to examine the economic implications of population aging in Japan as well as the policies designed to address it.2 The macroeconomic effects of demographic changes in the model are manifested through two main channels: (1) on the supply side, changes in the age structure have implications for labor supply, and (2) on the demand side, population aging has implications for aggregate consumption, saving, and investment. Introducing these features into a multi-country framework based on Multimod, the model is calibrated based on age-earnings and demographic data for Japan as well as for other industrial countries (01) as a group. To introduce a policy dimension, the model is further extended to incorporate a social security transfer scheme. This allows policy parameters affecting net taxes—i.e., taxes less transfers—to be directly incorporated into private sector behavior.

4. The consumption and saving behavior in the model flows from a modern life-cycle paradigm.3 Within this multi-cohort framework, younger agents tend to be net borrowers, reflecting the fact that permanent income exceeds current income; mature agents tend to be large net savers at the peak of their earnings potential; finally, the elderly also tend to save (albeit to a lesser degree), reflecting precautionary saving in the face of lifetime uncertainty and retirement.

5. Compared to many previous studies, the present analysis suggests somewhat smaller effects on saving rates from changing demographics. Much of the previous work is based on macroeconomic time series evidence and reduced-form coefficients from saving regressions on dependency ratios.4 These older studies tend to find very large negative effects on saving rates from increasing dependency rates.5 This paper takes a more structural approach, based on a model of overlapping agents whose behavior is more closely tied to the microeconomic evidence on household saving. In the case of Japan, as with many other countries, a stylized fact at the household level is that the elderly generally do not dissave.6 Consequently, population aging does not guarantee a large decline in aggregate saving, particularly when factors such as increasing longevity are also taken into account.

6. The results can be summarized as follows. In Japan, demographics are defined by a sustained decline in birth rates and an increase in longevity, leading to an aging of the population. The sharp decline in fertility rates is also responsible for a significant decline of Japan’s population. As a result of a contracting workforce, the level of real GDP is projected to fall (from a baseline with a constant workforce) by about 20 percent cumulatively over the next half-century or so. In terms of growth, annual GDP growth in Japan may be lower by about 0.5 percent for sometime as the economy settles to a long-run equilibrium with a permanently higher elderly dependency ratio and smaller supply of labor.

7. In per capita terms, GDP per person declines slightly in the long run (relative to baseline). Output falls in proportion to the contraction of labor—measured in efficiency units. However, the percent decline in effective labor supply is larger than the fall in the (adult) population, as the share of elderly increase. Investment and saving levels also decline with GDP through the adjustment process. The decline in investment reflects the desire to shed capital in the wake of the contraction in labor and output, though investment rates (as a share of GDP) remain unchanged. However, saving rates and, hence, the current account ratio increase slightly as the population ages. Despite a higher proportion of elderly who tend to save less, the increase in longevity and the decline in the inflow of young agents (who tend to have high consumption propensities) act to raise saving rates in Japan.

8. In terms of policy implications, the analysis highlights the importance of taking into account prospective changes in the macroeconomic environment when evaluating policies that address the challenges posed by population aging. An assessment of fiscal sustainability, for example, which focused only on the social security dimensions would miss an important component of the analysis if it ignored the macroeconomic implications of demographic changes for the fiscal accounts. Similarly, the endogenous response in private behavior to various policy changes should also be taken into account when examining policy reforms. On this last point, the simulation analysis suggests that changes in social security benefits can have a notable impact on private sector saving. In particular, a balanced decline in benefit and contribution rates is shown to boost private saving rates by nearly half of the reduction in benefit rates, as agents anticipate having to self-finance more of their consumption in retirement.

B. Demographic Trends

9. A central feature of Japan’s demographics is a long-run decline in fertility. In the postwar era, the total fertility rate-defined as the number of births per woman-experienced a sharp decline in a single ten-year span, falling from over 3½ births in 1950 to just 2 births in 1960 (Figure I.1). Since that time, the fertility rate has continued to decline generally, and now remains well below the replacement rate.7 The implications for the overall population of this dramatic fall in fertility is that the number of young adults or workers expected to arrive in the future (ignoring immigration) will decline significantly for some time to come.8

Figure I.1.
Figure I.1.

Japan-Total Fertility Rate, 1950–2050

(births per woman)

Citation: IMF Staff Country Reports 2000, 144; 10.5089/9781451820591.002.A001

Source: Takayama (1998) and World Bank

10. Redefining “births” as the inflow of new adults into the economy, Figure I.2 shows the historical and projected evolution of Japan’s “birth rate” b since 1960. As evident in the graph, the inflow of young adults as a share of the adult population has declined significantly over the past 40 years; moreover, even assuming a modest recovery in fertility rate over the next fifty years, the “birth rate” is projected to remain far below its historical levels well into the next century.9

Figure I.2.
Figure I.2.

Japan—“Birth” Rate & Population Growth, 1960–2150

(percent of adult population)

Citation: IMF Staff Country Reports 2000, 144; 10.5089/9781451820591.002.A001

Data Sources: OECD and World Bank

11. The long-run decline in Japan’s birth rate has two important demographic implications: a declining population and an aging population. As seen in Figure I.2, the decline in the “birth rate” b is associated with a decline in the population growth rate n. In fact, given that fertility rates have already fallen to such an extent, a declining adult population (n<0) can be expected for much of the 21st century.

12. With smaller cohorts of new adults arriving in the future, the relative share of working age people will diminish over time and the population’s average age will increase. The result is a sharp increase in the share of elderly in the population. Figure I.3. shows the dramatic rise in the ratio φ of Japan’s elderly dependents (age 65+) as a share of the adult population projected from the analytical model described in Annex I, as well as projections from Japan’s Ministry of Health and Welfare and the World Bank.

Figure I.3.
Figure I.3.

