Discrimination statistique à l’embauche des femmes et apprentissage des employeurs

Given a gender wage gap of 9 % at the beginning of working life, one may assume screening discrimination against women. To test its existence, we use a French survey on the beginnings of careers (Generation 98) which makes it possible to detail career paths over ten years. Based on the model developed by Altonji and Pierret, which reports on wage determination in the face of uncertainty of hiring and considers progressive learning of firms with time, wage equations on pooled data are implemented testing several specifications of the experience, taking also into account its endogeneity. We find that learning about productivity with experience would be more noticeable for women and would mainly occur through seniority while it would equally operate through previous employment experiences for men. Indeed, at the time of hiring, returns to education are higher for women and with career advancement, hard-to-observe correlates of productivity obtain a greater wage impact for women than for men, supporting the hypothesis of initial screening discrimination.