The Literature

5. There is a broad consensus in the literature that population aging in industrialized countries will reduce output growth and limit increases in economic welfare (see Kohl and O’Brien 1998, OECD 1998a for a survey). A shrinking population will be associated with lower levels of employment and thus output, although the impact on per-capita incomes could be mitigated by rising capital-labor ratios, productivity increases, and higher labor participation rates. However, to the extent that savings behavior is governed by life-cycle motives—that is, individuals save for retirement during their working life and dissave during retirement—most models predict saving rates would decline as the share of the elderly population increases (e.g., Meredith 1995). The ability of the government to stem this decline through higher public saving will be limited, owing to the increasing demand for higher spending on social services on the elderly.

6. The conclusion reached by most papers is that, in order to maintain fiscal sustainability and preserve intergenerational equity, net public pension benefits would need to be substantially cut (Chand and Jaeger 1996, OECD 1998a, b). Yashiro et al. (1997) also point out the need for substantial labor market reforms in order to achieve higher participation rates among married women, and reduce seniority wages that tend to act as a disincentive to hiring middle-aged and elderly workers.

7. However, in focussing largely on the broad macroeconomic picture, this strand of the literature has generally provided scant help in choosing between different reform options. In general equilibrium models, the fiscal sector is usually treated on a fairly aggregated basis, and demographic effects have largely been captured through a simple dependency ratio (e.g., Masson and Tryon 1990, Yashiro and Oishi 1997). As a result, these models tend to lack the level of detail needed to analyze the pros and cons of structural pension reform measures.²

8. By contrast, a more pragmatic strand of the literature has focussed on the implications of demographics for fiscal and pension policies, while largely neglecting the feedback between tax and spending policies and macroeconomic variables. A number of papers have projected government accounts on the basis of fixed assumptions for future output growth and interest rates. This has proved useful in illustrating the dimension of fiscal adjustments needed, but the results have suffered from the drawback that different policy measures may affect saving and investment incentives, labor supply decisions, and relative prices in quite different ways. For example, the question of whether to finance social security transfers through direct or indirect taxes could hardly be answered in such a framework. With that caveat, the results of a number of studies are summarized below. Although not necessarily comparable, they shed light on Japan’s fiscal challenge from different angles:

• The Ministry of Health and Welfare (MHW) regularly updates actuarial calculations to estimate the degree to which pension contributions would need to be raised to maintain solvency of the main components of the public pension scheme (see Takayama 1998). The latest results—taking account of the recent pension reform—indicate that contribution rates to the main public pension system will still need to increase by about 50 percent to maintain financial viability (see below).

• Chand and Jaeger (1996) estimated the size of pension liabilities for major industrial countries through 2050, based on World Bank demographic projections (Bos et ah 1994), albeit with a somewhat simplified social security model. As of 1995, the present value of Japan’s net pension liabilities (through 2050) was estimated around 110 percent of GDP, roughly at par with Germany and France. This result is within the range of Leibfritz et al. (1995) and Roseveare et al. (1996), who estimated the value of net pension liabilities for Japan between 50–200 percent of GDP, depending on different productivity and interest rate assumptions.

• A number of papers have followed in the tradition of Auerbach and Kotlikoff (1987) by applying generational accounting models to analyze the distributional consequences of fiscal policies. Takayama et al. (1998) and Takayama and Kitamura (1999) have identified large generational imbalances in Japan, with future generations expected to pay about 3–4 times more net taxes and social security contributions than the generation currently in retirement. Similar conclusions have been drawn by Hviding and Merette (1998).

Approach of this Paper

9. In this paper, both strands of the existing literature are effectively combined. Policy assumptions affecting general government and social security finances are first fed into a small fiscal model which—under given macroeconomic assumptions—provides a long-term projection of fiscal accounts.³ The results of the fiscal model are then fed into Multimod, which produces revised macroeconomic assumptions, and the process is repeated until a reasonable degree of convergence is achieved. This approach is used to generate a baseline model, which illustrates the likely fiscal and macroeconomic consequences of current government policies. Once a baseline scenario has been established, Multimod is used to analyze the effects of alternative fiscal policies relative to those assumed in the baseline simulation.

View Full Size

• Download Figure
• Download figure as PowerPoint slide

10. For the purpose of this paper, the standard Multimod framework was modified in two key points (see Chapter I by H. Faruqee for a fuller explanation).⁴ First, the saving and consumption equations of the Japan model were disaggregated for different age cohorts, using age-earnings profiles from official survey data. This prepared the ground for a more realistic simulation of the saving behavior of an aging population, compared to earlier studies that worked largely through the age-dependency ratio. Second, with retirement income a key factor in life-cycle models, behavioral equations for private agents have been updated to include social security parameters, which permits modeling the effects of social security policies on the real economy. For computational purposes, the Multimod framework also needed to be reduced to four major country blocks—Japan, the U.S., other industrial countries, and emerging and developing countries—which nevertheless still allowed for the simulation of capital flows generated by a shift in the distribution of saving and investment as a result of global aging.

