Ethnic diversity in Peruvian schools: Disentangling peer and class composition effects

Academics have thought of many ways to reduce the gaps in academic achievement for ethnic minorities, one of which is through manipulating the classroom composition. Many studies show that there are positive spill-overs or peer effects of sharing space with better performing students (Sacerdote, 2001). However, other studies find that students from ethnic minorities tend to be negatively affected by ethnically diverse classes, either because of racial stereotypes or segregation (Steele and Aronson, 1995). This study applies Lee’s (2007) linear-in-means model with fixed effects to a sample of approximately 140 thousand students from a national education census in Peru to disentangle endogenous peer effects from contextual peer effects, while solving for self-selection and the reflection problem. The study focuses on the effects of ethnic diversity on Peruvian indigenous students. Results suggest that the effect of class composition on ethnically diverse classes is driven by two distinct and opposing channels, one related to abilities spill-overs and one to stereotypes and segregation. Endogenous peer effects are negative, possibly due to incentives faced by teachers and school directors to focus on better students and neglect the poorer performing ones, thus enlarging the gap between them. In terms of ethnic composition, results are heterogeneous: indigenous students benefit from a large proportion of similar peers, but the effect is the opposite for non-indigenous students. The study concludes that the current composition of the average Peruvian class is suboptimal, and a larger proportion of indigenous students could help reduce the current gaps in achievement due to ethnicity.