Bryan Caplan  

How Staggering is the Briggs-Tabarrok Effect?

PRINT
Language and Social Justice... Bartender, Cashier, Cook, Jani...
Rumor has it that GMU's Justin Briggs and Alex Tabarrok have hammered the final nail into the NRA's coffin.  Zack Beauchamp of Think Progress explains:
A new study, coauthored by a libertarian-aligned economist, has found strong evidence that the spread of gun ownership around the United States is a threat to public health...

Tabarrok and his coauthor, Justin Briggs, put together a bunch of data on gun ownership and suicide. After controlling for a series of potentially confounding factors, Tabarrok and Briggs ran a series of regressions to establish any links between guns and suicides.

Their results were staggering.
Here is Tabarrok's summary* of the study's results:
Using a variety of techniques and data we estimate that a 1 percentage point increase in the household gun ownership rate leads to a .5 to .9% increase in suicides.
Note the lack of superlatives.  So how "staggering" is this Briggs-Tabarrok Effect? 

To answer, you need to know two additional facts.

1. The total number of suicides in the U.S. is roughly 40,000 per year (38,285 for 2011).

2. There are about 310 million people in the United States.

Thus, the Briggs-Tabarrok effect says that depriving 3,100,000 people of their guns (a 1 percentage-point decrease in the gun ownership rate) would save about 200-360 lives (.5*40,000=200; .9*40,000=360).  In ratio form, the Briggs-Tabarrok effect says that to prevent a single suicide, 8,600 to 15,500 people - the vast majority of whom are not suicidal - must lose their guns.

Is that a good deal?  A standard $7M value of life implies a critical value of gun ownership between $452 and $814 per person per year.  If the marginal person's value of gun ownership is less than that, gun deprivation passes the cost-benefit test.  If the marginal person's annual value of gun ownership exceeds that, gun deprivation fails the cost-benefit test.  Note that this is not a value per gun; it is a value per person of having any guns.

If, like me, you've never held a gun, this might sound like a no-brainer.  How could anyone value gun ownership so highly?  If that's what you think, though, you really need to get out of your Bubble.  About 40% of American households own guns.  Self-defense aside, firearms are the foundation for several of the most popular hobbies in America - shooting and hunting for starters.  Anyone who rails against "gun nuts" can hardly deny that many folks adore gun ownership.

Not convinced?  The Briggs-Tabarrok story implies that you can unilaterally cut your family's suicide rate by getting rid of your family's guns.  (The same does not hold, of course, for your family's murder rate).  So suppose every gun owner in America were fully aware of the Briggs-Tabarrok Effect.  How many of these gun owners would get rid of their guns in order to avoid an extra 0.01% chance of suicide for each person with access to their guns?  How many would deem such a risk "staggering"?  If you say, "Not many," you are conceding that most gun-owners value their guns more than a tiny risk to those they hold dear. 

But is a pure cost-benefit approach to gun suicides even appropriate?  Probably not.  Everyone makes fun-but-risky choices - on diet, lifestyle, and sex for starters.  The risks you take affect not only you, but the people who care about you.  Many are far riskier than the Briggs-Tabarrok Effect.  Yet almost thinks it's wrong to use cost-benefit analysis to veto these personal decisions: "My body, my choice."  So why single out gun owners for welfare-enhancing persecution? 

If you personally know a lot of gun owners, the Briggs-Tabarrok Effect should concern you.  Accessible guns probably do slightly increase the chance that someone you care about will kill himself.  So be more cautious and spread the word.  But if you don't personally know a lot of gun owners, you should mind your own business.  Gun owners reasonably discount sermons about "staggering" risks from people who utterly fail to appreciate their hobby.  Want to help people?  Focus your nudging on risky activities prevalent among the people you personally know.  You'll sound less intolerant, be more persuasive, and do more good.

* Beauchamp accurately quotes Tabarrok's initial claim that "A 1% increase in the household gun ownership rate leads to a .5 to .9% increase in suicides."  Tabarrok revised his post after I emailed him for clarification.


Comments and Sharing





COMMENTS (29 to date)
Skinny Nick writes:

It would be one thing if gun ownership was correlated with violence toward others and other generally bad social effects (it might be i dont know). But beyond not being a simple non-aggression principle issue, I don't think suicide can just be put in the same category of purely bad outcomes.

Aren't progressive the same folks that supported Dr. Jack and assisted suicide? Why is fundamentally always wrong or bad for someone to commit suicide? I doubt that a great deal of the people killing themselves are in great physical health with a long and happy life in front of them (which i assume the 7M valuation holds).

