ANNEX I The Fiscal Investment and Loan Program (FILP)21
39. The FILP provides investment financing for public policy purposes.22 Historically, most of the funds for the FILP program came from peoples’ deposits in the Postal Savings system and from the pension system assets. The program has been an important tool for fiscal management—it extends loans to government financial institutions, public corporations, local governments, and several special central government accounts (jointly referred to as “FILP agencies”). In recent years, the largest loan allocations have been for mortgage financing, small and medium businesses, and road construction. While not formally a part of the general government sector, the FILP annual plan is formulated in coordination with the budget process, and is submitted by the Diet together with the general account and special accounts budgets. Owing to its size, the FILP is often referred to as the “second budget”. Total outstanding FILP loans at the end of FY2001 were equivalent to 66.5 percent of GDP (FILP Report 2002).
40. Reform of the FILP was initiated in April 2001, aiming at a gradual alignment of the activities of the program with market principles. The key elements of the reform were to increase the role of direct market financing for FILP agencies and to initiate subsidy cost analysis of FILP projects.23 As part of the reforms, the compulsory transfer of deposits from the postal savings and pension systems to the FILP has been abolished, and the FILP has to finance itself by issuing bonds (which are issued in the same auction and have the same yields as JGBs). For an interim time period of seven years, the postal savings and pension systems are to continue to underwrite FILP bonds, but at successively lower levels. The recipients of FILP funds are required to issue “FILP-agency bonds” directly to the market, and to apply for FILP loans only if sufficient funding cannot be secured in the market. The yields on these bonds have stayed very close to the respective JGB yields, suggesting that the market perceives the FILP-agency bonds to be implicitly guaranteed by the government. To assess the prospective financial implications of the program and to improve its efficiency, subsidy cost analyses of all FILP agencies have been undertaken and published. The most recent official estimate of the future cost of the program to taxpayers is ¥7.5 trillion (FILP Report 2002).
41. The FILP constitutes a key source of government contingent liabilities, recent efforts at reform notwithstanding. Kikkawa et al (2000) find that the FILP agencies have been very optimistic in their revenue forecasts and therefore the estimated subsidy costs probably underestimate the likely losses. Doi and Hoshi (2002) evaluate the balance sheets of FILP agencies and provide estimates (in net present value terms) of expected taxpayer cost related to FILP loans. They estimate the likely losses on loans to government financial institutions and special corporations to be ¥3614 trillion (about 7 percent of GDP).24 These expected losses are the sum of the estimated under-provisioning for nonperforming loans, overvaluation of assets, and subsidy costs. However, Takahashi (“Economic Society Policy”, May 2003, Cabinet Office) has questioned the accuracy of these estimates.
ANNEX II Social Security Projections, 2003–2050
42. The projection model is built around a set of equations for the revenues and expenditures of the pension and medical systems. The “median” official 2002 population projections are used in the simulation—these assume a recovery of the average fertility rate from the current 1.36 to 1.39. Japan’s social security system is very complex and consists of a large number of schemes with different contribution and benefit rules, different shares of government participation, and a complicated system of cross-transfers. Modeling it requires a large number of simplifying assumptions and the projections therefore represent only broad trends in social security finances. As with any long-term projection exercise, the assumptions underlying these projections are subject to great uncertainty.
43. The simulation of the pension system was roughly calibrated against the MHLW’s actuarial projections of the Employment Pension Insurance (EPI) and National Pension systems. Contributions depend on changes in the contribution rate as a share of income and on the rate of increase in employee compensation in the case of EPI and the Mutual Aid Associations, and on the flat contribution amount in the case of the National Pension. Benefit payouts grow in proportion to the growth in the elderly population, and inflation, and partially depend on wages (since the starting level of EPI benefits depend on the average level of the employee’s past salary). An adjustment is made in attempt to account for the 2000 pension reform plan, which called for a 5 percent reduction of benefits for new retirees starting in 2004, and a gradual increase of the pension eligibility age from 60 to 65 (the adjustment is made so that the results broadly follow the MHLW’s pension expenditure projections). Contributions from the government are assumed to remain at 1/3 of the basic pension (this is relevant only for the deficit of the social security system, not for the projection of the general government total net debt, which is independent of variations in the government contribution share).
