JDN 2457306 EDT 16:44

The United States is unique among the First World in its rate of mass shootings. We have on average two per week, while most other countries have one every couple of years. Of course, part of that is our huge population; if you control for population we don't actually have a particularly high number of mass shooting deaths—and there is no country in the world where mass shootings are a major cause of death. Mass shootings are dramatic, but not actually something worth worrying about—like terrorism.

Shootings in general, on the other hand, are a major cause of death. In the United States, over 11,000 people were killed by shootings last year. That's not enough to put it in the top 10 causes of death, but it is in the top 20.

The US is also unique in our rate of gun homicides, with about 30 gun homicides per million people per year compared to 2 or 3 gun homicides per million people per year in countries like the UK, France, and Germany. (If you read homicide statistics they are usually reported as “per 100,000” to one decimal place; I think this is ridiculous both for having unnecessary decimals and for suggesting an unreasonable sample size. I always like to report in “per million”, which is just the same thing without the decimal point.) Among OECD countries we are second only to Mexico; Chile and Turkey are the only ones even close, and they are in the OECD but not generally considered First World countries.

Contrary to the talking point “Without guns, murderers would just use knives”, the overall homicide rate is also much higher in the US than most other First World countries. At about 47 homicides per million people per year, we actually seem more like an Eastern European Second World country like Latvia (47) or Estonia (50) than we do Western European First World countries like Germany (8) or France (10) or the UK (10). Even safer are countries like Japan (3) and Singapore (2). It may be that attempted murders are conducted with knives in these countries, but knives simply aren't as effective at killing people as guns are (which is kind of why guns were invented in the first place!).

It could be income inequality rather than guns; the US is also an outlier in our level of income inequality. I think the Gini coefficient is a good measure for this purpose, as it gives a sense of the gap between the upper-middle and lower-middle classes, where lower-middle class people are substantially more likely to be the perpetrators—or victims—of homicide. Gini coefficients range from 0% (everyone has exactly the same income) to 100% (one person has all the income); in a later post perhaps I can explain more of the details of how this calculation is made.

The United States has a Gini coefficient of 41%, comparable to Russia's 40%, Lithuania's 38%, at Latvia's 37%. Once again we look like a Second World country, so this theory is doing pretty well so far. But other First World countries actually vary a good deal in their Gini coefficients; the UK is at 38%, France is at 33%, Germany is at 31%—and yet none of these countries has a homicide rate even one-third as large as ours. Japan has roughly the same Gini coefficient as the UK and yet one third the homicide rate.

So maybe it's not inequality per se but actually poverty that leads to homicide; poor people are more likely than others to be both the perpetrators and victims of crime. By the UN definition of poverty—Third World poverty, less than $2 per day at purchasing power parity—the number is basically negligible in all First World countries. So instead we could use the national poverty lines, under which France has a poverty rate of 8%, the US has a poverty rate of 15%, the UK has a poverty rate of... wait... 16%, and Germany has a poverty rate of... 16%? The UK and Germany have more poverty than the US? Doesn't seem right, does it?

See, the problem with using national poverty lines is that different countries set them in different places. The US has a very stringent definition of poverty, so even though our poverty rate doesn't look that bad compared to other First World countries we actually have a lot more real poverty. I suspect we are actually calibrating our poverty line based on what makes us look normal rather than actually trying to sensibly measure poverty.

The poverty line in the US is currently $11,770 per year. The poverty line in France is fairly close to that, set at 40% of the median income, which comes to about $12,400 per year. But the poverty line in Germany is defined at 40% of the mean income, which comes to about $17,000 per year PPP. (France's poverty line is set 5% higher than ours; Germany's is 44% higher.) That's why it looks like Germany has more poverty than we do—they count people as poor that we wouldn't. Adjusting for this, it really does look like the US has about two or three times as much poverty as most other First World countries, which could go a long way toward explaining our high homicide rate as well.

But the United States is an outlier in another way: Gun ownership. We are in fact the only country in the world that has more guns than people. We have 113 guns per 100 people. France has 31 guns per 100 people, Germany has 30, while the UK has only 7 guns per 100 people. Japan has less than 1 gun per 100 people.

Even in the US, it's not that most people have guns—only 34% of Americans own guns. It's that people who own guns own a lot of guns—an average of 3.3 to be precise, and in fact this distribution is highly skewed as well, so most people who own guns have 1 or 2 while a small number of people own as many as 50.