11. There are however two kinds of limitations to the approach of this paper:

• From a conceptual point of view, the analysis focuses largely on the public finance sustainability angle of the aging problem, and does not explicitly search for an optimal intertemporal solution that distributes the financial burden in an equitable way across generations. Moreover, the role of private pension arrangements (e.g., corporate pension schemes and 401(k)-type pension plans) is not explicitly covered by this paper.

• On a technical note, the link between Multimod and the fiscal model is not seamless. Multimod has been developed primarily for analyzing the effects of macroeconomic shocks in a forward-looking/rational expectations framework, and does not lend itself easily to simulating actual levels of economic variables. Moreover, it is already a rather complex framework and adding a fully specified fiscal sector would have made the simulation unwieldy. Therefore, some macro parameters in the fiscal model were simulated independently, although projections for key variables, such as growth and saving, were obtained directly from Multimod.

The Social Security System

12. Japan’s social security system is a multi-tier system that covers public pensions, health insurance, and unemployment insurance (Figure II.I).⁵ First-tier schemes (the National Pension and National Health schemes) provide pension and health coverage on an equal basis for all residents. The second tier consists of occupational schemes—separate for private employees and public servants—that provide earnings-related pensions and expand somewhat on basic health benefits. There are also possibilities to take out higher insurance through supplementary (third-tier) schemes that are privately financed and managed (but benefit from a number of tax exemptions), and are typically offered by larger employers.

View Full Size

Japan: Social Security and Welfare Expenditure by Revenue, Scheme and Category, FY1997

(In trillions of yen)

Citation: IMF Staff Country Reports 2000, 144; 10.5089/9781451820591.002.A002

Source: Ministry of Health and Welfare.

• Download Figure
• Download figure as PowerPoint slide

Pensions

13. The National Pension scheme provides a flat-rate “basic” pension for all residents of age 65 and older, independent of past labor force status.⁶ Additional earnings-related pensions are provided from age 60 by the Employees’ Pension Insurance (EPI) for the bulk of private-sector employees, and by Mutual Aid Associations (MAAs) for public-sector workers. Benefits drawn under these schemes depend primarily on past monthly wages and the duration over which contributions were paid. Benefit increases are presently linked to the disposable income of active workers, with adjustments in the pension formula taking place every five years. During interim years, pensioners only receive cost-of-living adjustments. Contrary to most other industrial countries, pension benefits are virtually untaxed. The maximum replacement ratio for workers covered by earnings-related schemes is around 80 percent of net wages (Takayama 1998), which is high by international standards, although the average net replacement ratio for the elderly population as a whole is only around 53 percent (Yashiro et ah 1997). Compared to some of the larger European welfare states, pension expenditure in Japan is indeed still fairly low, reflecting limited coverage of the earnings-related pension scheme and the comparatively low level of basic pension benefits (Table II.2).

Table II.2.

Public Pension Expenditure, 1995

(In percent of GDP)

Source: OECD (1997).

¹FY 1998.

Table II.2.

Public Pension Expenditure, 1995

(In percent of GDP)

Canada	France	Germany	Italy	U.K.	U.S.	Japan1
5.2	10.6	11.1	13.3	4.5	4.1	7.3

Source: OECD (1997).

¹FY 1998.

View Table

Table II.2.

Public Pension Expenditure, 1995

(In percent of GDP)

Canada	France	Germany	Italy	U.K.	U.S.	Japan1
5.2	10.6	11.1	13.3	4.5	4.1	7.3

Source: OECD (1997).

¹FY 1998.

14. Pensions are financed from three sources, the largest of which are pension contributions. Persons not covered by any occupational schemes (the jobless, self-employed, farmers, students, and their spouses) are required to pay a flat-rate contribution of ¥13,300 per month to the National Pension scheme. Participants in earnings-related schemes pay a fixed share of monthly basic wages in contributions, which is matched by employer contributions.⁷ To finance basic pension benefits for members of occupational pension systems, the National Pension Insurance receives financial transfers from the EPI and MAAs in proportion to the number of beneficiaries that are covered by these schemes.

15. The other two financial sources consist of government transfers, which are currently set at one third of basic pension payments, and returns on public pension assets. Owing to large surpluses in the past, assets held by pension schemes have accumulated to about 50 percent of GDP—about 6 years worth of pension benefit payments, which is unparalleled among industrial countries (for comparison, public pension assets of the U.S. account for about 10 percent of GDP). These assets are largely invested through the Fiscal Investment and Loan Program (FILP), which provides loans to central and local governments, government financial institutions and public agencies (largely for the purpose of infrastructure investment), and holds large amounts of government securities. Owing to the nature of such investments, returns on pension assets have been comparatively low, dropping to below 4 percent in FY1998 from around 6–7 percent in the 1980s.

