Beschreibung:
In this paper we study whether and how the belief that the gender of politicians affects their competence on different issues influences electoral outcomes depending on the salience of those issues. We first propose a theoretical model of issue-specific gender bias in elections which can describe both the presence of a real comparative advantage ('kernel-of-truth' case, or stereotype) and the case of pure prejudice. We show that, if the bias exists, it influences electoral results and that its effect can be partially reversed by successful information transmission during the electoral campaign. We then empirically investigate the relation between issue salience and women's performance using US data on House and Senate elections. Estimates of issue salience are obtained using Google Trends data. Exploiting the longitudinal dimension of the dataset at district level and an IV strategy to rule out possible endogeneity, we show a positive correlation between the salience of those issues that are typically listed as feminine and women's electoral outcomes. We therefore conclude that a bias indeed exists. The average effect of the bias is sizable with respect to the share of votes for women candidates, even if not large enough to significantly increase the probability that women candidates win elections.