The Dose Response of Criminal Groups: Effects on Homicides and School Dropout

Job Market Paper link

Many places in the world have a substantial number of criminal groups. The total number of criminal groups is likely a key determinant of crime in these locations. I study the impact of the number of Mexican Drug Trafficking Organizations (DTOs) on homicides and school dropout in Mexican municipalities between 2006 and 2018. For identification, I develop a novel selection model using distance-based instrumental variables to estimate the causal effects. The instruments are based on DTOs’ distance to their existing base of operations. My selection model delivers interpretable treatment effects for each possible number of DTOs, while conventional two-stage least squares estimators do not. The selection model can be used to examine market structure effects for other licit industries as well. I find that as DTOs increase, there is a significant increase in homicides. Having 8 DTOs, the largest amount observed in my data, causes a 16-fold increase in homicides relative to the overall mean. Moreover, the relationship is increasing and convex, contrary to popular models of crime. Additionally, I find increased DTOs also increase local school dropout in middle school. These effects are concentrated amongst older and male students. I find no effects for younger or female students. The gendered and age-differentiated results suggest a recruitment channel from the DTOs, who rely on adolescent males for their workforce. Consistent with this mechanism, I find that as DTOs increase in number, the number of adolescent crimes with middle-school dropouts as perpetrators increase. Moreover, homicides with adolescent victims increase as well. This paper is the first to show the non-linear impacts of the number of criminal groups on crime, and how these non-linear impacts spill over into local schooling decisions.