Municipal consolidation and local government behavior: evidence from Japanese voting data on merger referenda

The costs and benefits of decentralization were first formalized by Oates (1972).

Several works have reported the role of key features influencing municipal consolidation in Japan, including population size and area, fiscal conditions, and heterogeneity of income (e.g., Hirota 2007; Miyashita and Nakazawa 2009; Miyazaki 2006; Nishikawa 2002; Weese 2008).

Brink (2004) also employed voting data on municipal partition to estimate key factors affecting municipal boundary reform. However, the study comprised only 24 observations owing to the lack of boundary reform cases in Sweden. Further, the length of the data employed ranged from 1978 to 1999, a period so long that socioeconomic conditions might have changed during that time. Additionally, the study used data pertaining to local referenda on municipality partitioning, in contrast with most empirical works that have studied the mergers of school districts or local governments.

Some studies state that Japanese unconditional grants are calculated following the nonarbitrary formula, and thus, they may be considered as lump sum transfers (Mochida 2006; Reschovsky 2007).

Note that though it is usual for unconditional grants to be reduced after consolidation due to the alteration of population size in related municipalities, the amounts of unconditional grants merged municipalities could take have remained unchanged for the grace period (10 years) granted after consolidation. Because this scheme is very similar to the assumption of fixed lump sum transfers in theoretical analysis, the theory of this study could be adequately tested by this estimation.

MIC is the abbreviation of the Ministry of Internal Affairs and Communications (henceforth, MIC).

See, for example, Akizuki (2001) and Mochida (2001, 2006).

Formerly, the range of tax rates local governments could set was limited. See, for example, Mochida (2006, p. 164).

The central government uses specific grants to guide the implementation of these projects through budget controls and as a means to stimulate the economy and reduce interregional income inequality.

A vertical fiscal imbalance indicates differences in the balance of tax revenues and expenditures between the central and local governments, respectively; a regional fiscal imbalance refers to the differences in fiscal capacities of local governments.

See Mochida (2006) and Reschovsky (2007).

See, for example, CLAIR (2009), MIC (2006a, p. 41), and Yokomichi (2007).

However, municipal mergers have not been made compulsory by law in Japan. Rather, consolidations have taken place more voluntarily than in the past (Yokomichi 2007).

The referendum data were originally constructed as follows. First, municipalities holding a consolidation referendum were identified using Digital Archives of Mergers (henceforth, DAM) (MIC 2006b). As the DAM only covers the cases in which a consolidation was realized after a referendum, the other local referendum cases were identified from the homepage(s) operated by the municipalities or merger consultation committees, and other documents describing the merger committees, such as newspaper articles. Second, based on the list created in the first step, information pertaining to the referenda was obtained from the sources used in the first step. All of the municipalities employed in the estimation are listed in Online Resource 1 with the referendum data, such as date of election, number of the other consolidation partners, and approval rates.

The total number of referenda (426) is close to that reported by the MIC (2010b) (418), and thus, almost all referenda cases are covered. The difference in numbers can be explained by the fact that the MIC counted some referenda as a single case when two or more referenda were held for an identical consolidation plan, whereas the author counted each referendum in such cases separately.

Outliers have some particular characteristics, such as consolidation including enclaves and isolated islands, where large amounts of subsidies are often supplied by the central government. Because such municipalities likely differ from others, they are omitted from the sample.

For the per unit cost depending on population, see, for example, Alesina et al. (2004), Andrews et al. (2002), Brasington (1999, 2003b), and Duncombe and Yinger (2007).

The effect of transfers on consolidation incentives is also ambiguous. As seen in the second term of (8), utility gain from consolidation for municipality \(i\) depends on the relative sizes of the populations, median incomes, and lump sum transfer across municipalities. Except for population, large or small relations in income and transfer between municipalities \(i\) and \(j\) are not assumed, thus implying that it is ambiguous whether utility gain arises from consolidation with lump sum transfers.

Figure 2 in Miyazaki (2013) depicted the log of expenditure against that of the population in 2000, suggesting that the log of per capita cost is U-shaped for Japan.

On the one hand, inspired by Ellingsen’s (1998) majority voting model of separation/consolidation, Brasington (2003a) split the sample into larger and smaller communities, and showed that a larger district, which could be the decisive voter after consolidation, preferred to merge.

Difference in residents’ preferences for public services is frequently reflected in socioeconomic elements like income and race.

See Appendix B in Miyazaki (2013) for calculation methods of the median income.

Hinnerich (2009) found that a local government with a strong tendency to free ride increases its per capita debt before annexation.

Besides, in the context of the political business cycle, Johansson (2003) showed that intergovernmental grants are used as a device to win votes.

      On the other hand, setting a grace period for unconditional grants may exert an effect on the relationship between the grants and consolidation preference. During the grace period, which continues for 10 years, merged municipalities can receive the same amount of unconditional grants as they would have had they not merged, while it is usually expected that unconditional grants are reduced after consolidation, because the grants are calculated taking into account economies of scale in public service provision. This means that during the grace period, the theoretical prediction that the amounts of grants municipalities take will remain unchanged irrespective of consolidation status holds, thus suggesting that this regression model, which is derived from the theory, makes it possible to examine the association between consolidation preference and lump sum grants.

Note, however, that both specific and unconditional grants are, to be precise, not lump sum grants. Specific-purpose grants in Japan are said to be decided in accordance with the allocation standard, which while being objective, is set by bureaucratic institutions (Reed 2001). Unconditional grants, or the Local Allocation Tax in Japan, play the role of the fiscal equality transfer system and are allocated according to a formula based on the difference in basic expenditure needs and fiscal capacities of local governments (e.g., Mochida 2006). Therefore, municipality consolidations have usually given rise to a reduction in unconditional grants, because fiscal expenditure needs are calculated considering economies of scale. As per the 1999 amendment to the SLMM, however, unconditional grants for those municipalities that had merged by FY 2006 were calculated as if consolidation had not taken place for a certain time (e.g., CLAIR 2009), implying that unconditional grants are forecasted as remaining unchanged after consolidation. Therefore, unconditional grants are more likely to have the properties of a lump sum grant than a specific grant.

See Table 3 in Miyazaki (2013) for annual descriptive statistics for approval rates of referenda.

Financial support for mergers was reviewed nearly annually from 1999 through 2006 in order to spur municipal mergers.

In effect, the national government had devised a plan to restrict the functions of small municipalities, as reported by Prof. Nishio of the Local Government Investigation Committee in 2002 (known as “Nishio shian” in Japanese). Besides, the national government reduced unconditional grants through a series of reforms in intergovernmental transfers during 2004–2006 (“Sanmi Ittai Kaikaku” in Japanese). It also changed the method to calculate grants for the scheme, wherein grants for small municipalities were substantially reduced. Moreover, a questionnaire survey for relevant persons, including mayors, provides evidence that some small municipalities are motivated to consolidate after knowing of the plan (e.g., Workshop on Regional System and Town and Village 2008, p. 40–41).

Akizuki (2001) called this Japanese central-local relationship “controlled decentralization,” under which the central government designs and controls many governmental functions, such as policing and education, by using various channels, such as personnel exchanges between national and local governments and distribution of grants. It is well known that although local expenditure accounts for about 60 % of general government expenditure in Japan, part of the local expenditure is not under the local government’s control, because part of local revenue is provided by the central government. The central government can stipulate usage of a part of local expenditure by law and provides guidance to the local administration (Akizuki 2001; Mochida 2006).

In Japanese, the creation of a new municipality is called “Shinsetsu gappei,” while the absorption of a municipality is called “Hennyu gappei.”