Floods and local incumbency disadvantage : a regression discontinuity design

Do voters punish political leaders in the aftermath of a natural disaster? Using a Regression Discontinuity Design, this article explores how floods affect local electoral results in Chile. Taking data from polling stations, our estimations show that, floods reduce the likelihood of incumbent candidates’ chances of winning in affected polling stations in a range from 11.3% to 22%, depending on the bandwidth considered. It also follows from our results that the greater the electoral competition, the greater the probability of losing the next race. The paper explores the electoral mechanism and finds that turnout is significantly lower in affected areas, pointing to voter dissatisfaction as an explanatory factor.