16. The authorities have taken steps to improve the solvency of the social security system. The recent pension reform bill (approved by the Diet in March 2000) contained provisions to cut lifetime pension benefits by around 20 percent for future retirees, particularly through a reduction in benefit levels by five percent for future retirees, a gradual increase in the age of eligibility for earnings-related pension payments to 65, beginning in 2013, and fully indexing EPI pension increases to the consumer price index instead of disposable income. At the same time, government transfers to the basic pension scheme are to be raised from one third to one half of basic pension benefits from 2004. Nevertheless, according to MHW calculations, the EPI contribution rate would still have to rise to 25 percent of monthly basic wages by 2025 (from the current level of 17¼ percent) to maintain solvency.⁸ Even with such an increase, EPI reserves would shrink to around 3½ years worth of benefit payments.

Health insurance

17. Health insurance coverage in Japan is universal, based on an organizational structure that is similar to the pension system. The National Health Insurance (NHI) scheme provides basic benefits for all residents, and there are numerous occupational schemes (including the Under the pension reform, the wage base for pension contributions is to be broadened to include bonus payments (which account for an average 20 percent of total worker compensation), but the contribution rate will be reduced to keep the change revenue-neutral. Employee’s Health Insurance, health insurance schemes run by the MAAs for public employees, and company-managed schemes mainly for employees of larger companies). Elderly persons (from the age of 70) are covered separately by an old-age insurance system.

18. Unlike the pension insurance, first- and second-tier health benefits are very similar. Japan has adopted a fee-for-service system, with most of the medical services provided by private hospitals and clinics. Access to providers does not depend on the type of insurance coverage, and insurance schemes differ only slightly with respect to required co-payments. Members of occupational schemes are typically reimbursed for 80 percent of medical expenses, while NHI members receive only 70 percent. Medical care for the elderly is almost fully covered, requiring only very small co-payments. Costs for medical services are regulated by a medical fee schedule (which is revised every two years), and drug prices are also standardized, with drug expenses being reimbursed in line with other medical expenses.

19. Compared to other major industrial countries, health expenditure in Japan has been fairly low relative to GDP (Table II.3). Although the system exerts only limited control on demand, and private suppliers have incentives to provide excessive quantities of service, health spending is on a level comparable to countries with nationalized health services, such as the U.K. While underlying health factors that have contributed to Japanese longevity probably also imply less need for medical services, strict price controls and increases in co-payments in recent years have also been effective in keeping overall expenditure low (Oxley et al. 1995). This is likely to change, however, as the Japanese population grows older. With persons over 65 consuming roughly four to five times as much in health services as younger persons, a strong rise in health expenditures can be anticipated.

Table II.3.

Public Health Expenditure, 1995

(In percent of GDP)

Source: OECD Health Data, 1997.

Table II.3.

Public Health Expenditure, 1995

(In percent of GDP)

			Canada	France	Germany	Italy	U.K.	U.S.	Japan
Total health expenditure			9.6	9.8	10.4	7.7	6.9	14.2	7.2
	Of which:
		Public expenditure	6.9	7.7	8.2	5.4	5.9	6.6	5.7

Source: OECD Health Data, 1997.

View Table

Table II.3.

Public Health Expenditure, 1995

(In percent of GDP)

			Canada	France	Germany	Italy	U.K.	U.S.	Japan
Total health expenditure			9.6	9.8	10.4	7.7	6.9	14.2	7.2
	Of which:
		Public expenditure	6.9	7.7	8.2	5.4	5.9	6.6	5.7

Source: OECD Health Data, 1997.

20. Financing of health benefits is provided to a substantial extent by public transfers, which differ across various schemes. On average, subsidies cover about half of total benefits for the National Health Insurance, one third of old-age medical care spending, and 12 percent of expenditure by occupational schemes—which translates into an average of 30 percent for public health expenditure as a whole. Since elderly people are exempt from health insurance contributions and pay only small co-payments, other insurance schemes together provide 70 percent of financing for old-age medical benefits through financial cross-transfers.

The State of Government Finances

21. The situation of Japan’s public finances has worsened dramatically during the 1990s. Tax revenues receded strongly after the collapse of the asset price bubble in the early 1990s, and expenditures have been driven upward by economic stimulus measures, above all public works spending, and by the costs of assuming the liabilities of a number of failed financial institutions (Mühleisen 2000). As a result, the general government deficit has reached an estimated 9 percent of GDP in FY1999 (excluding the social security surplus), and even if Japan’s cyclical position is taken account of, the structural deficit remains worse than that of other G-7 countries (Table II.4).

Table II.4.

General Government Finances, 1999

(In percent of GDP)

Source: IMF World Economic Outlook; April 2000; staff estimates.

¹Excluding social security.

Table II.4.

General Government Finances, 1999

(In percent of GDP)

	Canada	France	Germany	Italy	U.K.	U.S.	Japan
Actual balance	2.8	-1.8	-0.71	-1.9	0.3	0.01	-9.21
Structural balance	3.3	-0.8	-0.71	-0.5	0.1	0.21	-8.11
Gross debt	88.1	58.6	61.1	114.9	44.8	62.4	125.4
Net debt	56.7	49.0	52.4	108.8	39.0	50.6	38.1
Net debt, excluding social security	…	…	53.1	…	…	59.1	879

Source: IMF World Economic Outlook; April 2000; staff estimates.

¹Excluding social security.

View Table

Table II.4.