44. Medical expenditure forecasts reflect the expected increase in per capita expenditures as the population ages. Based on data from the MHLW, per capita medical expenditures for people aged 65 and above are about 5 times higher than per capita expenditure for people younger than 65 and this differential is assumed to remain. The growth rate of the price for medical services over the last decade has exceeded CPI growth by about 1 percentage point in Japan and this is assumed to continue in the future. The actual growth of medical costs will depend on regulatory developments in the pharmaceutical and medical service sectors, and on future changes in the medical insurance system rules. Contributions are assumed to grow in line with employee compensation (proxied by GDP growth). Under the current system, medical benefits for the elderly are financed by transfers from the government and from employment insurance schemes (the elderly make practically no contributions to any insurance scheme, except that under the new long-term care system, elderly insured persons will also have make a small annual contribution). A detailed long-term government projection of the medical system finances under a consistent set of macroeconomic assumptions is not available, but a comparison with published MHLW’s projections for year 2025 show that our medical expenditure forecast is rather conservative.
Alesina, A. and Perotti, R. (1996), “Fiscal Adjustment in OECD Countries—Composition and Macroeconomic Effects”, IMF Working Paper 96/70.
Alesina, A., Perotti, R. and Tavares, J. (1998), “The Political Economy of Fiscal Adjustments,” Brookings Papers on Economic Activity.
Braun A. and Kubota, K. (2000), “The Effect of Government Capital on Labor Productivity in Japan’s Prefectures,” Working Paper Presented at the University of Tokyo.
Callen, T. and Mühleisen, M. (2002), “Developments in the Banking and Life Insurance Sectors,” in Japan—Selected Issues, IMF SM/02/206.
Dekle, R. (2002b), “Population Aging in Japan: Its Impact on Future Saving, Investment, and Budget Deficits,” Working Paper, Department of Economics, USC.
Doi, T. and Hoshi, T. (2003), “Paying for the FILP” in Structural Impediments to Growth in Japan, Blomstrom, Corbett, Hayashi, and Kashyap (eds.), University of Chicago Press.
Faruqee, H. and Mühleisen, M. (2001), “Population Aging in Japan: Demographic Shock and Fiscal Sustainability”, IMF Working Paper 01/40.
Ihori, T. and Doi, T. (2002), “Sustainability Problem,” in Government Deficits and Fiscal Reform in Japan, edited by Ihori and Sato.
Ishikawa, T. (2002), “Population Decrease, Aging, and Japan’s Long-Term Economic Outlook to 2050,” November 2002, NLI Research Institute, Tokyo
Kashyap, A. (2002), “Sorting Out Japan’s Financial Crisis,” Federal Reserve Bank of Chicago Economic Perspectives, 4th quarter 2002.
Kikkawa, M., Takeshi, S., and Miyagawa (2002), “Soundness of the Fiscal Investment and Loan Program,” in Structural Problems of the Japanese Financial System, edited by M. Fukao, Tokyo: Japan Center for Economic Research, p. 41–59.
Takayama, N. and Y. Kitamura (1999), “Lessons from Generational Accounting in Japan”, American Economic Review, 89:2, pp. 171–80.
Yoshino, N. and Nakano, H. (1996), “Interregional Distribution and Productivity Effect of Public Investment,” Financial Review 41, p. 16–26.
Prepared by Dora Iakova (ext. 35365).
The net present value of unfunded pension liabilities was estimated at above 100 percent of GDP (Actuarial Report of the Employee Insurance Program, Ministry of Health, Labor, and Welfare, 1999). No estimate is available for the medical insurance system, although medical expenditure is expected rise much faster than pension expenditure in percent of GDP.
Based on national income accounts data, social security expenditure increased from 9.6 percent of GDP in FY1991 to 15 percent of GDP in FY2001.
As an example, the government had to assume the liabilities of the Japanese National Railway Settlement Corporation (JNRSC) amounting to 5.2 percent of GDP in 1998.
While it is difficult to estimate the size of the needed public injection, one indicative measure is the amount of uncovered loan losses at major and regional banks. Callen and Mühleisen (2002) estimate this amount to be between 1.5 and 5 percent of GDP. Market analysts’ estimates of net loan losses vary between 4 and 14 percent of GDP (Kashyap (2002), Table 2).
IMF, World Economic Outlook (September 2003) finds that the probability of crisis increases when public debt reaches the 25-50 percent of GDP range. Most other studies have looked at how debt crisis probabilities change when external debt-to-GDP increases—or example, IMF (2002) suggest that 40 percent of GDP is a useful threshold and Manasse et al (2003) find that countries with external debt higher than 50 percent of GDP are more likely to experience default episodes.
