Policies to reduce drunk driving do not appear to work. There are still many accidents caused by them, and threats of imprisonment (rarely credible) or loss of driving privileges (recanted when the offender needs to drive for a living) do not have the necessary bite. Indeed, drunk driving has only real consequences when an accident occurs. Are there better solutions?
An economist always looks whether there are market based solutions that would properly drive incentives. In particular, one would want (potential) drunk drivers to internalize the cost they exert on others with their behavior. That is, one would want to impose the appropriate tax. Steven Levitt and Jack Porter looked at this in their 2001 JPE piece. Considering only fatal crashes, they estimate that drivers with alcohol in their blood multiply the probability of a crash by 7, legally drunk (above (0.10%) ones even by 13. Now, using fatalities and traffic statistics, as well as measures of the statistical value of life, they then are argue that driving with alcohol should be taxed at US$0.15 a mile, the double for legally drunk, for the costs of fatalities to be covered. Of course, such a tax would be impossible to enforce. But one could fine people when caught, even when no accident is involved. That fine would amount to US$8000,given typical arrest rates. Now enforce that.