Thesis: This thesis studies the link between institutions, gender and politics. Three questions are studied: can institutions undo gender norms? Would institutions be more gender-egalitarian if they were headed by women? Why are women absent from positions of power?The first chapter of this thesis tests whether institutions can undo gender. In particular, we study the consequences of institutions on the perpetuation of gender norms. We study the norm according to which a woman should earn less than her husband. Using the German division as a natural experiment, we show that East German institutions have undone gender. East German women can earn more than their husband without increasing their number of housework hours, put their marriage at risk, or withdraw from the labor market. By contrast, the norm of higher male income and its consequences are still prevalent in the West.The second chapter studies whether institutions would be more gender-egalitarian if more women were heading them. In particular, I test whether female politicians have the same priorities than their male counterparts. The context studied is the French Parliament from 2001 to 2017. Using text analysis and quasi-experimental variations to randomize legislators' gender, this chapter shows that women are twice more likely to initiate women-related amendments in the Lower House. Women's issues constitute the key topic on which women are more active, followed by health and childhood issues whereas men are more active on military issues. I provide supporting evidence that these results are driven by the individual interest of legislators. Finally, I replicate these results in the Upper House by exploiting the introduction of a gender quota.The third chapter studies the reasons behind the underrepresentation of women in positions of power. I investigate whether the persistence of incumbents hinders female access to political positions when incumbents are predominantly men. I exploit regression discontinuity from close electoral races in French municipalities to randomize the eligibility of incumbent mayors for reelection. Despite a context increasingly favorable to the election of women, I find that the persistence of incumbents does not block female access to the position of mayor. I investigate the mechanisms and show that it is more difficult for a woman to replace a female incumbent than a male one.