For a paper I am working on, I am reviewing some of the econometric work that uses paired adjacent counties that straddle state borders to examine the effects of state-level policies. There is a growing number of such papers, many citing the paper by Dube, Lester, and Reich, 2010, on the effects of minimum wage legislation. One really good one is by Lynn Scholl, who is now at IAB, and is a chapter in her 2011 UC Berkeley School of Public Policy dissertation, Essays on Transportation Safety, Economics, and Policy, and estimates the effects of graduated drivers licenses on teen deaths in auto accidents. By using adjacent counties that straddle state borders, Scholl gets much better control group for teen driver conditions (especially for metro areas, the driving conditions on either side of the border are likely to be very similar and the be changing over time in similar ways). The results are striking.
I find that the strongest GDL programs, as rated by the Insurance Institute for Highway Safety, reduce teen driving related fatal crash rates by 25 to 34 percent and teen driving error related fatal crash rates by 34 to 45 percent. The most effective components of the GDL were early nighttime driving curfews beginning between 6pm to 10pm at night. Passenger restrictions had statistically significant effects only when controlling for the number of licensed teens on the road. For example, the zero to one passenger limit reduced quarterly county-level young teen driver involved fatal crash rates per 100,000 by -6.388 points, relative to a mean rate of 9.5 in state-periods without restrictions. These passenger limits were also highly effective at reducing nighttime crashes among teens, with a reduction in these crash rates of -5.909.
Having an 18 year old who has been driving for two years now, I am so happy California implemented this kind of graduated drivers license. Compared with my youth in Puerto Rico, where we drove in exactly the ways that increased the risk of killing our friends, the driving under the GDL program was, in my sample of one, probably much safer. It is nice to see Scholl’s work confirm and quantify that intuition. And a new research agenda opens up: why are some states not implementing this very effective and presumably fairly costless program.
*Another paper using the same county state-border technique is Rocco Huang, 2008, Evaluating the real effect of bank branching deregulation: Comparing contiguous counties across US state borders.