Japan-Elderly Dependency Ratio, 1960–2150

(percent of adult population)

Citation: IMF Staff Country Reports 2000, 144; 10.5089/9781451820591.002.A001

Sources: OED, World Bank, Japan Ministry of Health & Welfare, and IMF Staff Estimates

13. It should be noted that, while a decline in fertility is generally associated with an aging population, a contraction of the population need not occur, especially when mortality rates are also falling (i.e., longevity is increasing).10 Among other industrial countries as a group, for example, while population aging and a slowdown in population growth is expected, the population is not expected to shrink to any large extent. However, the more precipitous decline in fertility rates in Japan implies that a declining population will also figure as a prominent feature of its demographics.

C. Age-Earnings Profiles in Japan

14. A key component to the analysis of demographic effects is the nature of the age-earnings profile. Over the life cycle, individuals can expect a hump-shaped pattern to labor earnings. Initially, as agents join the workforce, they can expect a rising path of earnings, reflecting productivity gains that come from work experience and seniority wages that reward work service. Eventually, labor earnings level off and decline as agents move into retirement.

15. The profile that summarizes the life-cycle earnings path is important in determining both supply-side and demand-side implications of population aging. On the supply side, age-earnings profiles provide an indicator of the changes in relative productivity and (inelastic) labor supply that occur over an individual’s working life. On the demand side, the anticipated path of labor income influences the saving plans of consumers over their lifetimes. Changes in the demographic structure of the population can thus have important macroeconomic implications for aggregate saving and labor, stemming from these life-cycle effects.

16. To calibrate the model according to the life-cycle pattern of earnings, empirical age-earnings profiles for Japan are estimated using cross-sectional data on wage-based salaries by age group for the period 1970 to 1997. The earnings data are adjusted by labor force participation rates, reflecting the fact that the share of persons with zero earnings (i.e., those who are retired) varies across age groups. The age-earnings data thus represent the average earnings per person (not per worker) within each age category. The data points are shown in Figure I.4, normalized relative to per capita labor earnings of the youngest cohort.11

Figure I.4.
Figure I.4.

Japan-Age-Earnings Distributions, 1970–1997

(relative to youngest cohart)

Citation: IMF Staff Country Reports 2000, 144; 10.5089/9781451820591.002.A001

Sources: Japan Statistical Yearbook and IMF Staff Estimates

17. As shown by the data, relative earnings profiles have been stable over the sample period. This suggests that the underlying institutional and structural features that implicitly underpin the variation in relative labor earnings across age groups—such as seniority wages and retirement age—have also been fairly stable historically.12

18. In the analysis, the cross-sectional pattern between age and earnings that emerges from the data is taken as representative of the time-series pattern of an individual’s wage earnings over the course of his or her lifetime. Moreover, this relation between age and relative earnings is assumed to reflect the changes in relative productivity and labor supply that occurs over an agent’s life cycle.13 The profile essentially summarizes the nature and timing of the rise and fall in relative earnings that agents can expect as they become mature workers and then move (gradually) into retirement.

19. To characterize an agent’s life-cycle income profile, non-linear least squares (NLLS) estimation is used on the following equation:

r y ( s , t ) = a 1 e α 1 ( t s ) + a 2 e α 2 ( t s ) + ( 1 a 1 a 2 ) e α 3 ( t s ) ; ( 1 )

where ry denotes relative labor income, s and t represent cohort and time indices, respectively, and their difference determines the age of a particular cohort group. The restriction on the at terms reflects the normalization that the youngest cohort (i.e., when s = t) has relative income equal to unity. The fitted values of the estimated equation (1) are shown by the line in Figure I.4.14

20. The income-profile parameters in equation (1) enter the model in two distinct ways. On the supply side, the parameters directly affect the dynamics of aggregate labor supply—measured in efficiency units—since they reflect the relative productivity and labor supply of workers at different ages. On the demand side, since consumption is (partly) based on permanent income, these income-profile parameters also affect consumption-saving plans through their impact on human wealth—i.e., the present value of future labor income streams.15

21. Consumption/saving propensities do vary by age as consumers choose to smooth lifetime consumption in the face of life-cycle income. Younger agents tend to dissave or borrow if possible—i.e., if not liquidity constrained—to consume at levels commensurate with permanent income which exceeds current income.16 Older agents tend to save for retirement when labor earnings are relatively high. Meanwhile, the elderly tend to save at lower rates (falling to zero) in retirement as they largely consume out of asset (including annuity) income and transfers.

22. On this last point, unlike traditional life-cycle models—e.g., Diamond (1965) —the elderly do not dissave or run down financial assets in this model due to life-time uncertainty. Instead, with retirement, agents build up wealth to some target level as a precaution against the possibility of remaining alive without labor income.17 This behavioral feature allows the multi-cohort framework to avoid a common criticism of standard life-cycle models that posit large negative saving rates among retirees.18 In terms of empirical evidence, the model’s behavioral implications appear consistent with numerous studies at the household level that find scant evidence of dissaving among the elderly.19

Pension System

23. To complete the analytical framework, social security needs to be modeled. An advantage of the multi-cohort approach is that social security transfers can be straightforwardly included. In particular, a pension system can be introduced into the framework as follows. Consider the simple case of a lump-sum transfer scheme:

t r ( s , t ) = { α ( t ) ; s > j ( t ) + β ( t ) ; s j ( t ) ( 2 )

where tr(s, t) represent the transfers paid or received by individuals, depending on their age. Younger generations s > j(t) pay into the system, while older agents or pensioners sj(t) receive a benefit. For any transfer scheme, a full-financing condition can be written as follows:

t t r ( s , t ) N ( s , t ) d s = 0. ( 3 )

This general condition must hold for the transfer scheme to be deemed fully funded (i.e., no unfunded liabilities). Otherwise, there would exist a financing gap that would need to be covered through other revenues or government borrowing. The amount of the financing gap is determined by the following relation:

g a p ( t ) = β ( t ) φ ( t ) ( 1 φ ( t ) ) α ( t ) . ( 4 )

A positive gap would indicate a shortfall of financing relative to benefits. Full-financing—i.e., a zero gap—would require the well-know condition that the benefit-to-contribution must equal the support ratio, defined as the number of working-age persons relative to elderly dependents.

