The fiscal effects of cross-prefectural border municipal mergers: evidence from Japan using the synthetic control method

ABSTRACT

The purpose of this paper is to investigate whether social connectedness can mitigate the fiscal common pool problem in municipal mergers, as many studies have primarily focused on whether this problem occurs. I employ the synthetic control method by using higher-level government borders to identify the moderating effects of social connectedness. The findings reveal that subordinate partners in cross-prefectural border mergers significantly reduce their bonds and construction expenses just before the mergers. Merger partners with strong incentives to merge do not create fiscal common pool problems, which is the opposite of previous studies. The result might imply the potential effectiveness of merger support policies that prioritize social connectedness.

KEYWORDS:

• Cross–prefectural border municipal merger
• fiscal common pool problem
• social connectedness
• moderating effects
• synthetic control method

JEL CLASSIFICATION:

• H72
• H77

I. Introduction

Many studies investigate the fiscal common pool problem of municipal mergers (e.g. Blom-Hansen 2010; Fritz and Feld 2020; Goto, Sekgetle, and Kuramoto 2021; Hansen 2014; Hinnerich 2009; Hirota and Yunoue 2017, 2020; Jordahl and Liang 2010; Nakazawa 2016; Saarimaa and Tukiainen 2015). The fiscal common pool problem is related to the free-rider problem based on the concept of the law of $1/n$. Both public project and bond size tend to increase with a common pool size just before mergers because merged municipalities expect to share their fiscal burden after mergers and to be able to free-ride.¹ In Japan’s case, Hirota and Yunoue (2017, 2020) reveal that subordinate merger partners rapidly increase their construction expenses and bond issuances just before mergers, although those of the dominant merger partners decrease or do not increase.

The purpose of this paper is to investigate whether social connectedness, including historical, economic, and geographical ties, can mitigate the fiscal common pool problem in municipal mergers. Many studies have focused on whether the common pool problem occurs. Bhatti and Hansen (2011) emphasize the significance of social connectedness in the context of municipal mergers. In the case of heterogeneous merged municipalities, one would anticipate greater welfare losses. Moreover, there are relatively few studies that explore ways to prevent the common pool problem. For example, Askim and Houlberg (2022) reveal the moderating effects of social norms just before mergers.

In this paper, my focus is on cross-prefectural border municipal mergers in Japan, and I employ the synthetic control method by using higher-level government borders to identify the moderating effects of social connectedness. In many merger cases, municipalities choose neighbouring partners within the same higher-level governments because they expect to receive support including subsidies from these higher-level governments. Given the similarity of social connectedness among merged municipalities, it is not easy to reveal whether social connectedness mitigate the fiscal common pool problem just before mergers. To identify ways to mitigate the fiscal common problem in quasi-experimental approaches, the cross-prefectural border merger case provides a more suitable approach, aligning with the literature on social connectedness. From an identification strategy perspective, the synthetic control method is capable of generating appropriate counterfactuals, especially when dealing with a single treated unit.

This paper is organized as follows. Section II describes the background of Japan’s municipal mergers. Section III presents the empirical framework and results. Section IV concludes.

II. Background

In Japan, there are three layers of administrative systems: the central government and prefectural and municipal governments. Prefectures and municipalities are both local public entities that collaborate in local administration based on their respective roles and responsibilities. Prefectures serve as regional authorities that encompass municipalities and oversee broader regional administration. Each municipality falls within a specific prefecture.

The central government enacted the Special Municipal Mergers Law, which has been called a carrot-and-stick policy, to induce a large number of municipal mergers in FY1999.² The law provided that if municipalities decided to merge up to FY2005, the merged municipalities would receive special fiscal or other support from the central government, such as local allocation tax (LAT) grants, subsidies, and bonds.³ In contrast, for municipalities that decided not to merge, their grants, subsidies, and others would decrease.

Prefectures played many roles in the process of municipal mergers, including creating combinations of merger partners, establishing subsidies for mergers, setting up support headquarters, and dispatching prefectural staff, all based on national guidelines (e.g. Goto Yasuda Kinen Tokyo Toshi Kenkyusho 2013; Hirota and Yunoue 2014). Most prefectures adhered to the national guidelines for promoting municipal mergers. The reason is that many prefectures depend on intergovernmental transfers from the central government. During the period from FY1999 to FY2005, the number of municipalities decreased from 3,232 to 1,820.

