2006/7 Research & Review

Simulation Analysis of Japan's Fiscal Policies for Sustainable Public Debt Management - Reexamination of the Broda & Weinstein paper

DOI Takero
Former Faculty Fellow, RIETI / Associate Professor, Faculty of Economics, Keio University

Introduction

One of Japan's most pressing contemporary issues is restoring fiscal health to achieve a primary balance surplus by the early 2010s. The need to curtail government debt outstanding, which has reached to an unprecedented level, is well recognized.

Amid increasing concern over the sustainability of Japan's government debt, Broda and Weinstein (2005) [hereinafter referred to as B&W (2005)] put forward estimates of Japan's future fiscal conditions based on data from the System of National Accounts (SNA). They argued that Japan's government debt, when viewed in net liabilities, is not so large as to be dangerous and can be sustained by securing a viable level of the government revenue-to-gross domestic product (GDP) ratio.

In light of this perspective, this article intends to deepen the discussion on Japan's fiscal management in the future based on Doi (2006) that attempted to reexamine B&W (2005). Doi (2006) is produced as part of a RIETI research project titled, "Empirical Analysis of the Sustainability of Public Debt and Government Bond Management Policies."

Net or gross?

Prior to the reexamination, let me clarify how we define government debt. B&W (2005) views government debt in terms of net liabilities of the government. When discussing the debt balance of the Japanese government, whether the amount of debt should be measured in gross or net terms is often disputed. More specifically, the question raised here is whether we take the net amount (the amount of government liabilities net of financial assets held by the government) or the amount of gross government debt outstanding without offsetting financial assets.

The level of Japan's gross debt is, by far, the highest among industrialized nations. When viewed in net terms, however, Japan's debt level is comparable to Canada and Belgium and not conspicuously high. Based on this, some people even argue that Japan is not in urgent need of fiscal reconstruction.

But then, how should we understand gross and net liabilities? The answers depend on what financial resources the government uses for debt repayment. If the debt is repaid solely from future tax revenue without tapping into revenue on the sale of government-held financial assets, it would be reasonable to take the gross amount of debt outstanding as the government debt level. In this case, the net amount would be meaningless because the financial assets, used in the calculation of such an amount, are held for purposes other than debt repayment and thus revenue on their sale cannot be used for repayment.

For further clarification, the factors complicating the net-versus-gross arguments will be classified as three aspects. With the conclusion first, this is summarized as follows:

  1. Provided that the amount of future social security benefits (obligations) is unknown (or not taken into consideration), it would be appropriate to take the amount of gross government debt as the amount of general government debt that is consistent with future fiscal burden. Although social security funds hold reserves as an asset, it should be viewed in light of the accrual principle, that is, it should be assumed that a matching liability is being accrued in the form of benefit obligations. Indeed, these two - reserves as an asset and benefit obligations as a liability - will offset each other. However, the amount of gross government debt as ordinarily defined does not include the amount of such future social security benefit obligations.
  2. Provided that the amount of future social security benefits (obligations) is reasonably estimable and taken into consideration, the amount of general government debt consistent with future fiscal burden would be equal to the sum of the amount of net government debt outstanding and the estimated amount of future social security benefits (obligations). And this amount is also equal to the amount of gross government debt (which does not include social security benefit obligations).
  3. Of the assets held by the government, those held as matching assets for its liabilities in short-term government securities, i.e. financing bills (FBs), should be offset by liabilities. The other assets, however, include those held as a budget buffer to cope with future fiscal needs. Whether to use these assets for debt repayment in the future is subject to governmental policy decisions. Thus, it is not self-evident whether these assets should be offset by liabilities.

The above ideas can be illustrated with numerical examples for two different time frames, namely, Period I (current) and Period II (future). For the sake of simplicity, both the interest rate and rate of return are assumed to be zero. Also, local governments are omitted from the discussion here and it is assumed that the central government holds no financial assets.

Suppose that the social security funds receive social security premiums worth 20 in Period I and put all the money in reserve toward future social security benefit payments. That is, it was determined as of Period I that a fund worth 20 was to be set aside for social security benefit payments in Period II. Let us further suppose that it was decided (with no other intention) to invest 15 worth of the reserved fund in government bonds and the remaining 5 in equity securities. Let us also assume that the central government (without any other intention) issued bonds with a total face value of 60 to finance 60 in expenditures in Period I and that the social security funds underwrote a portion of these bonds worth 15 in total face value.

Now, let us see what happens to the general government, which is comprised of the central government and the social security funds, at the end of Period I. Looking at the general government's balance sheet as of the term end, the central government's issuance of bonds worth 60 is reported in the liabilities section while the social security funds' reserve worth 20 is recognized in the assets section. Thus, at first glance, it may appear that the net general government debt amounts to 40.

However, this particular 20 in the social security fund reserve is already designated, as of the end of Period I, for the payment of social security benefits in Period II. Therefore, under accrual basis accounting, social security benefit obligations worth that amount must be recognized in the liabilities section.

