Empirics of the median voter hypothesis in Japan

This paper empirically analyses for the first time the median voter hypothesis in Japan as a means of investigating whether or not Japanese prefectural finance reflects the preference of the median voter. The hypothesis is tested by estimating the demand functions of local public goods in each prefecture. As official data on the income of the median voter is unavailable in Japan, respective prefectural data is constructed using official data on income distribution and taxation. Reasonable intuitive interpretation of results indicates that the median voter hypothesis is supported in prefectural finances, and that voter preference affects the outcome of gubernatorial elections, i.e., a governor's reelection probability, by estimating a probit model. When considering the centralized prefectural government system in Japan, these results indicate that central government management of prefectural expenditures via inter-regional grants ultimately reflects jurisdictional median voter preference. First version received: December 1997/Final version received: February 1999