General Government Finances, 1999

(In percent of GDP)

	Canada	France	Germany	Italy	U.K.	U.S.	Japan
Actual balance	2.8	-1.8	-0.71	-1.9	0.3	0.01	-9.21
Structural balance	3.3	-0.8	-0.71	-0.5	0.1	0.21	-8.11
Gross debt	88.1	58.6	61.1	114.9	44.8	62.4	125.4
Net debt	56.7	49.0	52.4	108.8	39.0	50.6	38.1
Net debt, excluding social security	…	…	53.1	…	…	59.1	879

Source: IMF World Economic Outlook; April 2000; staff estimates.

¹Excluding social security.

22. Although Japan’s net general government debt still appears relatively small (at 40 percent of GDP in 1999), this figure includes assets owned by the social security system—accounting for 50 percent of GDP—which are more than offset by future pension claims. If social security assets are excluded, Japan’s debt situation is considerably worse than that of other industrialized countries. General government debt (excluding social security) has roughly doubled over the past decade, and markets have been concerned that the true level of public debt could be even higher, given the possibility that the government might have to cover contingent liabilities from loan guarantees and losses by public sector enterprises.

23. The current fiscal situation is clearly unsustainable and calls for strong adjustment measures in coming years. However, the problem for the authorities lies in reducing the fiscal deficit in a way that will not jeopardize the nascent economic recovery. In 1997, the last attempt at fiscal consolidation had to be abandoned as a rise in the consumption tax rate and a cut in public works spending were among the factors contributing to the renewed economic downturn. Although factors other than fiscal policy were at work on that occasion, the principle that fiscal consolidation needs to take account of evolving economic and financial conditions—including the robustness of the recovery—remains a valid one.

Demographic and Macroeconomic Background

25. The following simulations use demographic projections prepared by a World Bank team (Bos et al 1994) that also have been widely used in earlier studies. The projections imply that Japan’s population will reach its maximum of 128 million people in 2010. The old-age dependency ratio would plateau at around 45 percent between 2020 and 2030, before rising again to peak at 55 percent by 2050.¹⁰ More recent population projections, including by the Japanese authorities and the U.N., have assumed an even faster decline in fertility rates, which leaves the Bank’s projections at the optimistic end of the current forecast spectrum (Figure II.2). The projections have been retained, however, in part because they are based on the assumption that the population will converge toward a steady state, which is a necessary condition for undertaking the Multimod simulations. However, to avoid painting a too optimistic picture in the simulations, the age distribution has been slightly shifted towards the elderly (by assuming extended longevity), approximating the MHW’s old-age dependency ratio while maintaining population levels projected by the World Bank team.

View Full Size

Demographic Projections

Citation: IMF Staff Country Reports 2000, 144; 10.5089/9781451820591.002.A002

• Download Figure
• Download figure as PowerPoint slide

26. The overall macroeconomic outcome of the simulation is described in Chapter I. The results—which also incorporate the fiscal baseline assumptions explained below—show that, although Japan’s long-term saving rate would remain broadly unchanged compared to a stable demographic scenario, the output loss resulting from a shrinking labor force could still reach 15–20 percent in the steady state. The finding of a broadly unchanged saving rate is related to rising longevity, which causes individuals to save more in their productive phase, and a reduction in younger workers, which reduces the share of the population that dissaves. Finally, owing to rising demand for funds for capital investment, real interest rates would gradually rise toward a long-term level of around 3 percent.

27. While significant, the output loss projected by this simulation would only be a relative loss (compared to a scenario without a demographic shock), which would not necessarily imply absolute negative output growth in the future. Although the labor force would shrink significantly and higher labor force participation would be discouraged by rising wage deductions from taxes and social security contributions, this would be mitigated by an increase in the capital intensity of the production process (the baseline model predicts that the capital-labor ratio would roughly triple by 2050), and positive total factor productivity growth (the baseline scenario assumes TFP growth of about ¾ percent per year). However, even under relatively strong investment growth, potential output growth would peak at around 2 percent over the next decade and then decline over time towards its steady-state rate of ¾ percent over the next 50 years. Owing to a shrinking population, per-capita growth would average about 1 percent over that period.

Baseline Projection: Social Security Finances

28. The model for projecting the finances of social security is complicated by the large number of different schemes and the complex system of financial transfers (see Annex for details), but the simulation is based on a few main assumptions:

• Pension payments are modeled to increase largely in line with consumer prices and the number of retirees. EPI pension cuts would be implemented as foreseen by the recent pension reform bill, and other earnings-related pension benefits would follow a similar pattern.

• Pension contributions. As envisaged by the government, flat rate contributions to the National Pension scheme would increase by ¥800 in real terms every year between 2005 and 2020. Similarly, EPI pension contribution rates would initially rise in 5-year steps, reaching 27½ percent of monthly wages in 2025, although two further rounds of increases are assumed to bring the rate to 28½ percent by 2050, and 30 percent thereafter, to limit the drawdown of pension assets.¹¹ Contribution rates to other pension schemes would rise in line with the EPI.

