AN EMPIRICAL CHARACTERIZATION OF THE DYNAMIC EFFECTS OF POLICE SPENDING ON VIOLENT AND PROPERTY CRIME

Abstract

The study estimates the dynamic effects of shocks to police expenditures on measures of violent and property crime rates using annual U.S. state‐level data for the period 1960–2015. We employ a structural panel VAR model and achieve identification by imposing the restriction that police spending responds to structural shocks to crime with at least a lag of 1 year. Results indicate that a shock to police spending leads to (a) persistent and significant decreases in violent and property crime rates and (b) significant and persistent negative impacts on crime rates in periods of high crime but little impacts in periods of low crime. Variance decompositions show that shocks to police spending account for moderate to large proportions of the variability of U.S. state‐level crime rates. Our findings are robust across separate measures of violent and property crime rates, as well as to the inclusion of additional variables to the baseline panel VAR model. (JEL K42)