Neighborhoods are strong determinants of both economic opportunity and criminal activity. Does improving connectedness between segregated and unequal parts of a city predominantly import opportunity or export crime? We use a spatial general equilibrium framework to model individual decisions of where to work and whether to engage in criminal activity, with spillovers across the criminal and legitimate sectors. We match at the individual level various sources of administrative records from Medellin, Colombia, to construct a novel, granular dataset recording the origin and destination of both workers and criminals. We leverage the rollout of a cable car system to identify key parameters of the model, informing how changes in transportation costs causally affect the location and sector choices of workers and criminals. Our counterfactual exercises indicate that, when improving the connectedness of almost any neighborhood, overall criminal activity in the city is reduced, and total welfare is improved.

Written with Carlos Medina (Banco de la Republica), Anant Nyshadham (University of Michigan), Daniel Ramos (John’s Hopkins University), Jorge Tamayo (Harvard Business School), Audrey Tiew (University of California, Berkeley).