Monday, October 09, 2017

Gun Control and Homicide Rates

Some recent comment threads on Slate Star Codex, my favorite blog, have dealt with the always lively issue of gun control. One standard argument is "we know gun control laws work because the U.S., which has relatively few restrictions, has a much higher homicide rate than countries such as Canada or the U.K., which have much more restrictions."

One response sometimes offered is that there are other countries, such as Mexico and Brazil, with both restrictive laws and homicide rates much higher than in the U.S. That then gets into the question of what comparisons are more relevant, in what respects the U.K. is more like the U.S. than Brazil is.

An alternative approach, which I think more useful, is to ask whether the difference in homicide rates existed prior to the difference in regulation. The web makes that question much easier to answer than it would have been twenty years ago. 

In the case of the U.K., the answer is pretty clear. According to the Wiki page on Firearms Policy in the U.K., the first restrictive legislation was the pistol act of 1903, but it had little effect:
The Act was more or less ineffective, as anyone wishing to buy a pistol commercially merely had to purchase a licence on demand over the counter from a Post Office before doing so. In addition, it did not regulate private sales of such firearms.
The first  significant restriction was the Firearms Act of 1920. There were additional acts in 1937, 1968, 1988, 1997 and 2006.

The data on Homicide rates per 100,000:

Year U.S.  England&Wales Ratio
1900 1.2 0.96 1.3
1910 4.6 0.81 5.7
1920 6.8 0.83 8.2
1930 8.8 0.75 11.7
*1946 6.4 0.81 7.9
1950 4.6 0.79 5.8
1960 5.1 0.62 8.2
1970 7.9 0.69 11.4
1980 10.2 1.11 9.2
1990 9.4 1.09 8.6
2000 5.5 **1.71 3.2
2010 4.8 1.14 4.2

*No data for the U.K. 1940-1945
**The figure is for the U.K. rather than England and Wales

Looking at those data, it is hard to believe that the reason the U.K. has a lower homicide rate than the U.S. is restrictive legislation.

My point here is not that gun control doesn't (or does) work. I wouldn't be surprised if some restrictions on firearm ownership reduced the homicide rate, but if so, the effect on the U.S./U.K. ratio is lost in the noise. My point is rather that the sort of factoids that show up in this sort of argument, even when they are true, are rarely as solid evidence as those who offer them claim.

This would be a better post if I had a good example on the other side of the same debate. I don't, but perhaps someone reading this can offer one.


Lawrence Kesteloot said...

In the great book "Nine Crazy Ideas in Science" (, chapter 2 investigates gun/crime and concludes that gun laws (and gun ownership) have little effect on crime. (He's mostly debunking the idea that more guns means less crime.) David, I think you'd like the book, it seems like the kind of analysis you like to do here. Also, I met your son at Burning Man! Had a great time chatting with him in Kidsville.

Carl M. said...

Interesting how England is just as violent as before gun control.

Anonymous said...

Off topic.
We went from "There are frickin' Nazis in Virginia!" to "Rocket Man is gonna nuke the US!" to "Republicans will not end Obamacare!" to "NFL players are disrespecting the flag!" to "Accountants are going Postal!" to "American Progressives get angry because Secession is not allowed outside the US!" to "A rapist pawn has just lost his protection and has been sacrificed for some obscure reason we will never know."
It is always react, react, react.
Tedious. What are they trying to hide?

Tom Mazanec said...

What happened in the US between 1900 and 1910? The murder rate nearly quadrupled, then stayed within a factor of two or so for a century.

David Friedman said...


My guess is that it's a change in how the data was collected or defined, but I don't know.

Geech said...

I once read a book called Guns and Violence: The English Experience that did a similar comparison and reached the same conclusion.

One of the random digressions in that book indicated that British police were historically unarmed not due to low crime rates but to assuage fears of a police state.

montestruc said...

I am working on a paper on the subject, and have found the following.

First off crime data, what is recorded here are absolute numbers of incidents, not rates per population.

