School Choice, Skill Measures, and Graduation∗

This paper studies the effects of combining skill measures to construct the priority order of a centralized education market. We use data from Mexico City, where seat rationing relies solely on a one-shot exam score. We first show that admission to the most over-subscribed schools decreases graduation for marginally admitted students, but this effect is heterogeneous. It is decreasing in the one-shot exam for inframarginal students. It is negative for marginal students with low GPAs and boys but null for marginal students with high GPAs and girls. We then use a model of school choice and graduation that allows for match effects to study the equity and efficiency of counterfactual priority orders that combine the one-shot exam score and GPA with different weights. The larger the weight on GPA, the larger the share of girls and low-SES students that get access to the most over-subscribed schools. However, using roughly equal weight on both skill measures maximizes girls’ and low-SES students’ graduation rates at these schools