If one were to summarise in one sentence the results of the extensive econometric literature on deterrence that has developed since Becker (1968) published his economic theory of crime, that sentence might be something like this: The probability of punishment seems to be an important deterrent, whereas the evidence regarding the severity of formal punishment is much less conclusive [e.g., see Elliot (1977) and Rasmussen and Benson (1994) for literature reviews reaching this conclusion]. Precisely the same conclusions seem to apply to the crime of DUI. Factors that enhance the probability that a driver who drinks will be stopped by the police seem to significantly reduce drunk driving-related fatalities. This result seems to be quite robust. On the other hand, findings regarding laws that mandate various minimum levels of punishment are much less robust, although they seem to be significant for some specifications. One possible explanation of this is that when the chance of detection and/or arrest is small enough, increases in the severity of punishment may have no marginal effect on deterrence [Cook (1980)]. In this light, finding any law enforcement deterrent effect for drunk driving might, in fact, be surprising. After all, it does not seem that any sort of consistent, long-term, systematic law enforcement effort has been made against drunk driving. Certainly, many laws have been passed, but if police do not actively pursue offenders, such laws may have no long-term consequences. Policing agencies do occasionally develop short-term programs that target drunk drivers, setting up temporary sobriety checkpoints, temporarily increasing patrols during holidays or weekends, and so on, of course, but it seems that, in general, a DUI arrest is probably relatively unlikely unless an accident actually occurs [Benson et al. (forthcoming)]. At least in some jurisdictions, police are more likely to escort or drive a drunk person home than to arrest him, for instance, unless an accident occurred. The conclusions of this study are in contrast to the conventional wisdom that has developed from econometric studies focusing on alcohol taxes as a deterrent. The conclusions echo those of other recent studies [Kenkel (1996); Chaloupka and Wechsler (1996); Mast et al. (1998)], however: Specifically, it seems that direct law-enforcement efforts could be an effective way to significantly reduce drunk driving if policies were implemented to produce systematic, persistent, and consistent increases in the probability of being stopped and arrested for drunk driving (and perhaps in the severity of punishment). At the very least, such policies are deserving of much more attention than they have been getting. (Author/publisher)
Samenvatting