All merged municipalities chose to merge with neighbouring municipalities within the same prefectures except in one merger case.⁴ The reason is that there are not only cultural connectedness but also administrative or fiscal interrelationships between prefectural and municipal governments, embodied in links such as personnel exchange, intergovernmental transfers from prefectural government, and public projects. However, Yamaguchi village in Nagano Prefecture decided to merge with Nakatsugawa city in Gifu Prefecture in FY2004. Figure 1 shows the map of both Gifu and Nagano prefectures. The sole cross-prefectural merger case was Yamaguchi village in Nagano Prefecture. Nakatsugawa city was the dominant merger partner, with the largest population size in Gifu at 55,477 people in FY2003. The other six towns and villages in Gifu (Sakashita town, Kawakami village, Kashimo village, Tsukechi village, Fukuoka town, and Hirukawa village) were subordinate merger partners that had smaller population sizes, at less than one-fifth the size of the dominant merger partner. Yamaguchi village in Nagano Prefecture historically had an intense connection to Nakatsugawa city in Gifu Prefecture. In FY1958, Misaka village in Nagano decided to merge with Nakatsugawa city in Gifu across the prefectural border. Yamaguchi district in Nagano, which is a part of Misaka village in Nagano, did not merge, while the other three districts of Misaka village decided to merge with Nakatsugawa. According to the opinion survey of residents in FY2001, over seventy percent of residents hoped to merge with Nakatsugawa city across the prefecture’s border (e.g. Kitazaki 2005). Fiscal simulations by the Yamaguchi village government revealed better conditions under such a merger than under other potential mergers with neighbouring municipalities in Nagano. However, the Nagano prefectural mayor had a negative attitude towards the merger. Yasuo Tanaka, the prefectural mayor, first rejected the cross-prefectural merger proposal, but later allowed the merger after consulting with the minister of Internal Affairs and Communications.

Figure 1. Cross-prefectural municipal mergers.

III. Empirical framework and results

Synthetic control method

To reveal the fiscal common pool problem under cross-prefectural municipal mergers, I apply the synthetic control method developed by Abadie and Gardeazabal (2003) and Abadie, Diamond, and Hainmueller (2010, 2015). The method is to generate synthetic control values for a treated unit by calculating the weighted average of untreated control units from a donor pool. The method can consider time-varying confounding when there is a close fit between the treated and synthetic control paths and is similar to a combination of the matching and difference-in-differences methods. In detail, the pretreatment root mean square prediction error (RMSPE) of the treated outcomes and the counterfactual pretreatment outcomes are minimized as follows:

(1) $minRMSPE=min{\left(\sum _{t=1}^T\frac{{(y_{it}-{\hat{y}}_{it})}^2}{T}\right)}^{\frac{1}{2}}$(1)

where $y_{it}$ is the observed outcome of unit $i$ in pretreatment period $t$. ${\hat{y}}_{it}$ is the counterfactual outcome of $i$ in period $t$, given by the weighted average of untreated units from the donor pool. Suppose that ${\hat{y}}_{it}={\sum }_{j\ne i}{w}_j{y}_j$ and $\sum w_j=1$. The counterfactual weights $w$ chosen by the RMSPE are minimized (e.g. Abadie 2021; Gilchrist et al. 2022).

This paper employs the eight merged municipalities as treated units. Nakatsugawa city decided to merge with seven towns and villages of Gifu and Nagano prefectures in FY2004. The main treated unit in this paper is the cross-prefectural merger partner, Yamaguchi village in Nagano prefecture. In this merger case, Nakatsugawa city is the dominant partner. The other six towns and villages in Gifu and Yamaguchi village in Nagano are subordinate partners. According to Hirota and Yunoue (2017), subordinate partners rapidly increase their bonds, total expenditure, and construction work expenses a few years before mergers. This paper sets the treatment period from FY2001 to FY2003. The period in which the RMSPE should be minimized is from FY1980 until FY2000. Additionally, this paper employs 88 nonmerged municipalities in both Gifu and Nagano prefectures for the donor pool to calculate the synthetic control values.⁵

The main outcome variables are per capita bonds, total expenditure, and construction work expenses, following previous papers. Additionally, I employ the per capita prefectural subsidies because the different prefecture’s policies might affect the cross-prefectural partner’s behaviour. As predictors, I employ the population; the share of the population under 15; the share of the population over 65; the shares of the primary and tertiary industries; the shares of LAT grants, central subsidies, and prefectural subsidies in total revenue; and per capita accumulated debts. These predictors are almost the same as those in previous papers (e.g. Hirota and Yunoue (2017, 2020)). This paper uses Japan’s municipal data from FY1980 to FY2003, excluding the data from FY1981 to FY1984 due to data unavailability. The data on municipal governments are derived primarily from the Statistics of the Final Accounts of Municipal Governments, Survey on the Basic Register of Residents, and Population Census. Table 1 shows summary statistics of the cross-prefectural partner and donor municipalities.

