Treated Politicians, Treated Voters: A Natural Experiment of Fiscal Expenditure

• In Event: Election Timing: Causes and Consequences

Abstract

When an election is approaching, political parties and/or incumbents attempt to increase fiscal expenditure for their reelection objectives. Although this “political budget cycle” hypothesis is often claimed, any empirical test confronts methodological challenges, as the timing of an election is often a function of unobservable strategic calculation by the incumbent party, has no intra-country variation for any given time, and/or is correlated with a range of macroeconomic and/or public-opinion fluctuations. Japanese local elections provide a unique natural experiment to cope with these issues, because the timing of a municipal election can be regarded as a treatment naturally and randomly assigned. After confirming that nearly a hundred of the observable pretreatment variables are indeed well balanced between the treated and control groups, we show statistically significant and substantial increases in fiscal expenditure for pork-barrel projects during the fiscal year preceding an election of mayors but not that of assemblies because of budget decision making institution.

Authors

• Kentaro Fukumoto, Gakushuin University

• Yusaku Horiuchi, Dartmouth College

• Shoichiro Tanaka, Gakushuin University