Do you feel the heat around the corner? The effect of weather on crime

In this paper, we study the weather–crime relationship using a unique high-frequency, city-level data set for the USA with 2.4 mil. observations. In contrast to the existing literature often using daily data, we match hourly observations of weather and crime. Our results show that using daily observations overestimates the effect of temperature and underestimates the effect of precipitation on crime and leads to different conclusions about the significance of variables. In addition, we document evidence for a nonlinear relationship between weather variables and crime. Again, results differ greatly between daily and hourly observations.