John Donohue and I have weighed in again on the concealed-handgun debate. (You can read previous writings on this subject here, here, and here.) This time we have responded to an empirical article by Carlisle Moody and Thomas Marvell, who claim they are “confident” that “the evidence, such as it is, seems to support the hypothesis that the shall-issue law is generally beneficial with respect to its overall long-run effect on crime.”
We point out in our response that the M&M article might have been more credible if
1) The authors had interpreted the sign of the coefficients correctly:
They state that “14 states experienced cumulative benefits while 10 states experienced cumulative costs.” Unfortunately, they have the numbers backwards: 14 of the 24 states are shown in Moody and Marvell’s Table 10 to have cumulative costs; that is, according to their own estimates, RTC laws lead to higher crime costs for the majority of states!
2) The authors had not found an implausibly high crime reduction in Florida:
Specifically, Moody and Marvell’s analysis indicates that the overall impact of RTC laws on crime through 2000 has been to lower crime by $28 billion nationally. But the same table reveals that Florida’s RTC laws alone experienced a crime cost reduction through 2000 of almost $31 billion. In other words, across the 24 states that they analyze, they attribute a benefit of almost $31 billion to the Florida RTC law and estimate an overall harmful effect of about $3 billion of RTC laws across the other 23 jurisdictions. So much for “generally beneficial.”
Now let us pause to reflect on this finding for a moment. If you had an intervention that had a net harmful effect in 23 out of 24 jurisdictions, while at the same time you estimated a massive benefit from the same intervention in only one state, would you assert that the intervention was “generally beneficial”?
3) The authors had coded the underlying data correctly (see p. 51 of our response).
As for now, the best empirical evidence still does not support the “more guns, less crime hypothesis.”
I’m often asked to reconcile this conclusion with the results of my LoJack article. You see, in an article with Steve Levitt, we showed that LoJack seemed to have a large deterrence effect on auto theft. Thieves (especially pros relative to joy riders) are less likely to take cars in a city like Boston, where a sizable fraction of the cars have LoJack. Unobservable precautions, like LoJack and silent alarms, can deter crime generally because potential criminals don’t know at the point of committing the crime whether their particular victim is protected or not. In contrast, a precaution like the Club, which is observable to the potential thief, probably just shifts crime to other victims.
But if our LoJack paper is correct, why wouldn’t we expect to see a similar crime-reducing effect from concealed handguns? Concealed weapons can also be a type of unobservable precaution that can deter potential criminals from committing crimes. Why doesn’t my concealed-handgun empiricism find similar reductions in crime when state laws make it easier to carry concealed handguns?
A powerful answer to this question comes from none other than the New York Knicks basketball coach Mike D’Antoni. As reported in last week’s New York Times:
Mike D’Antoni was thrilled to see Nate Robinson break out of his shooting slump Wednesday, but not so pleased with his behavior. Robinson — who was on the bench at the time — celebrated a second-quarter David Lee dunk by skipping down the baseline and bumping the Suns’ Amare Stoudemire, who had fouled Lee on the play. Robinson was assessed a technical foul, his sixth of the season. “To be honest with you, that’s why you don’t have concealed weapons, because I’d have shot him at that point,” D’Antoni said wryly. “I do like his feistiness, but he just needs to channel it in the right way. And he knows that.”
D’Antoni knows that the problem with a concealed weapon is that it can easily change from a defensive “unobservable precaution” to an offensive weapon to commit crimes. People are not as likely to get angry and rip a LoJack out of their car to beat someone with it.