I made a graph of guns per 100 people and homicides per million people per year in First World countries, and you may notice something about that trendline:

guns and homicide

But the US is such an outlier that it actually seems like we should remove it (a bit odd to explain gun violence in the US with a model that excludes the US, but otherwise it's basically driving the whole regression!). The result is a much weaker, but still positive, relationship:

guns and homicide exclude US

Alternatively, we can also include some other countries with homicide rates comparable to or larger than the US; I think the most reasonable comparison seems to be with Second World countries such as Latvia, Estonia, Russia, and Kazakhstan; then what we get is a small negative relationship! But if you look at the scatter plot, you can see that it's really not doing much explaining:

guns and homicide Second World

I think the second graph is the most reasonable one; we're comparing similar countries with similar homicide rates, and what we find is a moderate effect: more guns are associated with somewhat more homicides. But really we should probably be controlling for more variables, like GDP, poverty, maybe even unemployment and urbanization. (Which I could do, theoretically; but sorry, I don't think I'm going to do a full multivariate panel regression with robustness analysis just for a blog post. For that kind of work I expect to be paid, or at least offered academic or publication credit.)

Does this mean that we need more gun control in order to stop homicide? Well... maybe. Using the Brady score as a measure of the strength of gun control laws, The Washington Post found basically no correlation between a state's gun control laws and its level of homicide. As far as I can tell this analysis was sound; Brady scores just don't seem to predict homicides. So if gun control is what we need, then Brady scores aren't doing a good job of measuring the strength of gun control. (Contrapositively, if Brady scores are a good measure of gun control, then gun control isn't good at reducing homicides.)

However, there is still a very good reason to reduce the availability of guns, and it's kind of the elephant in the room in all of this: Not homicide, but suicide. Going back to the original First World graph, I've now included suicide rates, to scale at deaths per million people per year:

guns and suicide

Nordic countries like Finland and Norway have higher suicide rates than we might expect because it's cold and dark—for the same reason Alaska also has a high suicide rate. Japan has a very high suicide rate because it has a culture of honorable suicide. Let's exclude those. The US doesn't have an outlier suicide rate, but it still does have an outlier gun ownership rate, so let's still exclude it as well. The same basic relationship holds, though it's slightly weaker:

guns and suicide excludes

Indeed, more rigorous research that controls for other variables supports this analysis: Across US states, higher availability of guns is strongly associated with higher rates of suicide. Rigorous international studies have found a similar relationship across countries.

Even Fox News (consider the previous italics both title and emphasis!) admits that higher rates of gun ownership are associated with higher rates of suicide!

That's one reason why I'm not too offended by the common turn to mental illness whenever there is a mass shooting. It's a totally inaccurate account of what causes mass shootings; but it's also a major problem around the world—and it's not an inaccurate account of what causes suicides. Basically there are two major risk factors for suicide: Depression is number one, and gun ownership is number two. Since suicides vastly outnumber homicides in almost every country (unlike homicide, suicide is in the top 10 causes of death in the US), I'm fine with talking about mental illness when we talk about guns—that is, as long as you're actually serious about talking about mental illness; John Oliver is, but I think he's right that most people actually aren't.

This result is rarely talked about, perhaps because it's counter-intuitive for many people. There is a common narrative that "they'll find another way to kill themselves”, but this is simply not accurate. It is well-documented that suicide rates go down whenever you remove an easy method of committing suicide from the population—be it by putting nets on tall bridges, adding safety interlocks to cars and ovens, or restricting access to guns. Most suicides are impulsive acts, not carefully planned decisions. Making it harder to commit suicide does reduce suicide.

And that, above all, is why we need gun control. It would probably reduce mass shootings (it certainly seemed to in Australia), but that isn't what's really important. The reason we need gun control is that it would reduce suicidebecause the most likely person for your gun to kill isn't a criminal or intruder, and it isn't a member of your family: it's you.

[all the graphs are my own work, made using the data I cited in the text.]

Patrick Julius

Patrick Julius

This email address is being protected from spambots. You need JavaScript enabled to view it.
Recent Articles
Inequality: Fantasy versus reality
Who pays the tax?
Why aren't we colonizing other planets?
We live in the cyberpunk future.
Should we vote for nincompoops?