Abstract
We document that local violence generates spillover effects beyond areas where violence takes place, via out-migration from violence-affected areas and peer exposure to violence. In the context of the Mexican war on drugs, we use violence-induced student migration as an exogenous source of variation in peer exposure to violence to estimate its effects on the academic performance of non-moving students in areas not directly affected by violence. Our results show that municipalities that face more violence experience higher rates of student out-migration. In receiving schools in areas not directly affected by violence, adding a new peer who was exposed to local violence to a class of 20 students decreases incumbents’ academic performance by 2 percent of a standard deviation. Negative effects are more pronounced among girls and high-achieving students.