The next Economics Seminar of Paris-Saclay will take place on Thursday, Mars 31 from 12h15 to 13h15. Nina Guyon (National University of Singapore) will present Desegregating Schools: Evidence from Middle School Closures in Deprived Neighborhoods.

This paper studies the effects of a desegregation policy consisting in closing down a middle school located in a deprived neighborhood and reallocating its students to other middle schools in the city. I analyze the direct effects on students from deprived neighborhoods (the “movers”), as well as the indirect peer effects on incumbent students in receiving schools (the “receivers”). In both cases, I make use of the staggered closure of middle schools in cities all over France as well as of the availability of control cities, and I compare cohorts of students before and after closure. Exhaustive administrative panel data at the student level allow me to account for potential selection of the movers into receiving middle schools as well as for a potential “rich flight” of the receivers by using the predicted middle school students should attend instead of the actual one. This also allows me to study how the direct and indirect effects of the policy vary with the proportion of new comers in the receiving schools. I find that a school closure leads to a decrease in the probability of dropping out of school after middle school for the movers that is driven by boys and students from low socioeconomic status (SES). Crucially, the probability of dropping out also decreases for the receivers for the same groups of students. The effects on the movers are observed despite a small increase in class size and while no effect on class size is found for the receivers. The effects on low-SES receivers are consistent with ranking effects being stronger than disruption effects. On the contrary, for high-SES receivers, I find a decrease in the probability of attending an academic high school that does not vary with the proportion of new comers. For them the data show evidence of a “rich flight” to the private system that can explain the previous effect while changes in classmates characteristics are unlikely to explain it.