Optimal top-n policy

A widely adopted affirmative action policy in the university sector is the “top-$n$” policy, where the universities guarantee admission to students in the top $n$ fraction of their class from every high school. The efficacy of the top-$n$ policy in university integration has been questioned because students strategically relocate to low-achieving high schools. We show that when minorities are less likely to strategize than others, the policy can even segregate minorities from the target university, compared to the school-blind policy. A suitably chosen eligibility requirement, featuring the minimum time students must spend at a high school in order to be eligible for top-$n$ admissions, can restore the efficacy of this policy. However, the most stringent requirement is not always optimal. The optimal requirement depends on the original distribution of students across high schools.