"Inefficiency and Self-Determination: Simulation-based Evidence from Meiji Japan"
Weese, Eric, Masayoshi Hayashi and Masashi Nishikawa
|Revised in March 2017 and May 2021.
We consider a popular theoretical model of jurisdiction formation where there is a tradeoff between efficiencies of scale and heterogeneity. We develop a maximum score estimation technique to determine the parameters of a central planner's payoff function given the way they partitioned a territory into jurisdictions. We apply this technique to historical data on a set of centralized boundary changes in Japan: walking distance appears to largely determine jurisdiction boundaries, with only small effects from land type and historical feudal ruler, and no effect of religion. We then assume that local villages shared these preference parameters emphasizing walking distance, and use binary integer programming to calculate core partitions for a decentralized coalition formation game based on this model. Core partitions exist with very high probability. In a counterfactual world in which there are no between-village income differences, these core partitions are extremely close to the partition that would be chosen by a utilitarian central planner. When actual cross-village income differences are used, however, sorting on income results in mergers that are both smaller and geographically discontiguous.