When this is properly reflected on the general government's balance sheet as of the end of Period I, total assets are 20, unchanged from the aforementioned amount, but total liabilities amount to 80 since 20 in social security benefits obligations, in addition to 60 in bond issuance, must be recognized. The amount of net liabilities calculated by offsetting these total assets and liabilities is 60, which is equal to the amount of gross government debt.

Firstly, gross government debt, as ordinarily defined, does not include social security benefit obligations. Therefore, if offset by the asset (reserve) held by the social security funds, the general government's liabilities (government bonds, in this example) that represent the fiscal burden set to be enforced in the future would be underestimated. The asset held by the social security funds would not help reduce the fiscal burden faced by the central government with respect to the redemption of the bonds issued because the social security funds' reserve is to be used for the future "redemption" of social security benefit obligations.

We will now consider what sort of fiscal burden would be incurred in Period II when the amount of social security benefits to be paid in the future (Period II) is specified. To begin with, it is assumed that the amount of tax burden required in Period II is 60 for the redemption of the bonds issued by the central government in Period I.

The social security funds are able to finance 20 for the payment of social security benefits from the existing assets (reserve) held.

Furthermore, if the exact estimate (20 in this example) of social security benefits in the future (Period II) had been available beforehand, as assumed in B&W (2005) and Doi (2006), the resulting amount of tax burden would be 40 (derived by deducting 20 from 60). This is equal to the amount of net government debt.

As such, the amount of fiscal burden in the future (Period II) is 60. And liabilities held by the general government, regardless of whether defined as 60 in gross government debt (excluding social security benefit obligations) or as a sum of 40 in net government debt and 20 in future social security benefits (obligations), amount to 60, equaling the amount of future fiscal burden. Therefore, it would be appropriate to use the amount of net government debt in combination with the estimated amount of future social security benefits (obligations) as a measurement of government debt only if such future obligations can be reasonably estimated. Based on this, it would be wrong to argue, without specifying the amount of social security benefit obligations to be fulfilled in the future, that Japan's fiscal conditions are not serious because the net government debt-to-GDP ratio is not so high.

Regarding the point discussed in (3) above, some financial assets (e.g. U.S. treasury bonds) are held as matching assets for the government's liabilities in FBs (particularly, those equivalent to former foreign exchange bills). Thus, it is reasonable to have these financial assets offset by liabilities. However, both assets and liabilities held by the Fiscal Loan Fund (for Fiscal Investment and Loan Program: FILP) are excluded because, in its nature, it is a governmental financial institution.

Yet the other financial assets may be those held either by the central or local government as a budget buffer for use at a time when unexpected needs arise (e.g. the fiscal adjustment fund reserved by local governments). That is, whether to use these assets for the purpose of debt redemption in the future, a reduction in future fiscal burden, is subject to the government's policy decisions and it is not self-evident whether these assets should be offset by liabilities.

Thus, Doi (2006) analyzes two possible cases: when financial assets held by the central and local governments are offset with liabilities (net government debt) and when such financial assets are not offset, which, in this article, is called the "case in which proceeds on the sale of government assets are not used for debt redemption."

Future direction of fiscal management

Next, let me introduce the findings from the analysis in Doi (2006), which is based on B&W (2005). For the purpose of this analysis, government expenditures are separated into three categories: 1) debt servicing costs, 2) transfers to the elderly aged 65 or above (provision of public pension benefits and medical benefits), and 3) other government expenditures (for younger generations). A situation is defined as fiscally sustainable if the nation is not presently (in period 0) bankrupt and it is deemed possible to manage fiscal operations in such a way that the ratio of government debt to GDP will return to the current level in period n. In this article, only the results obtained when period n is 2100 are presented.

Then, Doi (2006) calculates the ratio of government revenue to GDP ("tax revenue ratio") required for ensuring fiscal sustainability as defined above. In doing so, it is assumed hypothetically that this ratio can be maintained at the same level every year. Whether future fiscal operations can be managed without undermining fiscal sustainability is verified by examining the feasibility of the tax revenue ratio obtained.

In B&W (2005), conditions surrounding present fiscal management (at the initial point in time) and in the future have been set as follows. The GDP growth rate is assumed to be 2%. However, changing this rate to 0% or 1% would make little difference to the results. Interest rates are set at 0%-4% (0%-2% for the purpose of this article) above the nominal GDP growth rate. Future populations are based on date in "Population Projections for Japan (medium variant)" by the National Institute of Population and Social Security Research.

With respect to government expenditures, three scenarios (of which Case 1 scenario is omitted in this article) have been assumed. Case 2 is a scenario in which both per capita transfers to the elderly and per capita expenditures for the younger generations rise in proportion to GDP. Case 3 assumes that both per capita transfers to the elderly and per capita expenditures for the younger generations rise in proportion to per worker GDP.

In recent years, there has been substantial debate calling for suppression of the rise in social security expenditures. Thus, in addition to those described above, another scenario reflecting the direction suggested in the emerging debate has been analyzed in Doi (2006) in which social expenditures are assumed to rise in proportion to GDP growth adjusted for the effect of population aging ("aging-adjusted GDP growth" proposed by the Council of Economic and Fiscal Policy). This growth is calculated as the sum of the nominal GDP growth rate and the rate of the elderly population increase in a given year to the total population as of the preceding fiscal year. For this additional scenario, called Case 4, transfers to the elderly is assumed to rise at a rate equal to that of aging-adjusted GDP while per capita expenditures for the younger generations rise in proportion to per worker GDP, as in Case 3.

