Does violence against civilians depress voter turnout? Evidence from Bosnia and Herzegovina

We investigate the effect of violence against civilians on voting. Using data from elections in Bosnia and Herzegovina between 1990 and 2014 and exploiting variation in war intensity across municipalities, we estimate a negative impact on voter turnout. The effect is stable and persistent over twenty years after the war resolution. Our results are robust to the inclusion of pre- and post-war socioeconomic and political characteristics, to instrumental variable estimations based on terrain ruggedness, and to restricting the sample to voters who were too young to be selectively targeted. Distinguishing between civilian and military victims, we show that violence against civilians drives the negative effect. Next, we examine different mediating mechanisms including forced migration and demographic selection, ethnic composition, physical capital damage, post-conflict reconstruction, and labor market conditions. Our results support the hypothesis that violence affects voting through a “moral” dis-utility from showing allegiance to politics and the society by casting a vote. Using survey data, we show that respondents in more affected municipalities report lower generalized trust, trust in institutions, and voting.