Abstract
We study the effect of mobile police patrolling on crime, using: (a) a natural experiment that increased patrolling in 6,000 well-defined areas, by economically realistic amounts and under relatively normal circumstances, and (b) a high-frequency geocoded dataset based on the GPS signals transmitted by mobile police officers. We first document that the natural experiment increased patrolling by large relative amounts, both on certain days and, within those days, at certain hours. Using a panel dataset at the 5-minute level, we find that the increased police presence generated large immediate reductions in crime. However, dynamic deterrence is very low: crime rates quickly revert to baseline levels once patrols leave an area. As a result, the estimated effects using a panel dataset at the daily level have implied elasticities indistinguishable from zero. Our results show that the effectiveness of mobile police patrolling depends strongly on the size of the dynamic deterrence.
Invited by: Research Seminar Team
Local Organizer: Massimiliano G. Onorato