Do vouchers lead to sorting under random private-school selection? Evidence from the Milwaukee voucher program

This paper analyzes the impact of voucher design on student sorting in the application and enrollment phases of parental choice. More specifically, it investigates whether there are feasible ways of designing vouchers that can reduce or eliminate student sorting in these phases. Much of the existing literature investigates the question of sorting where private schools can screen students. However, the publicly funded U.S. voucher programs require private schools to accept all students unless oversubscribed and to pick students randomly if oversubscribed. This paper focuses on two crucial features of the Milwaukee voucher program?random private school selection and the absence of topping up of vouchers. In the context of a theoretical model, it argues that random private school selection coupled with the absence of topping up can preclude sorting by income in the application stage, although there is still sorting by ability. This design can avert sorting by ability in the enrollment stage, but revelation of new monetary costs (as has been the case in Milwaukee) can lead to sorting by income. Using a logit model and student-level data from the Milwaukee voucher program for 1990-94, the study finds strong and robust evidence that this indeed has been the case in reality. These findings have important implications for voucher design and policy.