Gun Violence

The sheer senseless brutality of the mass killings in Buffalo and Uvalde have set in motion a familiar rhetorical dance.

Advocates of stricter gun control accuse opponents of pandering to the gun lobby, or putting selfish interests ahead of the lives of innocents. Opponents respond by accusing advocates of trying to strip law-abiding citizens of their property and constitutionally protected rights. In less polite corners of the social media space, people are accused of being child killers on the one hand or gun-grabbing tyrants on the other. It’s a depressing situation, revealing an enormous cultural and conceptual gulf that seems impossible to bridge.

But we must to try and bridge it.

In this post I want to consider what kinds of gun control initiatives might be appealing, or at least acceptable, to a rural gun owner such as Bart Barber, the author of a twitter thread that has achieved wide circulation. But before I get there, it might be useful to discuss the manner in which statistics have been used and abused in this debate.

Consider, for example this piece from CNN, which claims as “indisputable fact” that “where there are more guns, there are more gun deaths.” The data offered in support of this claim are at the state level, where total gun deaths do indeed correlate positively and strongly with gun ownership, provided one pools together suicides and homicides. The article does indeed acknowledge that it is doing so, but the snippets on social media that repeat the claim often omit this fact.

The result is to create a misleading impression that gun violence cannot be effectively reduced without reducing gun ownership. While this claim may be true for suicides, there are other policies that could be effective for homicides, for reasons discussed below. And this is important to keep in mind, because substantially reducing gun ownership is not going to be possible in America for the foreseeable future.

The positive association between gun ownership and gun deaths at the state level is driven largely by suicides. The following chart shows gun deaths (suicides and homicides combined) and gun ownership, based on ownership data from the RAND Corporation and mortality data from the CDC (both for 2016, the last available year for ownership data). The positive relation claimed in the CNN article is clearly discernible:

But what if we look at homicide and suicide separately? Then you see that the positive association holds even more tightly for suicides (shown in red) but breaks down almost entirely for homicides (in yellow):

Here we see a cluster of states with very high ownership and very low gun homicide rates, comparable to homicide rates in states with the lowest rates of gun ownership. Each yellow dot has a corresponding red dot vertically above (or in rare cases below) it, with each pair corresponding to a given state. Montana, with the highest ownership rate, has among the highest suicide rates but among the lowest rates of gun homicide. Several states exhibit very similar patterns.

Four states stand out for different reasons: Maryland, Illinois, New Jersey, and Louisiana are highly unusual in having lower suicide than homicide rates.

One factor in accounting for these patterns is that homicides are much more likely to be committed with weapons transported from other states, and with guns that have been lost or stolen. Policies that make it harder to use a lost or stolen gun should have a negligible impact on ownership, but could have a significant impact on homicides.

What might these policies look like? Several are discussed in my book with Dan O’Flaherty, but I will mention two.

Most phones now use biometric recognition technologies that make it difficult for them to be used by others if lost or stolen. The same technologies exist for guns. In addition, guns can be restricted to use by legal owners using radio frequency identification tokens that enable communication with bracelets or watches. Encouraging, subsidizing, or mandating the use of smart guns could result in fewer homicides, as well as fewer accidental discharges, without meaningful reductions in ownership.

Another approach is to make legal owners liable for injuries or deaths that arise from the use of guns that have not been reported lost or stolen. The costs of insuring against these claims would create strong incentives for safe storage, while also encouraging the adoption of smart guns.

These policies will have negligible effects on gun suicides, which need to be addressed by other means. They would also have failed to prevent many of the mass killings that have made headlines in recent years. But mass killings are like airplane crashes—dramatic events that attract media attention and calls for action—while the more routine homicides that occur on a daily basis are like traffic fatalities, more numerous but easier to ignore. A focus on making lost and stolen weapons harder to procure and use may not have much impact on the mass casualty events, but they could make a real difference in reducing the kinds of killings that remain under the radar. And unlike the more muscular policies that seem to have support in polling but not in state referendum results, the targeting of lost and stolen weapons could actually find widespread acceptance that cuts across hardened ideological lines. Or so one would hope.

---

I’m updating this post in response to various comments online and below.

1. Jeff Cotner points out that RAND uses gun suicides as an input for the computation of the ownership measure, so the correlation with suicides could be spurious. This is also true of the correlation with the composite measure of gun deaths (suicides and homicides combined), another reason to avoid it. But the point in my post about the lack of correlation with homicides remains valid.

2. Jeff also asked for separate figures for homicide and suicide, which I have posted here and here.

3. Many have pointed out that smart guns are more expensive, which is true, and I shouldn’t have said that their adoption would be “without meaningful reductions in ownership.” There will be a price effect on demand. But I expect that these technologies will become cheaper as adoption grows, just as we have seen with solar power. The price differential can also be reduced by differential taxes, or by differential insurance premiums (if people are required to insure against unlawful use by others).

4. A commenter below notes that biometric recognition introduces a new failure point. This is especially true in wet conditions, which is why the use of backup systems such as radio frequency identification tokens could be important.

5. Robert VerBruggen points out that he and others have noted the absence of correlation between gun ownership and gun homicides many times before. This is not surprising, it took just an hour or so of working with publicly available data to generate the figures in this post. The more important message, I think, is that we should be able to reduce homicides without reducing ownership, if we focus on making sure guns remain in the hands of those who have purchased them legally for their own private use.