8 Comments on “Fiscal Decentralization in Japan: Does it Harden the Budgets of Local Governments?”

Keimeir Kaizuka, and Anne Krueger
Published Date:
July 2006
  • ShareShare
Show Summary Details

Comments by Daniel Citrin

The chapter by Professors Tajika and Yui provides a very nice overview of the system of local government finances in Japan, and offers a number of sensible suggestions on how to improve intergovernmental finances. Their careful documentation of the growing role of local governments underscores the importance of reform, both on efficiency grounds and to help address Japan’s long-term fiscal challenges. In my comments, I would like to touch on both issues, starting with the main principles which should guide reforms of intergovernmental fiscal relations, drawing on the international experience.

Efficient intergovernmental relations

Professors Tajika and Yui highlight two main avenues for reform to strengthen efficiency which I thought I would discuss further: first, the need to harden local government budget constraints; and second, the need to increase the transparency of the transfer system, particularly regarding its redistributive aspects. These priorities are broadly appropriate, and progress on these fronts would help underpin an efficient fiscal structure. My comments will not touch on their third main area of concern—the role of the Ministry of Home Affairs in setting local government priorities, although institutional changes will clearly be associated with reforms in this area.

The concern about the hardness (or lack thereof) of local government budget constraints is well-placed. Indeed the issue could be expanded into a broader theme of how best to improve accountability of local governments. International experience points to a number of guiding principles to increase the accountability of local government. Of course, there are wide differences in the design of intergovernmental relations across the world, reflecting both institutional set-ups and relative preferences with respect to the provision of public goods and redistribution. Nevertheless, well-functioning intergovernmental fiscal systems are based on an incentive structure that promotes fiscal discipline and efficient resource allocation. Sub-national governments are more likely to act responsibly if they bear the cost of their actions. Typically, this involves a degree of sub-national autonomy in deciding expenditure priorities and the authority to own-finance locally provided services at the margin. Local borrowing would be subject to market discipline, without implicit or explicit government guarantees, and information on fiscal outcomes at a local level would be available on a timely basis.

Against this backdrop, many observers agree that local government finances in Japan fall short of these objectives in a number of ways:

  • Expenditure priorities for a large share of local spending are decided at the central government level, and financing at the margin relies largely on non-earmarked transfers through the local allocation tax (LAT).

  • Local debt is perceived as effectively guaranteed by the central government, and constraints on borrowing by local governments are relatively weak in practice.

  • Public works subsidies are allocated in a non-transparent manner, rather than through explicit objective criteria.

  • Data on local government finances is released with a long delay, making it difficult to monitor budget execution in a timely manner.

The government’s ongoing reforms of local government finances—the so-called “Trinity reforms”—contain some elements which should improve local government accountability and, hence, the efficiency of government spending. The Trinity reforms aim to allow local authorities greater discretion in allocating funds, while limiting to some extent their overall resources; this will be achieved by reducing subsidies for specific local projects and LAT transfers, partially compensated by a transfer of tax resources from the central government.

While a full assessment of the Trinity reforms will need to await more time (the program was initiated in 2003 and is slated to end in 2006, with the bulk of measures just now underway), our initial view is that they are largely in line with the objective of increasing the flexibility of local governments to determine and finance expenditure priorities. However, the scale of the reform is modest relative to the size of local budgets: to date, subsidies have been reduced by a cumulative ¥0.6 trillion, compared with total annual subsidies of over ¥30 trillion, and ¥3 trillion in cumulative tax resources transferred, against about ¥35 trillion of annual local tax resources. Total local government spending in FY 2005 is projected to reach almost ¥85 trillion.

Another aspect of accountability surrounds the arrangements governing borrowing by local governments—an issue briefly touched on by the authors. Although conditions have been established under which local government bonds can be issued, these have not been effective in restraining local government borrowing, with gross local government debt rising from 15 percent of GDP in 1990 to around 40 percent of GDP in 2003. The rules governing transfers to local governments help support the debt build-up, as an increase in debt servicing costs tends to raise the amount of transfers received through the LAT formula. In addition, financial markets perceive that local government bonds carry a central government guarantee, and hence do not differentiate among local government borrowers. The fact that a large share of local government bonds are held by public entities (including through the Fiscal Investment and Loan Program, FILP) also serves to undermine market discipline. Reforms slated for FY 2006 to relax the system of permits for local bond issuance could help to address the perception that local debt is guaranteed by the government. But the next stage of reform could take these issues up further. To ensure that local governments at the margin bear the cost of raising funds, the central government should clearly announce that there are no explicit and implicit central government guarantees of local government bonds. As noted by others, market discipline will apply only once the yields on local government bonds differ across borrowers.

