Gender Differences in High-Stakes Performance and College Admission Policies

The Gale-Shapley algorithm is one of the most popular college allocation mechanism around the world. A crucial policy question in its setting is designing admission priorities for students, understanding how they disadvantage certain demographic groups, and whether these differences are related to differences in college performance potential (i.e., whether these differences are fair). Studying a policy change in Spain, we find a negative e_ect of increasing the wei0ght of standardized high-stakes exams on female college admission scores, driven by students expected to be at the top. The effect on admission scores does not affect enrolment, but the percentage of female students in the most selective degrees declines, along with their career prospects. Using data on college performance of pre-reform cohorts, we find that female students most likely to lose from the reform tend to do better in college than male students expected to benefit from the reform. The results show that rewarding high-stakes performance in selection processes may come along with gender differences unrelated to the determinants of subsequent performance.