Assessing the spatial effects of subway implementation on intraurban robberies occurring on buses in Salvador, Brazil

This study examines the impact of the subway system's implementation in Salvador on the spatial distribution of robberies on buses (hereafter, on-board bus robberies). The analysis used a panel dataset containing socioeconomic and on-board bus robbery data from 2012 to 2018. We used data from the Brazilian Demographic Census and information on crime from the Public Security Secretariat of the State of Bahia. Our primary identification strategy was the Spatial Differences-in-Differences method. The results show that areas with lower residential density but high levels of economic activity and mobility recorded higher robbery incidence, suggesting that ambient population dynamics, rather than residential concentration, drive exposure to on-board bus robberies. The implementation of the subway system was associated with an increase in on-board bus robberies in neighborhoods directly served by the subway (direct effect), and a reduction in these incidents in adjacent neighborhoods through spatial spillovers (indirect effect).