24. The case of payroll tax financing is straightforwardly extended. In that case, individual contributions α(s, t) would be age-dependent, determined by social security taxes paid on individual labor income—τy(s, t). In what follows, a social security transfer scheme along these lines is introduced into Multimod, so that net taxes—i.e., taxes less transfers—replace the previous treatment of government revenues; social security benefits and contributions (as ratios to GDP) are taken as exogenous policy parameters. In terms of behavior, these benefit and contribution rates mainly affect private behavior through their implications for permanent income (i.e., the present value of future income less net taxes).

D. Multimod Simulations

25. Incorporating demographic dynamics and social security into Multimod, this section conducts various simulations to assess the economic impact of a rising dependency ratio and to examine certain policies intended to address the challenges of population aging.20 To quantify the economic impact of demographic changes in Japan, a reference scenario is constructed where the population is assumed to be stationary in an initial steady state.21 Then treating Japan’s demographic projections as a shock and assuming unchanged policies, the economic effects of population decline and aging are simulated.22

26. Table I.1 and Figure I.5 summarize the economic effects of demographic changes in Japan.23 With a declining and aging population, the level of real GDP is projected to fall by about 20 percent cumulatively (from baseline) in the long run.24 The decline in GDP largely occurs between 2025 and 2075, when the demographic changes are most pronounced. In growth terms, annual GDP growth is lower by about 0.5 percent per year over this time period, before the economy settles to a longer-run equilibrium with a permanently higher elderly dependency ratio.25

Table I.1.

Simulated Effects of Population Aging in Japan

(Percent deviation from baseline; unless noted otherwise)

article image

Percentage point deviation from baseline; interest rates are long-term nominal rates.

Deviation from baseline value in percent of baseline GDP

Figure I.5
Figure I.5

Japan-Multimod Simulation Effects of Population Aging and Decline

Citation: IMF Staff Country Reports 2000, 144; 10.5089/9781451820591.002.A001

Source: Staff Estimates.

27. In per capita terms, GDP per adult declines slightly in the long run (relative to baseline) for the following reason. The percent decline in output is in line with the contraction of labor, measured in efficiency units. However, the percent decline in effective labor is larger than the fall in the number of workers, given the aging of the workforce and the differences in labor productivity and supply across age groups implicit in the age-earnings profile. Extending the simulation further out (i.e., closer to steady state) would show the decline in per capita GDP to be about 5 percent relative to baseline, as the output-labor ratio returns to its baseline level.26

28. Investment and saving levels (relative to baseline) also decline in the long run with GDP. The fall in investment reflects the desire to shed capital in the face of declining labor and output in the economy; the rate of investment (as a share of GDP) though is more or less unchanged. Saving rates, however, increase slightly as the population ages. Despite a higher proportion of elderly who tend to save less, the decline in the inflow of young agents (who tend to have high consumption propensities) and the increase in longevity act to raise saving rates.27 Consequently, the current account surplus increases, mainly reflecting the rise in private saving. Correspondingly, the real exchange rate would depreciate initially before appreciating in the long run with the accumulation of net foreign assets.28

Policy Simulation

29. The framework can also be used to examine the impact of various policies. In Japan, the debate over social security reforms has largely centered around the extent of benefit cuts and the methods for achieving sustainable financing. On the latter, concerns have arisen that the prospective increases in the payroll tax burden to finance future pension benefits could act as a disincentive to work.29

30. To examine some of the implications of pension reform, a dynamic simulation of changes in social security is considered. In particular, the effects of a permanent reduction in social security premiums and benefits are simulated. Contribution and benefits are both permanently reduced by 2 percent of GDP; the adjustment paths to this long-run reduction is kept the same for both the expenditure- and revenue-side of social security so that the direct effect on the overall fiscal balance is always zero.

31. Table I.2 and Figure I.6 show the simulated effects of a balanced reduction in social security contribution and benefit rates in Japan. A cut in these rates initially reduces consumption and output. Consumption remains below baseline for some period as agents increase their saving rates in response to the cuts. The initial slowdown in consumption underpins a decline in interest rates and some boost to investment. The saving effect is larger, however, and the current account ratio rises, Over the longer run, higher output and consumption levels are attained with the build-up of capital and net foreign assets.30

Table I.2.

Simulated Effects of Pension Reform in Japan

(Percent deviation from baseline; unless noted otherwise)

article image

Percentage point deviation from baseline; interest rates are long-term nominal rates.

Deviation from baseline value in percent of baseline GDP

Figure I.6.
Figure I.6.

Japan-Multimod Simulation Effects of Pension Reform

Citation: IMF Staff Country Reports 2000, 144; 10.5089/9781451820591.002.A001

Source: Staff Estimates.

32. The increase in saving rates can be understood as follows. The reform redistributes disposable income from pensioners who receive less transfer income to workers who face lower payroll taxes. This latter group generally have higher marginal saving propensities. In addition to this distributional aspect, saving across all age groups would generally rise as individuals, faced with less generous pension benefits now or in the future, need to save more for their own retirement and future consumption. Quantitatively, the simulations suggest that private saving increases by 0.8 percent of GDP, or nearly half the amount of the reduction in benefits (2 percent of GDP).31

Sensitivity Issues

33. Throughout the analysis, it has been assumed that the age-earnings profile remains stable in the face of demographic changes. Whether the stable historical relationship between age and relative earnings will prevail in the future, however, is an open question. With healthier seniors living longer and pension benefits possibly declining, older workers may decide to postpone retirement. A higher retirement age and similar considerations would tend to “flatten” the age-earnings profile and, hence, mitigate that economic impact of population aging.32 Intuitively, since the income profile largely determines age-specific characteristics in the analysis, a flatter profile suggests smaller differences across age groups and, thus, smaller economic implications from a changing age distribution.

