Seminario CEDE - Alvaro Riascos

In this paper, we estimate the causal impact of police presence on crime using a structural discrete-choice model of criminal location choice. Our approach combines credible quasi-experimental variation in police patrols—induced by emergency calls unrelated to crime—with a behavioral model of offender decision-making, allowing us to estimate deterrence effects and simulate policy counterfactuals. In our structural model, offenders choose locations based on potential rewards and the risk of apprehension, proxied by police presence. Using an instrumental variable for police deployment based on non-crime emergency calls, we find that increased police presence significantly deters crime. Our main results indicate that a 10% increase in patrol presence leads to roughly a 7–8% reduction in crime, an elasticity toward the upper end of prior estimates. The structural model reveals spatial heterogeneity in deterrence. According to our counterfactual simulations, reducing police response to non-crime emergency calls by 50% would lower total crime by about 7.1%—a reduction equivalent to increasing total police patrol time (comparable to the active force) by 11.2%. These findings extend reduced-form evidence by quantifying offenders’ responsiveness to police presence and by illustrating how alternative deployment strategies can enhance deterrence, with direct policy relevance for resource-constrained cities.