War on drugs, violence, and the share of low-income workers in Mexico

We analyse the average effects of increased violence generated by Joint Interventions (Operativos Conjuntos) within the so-called war on drugs at the municipal level in Mexico on the percentage of the working population earning twice the minimum wage or less. We implement a semiparametric difference-in-differences approach (Abadie 2005; Houngbedji 2016) by constructing a treatment dummy variable for the most violent municipalities of Mexican states treated by Joint Interventions. This approach uses covariates to adjust the differences between groups before the treatment through propensity scores. Consequently, assuming similar pretreatment characteristics in covariates, in the absence of the treatment, treated individuals would have a similar outcome relative to the nontreated group. After controlling for socioeconomic characteristics, results show an increase in the share of low-income workers in the most violent municipalities. Additionally, results show that the more violent the municipality is, the larger is the increase in the share of low-income workers. Our results are robust to changes in the sample and to changes in the construction of the treatment variable. Finally, we discuss some public policy implications.