Measuring Electoral Volatility in MMD/SNTV Systems : A Candidate-Level Analysis

Comparative politics scholars have spent a much time exploring the causes and consequences of electoral volatility, and the result is that much has been learned about what leads both established and emerging democracies to experience electoral volatility. While different measures and factors have been investigated in these studies, all measures use political parties as the nit of analysis and measure this phenomenon at the national level. For many electoral systems, this is an appropriate way of measuring electoral volatility, but for systems like MMD/SNTV used previously in Japan and Taiwan and open-list PR systems like that used in Brazil, such may understate the extent to which volatility occurs. To correct this problem, we explore electoral volatility at the district level by using the candidate as the unit of analysis, and, using data from postwar elections in Japan, we show that traditional measures do understate the extent to which electoral volatility occurs. Our empirical results show that candidate volatility is impacted by several factors including partisanship and the number of candidates