34. A countervailing effect, however, is the possibility that the extent of population aging and contraction may be understated. The demographic projections—taken as exogenous—are largely based on World Bank projections; these projections impose a stationary population by construction by 2150. However, in the case of Japan, the downward momentum in birth rates and population growth rates would require a significant stabilization and recovery for a stationary population to obtain within this time frame (see Figure I.1).

35. Another issue is total factor productivity growth (TFP) which is taken as exogenous. Some have argued that the transition to a smaller, older workforce in Japan will require adoption of labor-saving technologies and human capital-intensive production.33 These production technologies may stand to benefit more from dynamic efficiency gains associated with technological change and improvement. Thus, the possible structural shift in the economy may entail a rise in the overall rate of technological progress in the economy, allowing output and living standards to rise at higher rates than otherwise.

E. Conclusions

36. Demographic changes will be a defining feature in Japan for the foreseeable future. A sustained decline in fertility rates underlies a rapid aging and decline of Japan’s population that can be expected to continue well into the new century. This dramatic demographic shift will likely have profound social and economic implications. Using a multi-cohort modeling approach and the information in empirical age-earnings profiles, this analysis has sought to quantify some of the macroeconomic implications of demographic changes in Japan. The results of the dynamic simulations can be summarized as follows.

  • Population aging and decline in Japan will likely result in slower growth in output for some time. Absent any significant acceleration in total factor productivity, annual output growth in Japan would be lower by about one-half percent over the next half century or so as the workforce contracts. In per capita terms, GDP per person could also decline (relative to the case of no demographic changes) since the significant contraction of the population and workforce suggests an even larger decline in effective labor supply, given the rising share of elderly.

  • Saving rates and the current account ratio need not decline significantly with an aging population. Though a higher share of elderly would tend to reduce saving rates, other aspects of demographic changes in Japan, such as fewer young adults (and youth dependents) and increased longevity, would tend to counterbalance the negative effects of population aging on saving rates.

  • In terms of policy implications, the analysis highlights the importance of taking into account prospective changes in the macroeconomic environment when evaluating policies that address the challenges posed by population aging. Moreover, the potential impact of social security reforms on private sector behavior should also be incorporated. On this last point, the simulations show that reforms to social security benefits could have large effects on private saving. In particular, the model suggests that a balanced decline in benefit and contribution rates would boost private saving rates by nearly half of the reduction in benefit rates, as agents anticipate having to finance more of their own consumption in retirement.

ANNEX I Analytical Framework

1. In this section, some key components of the analytical framework are described, beginning with population dynamics.1 For the overall (adult) population, the basic law of motion is given by:

N . ( t ) N ( t ) = n ( t ) = b ( t ) p ( t ) , ( A 1 )

where N is the population level and n is the growth rate, b the “birth” rate—defined as the arrival rate of new adults—and p is the mortality or death rate;2 a dot over a variable denotes the derivative with respect to time. Integrating equation (A1) over time yields an expression (up to a constant of integration) for the size of the total population at any moment in time:

N ( t ) = e t n ( v ) d v = e t [ b ( v ) p ( v ) d v ] , ( A 2 )

Equation (A2) shows that population size evolves according to the accumulation of past changes to its growth rate (i.e., the difference between past birth and death rates), which determines the size of the current population as a growth factor times the size of the initial population.3

Dependency Ratio

2. To examine various demographic issues—e.g., population aging—it is useful to define a measure that characterizes the age distribution of the population. By summing up all cohorts above a certain age, an elderly dependency ratio can be defined as a proportion of the total population as follows:

φ ( t ) = j ( t ) N ( s , t ) N ( t ) d s ; 0 < φ < 1 , ( A 3 )

where ɸ measures the proportion of all individuals older than some threshold age—indexed by j(t). Assuming that this age definition does not change, the index j(t) moves with time to include new dependents, who have just reached the threshold age at each moment in time (i.e., j(t) = 1). In the case where birth rates are constant, it can be shown that the dependency ratio ɸ would also be constant.4 Otherwise, the dependency ratio evolves over time according to the time derivative of equation (A3):

φ ˙ ( t ) = N ( j ( t ) , t ) N ( t ) [ p ( t ) + n ( t ) ] φ ( t ) . ( A 4 )

3. At each moment in time, the change in the dependency ratio is determined by the relative size of new dependents, less the proportion of the elderly who die each period and accounting for growth in the population base —i.e., the scaling variable. Using equation (A4) and the paths for Japan’s birth rate and population growth rate shown in Figure I.2, the model can be used to project the evolution of the dependency ratio as shown in Figure I.3 in the text.5

ANNEX II Reference Scenarios

1. To calibrate the magnitude of demographic changes and their effects, one can construct for comparison purposes counterfactual scenarios wherein certain demographic variables remain unchanged. Using this reference scenario as an artificial baseline, one could then simulate the economic impact of Japan’s demographic projections, as determined by the behavioral features of the model and given unchanged policies.1 These simulation results are helpful in projecting—in model-consistent fashion—the future paths of key macroeconomic variables (e.g., interest and growth rates) under population aging when considering issues such as pension reforms.2

2. The counterfactual exercise is described here in the appendix and is done in two different ways. In the first simulation, a stationary population is used as the reference scenario to help identify the effects of both population contraction and aging; in the second simulation, the effects of population aging alone are isolated. Japan’s demographic dynamics contain components of both phenomena.

Population Contraction and Aging

3. To examine the effects of a declining and aging population and work force implied by Japan’s demographic projections, a counterfactual scenario is constructed where the birth and death rates are chosen so that the rise in the dependency ratio is curtailed somewhat and the population does not contract. Specifically, in the initial steady state, it is assumed the population is stationary with the birth rate constant at its 2000 level onward; the death rate is also set equal to this value (i.e., b-p = n = 0).