Turning now to the authors’ second main proposal, we agree that there is a clear need to improve the transparency of intergovernmental fiscal arrangements. There are a number of aspects to this issue.

  • First, there appears to be a great deal of discretion in the allocation of earmarked subsidies, which could result in an inefficient allocation of resources and decreasing returns to public investment (which is largely financed by local governments). Adopting more transparent rules for the allocation of public works subsidies, within a cost-benefit framework, could improve the efficiency of public spending.

  • Second, the determination of LAT transfers is based on a complex formula. The idea behind the formula—to direct funds towards regions whose fiscal needs exceed their resources—is sound, and has been implemented successfully in other industrial countries. However, the formula itself appears complex with heavy information requirements, which tends to obscure an assessment of its effectiveness.

A broader review of the LAT in time could seek ways to make the transfer system more transparent and simpler. This should involve rationalizing the calculation of local government fiscal needs with a view to meeting local needs in a cost-effective manner—perhaps by removing items which can be manipulated by local governments and so potentially introduce an element of moral hazard (for example, a local government could increase its LAT allocation by building an unnecessary road). Also, as pointed out by the authors, any reform of the LAT formula should avoid unduly limiting the incentive for local governments to increase their own tax resources (under the current formula, an increase in local revenues would eventually lead to reduced transfers).

Let me suggest a hard-nosed way of making the transfer system more transparent and efficient: the central government should undertake a comprehensive review of local government spending, including of public works projects. Transfers would be adjusted accordingly—possibly leading to a sharp cut if a significant share of current local government spending is of low value, as many observers believe. Local governments would have the right to raise taxes (or borrow without guarantee) to finance additional spending. Such arrangements would run into significant obstacles from a political economy standpoint. Still, reforms in this direction would be very desirable as they would make local governments more accountable to taxpayers and would likely facilitate economic restructuring and a more efficient allocation of resources.

Local government reform and fiscal consolidation

My comments so far have focused on ways to improve the efficiency of the intergovernmental fiscal system in Japan. I would like to conclude by setting these issues into the broader perspective of Japan’s overall fiscal challenges. In this context, the key question is: what role can the intergovernmental fiscal system play in advancing fiscal consolidation? At its broadest level, this question relates to issues which are almost metaphysical in nature: What services should be provided by the public sector? How high should taxes be? Which level of government should provide public goods? The Japanese government and society will need to confront these issues as fiscal consolidation advances and a tension between raising revenues and creating a more focused government potentially arises. Overall, the process of putting the national finances in order will have implications for the structure of local government finances. The principles elaborated in the chapter by Professors Tajika and Yui, and echoed in my comments, will be relevant in the adjustment process: reforms should aim to enhance government accountability and transparency.

At a more specific level, we can ask how the reform of local government finances can contribute to fiscal consolidation. Here, the main gains must come through increased efficiency (shifting tax and expenditure responsibilities from one level of government to another does not affect the finances of the general government). There is a broad consensus that a decentralization of expenditure responsibilities should in principle improve efficiency by promoting a closer correspondence between expenditure priorities and local needs and reducing waste. Again, steps to improve government accountability and the transparency of the transfer system are more likely to promote efficiency.

In conclusion, I would like to stress that the increasing focus of policy makers on local government finances is welcome. We are hopeful that, properly designed and implemented, local government reforms can play an important role in addressing Japan’s large fiscal challenges.

Comments by Masayoshi Hayashi

Tajika and Yui discuss several interrelated issues in the Japanese local public finance, and identify, among others, the following as problems of the Japanese local public finance. First, local tax rates as well as local expenditures are tightly controlled by the central government, which implies an inefficient level of local public services. Second, the system of general intergovernmental transfers called the local allocation tax (LAT) yields an “excessive” redistribution among regions, making the incentive for local governments to raise their own taxes less effective. Third, the total amount of transfers made through the LAT is determined as an outcome of a budgetary process at the central level, which leads to an expansion of local public expenditures. These views are the “mainstream” ones held by many Japanese economists. While Tajika and Yui also discuss other salient features of the Japanese local public finance system, I will focus my discussion on those three propositions given their importance in the current Japanese policy dialogues, and offer alternative (possibly unorthodox) views on the matters.

