Turning bias into leverage: the case for gender quotas

This paper explores how voter prejudice against women in politics can evolve over time through learning. Building on a Bayesian updating framework, we model voters as holding biased priors about female candidates’ competence and updating their beliefs based on the performance of elected women. The model predicts that gender quotas can accelerate this learning process by increasing the visibility of competent female politicians, and that the effect is stronger, i.e., learning is faster, in more biased contexts. We test these predictions using two institutional reforms in Italian municipal elections. First, we exploit a short-lived gender quota reform (1993-1995) and find that quotas had a persistent effect on women’s representation even after their removal. Their impact was stronger in municipalities with historically higher gender bias, proxied by referendum outcomes on abortion and divorce and the gender gap in education. Second, we examine the 2012 introduction of gender quotas and genderconditioned double preference voting in municipalities with over 5,000 inhabitants. Using a sharp regression discontinuity design that allows for heterogeneous effects, we f ind that the reform increased women’s representation in the first post-reform election and in the subsequent electoral cycle. Effects are larger in municipalities with higher pre-treatment prejudice, especially in the subsequent cycle. Overall, these results support our theoretical prediction that quota-driven exposure to female politicians reduces bias more quickly where initial prejudice is stronger.