Wildfire Smoke and Crime: Evidence from California

Air pollution imposes large and multi-dimensional social costs, yet its effect on criminal behavior—and the channel through which it operates—remains unsettled, in part because credible identification is difficult; existing instruments tie wind variation to a fixed pollution source or rely on satellite plume presence without modeling transport. We estimate the causal effect of fine particulate pollution (PM2.5) on crime across all 58 California counties from 2006 to 2020, instrumenting monitor-based PM2.5 with a novel downwind smoke-exposure measure that transports each NOAA Hazard Mapping System smoke plume along the observed wind field over its own footprint—an instrument whose pollution source location varies from day to day as fires ignite and extinguish. The instrument is strong, with each additional downwind day raising monthly mean PM2.5 by about 0.26 μg/m3 (first-stage Wald F ≈ 45). A 1 μg/m3 increase in monthly mean PM2.5 raises violent crime by roughly 1.3% and motor-vehicle theft by roughly 1.4%, with the violent-crime effect concentrated in homicide and aggravated assault and no detectable effect on robbery, larceny, burglary, or total property crime. The pattern is robust to the pollution measure (wildfire-attributable PM2.5), the outcome scale (rates per 100,000), and the instrument geometry. A countylevel heterogeneity analysis finds the violent-crime effect homogeneous across demographic and economic characteristics—consistent with an acute physiological channel rather than a socioeconomic one—while the most severe harms fall disproportionately on more urban, younger, and more-minority counties. These findings identify a previously unvalued externality of wildfire smoke—an increasingly climate-driven pollution shock—and strengthen the case for smoke mitigation.