Abstract
In this paper we study the relationship between air pollution and crime. We construct a
city-level, hourly data set with 2.4 million crimes and link each crime to data on pollution
and controls, e.g. weather variables. We study the effects of four pollutants (nitrogen dioxide,
sulphur dioxide, carbon monoxide, and ozone), as these have different neurotoxic effects.
Our identification strategy relies on using high dimensional fixed effects and exploiting hourly
variations.
We find that carbon monoxide has a positive effect on violent crimes and ozone has a negative
effect on property crimes. These relationships are linear and we only find heterogeneity in the
effect of ozone on property crimes across cities.