After watching my grandfather die recently from super ugly cancer at home, my whole family has really rethought this issue. There was no value and lots of trauma involved in dying in a "natural" fashion from fatal, painful illness. I would have hated for his theoretical death 2 months earlier via gunshot to be counted somehow as proof of a terrible outcome and one we can prevent via gun control.

Finch writes:

The Briggs-Tabarrok Effect usefully suggests you may want to make you guns less accessible in the event of an impulsive decision. If your guns are for hunting or shooting sports and not self defense, storing them at a gun club or cabin makes them inopportune if you're intoxicated on that one really bad day.

Further, the curious lack of effective substitution of methods (which further reinforces the idea that suicide is generally a really bad decision on a really bad day, and not some long-term desire to not be alive) suggests you might not want guns around if you have people in your household that might be at elevated risk of bad impulsive decisions, such as teenagers, alcoholics, drug users, or the mentally ill or depressed.

I like your analysis above and agree that this is not strong evidence of some need for gun control. But it is a really interesting and potentially useful result at the household level. If you are not the median household, it may have implications for you.

MG writes:

The cost benefit analysis assumes a standard value of a life. I suppose this value must assume that the living wants to continue doing so. Persuade me that this is the correct (and not the highest, if not irrelevant) value when dealing with suicide.

CMOT writes:

[Comment removed for irrelevance. --Econlib Ed.]

LemmusLemmus writes:

I don't think I understand your maths. The estimate refers to "the household gun ownership rate", yet you're using the population total as your denominator.

Andrew_FL writes:

If you believe in a person's right to decide what to do with their own life, the argument is not persuasive at all. Suicide should be as fundamental a right as the ability to do drugs, in the liberal mind, no? I would estimate that there is a high probability that many TP readers are in favor of drug legalization.*

It's intellectually disingenuous or dishonest for one to claim support for one as a fundamental right and the other not. If it is sincere it is incoherent and inconsistent thinking.

*For the record I hold no particular attachment to drugs being illegal although I don't place much priority on changing those laws. And I think suicide is wrong in a moral sense but it should not be something society writ large engages in trying to stop through force of law.

Paul Crowley writes:

Scott Alexander presents excellent arguments in favour of trying to prevent suicide. This doesn't counter what Bryan argues above, but it does answer some of the comments.

paulanz writes:

Shouldn't the next step be to compare this negative effect to the positive effects of gun ownership?

If Briggs-Tabarrok can look into the past and determine that X suicides would never have happened with out guns, surely the same can be done analyzing home invasions, riots, other situations where people WOULD have been killed if they didn't have a gun.

Has anyone been able to study and estimate lives saved by guns using decent data and proper analysis?

I'd like to see their research put in an even wider context.

Having said that, it seems like common sense that the availability of guns and ammunition increase the likelihood of a 'successful' suicide, grim as that thought may be.

Ari Tai writes:

Are there data comparing suicide rates in other societies where there are fewer guns? I seem to recall that once someone becomes suicidal, the choice of a means is secondary.

In Japan for instance there are news reports of people brewing up deadly gasses from cleaning supplies in their bathrooms in a hope for a quick and painless end. Bad news is they often take someone else with them. And for the more educated, making a suicide look like an accident (running into another car perhaps) has a better insurance pay-out.

Another instance of driving with the windshield painted black, looking out the rear view mirror for guidance (an abuse of statistics given the future is largely unknowable).

Tracy W writes:
Yet almost [no one?] thinks it's wrong to use cost-benefit analysis to veto these personal decisions: "My body, my choice."

My first response to this statement was "Can I come and live on your planet?
(Then I noticed it was missing a word, if the missing word is "everyone" then we are on the same planet.)

drycreekboy writes:

@Ari Tai

"I seem to recall that once someone becomes suicidal, the choice of a means is secondary."

The difference guns make, in theory, is that it makes it easier to successfully act on a short-term or transient suicidal impulse. Guns don't cause the attempt, they just make it more likely the attempt is successful.


You are absolutely correct that for people with a fixed, non-transient suicidal the availability of guns is a secondary concern.

Teasing out the causality is difficult for many reasons, not the least of which is that guns are a good "option" for people in both categories and anywhere in between.

As to your second point, the United States does not have the world's largest suicide rate. Japan, which has a gun-control regime that leaves even many European countries in the dust, is, as many have noted, has a rate nearly double ours. Other countries where private gun-ownership/violence is relatively rare (or even illegal) that lead us in suicides include: France, Belgium, China, South Korea, and get this, Cuba. The number one country is Greenland. Great Britain is only just behind us, and they've spent a generation making private ownership of firearms an order of magnitude more difficult than here. Other societies where firearm ownership/possession is common, though more regulated, that trail us in suicide are Switzerland and New Zealand.

http://en.wikipedia.org/wiki/List_of_countries_by_suicide_rate#cite_note-18

drycreekboy writes:

I followed the discussion at MR, and just found it frustrating. There were gun rights advocates (among whom I number) who wanted to deny the effect, and gun control advocates who wanted to over-interpret it.