4. Figure I.A1 shows the evolution for the Japan’s birth rate and population growth rate under the counterfactual baseline scenario as well as under the alternative scenario of population aging. Figure I.A2 shows the evolution of the elderly dependency ratio under both scenarios. Note that since the fertility rate has already declined (as part of history), some aging will also occur under the baseline scenario since the full effects on the dependency rate come only with a lag—i.e., the present dependency ratio is still below its long-run level even if the population is stationary from this point forward.

Figure I.A1.
Figure I.A1.

Japan-Birth Rate and Population Growth Rate, 1990–2150

(percent of adult population)

Citation: IMF Staff Country Reports 2000, 144; 10.5089/9781451820591.002.A001

Figure I.A2.
Figure I.A2.

Japan-Elderly Dependency Ratio, 1990–2150

(percent of adult population)

Citation: IMF Staff Country Reports 2000, 144; 10.5089/9781451820591.002.A001

5. Comparing the two scenarios, one sees that the demographic shock underlying the aging scenario involves a continued decline in the birth rate. Population growth is initially positive and higher under the aging scenario before declining and turning negative below the zero growth rate maintained throughout the baseline scenario. This suggests that the mortality rate is also lower initially (i.e., greater longevity) under the population aging scenario than in the case of a stationary population. The fact that population growth eventually goes to zero in the aging scenario at a lower long-run birth rate further requires that the long-run death rate also remain lower than in the baseline.

Population Aging

6. To better isolate the impact of a changing age distribution, a second counterfactual or baseline scenario can be constructed where the population growth rate is identical to the aging scenario, but where the rise in the dependency ratio is muted. This is done by assuming a higher birth rate in the second reference scenario than in the aging case; correspondingly, the death rate is sufficiently raised so that the same population growth rate obtains. The paths for the b and n in these two scenarios are shown in Figures I.A3 and I.A4.

Figure I.A3.
Figure I.A3.

Japan-Birth Rate and Population Grouth Rate, 1990–2150

(percent of adult population)

Citation: IMF Staff Country Reports 2000, 144; 10.5089/9781451820591.002.A001

Figure I.A4.
Figure I.A4.

Japan-Elderly Dependency Ratio, 1990–2150

(percent of adult population)

Citation: IMF Staff Country Reports 2000, 144; 10.5089/9781451820591.002.A001

7. Comparing these figures to those associated with the previous counterfactual scenario, one sees that the aggregative implications of a decline in the birth rate are removed in the second experiment, but the distributional implications are accentuated. In other words, the rise in the dependency ratio in the aging scenario is relatively much larger against the second counterfactual scenario than the first (Figure I.A4 vs. Figure I.A2), but the growth differences are completely negated (Figure I.A3).

8. The comparative effects of both counterfactual scenarios versus the aging scenario is summarized below in Table I.A1. The top half of the table shows the effects attributable to both population aging and decline relative to baseline; the bottom half of the table shows the effects due solely to aging. The results indicate that almost half of the long-run fall in GDP is due to an aging workforce and not just a shrinking workforce.

Table I.A1.

Comparative Effects of Demographic Changes in Japan

(percentage point deviation from baseline; unless noted otherwise)

article image

Percentage point deviation from baseline.

Deviation from baseline level in percent of baseline GDP.

9. In terms of saving behavior, because mortality falls (i.e., longevity rises) to a greater extent (see Figure I.A3) in the second simulation—i.e., the second reference scenario versus the aging scenario—private saving will rise more in this case. As the death rate p falls, the planning horizon lengthens and the effective discount rate declines. Longer horizons and more patience on the part of agents tend to raise their saving propensities. In the simulations, this can be seen by the larger initial fall in consumption and improvement in net exports in the bottom half of the table. In the long-run, though, consumption falls by less because national income (and output) are higher as net foreign assets are accumulated.

References

  • Abel, Andrew B., “Precautionary Saving and Accidental Bequests.” American Economic Review, Vol. 75, pp. 777 –91, 1985.

  • Attanasio, Orazio, James Banks, Costas Meghir, and Gugliemo Weber, 1995, “Humps and Bumps in Lifetime Consumption,” NBER Working Paper No. 5350.

    • Search Google Scholar
    • Export Citation
  • Blanchard, O., “Debt, Deficits and Finite Horizons,” Journal of Political Economy, Vol. 93, pp. 223 –7, 1985.

  • Blanchard, O., and S. Fischer, Lectures on Macroeconomics (Cambridge, MA: MIT Press), 1989.

  • Browning, Martin, and Annamaria Lusardi, “Household Saving: Micro Theories and Micro Facts,” Journal of Economic Literature, Vol. XXXIV, No. 4, (December), pp. 1797 –1855, 1996.

    • Search Google Scholar
    • Export Citation
  • Buiter, W.H., “Death, Birth, Productivity Growth and Debt Neutrality,” The Economic Journal, Vol. 98, pp. 279 –93, 1988.

  • Carroll, Christopher D., “Buffer Stock Saving and the Life Cycle/Permanent Income Hypothesis,” Quarterly Journal of Economics, Vol. 112, pp. 1 –57, 1997.

    • Search Google Scholar
    • Export Citation
  • Carroll, Christopher D., and Andrew A. Samwick, “The Nature of Precautionary Wealth,” Journal of Monetary Economics, Vol. 40, pp. 41 –71, 1997.

    • Search Google Scholar
    • Export Citation
  • Chand, Sheetal K., and Albert Jaeger, Aging Populations and Public Pension Schemes, IMF Occasional Paper No. 147, Washington, DC: International Monetary Fund, December, 1996.

    • Search Google Scholar
    • Export Citation
  • Cutler, David M., James M. Poterba, Louise M. Sheiner, and Lawrence H. Summers, “An Aging Society: Opportunity or Challenge?” Brookings Papers on Economic Activity, pp. 1 –73, 1990.

