Association for Public Policy Analysis & Management

Each year, millions of Americans are subject to police-initiated civilian stops. In recent years, the utility of these interactions has been debated, and calls for law enforcement reform, accountability, and oversight have grown widespread. However, there is little causal evidence as to the impacts of civilian stop programs. We first measure the extent to which New York City police commanders have direct effects on civilian stops. We then use a quasi-experiment based on naturally-occurring movement of commanders across geographic regions (precincts) to estimate the impact of a commander’s civilian stop propensity on crime and other indicators of neighborhood health. Borrowing from methods used in the teacher value-added literature, we find a wide distribution of commander effects, even after conditioning on observable neighborhood characteristics, baseline crime rates, and baseline policing tactics. We find that a commander’s civilian stop propensity is highly predictive of future stops when commanders move across precincts; that is, we find a high-degree of portability of commander effects, indicating that police commanders transfer their tactical preferences across settings and officers. Importantly, we also find that commander switches are not correlated with crime or tactical trends occurring within the precinct prior to the commander’s arrival. Finally, we estimate the reduced form impact of these commander effects on outcomes such as crime, homicide, and overdose rates. In particular, we find that a 1 SD increase in a commander's stop effect leads to a 5% increase in monthly drug arrests. However, we find a precisely-estimated zero effect of a commander's stop propensity on violent crimes. We find no evidence of crime spillovers to adjacent precincts. Taken together, these results provide some of the first convincing causal evidence of the impact of New York City’s stop and frisk program on crime, and suggest that police commanders play an important role in implementing patrol strategy.