To me it's no surprise that Tabarrok and Branch found a relationship. Guns are a very effective way to kill yourself, and any society where there are lots of them means that they are going to be the instrument in a great many successful suicides. Yet the effect is obviously small when calculated on a household or per capita basis. If you cross-reference the Gallup numbers with Census data you see you'd have to get a net increase of 470K households owning guns to get this effect. I wonder what would happen to the number of drunk driving fatalities if you had a 1 % increase in the number of people who consumed x amount of alcohol a week? Ditto for a 1 % increase in the number of Five Guys/Hardees/Carl's Jr franchises and fatal heart attacks. I bet the effects would be similarly robust, but similarly modest on a per capita basis.

Another issue, since the paper is gated I don't know if they did any research on how these rates changed over time. Gun ownership, as a percentage of households, is considerably less today than it was even a generation ago (an artifact, possibly, in the rise of female-headed households). Are suicides higher or lower today than say, the 1950s, or certain benchmarks like the National Firearms Act of 1934 or the Gun Control Act of 1968?

I'm not a trained econometrician by any means, but an historical understanding might give us a better idea of just how tight the causal relationship is. As other countries show, stringent gun control is no barrier to suicide rates much higher than ours.

Mike Rulle writes:

This study would be perfect grist for Deirdre McCloskey's "The Cult of Significance". The study is absurd on its face. We live in a country of 310 million people with massively diverse backgrounds and circumstances. These authors believe they can "hold constant" all potential confounding variables----itself a changing set of phenomena---to determine we would "prevent" 300 suicides if no one had guns? I wonder if they "Monte Carlo'd" all the potential counterfactuals caused by no guns.


I bet the study also had a p-value of .001.

Mike Rulle writes:

Update of my last post.

300 saved suicides per 1% decrease in gun ownership---not no guns. My comment otherwise stays the same.

OneEyedMan writes:

Like, LemmusLemmus, I am confused. The elasticity is with respect to the households that own guns, which is something like 150 million people (likely they're bigger households than the average). The argument still stands but the magnitudes are different.

Rewriting the paragraph with these numbers...Thus the Briggs-Tabarrok effect says that depriving 1,500,000 people of their guns (a 1 percentage-point decrease in the gun ownership rate among _households_) would save about 200-360 lives (as before). In ratio form, the Briggs-Tabarrok effect says that to prevent a single suicide, 4,150 to 7,500 people - the vast majority of whom are not suicidal - must lose their guns.

Silas Barta writes:
The Briggs-Tabarrok story implies that you can unilaterally cut your family's suicide rate by getting rid of your family's guns.

That's nothing! You can cut your own suicide rate 100% by not killing yourself!

Arthur_500 writes:

Statistics can prove anything! I know, the first day of statistics we are told that we disprove the negative and we prove nothing.

However, if you construct your study carefully you can prove anything. So what is correct and what is wrong about this study?

Obviously, according to the study, Americans are different from the rest of the world. In places where firearms are not available, such as in Japan, there is a higher percentage of suicides in the population than in the US. It must be American gun ownership that drives the Japanese to commit suicide.

In places where the rates of gun access are lower but still available the US pretty much has the same rates of suicide. So where is the difference in the US that necessitates further research into firearm ownership?

Assuming that the purpose of this study is to prove that firearms ownership causes suicide and therefore firearms need to be removed from society we must recognize that this is not a scientific study at all as the conclusions were there from the start. All we have done is seen the use of mathematics to prove that guns are bad, spoons cause fat, etc.

So we get to the conclusion that if you put everyone in jail then only the bad guys will be on the street and they can easily be rounded up and put in jail. Great study. I wonder if my tax dollars went into the development of this piece.

Now what is the relationship of vehicle ownership to death? Motorcycle ownership to death? We already went through the same relationship with bicycle helmets and brain injury (negligible). I guess what we realize is that math can prove that living is risky and no one gets out alive.

David R. Henderson writes:

@Bryan,
3rd last paragraph: "Yet almost thinks” should be “Yet almost everyone thinks"

john writes:

I agree w @paulanz. Extensive data and analysis should laid out to see the real effects

MingoV writes:

There's another factor that has not been considered. Gun-related suicides usually are easy to distinguish from accidents. Many other suicide attempts are misclassified as accidents or are not identified as suicides at all. If suicidal people do not have access to guns, they will use other methods with decreased likelihoods of being classified as suicides. The number of suicides will not change, but the number of identified suicides will decline.

