We are grateful to Adam Posen for his helpful comments. All errors and opinions contained herein are solely our own, on our responsibility.
Since the asset-price bubble burst in 1990, the Japanese government has repeatedly announced fiscal measures to boost the economy. The government claims that discretionary fiscal spending from fiscal year 1992 to 1996 (although there was no stimulus measure in 1996) amounted to more than 65 trillion yen1: half of that supposed stimulus went to public works, such as construction and infrastructure, and 20 percent of it took the form of tax cuts (see table 1).2 Despite these economic measures, Japanese economic growth has been stagnant for more than 6 years, except for 1996, when the Japanese economy grew by approximately 3 percent.3 Why haven't the Japanese fiscal packages worked? The large part of the answer, which we address in this paper, has been the biggest part of the announced packages, namely, public spending.
|Table 1 Economic Stimulus Packages, 1992-96 (billion yen)|
|Note: Public works includes general public works, new social infrastructure maintenance, public building maintenance, and purchase of land for future public works.|
|a. Local solo represents public works conducted solely by local governments.
b. The tax cut cannot be separated into central and local for each package because of lack of data. Moreover, the numbers for fiscal years show the cumulative effect and do not necessarily match the sum of each package because they depend on when the tax cut was effective.
c. Other includes private investment initiatives such as increases in loans to medium and small firms and housing loans.
|Source: Nikkei Shinbun (various issues); Economic Indicators (various issues), Dai-ichi Life Research Institute.|
Our main finding is that, from 1992 to 1996, real discretionary fiscal spending was merely 21 trillion yen in public works. This is about two-thirds of the officially announced public works spending during the period. Meanwhile, tax revenues declined by about 33 trillion, which is approximately equivalent the total officially announced tax cuts. A closer look reveals that the local governments, not the central government, failed to spend announced discretionary fiscal spending. In addition, there is evidence that much of the public work was geared toward projects of dubious significance. However, the key to solving the mystery of Japanese fiscal packages is the awkward and fettering fiscal relationship between the central and local governments.
The Japanese Fiscal System and Underlying Problems
The Japanese government consists of the central and more than 3,000 local governments. The local government system has 4 layers and includes 47 prefectures, 13 metropolitan cities, 654 cities, and 2,566 towns and villages. By Japanese law, the central and local governments are financially independent, but in reality, the central government tightly controls local budgets.
The Local Finance Law allows the central government to control local governments' revenues (Ishi 1989): local governments must receive approval from the Ministry of Home Affairs to change tax codes or tax rates and to issue local government bonds (Chiho sai). However, the central government does not dictate local governments' expenditures; although many announced economic measures include "public works conducted solely by local governments," the central government can only request that local governments increase their public works spending-it cannot force them to do so. Therefore, there is no guarantee that all allocated funds are actually spent.
This peculiar system leads to a disparity in the local and the central governments' balance sheets and impairs the autonomy of the local government. In 1996, only 143 of more than 3,000 local governments (4.3 percent) were able to balance their budgets without aid from the central government. In all other cases, the central government intervened to make up for the deficit.
In addition, while the central government collected 55 percent of the total government revenue in recent years, it is the local governments that are directed to conduct the most of the spending (see figure 1). This inconsistency made it necessary for the central government to establish tax grants, thereby, impairing the independence of local governments.
Figure 1 Fiscal Connection Between Central and Local
As does the central government, local governments have two types of revenue resources: ordinary and special. The ordinary revenues include local tax revenue (Chiho zei), unconditional tax grants from the central government (Chiho-Kohu zei), and a local shared tax (Joyo zei).4 The shares of these sources in total revenue are 42.5, 19.7, and 1.3 percent, respectively, in 1997. Special revenues include conditional tax grants from the central government (Kokko-Shisyutsu-kin), profits from public sector enterprises, and local government bonds. Respectively, these sources are 15.2, 7.4, and 13.9 percent of total revenue.
There are local tax levies on property as well as personal and corporate income. However, it is the central government, not the local governments, which decide the basic local tax rates, and sets them across regions.
There are roughly six categories of taxes: road, petroleum gas, special tonnage, motor vehicle, aviation fuel, and sales tax. Except for the sales tax, each tax is spent on projects related to the tax category.5 This system allows the central government to match and control the allocation and redistribution of resources.
Unconditional tax grants from the central government are essentially tax subsidies. The amount is calculated by the central government using a predetermined formula that incorporates current economic conditions. The central government estimates the tax revenues and expenditures of local governments for the fiscal year. If the estimated revenues are less than the estimated expenditures, the central government provides subsidies in the form of unconditional tax grants.
In the special tax category, the central government gives conditional tax grants, which are tied to specific types of expenditures such as education, social welfare, and public capital outlays. On average, half of the conditional tax grants are spent on public capital investment.
