Gender discrimination and local elections.

Summary While the under-representation of women in politics is widely acknowledged, the reasons for this under-representation are still imperfectly understood and give rise to much debate. Among the possible explanations, the hypothesis that voters discriminate against women is frequently put forward, but not often substantiated. This study attempts to test this hypothesis in the case of France. To do so, we exploit a novel natural experiment that took place in France in March 2015 during the departmental elections. For the first time, candidates did not have to run alone but in pairs, necessarily composed of one man and one woman. The law also required that the order of appearance on the ballot be based on alphabetical order, which led to half of the pairs having a woman in the first position and vice versa. This historic change in the electoral process may have led some voters to believe that the candidate in first place on the ballot received more prerogatives than the person in second place, and thus to pay more attention to him or her. This reform provides an ideal analytical framework for assessing the presence of gender discrimination and analyzing its determinants: by comparing the votes received by pairs with a woman in the first and second position, and insofar as the gender of the first candidate is random, we are able to identify precisely the existence of discrimination by voters against women. In the end, we identify substantial discrimination against female candidates affiliated with right-wing parties, which affected the outcome of the election. We also show that voters' gender biases depend not only on the amount of information available on ballots, but also on the discrimination that exists in the local labor market.