Tight central control

The central government in Japan indeed controls a substantive part of local expenditures as well as local taxes. When we consider the central command and control of local governments, however, it is important to consider why such a system exists. Japan is a unitary country governed by a Constitution which defines rights and obligations,1 which of course are applicable to all Japanese wherever they may reside within the country: horizontal equity dictates that all persons should be treated comparably no matter where they reside. Modern constitutional rights and obligations concerns redistribution, which makes modern governments largely institutions for redistribution (Boadway and Keen, 2000). It is mainly local governments that actually implement such constitutional rights and obligations through their provision of public services and programs. It then seems very natural that national laws mandate2 local governments to ensure that their services and programs meet national standards so that every Japanese national can secure their entitlements guaranteed by the Constitution.

There seems to be central control on local taxes as well. In fact, the standard local tax rates are set by the Local Public Finance Law (LPFL), a national law legislated by the national parliament. However, the local taxes listed in the law do not exhaust all the taxes available to local governments. For example, the city planning tax is not listed in the LPFL and constitutes more than a negligible proportion of local budgets. Furthermore, local governments are in fact free to choose the rates of the taxes listed in the LPFL within some limits. Nonetheless, as Tajika and Yui point out, we rarely see local tax rates set above the standard rates. They argue this is because of collusion among local governments coordinated by the Ministry of Internal Affairs and Communication (MIC). I, however, would argue that this has more to do with the political concerns of local politicians. Recall that a majority of local revenues is spent for redistributive expenditures whose standards are set by the national government. As such, raising tax rates would be unpopular since taxpayers may not get back the benefits of their raised taxes through changes in local expenditures. In addition, even if local government colluded as Tajika and Yui argue, collusion might not be a bad thing at all, if we allow for the possibility of the adverse effects of tax competition.

In any case, we could argue that local governments in Japan are controlled by the central government to a large extent in relation to expenditures. An application of the decentralization theorem (Oates, 1972) then implies that local public expenditures in Japan are inefficient, as Tajika and Yui argue. However, recall that local expenditures in Japan are largely redistributive. As I argued elsewhere (Hayashi, 2004), the decentralization theorem fails to the extent that local expenditures are redistributive and that redistribution is a national concern.

Excessive transfers and incentive problems

Another popular view is that intergovernmental transfers in Japan are so “excessive” that local governments are less motivated to raise their own taxes. Let us consider the “excessive” part.

Excessive transfers

It is widely held among Japanese economists that, as Tajika and Yui argue, local governments “spend far more in per capita terms than their richer counterparts” thanks to excessive transfers. This tendency of “excessive” transfers has been found in a series of studies that compared the distributions of per capita revenues before and after the transfers were made (for example, Ishi et al., 1982; Kaizuka et al., 1987; Hayashi, 1987; Nakai, 1988).

But, the fact that the recipients of intergovernmental transfers obtain more per capita revenues than richer localities does not necessarily imply that the transfers are excessive. Most of local expenditures are mandated by the central government, and those mandates are funded partially in rich localities and fully in poor localities.3 In fact, as Tajika and Yui show with OECD data, Japanese local governments may be doing much more than most of the other OECD counterparts. While the share of local revenues (26 percent) is higher than the OECD average (21.9 percent), the share of local expenditures (40.7 percent) is even higher than the average (32.2 percent). And, if social security funds are excluded, the local share hikes up to 61 percent. Since local expenditures are more or less redistributive, per capita expenditures will tend to be higher in regions where the number of disadvantaged individuals is larger. In addition, since poorer regions are disadvantaged in terms of physical environments (for example, mountainous areas) and demographics (for example, higher ratio of the elderly), the unit cost of public services in those regions are expected to be higher (for example, Hayashi, 2004). It is then no surprise that poorer localities tend to spend more on a per capita basis.

Therefore it may be misleading to only look at the distribution of per capita revenue after transfers when we evaluate a proper level of intergovernmental transfers. The point is to set up a baseline expenditure that appropriately accounts for local characteristics that affect the expenditure level, and the social/redistributive criteria that are set at the national level. Without such baseline expenditure, I am not sure if we could say whether or not local expenditures are “excessive.” In any case, no serious empirical studies have been conducted so far to probe such baseline or standard expenditure levels.