    • Search Google Scholar
    • Export Citation
  • Davies, James, “Uncertain Lifetimes, Consumption and Dissaving in Retirement,” Journal of Political Economy, Vol. 89, pp. 561 –78, 1981.

    • Search Google Scholar
    • Export Citation
  • Diamond, P. A., “National Debt in a Neoclassical Growth Model,” American Economic Review, Vol. 55, pp. 1126 –50, 1965.

  • Faruqee, Hamid, Douglas Laxton, and Steven Symansky, “Government Debt, Life-Cycle Income and Liquidity Constraints: Beyond Approximate Ricardian Equivalence,” IMF Staff Papers, Vol. 44 (September), pp. 374 –82, 1997.

    • Search Google Scholar
    • Export Citation
  • Faruqee, Hamid and Aasim Husain, “Saving Trends in Southeast Asia: A Cross-Country Analysis,” Asian Economic Journal Vol. 12, No. 3, pp. 195 –217, 1998.

    • Search Google Scholar
    • Export Citation
  • Faruqee, Hamid and Douglas Laxton, “Life-Cycles, Dynasties, Saving: Implications for Closed and Small Open Economies,” IMF Working Paper, forthcoming (July), 2000.

    • Search Google Scholar
    • Export Citation
  • Feldstein, Martin S., “International Differences in Social Security and Saving,” Journal of Public Economics, Vol. 14 (October), pp. 225 –44, 1980.

    • Search Google Scholar
    • Export Citation
  • Flavin, M.A., “Excess Sensitivity of Consumption to Changing Expectations of Future Income,” Journal of Political Economy, Vol. 89, pp. 974 –1009, 1981.

    • Search Google Scholar
    • Export Citation
  • Hayashi, Fumio, “Why are Japan’s Saving Rates So Apparently High,” in S. Fischer, ed., NBER Macroeconomics Annual (Cambridge, MA: M.I.T. Press), pp. 147 –210, 1986.

    • Search Google Scholar
    • Export Citation
  • Horioka, Charles Yuji, “The Determination of Japan’s Savings Rate: The Impact of the Age Structure of the Population and Other Factors,” Economic Studies Quarterly, Vol. 42 (September), pp. 237 –53, 1991.

    • Search Google Scholar
    • Export Citation
  • Hurd, Michael D. and Naohiro Yashiro eds. The Economic Effects of Aging in the United States and Japan, (Chicago: University of Chicago Press), 1997.

    • Search Google Scholar
    • Export Citation
  • Jappelli, T., “Who is Credit Constrained in the U.S. Economy?” Quarterly Journal of Economics, Vol. 105, pp. 219 –34, 1990.

  • Jappelli, T., and M. Pagano, “Consumption and Capital Market Imperfections: An International Comparison,” American Economic Review, Vol. 79, pp. 1088 –105, 1989.

    • Search Google Scholar
    • Export Citation
  • Kosai, Yutaka, Jun Saito and Naohiro Yashiro, “Declining Population and Sustained Economic Growth: Can They Coexist?,” American Economic Review, Papers and Proceedings, Vol. 88 (May), pp. 412 –16, 1998.

    • Search Google Scholar
    • Export Citation
  • Kotlikoff, Laurence J. and Lawrence Summers, “The Role of Intergenerational Transfers in Aggregate Capital Accumulation,” Journal of Political Economy, (August), pp. 706 –732, 1981.

    • Search Google Scholar
    • Export Citation
  • Laxton, D., and others, Multimod Mark III: The Core Dynamic and Steady-State Models, Occasional Paper 164, International Monetary Fund, Washington, D.C., 1998.

    • Search Google Scholar
    • Export Citation
  • Masson, Paul R. and Ralph W. Tryon, “Macroeconomic Effects of Projected Population Aging in Industrial Countries,” IMF Staff Papers, Vol. 37 (September), pp. 453 –85, 1990.

    • Search Google Scholar
    • Export Citation
  • Masson, Paul R., Tam Bayoumi and Hossein Samiei, “Saving Behavior in Industrial and Developing Countries,” Staff Studies for the World Economic Outlook, (September), (Washington D.C.: International Monetary Fund), 1995.

    • Search Google Scholar
    • Export Citation
  • Meredith, Guy, “Demographic Changes and Household Saving in Japan,” and “Alternative Long-Run Scenarios,” in Ulrich Baumgartner and Guy Meredith, eds., Saving Behavior and the Asset Price ‘Bubble’ in Japan, IMF Occasional Paper No. 124 (April), (Washington D.C.: International Monetary Fund), 1995.

    • Search Google Scholar
    • Export Citation
  • Shibuya, Hiroshi, “Japan’s Household Saving Rate: An Application of the Life Cycle Hypothesis,” IMF Working Paper No. WP/87/15 (March), 1987.

    • Search Google Scholar
    • Export Citation
  • Takayama, Noriyuki, The Morning After in Japan: Its Declining Population, Too Generous Pensions, and Weakened Economy. (Tokyo: Maruzen Co., Ltd), 1998.

    • Search Google Scholar
    • Export Citation
  • Weil, P., “Overlapping Families of Infinitely-Lived Agents,” Journal of Public Economics,, Vol. 38, pp. 183 –98, 1989.

  • Weil, David N., “The Saving of the Elderly in Micro and Macro Data,” Quarterly Journal of Economics, Vol. 109 (February), pp. 55 –81, 1994.

    • Search Google Scholar
    • Export Citation
  • World Bank, Averting the Old Age Crisis, (Oxford: Oxford University Press), 1994.

  • Yaari, M.E., “Uncertain Lifetimes, Life Insurance, and the Theory of the Consumer,” Review of Economic Studies, Vol. 32, pp. 137 –50, 1965.

    • Search Google Scholar
    • Export Citation
  • Yashiro, Naohiro and A.S. Oishi, “Population Aging and the Savings-Investment Balance in Japan,” in Michael D. Hurd and Naohiro Yashiro, eds., The Economic Effects of Aging in the United States and Japan, (Chicago: University of Chicago Press), 1997.