As an example of the potential strength of market discipline, long-term interest rates recorded their largest monthly increase (105 bps) in the past 20 years in December 1998, when the Ministry of Finance announced the reduction of JGB purchases by the Trust Fund Bureau.
The calculation of the deficit of the central and local governments includes transfers to the social security system, but excludes the deficit of the social security system. In the simulation we assume that the adjustment is achieved by a combination of tax increases and expenditure reductions.
The Cabinet Office used 1 percent potential growth rate for its medium term fiscal estimates (“2002 Reference Estimates”). The Bank of Japan (Quarterly Bulletin, May 2003) also estimates the current potential output growth rate to be about 1 percent. However, a 1 percent real growth rate in the longer term may turn out to be a rather optimistic assumption, given the expected rapid decline in the working age population over the next 50 years. For instance, a significantly more pessimistic scenario is presented by Ishikawa (2002): accounting for the expected decline in the labor force and assuming a total factor productivity growth rate of 0.7 percent, he estimates that real GDP growth will decline to minus 0.4 percent by 2050.
The simulation is done in terms of net debt for analytical convenience since using gross debt would require specific assumptions on asset accumulation. Whether gross or net debt is the focus of analysis depends on the quality and liquidity of government assets. About a third of Japan’s government assets are in the form of liquid bonds and cash deposits. However, a large share of the assets are invested in long-term projects through the FILP system and are highly illiquid. Therefore, most assets may not be easily used to reduce gross debt.
For comparison, the debt-stabilizing adjustment in the primary balance of the general government is higher than the adjustments assumed in the baseline and pension reform scenarios of the last section by 4 percentage points and 2 percentage points of GDP respectively.
Strengthening the institutional framework for fiscal policy implementation can help sustain fiscal consolidation efforts. Fiscal adjustment in many OECD countries in the 1990s was accompanied by institutional reforms, including the formal adoption of fiscal rules, such as expenditure ceilings or balanced budget rules. For fiscal rules to be credible, they may need to be sufficiently flexible to accommodate automatic stabilizers and allow responses to unexpected negative shocks.
Alesina and Perrotti (1996) find mat in OECD countries adjustments that have relied mainly on current expenditure reductions have typically been more successful, and such adjustments have also tended to be expansionary.
See Doi (1998) and Yoshino and Nakano (1996). Kondo (2002) finds that rates of return differ significantly for different types of public capital and are relatively low in road construction and agriculture.
Braun and Kubota (2000) show that there is a strong negative correlation between the per capita income of prefectures and the level of public capital they receive. EPA (1997) estimates that the return to social capital in cities is roughly twice the rate of return in rural areas. The high level of investment in poorer areas has not stimulated self-sustaining growth as hoped. Instead, the areas have become heavily dependent on public construction.
Source: “Annual Report on the Japanese Economy and Public Finance, 2001-2002,” Cabinet Office, Government of Japan, page 165.
The 1998 downturn was widely ascribed to the 1997 VAT rate increase (from 3 to 5 percent), although there is no empirical evidence to that effect. The financial crisis that developed shortly after the VAT increase was a more likely cause of the economic problems. Short-term intertemporal consumption substitution may have also played some role.
Some academics also call for more radical changes of the system, such as moving to defined contribution plans (the public pension system in Japan is a defined benefit plan), or indexing benefits to aggregate macroeconomic variables (to account for the decline in population and possible decline in the potential growth rate). See Takayama (2003). The MHLW has also proposed a version of benefit indexation to macroeconomic variables.
In addition, currently about 37 percent of all participants do not make the required contributions to the National Pension system, and increasing the contribution amount may lead to a rising share of nonpayers.
Most of the increase is due to an expected rise in medical and long-term care expenditures. The finances of all the major schemes in the social security system—medical, pension, long-term care and unemployment insurance—have been deteriorating in recent years and all the schemes are currently either running deficits or are close to running deficits. Only the pension schemes have assets which can be used to cover the deficits. Therefore, unless benefits are reduced, substantial revenue measures would be required to meet rising expenditures.
Sanjay Kalra contributed to this Annex.
The subsidy cost of a project is the estimated total amount of subsidies, financial assistance and grants-in-aid to be provided by the treasury until the project’s completion. Subsidy cost analyses of FILP agencies are published by the Ministry of Finance (2002).
The authors also estimate that local governments would not be able to repay FILP loans equivalent to about 8 percent of GDP. However, these obligations should already be accounted for in the general government liabilities and therefore do not represent contingent liabilities.