Incentive effects

Replicating their own argument (Tajika et al., 2001), Tajika and Yui suggest that the LAT weakens local effort to raise own taxes. They argue that an increase in own revenues T leads to a reduction in transfer receipts from the LAT system since the LAT follows a gap-filling formula. It is yet another popular view among Japanese economists that this causes a typical “hold-up problem” that induces localities to raise inadequate revenues (for example, Akai et al., 2003).

Note that in their argument local revenues T are the choice variables. However, local revenue T may not be the actual choice variable. If there is a single tax base b and a unit tax rate t, local revenue would be given as T = t·b(t,s) where s is an additional policy variable that affects the volume of the base. Formulated as such, what matters would be the choices of t and s. Then, the following analysis would be possible.

The amount transferred through the LAT is roughly given as D - 0.75τb(t,s) where D is basic fiscal needs, and τ is standard tax rate set in the LPFL. The important point is that the rate τ is fixed and independent of actual tax rate r.4 Given this formula, if әb/әt < 0 (a standard assumption), r tends to be higher than the optimal level (for example, Smart, 1998). That is, the LAT may increase tax effort measured by the level of tax rate. If we define s as an index of activity to increase the tax base (әb/әs > 0), the same logic now implies that s would be too low. That is—as Tajika and Yui argue—the LAT may suppress tax effort. However, note also that the central government at the same time provides local governments with subsidies to promote local industries. This should in effect function as a matching subsidy to increase s. Therefore, we cannot say a priori whether s is too low or too high.5

Expansion of local expenditures in the 1990s

My view is therefore that the LAT may not affect tax effort of local governments as many Japanese economists argue. However, I do agree with the authors on the third point. That is, the total amount of transfers made through the LAT is determined as an outcome of a budgetary process at the central level, which leads to an expansion of local public expenditures. Indeed, since intergovernmental transfers occupy a large portion of central expenditures, determining the amounts of transfers needed is an indispensable part of planning the central budget. And I argue that this budgeting process, rather than the claimed “hold-up problem” of the LAT, may be one of the major culprits that led to the expansion of local expenditures in the 1990s. It is then instructive to take a closer look at the process than Tajika and Yui describe.

The central government annually drafts the Local Public Finance Program (LPFP) where projections for “standard” levels of local revenues and expenditures are made in order to allocate the public funds necessary to finance local expenditures. The LPFP has the function of determining the aggregate amount of the LAT that is necessary to sustain the “standard” local expenditures. A rough description of the LPFP may be given as follows. Expenditure projections are made for a list of items which include: (1) salaries and wages of local public officials; (2) current operating expenditures; (3) capital spending; and (4) debt service. Note that the projections are for “standard” (not actual) spending that the central government deems appropriate.6 It may also be worth noting that “standard” expenditures also account for discretionary components that can be expansionary or contractionary.

Revenue projections are made for local taxes, specific grants to local government (excluding the LAT) and local public bond issues. Note that the LPFP only considers local taxes listed in the LPFL, and revenues are projected with the use of forecasted changes in relevant tax bases with corresponding statutory “standard” rates. The total revenue projection is compared to the total expenditure projection. Given the Japanese fiscal structure, the latter value exceeds the former, yielding a gap for the LAT to fill. Since the funds for the LAT are fixed proportions of five national taxes,7 the gap is not always filled by such funding sources. If the funds do not cover the difference, the central government fills the remaining gap by: (a) inducing local governments to issue additional local bonds; (b) increasing the proportions with which the national taxes are earmarked for the LAT; (c) transferring additional funds (Special Fund Transfers) from the central budget for the LAT disbursements; and (d) issuing additional bonds at the national level. The last two measures, (c) and (d), are made in the Special Account from which the LAT is distributed among local governments.

Since the “standard” expenditures also account for discretionary components, the central government can adjust the amount of the LAT disbursement depending on policy objectives at the national level. I agree with Tajika and Yui that the LAT has been affected by policy discretion, but my view would depart from theirs in terms of sources of such arbitrariness. Tajika and Yui argue that local governments intentionally expand their expenditures, anticipating that such an increase would lead to an increase in the local demand estimated in the LPFP, and therefore, an increase in the LAT disbursements. While this so-called “soft budget” type is currently popular among leading economists in Japan (for example, Akai et al., 2003), I would argue that the major sources of the expansion may be found at the national level, rather than locally.