Table 1. Summary statistics.

	Yamaguchi (Cross-pref.)		Donors
	Mean		Mean
Variable	1985	2003	1985	2003
Outcomes
Bonds (million yen)	62.791	113.778	36.743	94.646
Total expenditure (million yen)	546.047	793.412	395.760	691.439
Construction works (million yen)	132.093	116.724	133.059	156.741
Predictors
Population (hundred)	21.500	20.540	145.238	147.994
Pop.15	20.047	12.853	19.852	13.770
Pop.65	18.093	29.163	15.524	26.857
Primary industry	26.600	17.000	21.701	14.618
Tertiary industry	38.238	50.700	37.597	47.351
LAT grants	35.851	50.232	25.706	29.299
Central subsidies	15.270	3.261	7.501	4.721
Prefectural subsidies	6.805	3.991	8.938	5.519
Accumulated debts	554.419	1171.442	322.738	877.252

Empirical results

Figure 2 shows the results of the per capita bonds of the cross-prefectural partner, that is, Yamaguchi village in Nagano Prefecture. Additionally, I apply the placebo tests suggested by Abadie, Diamond, and Hainmueller (2010, 2015). The placebo tests calculate the synthetic control of each municipality in the donor pool one by one, excluding the real treated units. The average placebo effects show the average gap between the placebo-treated units and synthetic control values. Moreover, the post/pre RMSPE ratios show the ratios of the post RMSPE to the pre RMSPE. The higher ratios indicate that the treated values are largely different from the synthetic control values in the posttreatment period.

Figure 2. Synthetic control of cross-prefectural merger partner’s bonds.

The top left side of the figure shows the results of both the treated and the synthetic control values. The colour part of the synthetic control values is the 95% confidence interval.⁶ The bonds of the cross-prefectural partner are lower than the synthetic control values and are outside the 95% confidence interval. The cross-prefectural partner decreases its bonds just before the merger. The result is the opposite of findings from previous papers. The top right side of the figure shows the placebo tests that show the differences between the placebo-treated and synthetic control values. The cross-prefectural partner’s bonds decrease in the posttreatment period, while there are no differences in the pretreatment period. The bottom left side shows the average placebo effects that indicate the average gaps between the placebo-treated and synthetic control values. The cross-prefectural partner clearly decreases its bond issuances in the posttreatment period. Moreover, the bottom right side shows the histogram of the post/pre RMSPE ratios.

(note) The outcome variable is the per capita bonds of the cross-prefectural merger partner. The monetary unit is 1 million Japanese yen, which is approximately 10,000 dollars at an exchange rate of 100 yen to 1 US dollar. The colour part of the synthetic control units is the 95% confidence interval. ‘Average placebo effects’ are the average gaps of both the placebo-treated and placebo synthetic values. ‘Post/pre RMSPE ratios’ are the ratios of the post RMSPE to pre RMSPE. The post/pre RMSPE ratio of the cross-prefectural partner is 4.222, which has the fifth-highest ratio out of 89 municipalities with a p value of 0.058 ( = 5th/89 units).

Table 2 shows the summary results of the synthetic control method on bonds. The average gap shows the average gap between the treated and synthetic control values. The average gap of the cross-prefectural partner is approximately −157.62 million Japanese yen, and the average gap ratio is approximately −52.8%. The bond reduction rate is largely higher than that of other treated units, including the dominant partner and other subordinate partners in Gifu Prefecture.

Table 2. Summary results of bonds.

Bonds	Avg. gap	Avg. gap (%)	Post/pre RMSPE	p value
Yamaguchi (Cross-pref.)	−157.619	−52.8	4.222	0.058
Nakatsugawa (Dominant)	−11.203	−23.9	1.486	0.395
Sakashita	−33.566	−25.7	1.416	0.442
Kawaue	20.581	20.6	1.980	0.244
Kashimo	29.226	28.0	1.066	0.605
Tsukechi	−6.502	−10.2	0.524	0.907
Fukuoka	−0.895	−2.6	0.652	0.849
Hirukawa	18.048	63.0	1.367	0.477

Table 3 shows the summary results of additional outcomes, including per capita total expenditure, construction work expenses, and prefectural subsidies. Regarding total expenditure, the average gap ratio is approximately −17.6%. Regarding construction work expenses, the average gap ratio is approximately −48%. The cross-prefectural character of the merger prevents the occurrence of the fiscal common pool problem under municipal mergers. This result differs from the findings of previous studies. The average gap ratio of per capita prefectural subsidies is −51.5%. The rate of reduction in prefectural subsidies is higher for the subordinate partners in cross-prefectural mergers than for other partners.

