Modeling Theories of Women's Underrepresentation in Elections

Research on women candidates in American elections uncovers four key facts: Women (i) are underrepresented among candidates, (ii) are underrepresented among office holders, (iii) perform better in office, and (iv) win open seats at equal rates to men. Scholars offer two types of explanations: Women are less willing to run than men, due to differential costs or a gap in self-perceived qualification, or voters discriminate at the ballot box. We formally model these mechanisms. Lower willingness to run predicts the first three facts but not the fourth. Voter discrimination at the ballot box predicts the first three facts and creates competing effects with respect to the fourth. Thus, the major stylized facts cannot be explained without voter discrimination, whether overt or more subtle. We explore whether a close-election regression discontinuity distinguishes the mechanisms; surprisingly, it does not.