    • Search Google Scholar
    • Export Citation
  • Yashiro, Naohiro, Takashi Oshio, and Mantaro Matsuya “Macroeconomic and Fiscal Impacts of Japan’s Aging Population with Specific Reference to Pension Reforms,” Economic Research Institute and Economic Planning Agency Working Paper, 1997.

    • Search Google Scholar
    • Export Citation
1

The government provides several types of social security benefits through the following agencies: Employee Pension Insurance (EPI) for private sector employees, Mutual Aid Associations (MAAs) for public sector employees, and National Pension (NP) and National Health (NH) systems for self-employed and agricultural workers. Universal national health coverage in Japan was instituted in 1961.

2

A more thorough examination of alternative fiscal policies and social security reforms is discussed in Chapter II by M. Mühleisen.

4

Cutler et al. (1990) take a third approach to population aging using the Ramsey optimal growth model, augmented for a changing support ratio—i.e., the share of effective workers to consumers. They find that saving (and investment) rates generally decline with population aging, but that short-run saving and saving net of investment tend to rise in the country aging more rapidly.

5

See, for example, Meredith (1995) for a summary of the macroeconomic evidence on demographics and saving with an analytical application to Japan; see Yashiro and Oishi (1997) for a review of studies on the implications for the saving-investment balance in Japan.

6

See Hayashi (1986). See Weil (1989) for a general discussion of the tension between the saving implications of demographics from the macroeconomic and microeconomic evidence.

7

Takayama (1998) cites three factors behind the decline in fertility rates in Japan: first, a decrease in the salary gap between men and women; second, difficulties reconciling work with child rearing; and third, a generous social security system to provide adequate living standards for the elderly.

8

The World Bank projections call for Japan’s fertility rate to stabilize and recover to the replacement rate by the middle of the century. This recovery is needed to (eventually) obtain a stationary population.

9

To be consistent with the model described later, the focus here is on the adult population (i.e., age 15+), and the birth rate is defined as the arrival rate of new adults or workers. Correspondingly, growth rates refer to the percent change of the adult population. In Figure I.2, historical data and projections for the “birth” rate are constructed from data from the OECD and World Bank on the youngest adult cohort group—i.e., 15–24 year olds—to yield the share of new adults arriving each period; this series resembles movements in the more conventional fertility rate shown in Figure I.1 with a 20-year lag.

10

World Bank demographic statistics for Japan project an increase in life expectancy (at birth) from 81 years presently to 84 years by 2050.

11

Data are based on contract wages by age, adjusted for labor force participation rates, as reported in the Statistical Yearbook; the earnings data are combined for both male and female workers for the following age categories: 15–24, 25–29, 30–34, 35–39, 40, 45–49, 50–54, 55–59,60–64, 65+. Midpoints of each age group were used in estimating the structural time-series equation relating earning to age. Ministry of Labor data on earnings—which include contract wages and bonus wages—would slightly raise relative earnings for the middle-age groups (age 35–54), but otherwise yields a very similar profile.

12

Whether the historical age-earning profiles will be a reliable guide to the future, particularly in the face of dramatic demographic changes, is an open question. We return to this issue when discussing the sensitivity issues later.

13

An issue is whether the empirical relation between age and relative earnings fully reflect differences in productivity and labor supply. The presence of seniority wages, for example, may be more a matter of prestige than productivity and may overstate the gains in productivity that accompany the rise in earnings of mature workers. On this score, however, it should be noted that the estimated earnings profile (solid line) in Figure I.4 tends to place the peak of relative earnings at an earlier age, when the productivity peak is perhaps more likely to occur.

14

NLLS estimates yield the following results (corrected standard errors appear in parentheses): α1=0.073**(0.001);α2=0.096**(0.002)α3=0.085**(0.001)R¯2=0.92,D.W.=1.31,S.E.E.=0.18;

article image

where *(**) indicates significance at the 5 (1) percent level. The α1 coefficients were obtained through grid search 1 = α2 = 200).

15

The paper describing the complete theoretical model is available from the author upon request.

16

See Jappelli (1990) for evidence suggesting that younger agents—who tend to be relatively asset and income poor—are the segment of the population most likely to face borrowing constraints. Liquidity constraints are included in the model to also reflect the fact that consumption tends to be somewhat hump-shaped over the life cycle; see Attanasio et al. (1995). This assumption also allows consumption to display some “excess sensitivity” to disposable income, reflecting the fact that some agents are unable to borrow and must consume out of current income; see for example Flavin (1981).

17

See Davies (1981) and Abel (1985). Kotlikoff and Summers (1981) argue that retirement saving in traditional life-cycle models appears insufficient to explain the amount of wealth accumulation and the fact that 80 percent of wealth is inherited in the United States. But as Abel (1985) shows, a life-cycle model with precautionary saving and accidental bequests can largely address these issues. Liquidity constraints can also be shown to further augment the amount of capital accumulation in the model.

18

Hayashi (1986) argues that positive saving among the elderly in Japan favors models with bequest motives over the life-cycle approach. Yoshiro et al. (1997) argues, however, that in the case of Japan the micro data tend to overstate the positive saving rates of the elderly by under-representing poorer agents who reside with younger family members.

19

For a recent review of household saving, see Browning and Lusardi (1996) and the references cited therein.

20

See Laxton et al. (1998) for a description of the Mark III vintage of Multimod.

21

The annex describes in detail the reference scenario as well as an alternative counterfactual scenario and its comparative implications for the results.

22

The public debt path is taken as given and social security benefit and contribution ratios (to GDP) are fixed. This is done to isolate the impact of population aging on the macroeconomy through its direct implications for saving, investment and labor supply. The additional implications through the fiscal accounts and social security are taken up in Chapter II.

