Different Voting Behavior in National and Local Elections: Analysis of Voter Turnout in Concurrent Elections

Abstract

Why does voting behavior vary in national and local elections? This is a key question as the concept of scale becomes increasingly important in electoral geography. To clarify this question, this study analyzed voter turnout in national elections when local elections were held on the same day. Although studies on concurrent elections have been conducted previously, few have focused on cases in which different levels of elections were held simultaneously. Previous studies also did not clarify whether there are regional differences among the effects of concurrent elections.

This study used difference-in-differences (DiD) estimation to analyze whether voter turnout in national elections increases when they are held concurrently with local ones. The outcome variable was voter turnout in Upper House elections held during 2004–2013 in Japan. The treatment variable was whether local elections were held concurrently with Upper House elections in each municipality. In order to make the treatment and control groups comparable, propensity score matching was utilized before DiD estimation.

The result of DiD estimation showed that voter turnout in Upper House elections increased by 6.6 percentage points when local elections were held on the same day. This estimation result is reliable because the event study suggested that the parallel trends assumption is generally reasonable. It was also found that the effect of municipal concurrent elections was greater than that of prefectural elections. Furthermore, the effect of municipal concurrent elections became more significant in rural areas. Detailed data from Hyogo prefecture showed that municipal elections record higher voter turnout than Upper House elections, especially in rural areas. This “turnout twist” phenomenon explains why the effect of municipal concurrent elections varies between urban and rural areas.

Examining three components of the effect of concurrent elections (i.e., electoral mobilization, cost sharing, and psychological stimulus), the results indicate that the level of interest and mobilization in national and local elections generate regional differences in the effect of concurrent elections. The level of interest and mobilization in national and local elections varies between urban and rural areas. It is argued that the level of interest in each election and electoral mobilization are important factors that contribute to different voting behavior in national and local elections in Japan.