Seminarios

Abstract

This paper provides experimental evidence of the overall impact of an after-school program on students' outcomes, and of peer effects in that context. Participants were between 10-16 years old and enrolled in public schools in El Salvador. I find that the program reduced reports of bad behavior by 0.17 standard deviations, school absenteeism by 23%, and increased school grades by 0.11-0.13 standard deviations. Changes in highly violent students mainly drove the results. Regarding group composition, I find that mixing students with different propensities for violence was better than segregating them. Moreover, there is an interaction between the group composition and individual baseline propensity to violence; more violent participants have greater effects on their academic outcomes and behavior when they are assigned to heterogeneous groups, supporting the rainbow model of peer effects. Additionally, it finds positive social spillover effects for non-enrolled children exposed to treated students.