Technological improvement and the externality to drunk-driving

Levitt and Porter (Journal of Political Economy, 2001) demonstrate that both the fraction of drunk-drivers and their relative risk of a fatal crash can be identified using information about the blood alcohol content of drivers involved in fatal two-car accidents. Applying their approach to data on fatal accidents in the United States from 1983 to 1990, they estimate that drivers with a blood alcohol content above 0.10 are 13 times more likely to cause a fatal accident and that the Pigouvian tax should be set at $8,000 per arrest to fully internalize the externality.

Given the many improvements in both vehicle safety technology and road design, in this paper, we repeat Levitt and Porter’s estimation for the two driver types drunk and sober with a more recent FARS sample period.  Across a number of model specifications, our results suggest that the relative risk of drunk to sober drivers was approximately 50% greater during 2000-2009 compared to 1983-1993.

One possible explanation is that since we do not account for non-fatal accidents, a greater proportion of sober driver’s accidents that would have been fatal in the earlier period were non-fatal in the later period, relative to drunk drivers.  This would be consistent with technological innovations in automobile safety that were more effective with sober drivers. For example, air bags that automatically deploy upon impact can reduce the risk of death if drivers are wearing a seatbelt. If sober drivers were more likely to wear seat belts, this technology would have a greater impact on the fatality rate of sober drivers. This hypothesis is tested using the FARS data and implications for public policy design are explored.