Closeness and Turnout in Runoff Elections

This study aims at contributing to the literature on the effect of political competition on voting turnout. I test the Downsian Closeness Hypothesis (DCH) using data on runoffs elections in Hungary. This research design allows for estimating the effect of expected closeness using the variation of margins in the first round races to avoid the endogeneity problem inherent in ex post measures of closeness. A simple statistical model of turnout in a runoff election is described and estimated in terms of a Generalized Linear Model (GLM), recognizing that the dependent variable is a fraction. The findings of the paper are consistent with the DCH: increases in margins between two parties in the first round significantly decrease turnout in the second, even when turnout in the first round is controlled for. This is in line with the theoretical considerations of the DCH but contrary to a large part of the existing empirical literature. The estimates of closeness are substantially greater than in previous papers and exhibits stability across models. In particular the estimates of the GLM are remarkably similar to those obtained with standard linear models, suggesting that the effect of closeness is quite well approximated by the latter modeling approach.