Abstract: This paper develops a theoretical model and exploits a natural experiment to identify peer effects in the college classroom. We build a model in which classmates have both learning and information externalities, and derive predictions for peer effects on college and labor market outcomes. We test these predictions in a regression discontinuity design based on a unique "tracking" admission system at a flagship university in Colombia. We find that a large increase in mean classmate ability reduced students' grades, graduation rates, and post-college earnings, which suggests a negative learning externality. The earnings effects were partially offset by a peer informational benefit that faded out over time.
Friday, February 22, 2019
3:30pm – 5:00pm