Table 3. Summary results of additional outcomes.

Total expenditure	Avg. gap	Avg. gap (%)	Post/pre RMSPE	p value
Yamaguchi (Cross-pref.)	−192.183	−17.6	2.741	0.042
Nakatsugawa (Dominant)	−23.337	−6.2	2.256	0.099
Sakashita	0.816	0.1	0.225	0.986
Kashimo	69.078	9.1	0.995	0.634
Tsukechi	−8.367	−1.5	0.556	0.887
Fukuoka	−44.128	−7.2	2.336	0.085
Hirukawa	−73.378	−10.2	1.153	0.493
Construction works
Yamaguchi (Cross-pref.)	−151.305	−48.0	1.857	0.093
Nakatsugawa (Dominant)	−13.570	−17.8	1.850	0.105
Sakashita	−23.045	−18.0	1.129	0.488
Kashimo	42.697	17.2	0.789	0.744
Tsukechi	2.816	2.1	0.527	0.919
Fukuoka	−14.844	−4.7	1.573	0.209
Hirukawa	−23.799	−10.2	1.501	0.267
Prefectural subsidies
Yamaguchi (Cross-pref.)	−54.431	−51.5	1.735	0.208
Nakatsugawa (Dominant)	−3.968	−16.6	1.797	0.188
Sakashita	−7.320	−18.1	0.690	0.667
Kawaue	−28.717	4.2	2.582	0.063
Kashimo	44.527	43.3	1.563	0.250
Tsukechi	−3.688	−6.7	0.788	0.594
Fukuoka	−6.767	−11.2	0.552	0.792
Hirukawa	−10.454	−16.4	1.449	0.333

IV. Conclusion

This paper investigates whether social connectedness can mitigate the fiscal common pool problem in municipal mergers. I reveal that the subordinate partners in cross-prefectural border mergers clearly decrease their bonds and construction work expenses just before the mergers. The result is the opposite of findings from previous studies.

In Japan’s case, all merged municipalities chose to merge through the carrot-and-stick policy combining fiscal support or the reduction of intergovernmental transfers from the central government. However, the cross-prefectural border partner decreases their bonds and construction works just before the mergers. The cross-prefectural border partner might place great emphasis on social connectedness including historical, economic, and geographical connectedness. The result might imply the potential effectiveness of merger support policies that prioritize social connectedness. The reason is that merger partners with strong incentives to merge do not create fiscal common pool problems.

Acknowledgements

The author acknowledges financial support received from the JSPS (Nos. 20H01504, 20H01450, 22K01542, 22H00856)

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

The work was supported by the Japan Society for the Promotion of Science [20H01450,20H01504,22H00856,22K01542].

Notes

¹ Askim et al. (2020) reveal that government agency facing a termination threat reacts by a last-minute flurry of spending when faced with a termination threat.

² Strebel (2018) examines the driving factors behind bottom-up mergers in the Swiss context and concentrated on financial incentives. However, in the case of Japan, the central government implemented not only a ‘carrot policy’, which includes grants, but also a ‘stick policy’ towards municipalities. Japan’s merger approach may lie somewhere in between a top-down and a bottom-up approach. See, Goto Yasuda Kinen Tokyo Toshi Kenkyusho (2013).

³ The amount of LAT grants provided to each municipality is determined based on the municipal fiscal shortfall calculated by the central government. Because LAT grants are general-purpose lump-sum grants, most municipalities prefer them to specific subsidies. For more details, see Ihori (2008), and Hirota and Yunoue (2017, 2020).

⁴ To my knowledge, there is no cross-higher level government border merger case.

⁵ As alternative donor pools, the paper employs 193 other nonmerged municipalities for donors. The 193 municipalities belong to neighbouring prefectures, including Toyama, Yamanashi, Nagano, Gifu, and Shizuoka prefectures. Additionally, the paper employs 95 merged municipalities belong to Nagano and Gifu prefectures. The results obtained are similar to those in this paper and are available upon request.

⁶ For more information on the confidence interval, see Firpo and Possebom (2018) and Abadie (2021).