The Effect of Police on Crime: Evidence from 10,000 Discontinuities

Abstract: Existing literature has found large negative effects of police patrolling on crime, typically using natural variation in the aftermath of terrorist attacks or experimental variation targeting hotspot areas. It is unclear, however, to what extent these effects extend to more general settings in terms of the day-to-day work of the police. In this paper, we estimate the effect of mobile police patrolling on crime in residential areas. To do this, we take advantage of a policy introduced by Essex police to ratchet up police presence in a 200 metre radius around the scene of a burglary, starting on the Thursday following the burglary and lasting for seven days. We find that this large and exogenous variation in police presence was not associated with a decrease in crime during the intervention week. We then use high-frequency police location data to study whether the presence of a police officer in an area has a contemporaneous negative effect on crime. We find that it does, but also that this deterrence effect is very short-lived.