Turnout twist : higher voter turnout in lower-level elections

Abstract

American and European political scientists have claimed that subnational elections record lower voter turnout than national elections in most democracies. Contrary to this conventional wisdom, Japanese municipal elections often record considerably higher voter turnout than national elections, particularly in small towns and villages. Why is Japan different from most other democracies? Is Japan the only exception? What explains such exceptionally high local turnout in rural Japan? Under what conditions do lower-level elections produce higher voter turnout? By examining these questions, this dissertation aims to improve our understanding of incentives affecting electoral participation both at national and subnational levels. First, I show that Japan is not the only exception. There are some other important cases of higher voter turnout in lower-level elections, which I define as the "turnout twist" phenomenon. They can be found in Australia, Canada, Finland, France, India, Italy, Northern Ireland, Spain, and Switzerland. Second, I hypothesize that relative voter turnout in subnational vs. national elections is determined by the relative magnitudes of how much is at stake and how much votes count ("vote significance") in these elections. In lower-level elections, electoral outcomes may less significantly affect public policies, but citizens may be able to affect the electoral outcomes more significantly by voting. Therefore, if the effect of vote significance outweighs the effect of stake, "turnout twist" is a logically possible consequence.

(cont.) I statistically test this model using cross-national data, cross-municipality Japanese data, and individual-level survey data from Japan. The results show that the model has impressive explanatory power. Finally, based on findings from my field research, I show that the electoral system used in Japanese municipal assembly elections (i.e., the single non-transferable vote system with an at-large district) produces very small vote margins between candidates, and thus gives a strong incentive to both voters and candidates to think hard about how much votes count.