23

The table shows the impact of the shock on the Japan block of the model in isolation—i.e., treating world variables as exogenous and without further feedback. Examining the effects of demographic changes on Japan’s variables in a multi-country setting (with feedback effects) would yield very similar results. Adding demographic dynamics simultaneously for the other industrial countries, however, will change some of the implications of the model, particularly for external variables. These difference are noted later.

24

Initially, output and investment rise owing to the fact that effective labor supply is growing at the outset relative to the stationary population assumed in the baseline. Significant aging (relative to baseline) also does not occur till after 2015. An increase in effective labor supply—somewhat akin to a positive shock to labor productivity—tends to reduce the interest rate and boost saving and investment. But, eventually, the decline in effective labor with population aging and contraction (similar to a negative productivity shock) lead to a rise in interest rates as saving and investment levels decline with output.

25

With inflation targeting, the impact of aging on inflation is negligible. Under a money targeting policy rule, inflation and nominal interest rates tend to rise by about 0.5 percent during the period of slower growth. In this latter case, though, the path of nominal interest rates and nominal exchange rates would look somewhat different.

26

With the initial decline in interest rates, the economy experiences some capital deepening (i.e., rise in the capital-labor ratio), and thus, a rise in GDP per effective unit of labor. But because aggregate productivity is exogenous and the interest rate is fixed (equal to the world rate) in the long run, the capital-labor ratio is predetermined in the long-run by the production function.

27

An increase in longevity—i.e., a decline in the mortality rates in the model—increases agents’ planning horizons and lowers their effective discount rate. Longer horizons and more patience tend to raise saving rates. See Annex II for a further discussion. Other parameters affecting consumption (e.g., liquidity constraints) could also have some bearing on the saving implications of demographics. For example, fewer liquidity constraints—i.e., more dissaving/borrowing by younger agents—would tend to magnify the increase in aggregate saving rates from population aging.

28

In multi-country simulations, the results for Japan are broadly similar to those shown in Table I.1. With the single country simulation (i.e., small open economy assumption), the world interest rate is fixed; hence, the domestic interest rate in the long run is also fixed (equal to the world rate) by interest rate parity. In the multi-country case, however, the world interest rate is endogenous and tends to decline permanently with the world demographic shock. The reason is that world population growth is positive (for some time) with the shock; the increase in labor supply spurs saving and investment abroad and leads to a decline in the world interest rate relative to the baseline. For Japan, domestic rates still tend to rise during the adjustment phase with aging, but then converge to the lower long-run level of interest rates internationally which mitigates somewhat the fall in output in Japan.

29

See Chapter II for a fuller discussion of social security and the menu of alternative reform options. Note also that labor supply is assumed to be inelastic (though age-varying) here and hence unresponsive to policy changes. In Chapter II, labor supply effects are also considered.

30

Multi-country simulations of this shock yield very similar findings. Note that the simulation in Table I.2 ignores possible labor supply gains surrounding a cut in payroll taxes. This case is taken up in Chapter II.

31

The saving effects shown are close to those implied by cross-country estimates reported in Feldstein (1980).

32

With a rapidly aging workforce, firms may find it more difficult to reward workers with seniority wages which could affect the relative earnings profile. But in a sense, this is consistent with the notion that seniority wages may not necessarily reflect higher productivity of older workers. Hence, the empirical age-earnings profile may already understate the decline in relative productivity of workers at advanced stages of their earnings cycle, and thus, may understate the labor supply effects of an aging workforce.

1

The paper describing the complete theoretical model is available from the author upon request.

2

See Buiter (1988) for the case of constant population growth with fixed birth and death rates. In that case, the number of agents belonging to a generation s at time s (i.e., at the time they are bora), as a proportion of the contemporaneous population, is given by N(s, s) = bN(s); the number of survivors from that cohort at time t ≥s is then given by N(s, t) = bN(s) e-p(t-s), where p) is the common death rate facing all agents.

3

Note that the size of the population, which until now has been defined in relative terms—vis-à-vis a reference population, can also be defined in aggregative terms—as the sum of all existing individuals across all generations (indexed by s): N(t)=N(s,t)ds.

4

With steady population growth, the dependency ratio would settle down to its long-run value: φ¯=eb¯Δ, where Δ ≡ t - j(t) and b¯ are constants.

5

Note that the empirical death rate p is determined implicitly by the difference between b and n. Because the model assumes a common death rate independent of age, the model tends to overpredict the share of elderly in the population. To compensate, a uniform adjustment factor of +0.5% was added to the birth rate series shown in Figure I.1; this size of this increase in the inflow of young adults broadly offsets the distributional implications of a common death rate across cohorts. To match the population growth rate, p also reflects this uniform adjustment.

1

The basic simulations are conducted in a model without pensions to isolate the economic effects of population aging alone. A pension transfer scheme is later introduced when examining policy shocks.

2

See Chapter II by M. Mühleisen.

  • Collapse
  • Expand
Japan: Selected Issues
Author:
International Monetary Fund
  • Figure I.1.

    Japan-Total Fertility Rate, 1950–2050

    (births per woman)

  • Figure I.2.

    Japan—“Birth” Rate & Population Growth, 1960–2150

    (percent of adult population)

  • Figure I.3.

    Japan-Elderly Dependency Ratio, 1960–2150

    (percent of adult population)

  • Figure I.4.

    Japan-Age-Earnings Distributions, 1970–1997

    (relative to youngest cohart)

  • Figure I.5

    Japan-Multimod Simulation Effects of Population Aging and Decline

  • Figure I.6.

    Japan-Multimod Simulation Effects of Pension Reform

  • Figure I.A1.

    Japan-Birth Rate and Population Growth Rate, 1990–2150

    (percent of adult population)

  • Figure I.A2.

    Japan-Elderly Dependency Ratio, 1990–2150

    (percent of adult population)

  • Figure I.A3.

    Japan-Birth Rate and Population Grouth Rate, 1990–2150

    (percent of adult population)

  • Figure I.A4.

    Japan-Elderly Dependency Ratio, 1990–2150

    (percent of adult population)