A New Look at the Effect of DUI Enforcement on Alcohol Impaired Accidents and Fatalities: A Dynamic Panel Estimation

Economists have long been interested in crime.  Since Becker's seminal work in 1968, a fundamental assumption in economic models of crime is that crime is rational, and economic agents are believed to respond to incentives.  Over the last 30 years, one of the most costly and highly studied criminal activities is drinking and driving.  For context, since 1982 an average of 19,168 motorists die in automobile accidents attributed to impaired driving each year, which is an average of 45.07% of all traffic fatalities.  To mitigate these significant social costs a number of laws have been put in place and there is no doubt that overall drinking and driving has been deterred.  However, there is very little empirical evidence that one of the most costly deterrent measures, increased enforcement, actually reduces fatalities.  In this paper we reevaluate the impact of enforcement on impaired driving using a unique, micro-level drinking and driving dataset from the state of Washington.  Specifically, we have constructed a panel level dataset using data from a number of Washington state agencies including: the Washington Traffic Safety Commission, the Washington Association of Sheriffs and Police Chiefs, and the Washington State Patrol.  A novel feature of this dataset is a unique variable to proxy for DUI enforcement, the actual blood alcohol concentration (BAC) for each arrested motorist.  This richer source of information is valuable because it allows us to break enforcement into multiple groups based on impairment levels (unimpaired, marginally impaired and heavily impaired motorists) and thus disentangle the magnitude of enforcement and the underlying propensity to drinking and drive.  Preliminary results using a dynamic panel model suggest that analysis using standard arrest data do not find an impact of enforcement on fatalities, but conditioning on impairment levels we find that enforcement does deter marginally impaired motorists.