Another for the “unintended consequences” file
Washington, DC, has a juvenile curfew law. Anyone “under the age of 17 cannot remain in or on a street, park or other outdoor public place, in a vehicle or on the premises of any establishment within the District of Columbia during curfew hours.” There are exemptions for juveniles accompanied by a parent and for travel for jobs (no detours allowed.)
Curfew laws keep some juveniles off the streets during curfew hours, but which ones? The criminals seem the least likely to be deterred, and with fewer people on the street, perhaps the criminals are emboldened.
The DC curfew switches from midnight to 11 pm on Sept 1 of every year. In a working paper, Jennifer L. Doleac and Jillian Carr test the effect of DCs juvenile curfew on gun violence by looking at the number of gunshots heard in the 11pm to midnight “switching hour” just before and just after Sept 1. From a summary:
The September 1 change provides a clean natural experiment. If curfews reduce gun violence, then when the curfew shifts to 11:00 p.m. rather than midnight, gunfire between 11:00 p.m. and midnight should go down. Does it?
Just the opposite. Using data on gunfire incidents from ShotSpotter (acoustic gunshot sensors that cover the most violent neighborhoods in D.C.), we find that after the curfew switches from midnight to 11:00 p.m., the number of gunshot incidents increases by 150 percent during the 11:00 p.m. hour.
This amounts to 7 additional gunfire incidents city-wide per week, during that hour alone. Jane Jacobs was right: the deterrent effect of having lots of people out on the streets is powerful. This makes juvenile curfew policies counter-productive.
The use of ShotSpotter data is innovative and avoids some problems with issues of police enforcement. Calls to 911, however, don’t show the same pattern as the ShotSpotter data, which is worrying.
I’d also like to see more information on the proposed mechanism. Is it really the case that significantly fewer people are out on the streets at say 11:30 pm after the curfew has been lowered to 11 pm than when the curfew was set at midnight? The curfew only directly affects people under 17 and, as noted above, there are quite a few exemptions. Also what are the ages of those typically arrested on the basis of ShotSpotter alerts?
By the way, on a typical day in DC there are almost 15 gunfire incidents heard by ShotSpotter (data here, the authors report 8 but that may be from a restricted sample). A lot of gunfire is heard around a handful of schools. The ShotSpotter system is quite accurate. Although it misses some shots it distinguishes shots from car backfires better than people do. I also found this note from the Washington Post amusing, in a frightening way:
About a third of detected gunshot incidents in the city happen on New Year’s Eve or around July 4. Officials explain the high rate as celebratory gunfire.
Alex Tabarrok is a professor of economics at George Mason University. He blogs at Marginal Revolution with Tyler Cowen.