In the 1980s, the U.S. wanted Japan to increase its domestic demand so that the U.S. trade deficit against Japan would shrink. The two countries held the Structural Impediments Initiative negotiations which led to a commitment made in 1993 by Japan to execute the ¥430 trillion public works plan, later expanded to ¥630 trillion,8 within the next ten years (for example, Schoppa, 1993, 1997; Ikegami, 1998). This commitment was bolstered by the post-bubble recessions that started in the early 1990s and intensified i the late 1990s, which necessitated counter-cyclical fiscal policy. Especially, public works are a frequently employed instrument for expansionary policy in Japan. However, since local governments execute most of public expenditures as we saw, the central government had to “mobilize” local governments to fulfill the commitment made with the U.S. (for example, Pascha and Robaschik, 2001).

Recall that the local “standard” expenditures in the LPFP also take account of discretionary components that correspond to expansionary policies for public works. When expansionary policies are undertaken, local expenditure projections in the LPFP are accordingly inflated, creating an additional fiscal gap to be filled. This gap, traditionally filled by matching grants from the central budget, was covered by an increase in LAT disbursements in the mid-1980s and more typically in the 1990s. More specifically, the LAT system was adjusted so that more public works yields increases in LAT disbursements. While an increase in the specific grants yields a one-to-one increase in the central budget, an increase in LAT disbursements does not have such an effect since the disbursements are processed through the LAT special accounts. It was a political commitment in the 1980s to reduce the national deficits which had been accumulated since the 1970s. Since an increase in matching subsidies implies an expansion of the central budget, in this sense the LAT seems to be an attractive instrument for myopic central policy makers who want to restrict the central budgets but, at the same time, want to undertake an expansionary economic policy.

The LAT did expand local expenditures. But I am not sure if that is because the LAT makes local governments less disciplined, as the authors argue. The expenditure expansion was exactly what the central government wanted local governments to do. In fact, as Tajika and Yui show in Figure 7.3, local expenditures have been reduced since 1999, although nothing essential has changed in the LAT system. This may constitute further evidence that the central government does control local expenditures. Then, the issue of fiscal discipline may be more relevant for the central government than for local governments.

Concluding remarks

My discussion on the topics covered by Tajika and Yui may highlight the points that we should emphasize more than the current literature does, including work by Tajika and Yui. First, we should pay more attention to the design of appropriate fiscal relations between levels of government. A large amount of intergovernmental transfers might be a simple reflection of a large vertical fiscal gap in Japan. Currently, Japan’s local governments may have too large a mandate to fulfill. Without delineating proper expenditure roles and revenue capacities of local governments, it may not be possible to evaluate the current system of local public finance. Second, more attention should be paid to the decision making process at the central level. The LAT system is unique in the sense that its total amount and disbursements are determined through negotiations among the central ministries—not through a simple mechanical formula, with substantive interdependence among other types of central grants. This aspect of the Japanese system is theoretically analyzed by Sato (2001) and, of course, discussed here by Tajika and Yui. Further analysis would definitely be desirable.


For example, Article 25 of the Constitution states: “All people shall have the right to maintain the minimum standards of wholesome and cultured living. In all spheres of life, the State shall use its endeavors for the promotion and extension of social welfare and security, and of public health.” Also, Article 26 maintains that “All people shall have the right to receive an equal education correspondent to their ability, as provided by law. All people shall be obligated to have all boys and girls under their protection receive ordinary education as provided for by law. Such compulsory education shall be free.”

For example, the Daily Life Security Law (1950, Law No. 144) requires specific types of municipality (that is, cities) to set up welfare offices to provide public assistance. The welfare offices, which are within the administrative structure of local governments, follow the standards set by the national government. Towns and villages are not required to set up welfare offices. In the case of towns and villages that do not set up such offices, prefectures assume the role of providing public assistance.

The national government partially covers the expense with central government subsidies. While the rest is covered by local taxes in rich localities, the Local Allocation Tax, that is, general-purpose grants determined on a gap-filling formula, takes care of the difference in poor localities.

