The Probability of Arrest for Drunk Driving: An Empirical Investigation of a Predator-Prey Model

Abstract

We measure the probability of arrest for drunk driving as the number of arrests in Iowa counties divided by the number of incidents imputed from CDC survey data. Both the number of officers per-capita and their allocation are statistically significant determinants of the probability of arrest, but their elasticities are small. Our rough imputation indicates a comparatively small benefit from hiring an additional officer. This may help explain why arrest rates are so low and suggests that policies that reduce excessive drinking, raise the return to police efforts, and reduce recidivism may be more effective than simply hiring additional officers. (H12; K42; R41)