• Health spending. Rising longevity, more intensive treatment options, and a possibly growing demand-supply gap are estimated to push up real per-patient medical expenditure almost twofold over the next decades, absent a determined effort to contain health care costs. The introduction of nursing care insurance and an increase in co-payments could somewhat dampen the upward trend in medical expenses per elderly person, but these are nevertheless expected to almost double over the forecasting period. Health care contributions are assumed to rise to the extent necessary to maintain financial balance of the health insurance system, as they have in the past.

29. On the basis of these assumptions, the social security system is expected to remain solvent throughout the projection period (Figure II.3, Table II.5). Overall pension expenditure would stabilize at around 12 percent of GDP by 2020 (up from TA percent in 1998), partly owing to a slower increase of real per-retiree benefits as a result of the planned increase in the EPI retirement age. With the planned contribution rate increases, assets held by the pension schemes are projected to remain stable relative to GDP through 2020. Although reserve ratios (assets over annual benefit payments) would temporarily fall during this period, a further increase in contribution rates by 2025 would ensure sufficient asset accumulation to maintain solvency through the peak of the aging process in mid-century.

View Full Size

Japan: Social Security Operations, 1970–2070

Citation: IMF Staff Country Reports 2000, 144; 10.5089/9781451820591.002.A002

Source: Economic Planning Agency, National Income Accounts; and stuff calculations.¹ Total revenue divided by aggregate wage payments.² Benefits received by the elderly are divided by the number of persons aged 65 and above.

• Download Figure
• Download figure as PowerPoint slide

Table II.5.

Projected Social Security Operations, 1998–2050

(in percent of GDP)

Source: Economic Planning Agency; and staff projections.

Table II.5.

Projected Social Security Operations, 1998–2050

(in percent of GDP)

		Act.	Proj.	Proj.	Proj.	Proj.	Proj.
		1998	2010	2020	2030	2040	2050
Premium revenue		10.9	12.4	14.4	14.6	14.4	13.9
	o/w: Pensions	6.9	8.0	9.2	8.9	8.3	7.9
	o/w: Health insurance	3.7	4.2	4.9	5.4	5.8	5.7
Government transfers		2.8	4.4	5.6	6.1	7.0	7.6
Property income		1.8	1.7	2.1	2.5	2.9	2.5
Benefit payments		12.9	16.3	20.0	20.6	22.8	23.7
	Of which: Pensions	7.3	9.4	11.6	11.5	12.6	13.0
	Of which: Health insurance	5.1	6.5	8.0	8.9	10.0	10.5
Other expenditure		0.5	0.5	0.5	0.5	0.5	0.5
Balance		2.1	1.7	1.6	2.1	0.9	-0.2
Memorandum item:
Social security assets		50.0	49.6	51.2	59.8	64.9	57.8

Source: Economic Planning Agency; and staff projections.

View Table

Table II.5.

Projected Social Security Operations, 1998–2050

(in percent of GDP)

		Act.	Proj.	Proj.	Proj.	Proj.	Proj.
		1998	2010	2020	2030	2040	2050
Premium revenue		10.9	12.4	14.4	14.6	14.4	13.9
	o/w: Pensions	6.9	8.0	9.2	8.9	8.3	7.9
	o/w: Health insurance	3.7	4.2	4.9	5.4	5.8	5.7
Government transfers		2.8	4.4	5.6	6.1	7.0	7.6
Property income		1.8	1.7	2.1	2.5	2.9	2.5
Benefit payments		12.9	16.3	20.0	20.6	22.8	23.7
	Of which: Pensions	7.3	9.4	11.6	11.5	12.6	13.0
	Of which: Health insurance	5.1	6.5	8.0	8.9	10.0	10.5
Other expenditure		0.5	0.5	0.5	0.5	0.5	0.5
Balance		2.1	1.7	1.6	2.1	0.9	-0.2
Memorandum item:
Social security assets		50.0	49.6	51.2	59.8	64.9	57.8

Source: Economic Planning Agency; and staff projections.

30. Public health spending would rise from 5 percent in GDP in 1998 to 8 percent of GDP in 2020 and to more than 10 percent of GDP by 2050 (comparable to OECD 1997 projections). Health insurance contributions would have to increase to keep up with rising public expenditures. From the current level of around 8 percent of wages, contributions would be expected to rise to 11 percent over the next 25 years, before peaking at 13 percent by the end of the forecasting period.

31. The impact of population aging on the fiscal accounts is apparent from the fairly drastic increase in government transfers implied by these projections. Under present formulas, transfer payments would rise from currently 2¾ percent of GDP to 514 percent of GDP in 2020 and more than 6 percent of GDP in 2050, by which time the bulk would be spent on health benefits. While these transfers would ensure that the overall balance of social security would remain positive for most of the projection period—preventing a drawdown of assets until the demographic peak around 2050—they would clearly impose a heavy burden on the overall fiscal situation.

Baseline Projection: General Government Finances

32. The simulations for general government finances are guided by a gradual approach to fiscal consolidation, represented by two operational policy targets:¹²

• Stabilize net debt within 10 years. The simulation assumes that the government would embark on a sustained path of fiscal consolidation to stabilize net debt (excluding social security assets) at around 120 percent of GDP by 2010.