In addition, taxes that are not listed in the LPFL are excluded from the basic fiscal revenue (= τ·b(t,s)). Such taxes, like city-planning tax, therefore do not affect the transfers received through the LAT.

A survey shows that local governments prioritize the industrial and commercial bases over extending their tax revenues (INITIA Consulting, 2003). This then may imply that the level of s should not be too low.

For example, the wages and salaries are made comparable to those of the central government officials. The subsidized components of current operating expenditures and capital spending are calculated using subsidy rates and unit costs set by relevant national laws and the central budget. Debt service is accounted only for limited types of local bonds specified by the central government (for example, those listed in the Local Borrowing Program).

The ratios of the five taxes are 32 percent for personal income tax and liquor tax, 35.8 percent for corporate income tax, 25 percent for tobacco tax, and 29.5 percent for the consumption tax.

These are huge numbers since Japanese GDP is about ¥500 trillion.


    AkaiN.; SatoM. and YamashitaK.2003Chihokofuzei no keizaigaku [The Economics of the Local Allocation Tax]Tokyo: Yuhikaku.

    BoadwayR. and KeenM.2000“Redistribution,” in A. B.Atkinson and F.Bourguignoneds.Handbook of Income Distribution 1Amsterdam: Elsevier pp. 677-789.

    HayashiM.2004“Redistribution, Decentralization and Community Sharing,”paper presented at the International Symposium on Fiscal Decentralization in Asia Revisited Hitotsubashi University November 5-6.

    HayashiY.1987Gendai zaisei no saibunpai kozo [Redistributive Structure of Modern Public Finance]Tokyo: Yuhikaku.

    IkegamiT.1998“Ippanzaigensyugi no genkaito aratana ippanzaigensyugi” [“The Limits and the New Aspects of the General Revenue Doctrine”] in N.Jin-no and M.Kanekoeds.Chiho ni zaigen wo [Expanding Local Tax Bases]Tokyo: Toyo keizai shinpo sya pp. 77-133.

    INITIA Consulting Inc.2003Chihojichitai ni yoru kigyo yuchi nit suite no anketo chosa hokokusyo [Survey on Local Attraction of Business]Tokyo: INITIA Consulting Inc.

    IshiH.; KoizumiI; HasegawaT.; HataK. and YamamotoY.1982“Chiikikan ni okeru jueki to futan nokiketsu to saibunpai koka” [“Regional Benefits, Regional Burdens and Interregional Redistribution”]Keizai BunsekiVol. 86 pp. 1-187.

    KaizukaK.; HonmaM.; TakabayashiK.; NagamineJ. and FukumaK.1987“Chihokofuzei no kino to sono hyoka, part II” [“The Evaluation of the Function of the Local Allocation Tax: Part II”]Financial ReviewVol. 5 pp. 9-26.

    NakaiH.1988Gendai zaisei no suryo bunseki [A Quantitative Analysis of Fiscal Burdens]Tokyo: Yuhikaku.

    OatesW. E.1972Fiscal Federalism,New YorkHarcourt Brace Jovanovich. PaschaW. and RobaschikR.2001“The Role of Japanese Local Governments in Stabilisation Policy.”Paper presented at the Second International Convention of Asia Scholars BerlinAugust9-12.

    SatoM.2001“Intergovernmental Transfers, Governance Structure and Fiscal Decentralization,”Japanese Economic ReviewVol. 53No. 1 pp. 55-76.

    SchoppaL. J.1993“Two Level Games and Bargaining Outcomes: Why Gaiatsu Succeeds in Japan in Some Cases and Not Others,”International OrganizationVol. 47No. 3 pp. 353-86.

    SchoppaL. J.1997Bargaining with Japan: What American Pressure Can and Cannot DoNew York: Columbia University Press.

    SmartM.1998“Taxation, Incentives and Deadweight Loss in a System of Intergovernmental Transfers,”Canadian Journal of EconomicsVol. 31No. 1 pp. 189-206.

    TajikaE.; YuiY. and SatoM.2001“Chihokofuzei no naniga mondai ka: Yurumu chiho no zaisei kiritsu to sogai sareru zaisei kaizen doryoku” [“The Issues of the Local Allocation Tax: Weakening Incentives for Fiscal Reconstruction”]Zeikei TsushinVol. 56No. 12 pp. 23-33.

    Other Resources Citing This Publication