In a new working paper from the National Bureau of Economic Research, visiting Harvard economist Jeffrey Miron and Yale Law School student Elina Tetelbaum cast doubt on the conventional wisdom that pressuring states to raise their drinking ages to 21 has substantially reduced traffic fatalities. Overall, they find, adoption of the higher drinking age is associated with an 8 percent drop in traffic fatalities among 18-to-20-year-olds (taking into account vehicle miles traveled), but this effect is concentrated in states that raised their drinking ages on their own, prior to the 1984 law that threatened to withhold highway funds from states that continued allowing people younger than 21 to drink. Even in the early-adopting states, Miron and Tetelbaum say, the drop in fatalities may be partly due to other policies aimed at reducing crashes that were adopted around the same time.
Data from the Monitoring the Future Study (an annual survey of high school students) indicate that the drinking age has, at most, a modest effect on alcohol consumption. A drinking age of 18, Miron and Tetelbaum find, "is associated with an almost 4% increase in drinking participation rates, and approximately a 3% increase in heavy episodic drinking rates." Again, this association is seen mainly in states that raised their drinking ages before 1984, and it may be partly or entirely due to increased underreporting of newly illegal drinking. Notably, the drop in reported drinking found in the Monitoring the Future survey is not accompanied by a drop in reports of alcohol-related accidents.
The legal drinking age of 21 "fails to have the fatality-reducing effects that previous papers have reported," Miron and Tetelbaum conclude. While "most of the variation in the [drinking age] occurred in the 1980s," they note, traffic fatality rates have been "decreasing steadily since 1969," probably due to factors such as improved auto safety standards, medical advances, and better driver training.
[Thanks to Andrew Grossman for the tip.]