• Reduce debt levels gradually in subsequent years. The pace of overall fiscal consolidation is assumed to slow once the demographic transition fully gets underway after 2010. Nevertheless, further measures are assumed to be implemented in order to offset rising social security transfers and avoid substantial crowding-out of private investment over the long-term. It is assumed that general government debt (excluding social security assets) would be reduced to roughly 60 percent of GDP by 2070 compared to 120 percent in 2010 and almost 90 percent of GDP today.

33. The first policy target implies that the general government deficit (excluding social security) would need to be reduced from 9½ percent of GDP in FY2000 to 2½ percent of GDP within 10 years (Figure II.4). To achieve this reduction of 7 percentage points, it is assumed that the government would pursue the following strategy:

• It is likely that public works spending would be substantially cut following the onset of recovery, given the questionable return from such investment in the past. Public investment is assumed to fall by 3½ percent of GDP to reach the G-7 average of 4 percent of GDP by 2010. The withdrawal of stimulus would be somewhat mitigated by a Ricardian reaction of the private sector (which also appears to have limited the effectiveness of such policies in recent years).

• While interest payments and social security transfers would rise by a combined 2½ percent of GDP, spending cuts as well as administrative reforms currently underway could affect other current expenditure (e.g., payroll spending, employment subsidies), although the room for overall reductions is limited due to rising costs for public servants’ pensions which are borne directly by the government. Savings in that area are projected at 1 percent of GDP, which would leave other current expenditure higher relative to GDP than during most of the 1990s.

• On the revenue side, direct tax collections could be boosted by a cyclical recovery in income tax revenues (although a pickup in corporate tax revenue would lag for several years, owing to loss carry-forwards) and by base broadening measures. Although measures to expand the tax base could perhaps yield as much as 10 percent of GDP in revenue collections according to OECD estimates, the baseline only assumes additional revenue of 1-1½ percent as a result of such efforts, owing to potentially large political resistance. Overall, direct tax revenue is assumed to rise by about 3 percent of GDP through 2010, which would be comparable to revenue levels in the mid-1980s.

View Full Size

Japan: General Government Operations, 1970–2070

Citation: IMF Staff Country Reports 2000, 144; 10.5089/9781451820591.002.A002

Source: Economic Planning Agency, National Income Accounts; and stuff calculations.

• Download Figure
• Download figure as PowerPoint slide

34. Under these assumptions, the increase in indirect tax revenue required to achieve the debt target would be around 2 percent of GDP, which would be equivalent to a hike in the consumption tax rate from 5 percent to 10 percent by 2010. In view of the 1997 experience, this would be a rather significant adjustment, that would need to be phased in carefully over the full 10-year horizon.

35. Achieving the assumed long-term policy target—a reduction in net general government debt (excluding social security) to about 60 percent of GDP by 2070—would require additional adjustment measures of 4 percent of GDP. This is assumed to be achieved largely through further indirect tax increases. First, following the adjustment measures through 2010, there appears to be limited room for further expenditure cuts, especially as welfare obligations are likely to remain high. Second, income tax measures would be made more difficult by the continued rise in social security taxes. Indeed, income tax revenues are likely to decline somewhat relative to GDP, owing to a drop in the overall share of labor income, an increase in the share of older workers in the labor force, and increases in alternative work arrangements (such as part-time work). As a result, indirect tax revenue would need to steadily increase from 10 percent in 2010 to a peak of around 14 percent in 2050, consistent with a rise in the consumption tax rate to 24¼ percent, which would be larger than today’s VAT levels in other major industrial countries.

Sensitivity Analysis

36. For obvious reasons, the results of the baseline scenario are at best indicative. Given the long forecasting period, the simulation outcome is highly sensitive to the underlying assumptions, and the focus should thus be more on broad trends in macroeconomic variables rather than a single scenario itself. To help in assessing such trends, the results of two different simulations are presented in Table II.6, involving one scenario with more pessimistic demographic projections, and a second one with stronger productivity growth:

• Weaker demographics. The population projection by Bos et al. (1994) used in the baseline scenario projects a recovery in the fertility rate to 1.6 or more over the next 25 years, compared to its 1999 level of 1.34. However, the birth decline in recent decades appears to have been mostly related to an increase in the marriage age as more women enter into career streams, and there is little evidence so far to suggest that a reversal in this trend is about to occur (Yashiro et al. 1997). Therefore, this scenario assumes the fertility rate to stabilize at around 1.3 percent, which would result in a continuous population decline toward the steady-state level of 80 million, and an increase in the age dependency ratio to 80 percent by 2050 (20 percentage points higher than in the baseline model).

• Technology change. A second scenario assumes TFP growth at 1 percent, or 20 basis points higher than the historic average used in the baseline scenario. As the Japanese economy is likely to become more capital intensive, particularly through investment in new technologies, and with a gradually advancing structural reform process, it would not be unrealistic to expect TFP growth to accelerate in the future.

Table II.6.

Sensitivity Analysis: Social Security Indicators, 2050

(In percent)

Source: Staff calculations.