The central government also controls the local governments' bond issuance. By local the Finance Law, local governments cannot issue bonds without permission from the Ministry of Home Affairs. Furthermore, the law dictates that local bonds should be long-term (more than one fiscal year) and used only for public utility enterprises and capital investment. Before the Ministry of Home Affairs grants permission to issue local government bonds, it carefully reviews the financial position of the local government; if the ministry believes that the local government's financial position is not sound, it does not allow the bonds to be issued. This system leaves local governments financially dependent on the central government and impairs their ability to control their financial position.
Moreover, the majority of local government bonds are held by the Fiscal Investment and Loan Program (FILP) and by local banks that are tightly controlled by the Ministry of Finance. The permission from the Ministry of Home Affairs to issue bonds is a de facto central government guarantee on those bonds. Therefore, a market mechanism to evaluate the costs and benefits of public works does not exist. It is not to hard imagine that the local governments, which were asked to spend more and did not have to worry about the source of revenue, were likely to conduct unprofitable and unnecessary public works.
In addition, the share of public works solely conducted by local government has increased significantly. Figure 2 demonstrates that pubic works conducted by local governments quadrupled in the last 15 years while joint public works and public works conducted by the central government increased only marginally. This means that local governments have been given a heavier burden of public works spending but have not been given control over their own tax revenue.
Figure 2 Public Works by Type
Discretionary Fiscal Spending
To investigate the size of fiscal stimulus, we collected data from 1965 to 1996 on the initial budget, on actual revenue and spending for the central and local governments, and on the supplemental budget for the central government.6
The supplemental budget is usually announced in the Fall and Spring to adjust for tax revenues that fluctuate due to economic conditions and to accommodate discretionary fiscal policy. After the fiscal stimulus package passes the congress, the government (primarily the Ministry of Finance) puts the package into the budget.
To measure the real spending of a fiscal package, we compared the initial budget and actual spending. If the supplemental budget is completely spent within the year, the gap between the initial budget and the actual spending is roughly equal to the size of the fiscal package. Realistically, the supplemental budget cannot completely be spent within the year, and part of it is often forwarded to the next year. However, the forwarded budget is not included in the initial budget of the next year. Therefore, the gap between initial budget and actual spending roughly equals discretionary spending. To avoid the intricacies of a forwarded budget, the comparison was applied only to the cumulative amounts of the packages between 1992 to 1996. Although these gaps also include other cost fluctuations, such as cost of construction and land prices, we assume that such cost fluctuations are random and small enough as to be irrelevant.
Real Fiscal Stimulus
The most interesting result from the above analysis is that while the central government spent as much as it said it would (about 25 trillion yen for 1992 to 1996), the local governments spent much less than the central government projected (about 10 trillion yen less) (see table 2. In fact, the local governments spent 3.7 trillion yen less than their initial budget during 1992-96). That is, the local governments not only failed to spend the supplementary budget (6 trillion excluding the local governments' share of joint projects), but also spent less than their initial plans indicated. The cumulative effect is that the local governments reduced capital spending by about 10 trillion yen from the initially planned spending for 1992-96. Indeed, this 10 trillion yen was equal to the difference between the officially announced increase in public works and the actual increase. According to the government-announced economic measures, public works spending should have increased by 31 trillion yen. The balance sheet shows that the spending increased by only 21 trillion yen.
|Table 2 Actual Impact of Stimulus Packages on Central and Local Government Budget, 1992-96 (billion yen)|
|Actual public expenditure caused by fiscal stimulusa||Changes in tax revenue caused by fiscal stimulus|
|Types of construction|
|Total||Centralb||Total||Solo||Joint||Supportc||Disaster||Total||Central||Local||Tax grantse||Impact from changes in tax coded (central government only)|
|a. Actual public expenditure caused by fiscal stimulus was the difference between actual expenditure and the initial budget.
b. Central government figures exclude subsidies to local government.
c. Support is the local government's share of public works conducted solely by the central government.
d. Figures indicate changes in tax revenue due to changes in tax code and are estimated by the Ministry of Finance and presented in the Monthly Economic Statistics.
e. Tax grants include both conditional and unconditional tax grants from the central to local governments.
|Source: Japan Ministry of Finance (1997); Japan Ministry of Home Affairs (1998).|
Productivity of Public Investment
Another criticism against Japanese economic measures and traditional public works spending is that the quality of public works is low. If the quality of public investment has declined, its impacts should have declined as well.
Many economists have studied the productivity of public investment. Some have suggested that slow economic growth of the United States during the 1970s and 1980s was partly caused by smaller public capital spending (Aschauer 1989). However, in Japan, the Economic Planning Agency (EPA) determined that the multiplier on public spending was about one in the 1990s. Public investment increased by 15 percent from 1992 to 1993, while GDP only grew by 1 percent, underscoring how low the multiplier is.
