The Productivity of Public Capital: Evidence from the 1994 Electoral Reform of Japan

This paper attempts to estimate the causal effect of public capital stock on production using Japanese prefectural data. We first articulate the difficulty of consistently estimating the regional-level production function with public capital due to the endogeneity of the public capital stock amount. As the central government allocates most of the public capital across regions in Japan, the stock amount of public capital could be endogenous because it could be allocated to either booming regions to support private activity or to stagnating regions to help them become more productive. The endogeneity of public capital is more serious when local governments make decisions regarding public capital investments, as in the US, because such decisions are directly affected by local governments' budgetary constraints. We need an exogenous variation of public capital investment across regions in order to estimate the causal effect of public capital on production. Japan's electoral reform in 1994 offers an exogenous variation of this sort. The reform drastically changed the distribution of political representation in the Lower House across regions, and it accordingly changed the allocation of public capital across regions as well. The productivity of public capital based on this natural experimental identification strategy indicates higher productivity due to public capital than indicated by the OLS estimation.