Setting an example? Spillover effects of Peruvian magnet schools

In this paper we use a Multi-Cutoff Fuzzy Regression Discontinuity Design to evaluate spillover effects of students enrolled into Peruvian public magnet schools, Colegios de Alto Rendimiento (COAR), on educational outcomes of younger students in their schools of origin. Using administrative data from the Ministry of Education for 2016, we find that having at least one student admitted in a COAR school causes some negative spillover effects on math test scores of students from the following cohort. No evidence of statistically significant results is found for verbal and history test scores, nor for self-reported educational expectations. We discuss potential causes and reasons that may explain our results.