Although the EPA explains that this occurred because private investment in the same period was significantly lower because of the burst of the economic bubbles, many others suggested that productivity of public investment have decreased. To examine this, we estimated the impact of public investment stock on productivity on private capital (see table 3). The results indicate that public investment stock is not relevant in determining the productivity of private capital from 1970 to 1996. This implies that spillover from public capital investment has been negligible.
|Table 3 Quality of Public Investment: Regression for Private Capital Productivity|
|R2 = 0.986
Adjusted R2 = 0.985
F(2, 24) = 860.25
Number of observations = 27
Dependent variable: Private capital productivitya
|Public capital stockc||0.113||0.859|
|a. Private capital productivity is measured by nominal GDP divided by private capital stock. Using the natural log transformation, it is expressed as ln(GDP) - ln(private capital stock).
b. Labor force is measured by the number of workers normalized by private capital stock. Using the natural log transformation, it is expressed as ln(labor) - ln(private capital stock).
c. Public capital stock is measured by public capital stock normalized by private capital stock. Using the natural log transformation, it is expressed as ln(public capital stock) - ln(private capital stock).
Sources: Ministry of Finance Statistics Monthly (various issues), Ministry of Finance; Japan Annual Statistics (Nihon Tokei Nenpou) (various years), Ministry of Coordination and Arrangement.
The goal of this study was to address the effectiveness of public spending, which has been the largest part of fiscal stimulus packages. Our results indicate that public spending has been largely ineffective because:
1. Japanese local governments have failed to spend announced discretionary spending on public works.
2. The quality of public works spending has been low, that is, public investment did not generate productivity growth.
The underlying causes are the rigid relations between the central and local governments. The following may prove effective in addressing these issues:
1. Create markets for local bonds.
2. Create a system to evaluate potential returns and effectiveness of public investment projects.
3. Promote financial independence and accountability of local governments.
Of course, there are areas that need to be examined further such as inflexible spending patterns. The amount spent on different public works hasn't changed over the last 15 year period. This may provide more evidence for the claim that public works spending has been unproductive. Furthermore, there is a concern over the increase in local governments' debt. Local governments' debt was 103 trillion yen at the end of fiscal year 1996, which is almost twice the size of ordinary tax revenue of the local government (the sum of local tax, local shared tax, and unconditional tax grants from the central government). Consequently, the increase in interest payments dried up the available fiscal fund.
Aschauer, D. Alan. 1989. Is Public Expenditure Productive? Journal of Monetary Economics 23: 177-200.
Auspitz, J. Lee, Wojciech W. Gasparski, Marek K. Mlicki, and Klemens Szaniawski, eds. 1992. Praxiologies and The Philosophy of Economics: The International Annual of Practical Philosophy and Methodology 1. New Brunswick, NJ: Transaction Publishers.
Dai-Ichi Life Research Institute. Various years. Economic Indicators. Tokyo, Japan: Dai-Ichi Life Research Institute.
Economic Planning Agency. 1997. Economic White Paper. Tokyo, Japan: Japan Ministry of Finance.
Economic Planning Agency. Various years. Annual Report on National Account. Tokyo, Japan: Ministry of Finance.
Giorno, Claude, Pete Richardson, Deborah Roseveare, and Paul van den Noord. 1995. Potential Output, Output Gaps and Structural Budget Balances. OECD Economic Studies No. 24. Paris, France: OECD.
Holtz-Eakin, Douglas. 1992. Public-Sector Capital And the Productivity Puzzle. NBER Working Papers Series, no. 4122. Cambridge, MA: National Bureau of Economic Research.
Ishi, Hiromitsu. 1989. The Japanese Tax System. New York: Oxford University Press.
Kawasaki, Kenichi. 1996. Development of the ERI Compact Model. Tokyo, Japan: Economic Research Institute, Japan Economic Planning Agency.
Ministry of Coordination and Management. Various years. Japan Annual Statistics. Tokyo, Japan: Japan Ministry of Finance.
Ministry of Finance. 1992-97. Monthly Economic Statistics. Tokyo, Japan: Japan Ministry of Finance.
Ministry of Finance. 1997. Fiscal Statistics. Tokyo, Japan: Japan Ministry of Finance.
Ministry of Home Affairs. 1998. Local Budget White Paper. Tokyo, Japan: Japan Ministry of Finance.
Posen, Adam. 1998. Restoring Japan's Economic Growth. Washington: Institute for International Economics.
Samuelson, Paul A., and William D. Nordhaus. 1995. Economics. New York: McGraw-Hill, Inc.
Tamura, Yoshio. 1997. Japan's Fiscal System. Tokyo, Japan: New Toyo Keizai Co.
1. It is widely acknowledged that the effects of fiscal stimulus package should be measured with "Mamizu," which was about 22 trillion yen out of a 65 trillion total package. Mamizu is calculated as the total stimulus package minus spending in the following year, purchase of land for future public works, and other asset transfers and represents the amount that potentially increases GDP.
5. For example, road, petroleum gas, and motor vehicle taxes should be spent on maintaining and constructing roads, aviation fuel taxes should be spent on maintaining airports, and special tonnage taxes should be spent on maintaining ports.