Table II.6.

Sensitivity Analysis: Social Security Indicators, 2050

(In percent)

		Baseline	Weaker Demographics	Higher TFP Growth
Real benefit levels per beneficiary (2000 = 100) Pension (average)		126.3	108.6	130.1
	General health insurance	178.3	162.1	180.1
	Old-age medical care	196.6	168.4	196.4
Contribution rates
	Pensions (EPI)	28.5	35.0	28.2
	Health (average)	13.0	15.1	12.3
Government transfers to social security (percent of GDP)		7.6	8.4	6.4
	Consumption tax rate	24.6	29.4	17.8
Memorandum item:
Average per-capita growth rate, 2000–50		0.9	0.8	1.1

Source: Staff calculations.

View Table

Table II.6.

Sensitivity Analysis: Social Security Indicators, 2050

(In percent)

		Baseline	Weaker Demographics	Higher TFP Growth
Real benefit levels per beneficiary (2000 = 100) Pension (average)		126.3	108.6	130.1
	General health insurance	178.3	162.1	180.1
	Old-age medical care	196.6	168.4	196.4
Contribution rates
	Pensions (EPI)	28.5	35.0	28.2
	Health (average)	13.0	15.1	12.3
Government transfers to social security (percent of GDP)		7.6	8.4	6.4
	Consumption tax rate	24.6	29.4	17.8
Memorandum item:
Average per-capita growth rate, 2000–50		0.9	0.8	1.1

Source: Staff calculations.

37. The results show that the financial consequences of weaker demographics could indeed be severe, and they also highlight the importance of growth for preserving Japan’s social security system. Under more pessimistic population projections, both benefit cuts and further contribution increases would be necessary (relative to the baseline), especially in the pension system which depends to a lesser degree on government transfers. Even if pension benefits would be kept almost unchanged from their current levels in real terms, the EPI contribution rate would need to be raised to 35 percent. Similarly, despite lower real health care benefits, medical contribution rates would be higher than in the baseline, and the consumption tax rate would also need to be increased to achieve the assumed long-term fiscal target. However, the high consumption tax rate of 29 percent indicates that absent a more severe effort to curb medical expenditure (which would imply lower government transfers), it would become almost impossible to finance public health insurance in its current form. By contrast, higher TFP growth would imply that the benefit levels of the baseline scenario could be maintained at reduced contribution levels as well as lower government transfers, implying that the consumption tax rate could also remain at significantly lower levels than in the baseline model.

Speed and Composition of Fiscal Consolidation

39. One option for the government would be to achieve debt stabilization faster than the 10-year period assumed in the baseline scenario. While the benefits of such a policy are obvious in theory—including positive debt dynamics and higher public saving—there could be a drawback if strong fiscal adjustment measures stalled the recovery, making eventual fiscal consolidation all the more difficult to sustain. Moreover, multiplier analysis suggests that cuts in public investment could have particularly strong negative effects on growth, whereas tax measures would tend to affect output on a smaller scale. Thus, there is not only a question regarding the speed of fiscal adjustment, but also the composition of fiscal measures.

40. To illustrate the macroeconomic trade-offs, three simulations have been run under the assumption that general government debt is stabilized over a period of 5 years. Compared to the baseline scenario, the stronger fiscal path is assumed to result from the following policy measures: (i) cuts in public investment; (ii) an increase in the consumption tax rate; and (iii) direct tax base broadening.¹³ Faster consolidation would stabilize debt at a lower level than in the baseline, and it is assumed that this positive debt differential is maintained through the end of the forecasting period. All other assumptions (regarding social security parameters, technology, etc.) remain unchanged from the baseline scenario.

41. As expected, faster consolidation would result in sustained gains to economic output over the medium-term, although at the cost of short-term losses (Figure II.5). Real output would decline relative to the baseline scenario for the first 3–4 years of the adjustment The case of higher direct tax rates was also considered, but produced an inferior outcome compared to both consumption tax increases and base broadening measures. process, but rise above baseline output for the rest of the forecasting period. The largest long-term gains would be achieved in the case of public investment cuts, but these would also entail the largest short-term output losses, with real GDP being 1¼ percentage points below baseline output in the first year of the adjustment process.¹⁴ The cases of tax base broadening and consumption tax increases would imply a much smaller initial output loss as public investment would be kept unchanged, but long-term gains would also be smaller, owing to the distortionary effects of higher taxes.

View Full Size

Japan: Macroeconomic Effects of Fiscal Consolidation Strategies

(Difference to baseline scenario, in percent)

Citation: IMF Staff Country Reports 2000, 144; 10.5089/9781451820591.002.A002

Source: Staff calculations.¹ Real financial assets plus discounted real income over the projection period.

• Download Figure
• Download figure as PowerPoint slide

42. Long-term gains in real private consumption and private wealth (here used as proxy measures for economic welfare), shown in Figure II.5, also highlight the beneficial role of public investment cuts. Both measures increase immediately relative to the baseline scenario, whereas the other two strategies result in lower consumption for most of the time, as well as significantly smaller wealth gains.¹⁵ This reflects the fact that tax increases either directly reduce disposable labor income (in the case of direct taxes), or decrease future real incomes through an increase in consumer prices (in the case of consumption taxes).

