ASIA Pacific Economic Papers No. 379, 2009
Does voter turnout affect policy outcomes? This long-standing question has been re-visited recently with close empirical scrutiny. These studies, however, commonly suffer from a problem of omitting variables correlated with both causal and outcome variables: specifically, immeasurable political interactions between politicians, organised groups, and individual voters. We address this problem by exploiting an instrument based on new data measuring the amount of rainfall on the voting day. It is a valid instrument not only because it is correlated with voter turnout and uncorrelated with politically relevant omitted variables, but also because it is expected to satisfy the assumption of homogeneous partial effects, which has not been carefully examined in previous studies that took advantage of instrumental variables. Using a large, municipality-level data set from Japan, we show that the turnout effect on the amounts of intergovernmental fiscal transfers is indeed significant, positive, and large.