Each table provided herein shows the tax revenue ratios required for ensuring fiscal sustainability when government expenditures and interest rates are set as described above. The top table shows the results provided in B&W (2005) (the "B&W case"). Based on these results, B&W (2005) concludes that Japan's fiscal deficits are not in a critical condition because the resulting tax revenue ratios, though slightly above 32.2%, the average ratio of government revenue to GDP for the period spanning 1980-2000, are not unrealistic and deemed to be comfortably sustainable by implementing a reasonable tax increase in the future.

Tables: Results of estimates
Tax revenue ratios required for fiscal sustainability

Doi (2006) has performed additional simulations to reexamine the conclusion derived by B&W (2005). The results of the reexamination (the "benchmark case") are shown in the second table from the top. The amount of net government debt for this case is the same as that used in the B&W case. The simulations for the benchmark case have almost exactly reproduced the results in the B&W case although the resulting figures of these two cases, as shown in the respective tables, slightly differ due to minor differences in the estimates of future government expenditures.

Furthermore, Doi (2006) has modified some of the assumptions used in the B&W case. First, the base year for estimates is changed to fiscal 2005 so as to reflect the latest fiscal conditions. Consequently, the government debt-to-GDP ratio is somewhat higher and the remaining period to achieve a primary balance surplus is three years shorter than in the benchmark case. At the same time, however, this update also includes the incorporation of the effect of austere fiscal policies under the Koizumi administration. Government expenditures excluding debt servicing costs, which peaked at 35.2% in 2002 when measured as a percentage of GDP, have since fallen to 34.3% in 2005 and this is reflected in the updated estimates of future fiscal expenditures.

This case, for which all the assumptions other than those described above are identical to those in the benchmark case, is called the "updated case." The resulting figures are shown in the third table from the top. For instance, when the interest rate is 4% and fiscal expenditures follow the Case 4 scenario, the tax revenue ratio required for ensuring fiscal sustainability would be 36.2% for the updated case (and 35.5% for the benchmark case).

Next, the second table from the bottom shows the results of the modified updated case, in which assumptions are identical to those of the updated case except that revenue on the sale of government assets is not to be used for debt redemption. The net government debt-to-GDP ratio, which stood at 78.3% as of the end of 2004, is pushed up to 108.9% in this modified updated case. Tax revenue ratios required in this case are generally higher than those for the benchmark case. For instance, when the interest rate is 4% and fiscal expenditures follow the Case 4 scenario, the tax revenue ratio required for ensuring fiscal sustainability would be 36.8%, higher than the corresponding rate in the benchmark case, indicating the need to implement a substantial tax increase from the current level. However, provided that the ratio of primary surplus to GDP is to be maintained around 2% in the first half of the 21st century and then 2%-3.5% in the last half, the net government debt-to-GDP ratio would peak out at around 131% and the situation would be sustainable.

Finally, the impact of delaying the tax increase is analyzed. This "delayed tax increase case," is based on the modified updated case but with further assumptions concerning taxes. Specifically, it is assumed that no tax increase is to be implemented from 2005-2009 and the tax revenue ratio is to be maintained at the 2005 level (30.9%) during that period, and then the tax rate is to be raised from 2010 through 2100.

The results of this final case are shown in the bottom table. Figures presented in this table are higher than those in the second table from the bottom. Although the tax increase is delayed by only five years, this translates into a 1% increase in tax burden measured as a percentage of GDP because fiscal deficits accumulated in the five years would have to be curbed in the period five years shorter than in the updated case.

In this delayed tax increase case, provided that the interest rate is 4% and fiscal expenditures follow the Case 4 scenario, the tax revenue ratio required for ensuring fiscal sustainability would be 37.5%. This would require a substantial tax increase from the current level. Still, it is attainable if the primary surplus-to-GDP ratio can be maintained at around 2.5% in the first half of the 21st century and 2.5%-4.2% in the last half. However, given the relatively high surplus ratios that must be maintained in the last half of the 21st century, it would be uncertain whether the administrative authorities can resist political pressure to increase fiscal expenditures. A delay of only five years may not sound significant but this is what such a delay would entail.

In summary, if revenue on the sale of government assets is not used for debt redemption, the tax revenue ratio would have to be raised to a level almost 1% higher than proposed in B&W (2005). This translates into an additional hike of approximately 2% in the consumption tax rate. The findings from this analysis also suggest that delaying the tax increase in the course of restoring fiscal soundness would result in additional fiscal burden in the future; approximately 1% in the tax revenue ratio, which is equivalent to about 2% in the consumption tax rate. As such, while it is imperative to cut expenditures on unwanted, non-urgent projects, it is also necessary to assure that the timing of implementing a necessary tax increase will not be delayed without good reason.

>> Original text in Japanese

Reference(s)

October 10, 2006