43. Three main policy conclusions can be drawn from these findings:

• A shortening of the fiscal adjustment horizon ceteris paribus implies short-term output costs—suggesting that sharp tightening measures should be held back until private demand has reached sustainable levels.

• The risk to short-term output growth is largest in the case of investment cuts, although these would also generate the strongest long-term welfare gains. From an economic perspective, tax increases would provide a less risky instrument to achieve fiscal consolidation in the initial phase of an upswing. Investment cuts could then be phased in more gradually, depending on the progress of the recovery.¹⁶

• In terms of taxation measures, the simulations indicate that a consumption tax increase would have somewhat less beneficial effects than a broadening of the direct tax base. This highlights the fact that the VAT—although generally considered to be close to an optimal tax—does carry distortionary effects as it introduces a wedge between producer and consumer prices that is captured in Multimod. The model suggests that efficiency losses from base broadening (which could result from lower tax exemptions) are smaller, although only by a small margin.

How to Address the Aging Problem?

44. The debate over social security reform in Japan has focussed mainly on two questions. First, to what extent should future benefits be cut, and second, how to finance the rise in social security benefits associated with population aging? These questions, however, are related as it is generally agreed that significantly higher payroll taxes (as in the baseline scenario) would have undesirable effects on work incentives and labor costs, which indicates the need either to reduce future benefit levels or to examine alternative financing means, including higher indirect taxes:

• Benefit cuts. There have been various suggestions in the literature and in the public arena to cut net benefits, including through cuts in the gross replacement ratio, higher taxation of pension incomes, and an increase in the retirement age. Potentially large savings could be achieved by shifting all remaining pension components from wage to price indexation (including National Pension benefits and entry levels of earnings-related pensions; Yashiro et al. 1997). An increase in the retirement age would also contribute to overall labor supply and output, but this effect is likely to be small as the participation rate of elderly Japanese in the labor force is already quite high.¹⁷

• Higher consumption taxes. Raising indirect taxes would avoid some of the disadvantages associated with higher pension contributions (and health insurance premiums) and spread the burden of financing social security benefits over the general population (including pensioners themselves). Therefore, there have been suggestions to increase transfers from the general government to the pension system, financed through higher consumption tax revenue.

45. This debate is reflected in the following two simulations. First, one scenario assumes a reduction of social security benefits by 2 percent of GDP over a period of 20 years relative to the baseline scenario, which would allow a commensurate decline in social security contributions (for simplicity, the cuts could be in either pension or health benefits, or both). The second scenario keeps social security benefits unchanged relative to the baseline, but assumes that consumption taxes would be raised to finance a reduction in social security contributions identical to that in the first scenario. In this way, the scenarios analyze the trade-off between the level of social security benefits and consumption tax increases that appears to be at the heart of the pension debate in Japan.¹⁸

46. The results suggest that benefit reductions, coupled with a decline in social security contributions, would clearly offer the most beneficial reform alternative with regard to both growth and economic welfare measures (Figure II.6).¹⁹ The chart shows that both simulated reforms would result in higher long-term growth relative to the baseline, although the effects would be larger in the case of benefit reductions. While benefit cuts would imply a short-term output decline (owing to a drop in consumption as forward-looking agents would immediately increase saving for retirement age), the positive long-term effects on output would be substantially larger, owing to a fall in interest rates that would provide a boost to investment.²⁰ In the case of a consumption tax increase, private saving would decline relative to baseline, since the pension financing would be partly shifted from workers to pensioners who generally have a lower marginal propensity to save. Higher growth over the long term would thus mainly be an effect of increased labor supply as a consequence of lower payroll taxes.²¹

View Full Size

Japan: Macroeconomic Effects of Social Security Reform

(Difference to baseline scenario, in percent)

Citation: IMF Staff Country Reports 2000, 144; 10.5089/9781451820591.002.A002

Source: Staff calculations.¹ Real financial assets plus discounted real income over the projection period.

• Download Figure
• Download figure as PowerPoint slide

47. In contrast to benefit cuts, a reduction in social security contributions financed by a consumption tax increase would have mixed effects on private consumption and wealth. The initial decline in saving implies that interest rates rise and that there is less of a buildup in capital stock compared to the baseline (which translates into a valuation loss of financial assets and thus lower wealth). Moreover, the increase in interest rates leads elderly consumers—who are less liquidity constrained—to increase consumption, thus further reducing aggregate saving.²² The resulting decline in the current account surplus also depresses financial wealth through a drawdown of foreign assets. Eventually, the fall in wealth would imply that consumption would drop relative to baseline, although this would only occur towards the end of the projection period.

48. Although the results may depend to some extent on the particular specification of the model, the main policy conclusion is that a reduction of social security benefits would generate lasting output and welfare gains, although at the risk of some small short-term output losses that would need to be minimized through careful phasing. By contrast, financing social security through an increase in the consumption tax would also result in output gains relative to payroll tax financing, but beneficial welfare effects would not be permanent.