Mr. Econotarian writes:

The NRA handgun safety course tells you that if you have small children in your house or someone who is depressed, that it is very important to safeguard your firearm. I think everyone agrees with that.

I had a friend who ended up committing suicide by a shotgun purchased for skeet shooting. Only after he died did we find out from that family that he was deeply depressed, and had attempted suicide several other times (including vehicular carbon monoxide). If we knew that, I wonder if we could have encouraged him not to keep his gun in his apartment. But then I also recognize that he would have eventually jumped off a bridge or something if his depression continued. And there was no Internet back then, so knowledge of the helium technique was not widely known.

It should be noted that the guns often used in suicide are not the guns most politicians like to complain about. 20 round magazines will not help you shoot yourself more than a single-shot shotgun. A revolver will do as well as an automatic, and is cheaper.

Andrew_FL writes:

@drycreekboy-"More regulated" is a funny way to describe gun ownership being mandatory. (In Switzerland, anyway).

@Paul Crowley-I'm in favor of people helping suicidal individuals. I don't think that conflicts with viewing the decision as ultimately a fundamental right.

I seem to recall reading somewhere, incidentally, that the most common means of committing suicide varied a great deal depending on sex. Suicidal women favored poison, I think men guns. I can't remember. If so I would guess those "prevented suicides" would, in theory, be mostly men. Hm.

At any rate it's wort noting that if "we" decided suicide is a societal problem that society writ large needs to handle, it ought to be that "we" address the issue directly, rather than try to deal with it at the margin with some largely impotent means involving massive violations of the rights of vast numbers of non-suicidal individuals.

yarbel writes:

Owning a gun creates an externality. The reasoning in the post is that the risk is very small and people internalize it: "How many of these gun owners would get rid of their guns in order to avoid an extra 0.01% chance of suicide for each person with access to their guns? "

But it stands to reason that there would be some correlation between not caring enough about the externality and having a family member who's prone to commit suicide. If so, I wouldn't count on people to make the right judgement call and the hypothetical may not work. (admittedly, this objection does not negate the cost-benefit analysis).

drycreekboy writes:

@Andrew_FL

"@drycreekboy-"More regulated" is a funny way to describe gun ownership being mandatory. (In Switzerland, anyway)."

Point well taken, I'm just trying to be charitable to the gun-control side. *Automatic rifle possession* is indeed mandatory in Switzerland if you are of a military age. However, you are required by law to store the weapon safely and so forth. Who knows how closely that law is enforced or obeyed.

My understanding is that after you age out of military service you can buy your service weapon, but they require it to made semi-auto. Don't know if there is a mag-capacity limit or not (something else simple to get around if you want).

drycreekboy writes:

@ Mr. Econotarian:

"It should be noted that the guns often used in suicide are not the guns most politicians like to complain about. 20 round magazines will not help you shoot yourself more than a single-shot shotgun. A revolver will do as well as an automatic, and is cheaper."

That's a very important point in a country with ~300 million private firearms. The effect in the paper would presumably extend as much to an antique dueling pistol as to Glock 21. By the time we managed to reduce firearm ownership by enough make a significant difference, say >5 %, what intervening changes in society might have driven the suicide rate up or down. How long would that take to accomplish? The costs of the time and trouble of achieving the effect ought to figure into any CBA.

JdL writes:

If someone wants to kill himself, is doing so a bad thing? The author doesn't address this point.

Don writes:

You do know that the attempt to remove firearms from many gun owners will cause a spike in homicides, right? Won't this cancel out any number of lives saved from suicide?

MrMark writes:

Read a study about suicide rates in Canada post their increased restrictions on firearms ownership. Gun related suicides did fall, while suicide by hanging rose. The overall rate did not change substantially. Perhaps they should outlaw rope as well.

Wheylous writes:

I know I'm late to the game, and someone might have beaten me already, but it seems that the meaningful, fundamental factor behind suicides isn't gun ownership, but rather personal life circumstances and mental health. Trying to reduce the incidence of suicide by restricting guns is like trying to reduce the rate of homelessness by smashing their shopping carts so they leave - a project that was personally undertaken by a Hawaiian lawmaker, no joke: http://www.cnn.com/2013/11/20/us/hawaii-homeless-lawmaker-carts/

It's not reasonable to think that increased gun ownership is the causal factor behind suicide, but only a facilitating factor. The real cause of suicide is probably poor mental health, stress, and poor life conditions. Address the root, not the symptom.

Comments for this entry have been closed
Return to top