Crime, Spatial Reallocation, and Employment Dynamics: Evidence on Firm-Size Heterogeneity from Mexico City

This paper examines the relationship between neighborhood crime and firm employment in Mexico City. Using a balanced panel of 2,354 Basic Geostatistical Areas (AGEBs) for 2017--2022, we estimate Spatial Durbin Models with AGEB fixed effects, lagged and winsorized crime variables, and differential trends by dominant firm-size stratum, decomposing total effects into direct and indirect components. At the aggregate level, robberies are negatively associated with employment in the AGEBs where they occur, alongside positive spillovers toward neighboring areas. This aggregate pattern conceals substantial heterogeneity: large firms (101 or more employees) are the only stratum with a negative and significant direct effect, whereas small firms (0--10 employees) show a positive direct effect, consistent with structural immobility and the local reallocation of displaced demand. Medium-sized firms show negative but imprecisely estimated effects. A dynamic specification preserves these signs but is weakly identified once persistence is absorbed. Crime appears to reallocate employment across the firm-size distribution and across space rather than uniformly contracting local labor demand.