Using admission lotteries to estimate heterogeneous effects of elite schools

This paper studies the effects of enrollment in an elite school on students’ achievement. We use that elite schools in Amsterdam are often oversubscribed and admission is based on lotteries. Our results show that elite schools have negative effects on achievement of students who just qualify for the highest academic track and positive effects on achievement of students from the top of the baseline ability distribution. These results reconcile contrasting findings from previous studies that use regression discontinuity designs. We also find that value-added estimates of the effects of elite schools are severely biased.