This paper estimates how short-term changes in police presence affect crime using a high-resolution patrol-car location database for Dallas in 2009. The analysis exploits routine patrol behavior to isolate causal effects of police absence on local crime.
📍 Where the patrol data come from
- A unique dataset recording the exact GPS locations of Dallas Police Department patrol cars continuously throughout 2009.
- Location observations are used to measure how much time officers spent in each beat (police coverage area) at fine spatial and temporal scales.
🔧 How causality is identified
- Officer locations are endogenous because police are often drawn to crime, so a strategy to recover causal effects is required.
- The instrument leverages instances when officers respond to calls outside their allocated coverage beat. Those responses temporarily reduce presence in the abandoned beat and are plausibly driven by the police objective of minimizing response times rather than by contemporaneous crime in the abandoned beat.
- This outside-call response variable provides a plausibly exogenous shift in local police presence that isolates the effect of short-term absence on crime.
🔎 Key findings
- A 10% decrease in police presence at a location is associated with a 7% increase in crime.
- Results illuminate the "black box" of policing by showing that routine, short-term changes in patrol patterns materially influence criminal behavior.
💡 Why this matters
- Findings imply that modest and routine redeployments of patrol resources—driven by response-time goals—can meaningfully affect crime rates.
- Evidence is directly relevant for police deployment policies, response-routing practices, and cost-benefit assessments of rapid-response strategies.