You need to download several spreadsheets of data from this site, look at the years covered, note that the numbers of crimes are reported, not the per 100,000 population data. You need to calculate that so you need population data that you can get here:

We only need from 1900 to 2000, or in the case of the crime of “violence against the person” (VAP), till 1998, as the British changed how that crime was recorded that year such that doing reasonable analysis of that became problematic. This being because they changed a significant number of definitions as to what was and was not VAP, making comparison across that time boundary problematic.

So my paper will not involve data from after 1998, I am looking at the effects of gun control laws in England & Wales over the 20th century. I am treating VAP data it as if the century ends in 1998.

Significant gun laws were enacted in 1920, 1937, 1968 and 1988.

I have calculated the rates of murder, rape, robbery, and “violence against the person” VAP per 100,000 population per year.

If you want I can email you the graphs.

David Friedman said...


Please do.

Anonymous said...

Just spit-balling, but with the big peaks in the 30's and 70's, wouldn't that work out fairly well with drug/alcohol prohibition? Alcohol in the 30's and drugs in the 70's?

RJ Miller said...

Something to always be ready to share when UK crime statistics are brought up:

They don't record a murder as an actual murder until someone is convicted for the killing. I think by that standard our murder rate would appear to be only 60% or so of what it really is because that's our national clearance rate:

Naturally some of the most violent cities have lower clearance rates, something worth looking into when it comes to crime reduction:

Anyhow, that difference in definition that the UK has really throws a wrench into the idea that people can draw conclusions about gun laws strictly from reported murder rates alone.

David Friedman said...

RJ Miller:

I have seen that claim but have not seen any support for it. I followed up the "a report to a select committee of Parliment." It's a memorandum by Colin Greenwood as an appendix to the minutes of evidence, not an official government statement of any sort. Colin Greenwood is a retired police officer critical of U.K. gun regulation.

Do you have any official statement by the U.K. government supporting the claim? I have not seen one so suspect it is a bogus factoid, of the sort that often shows up on controversial issues.

RJ Miller said...

I do, but I would have to agree it's buried pretty deep:

It's alluded to in pages 16-18 of the document, especially in the first paragraph of box 1.1:

"It is important to note that there are issues surrounding the comparability of international homicide data. There are different definitions of homicide between countries, although definitions vary less than for some other types of crimes. Furthermore, there are differing points in criminal justice systems at which homicides are recorded, i.e. when the offence is discovered or following further investigation."

(emphasis mine)

I think there may be other official sources that back up Greenwood's claim but the Home Office Statistical Bulletin doc above is the most readily available.

In summary, it looks like murder rates between countries suffers from a bit of a "base rates" issue for many reasons that go beyond even the conviction status of a crime.

David Friedman said...


That passage says there are differences. It does not say "They don't record a murder as an actual murder until someone is convicted for the killing," which was your claim. It does not imply it --"following further investigation" is not the same thing as "until there is a conviction." Distinguishing a homicide from an accidental death sometimes requires investigation.

If that's the best support you can find, I think it's reasonably clear that the claim is false--it would be a big enough difference between the way everyone else defines the homicide rate so that if true it would show up.

Look at it from the other side. This is the sort of claim that people critical of gun regulation would want to believe, because it claims to undercut one standard argument by the other side--that the U.K. has gun regulation and low murder rates. Given that, once someone makes the claim it will be spread by people who want to believe it and are not very careful to make sure that what they believe is true.

As I think you have just demonstrated.

Tibor said...

Sorry for the late comment (today, I remembered that you also sometimes still post something on your own blog :-) ), but one other counterexample is the Czech republic. It goes in the opposite direction than Brazil or Mexico - liberal gun laws (second only to some US states), very low homicide rates (and crime in general). As you mention, a problem of these sort of comparisons is that the countries differ in more than just the gun laws. But since a lot of people seem to believe these laws are the chief cause of higher/lower homicide rates, I think it is still worth mentioning.

You still need a gun licence which, however, is a shall-issue (even a defence licence is) as long as you meet a couple of rather reasonable criteria:
1. No history of certain mental illnesses (basically to reduce the risk of crazy shooters)

2. Clean criminal record, or at least clean for a number of years. The required number of "clean" years increases with the severity of the previous crime and is infinite for crimes such as intentional homicide, terrorism, treason.

3. Age of at leats 21 (with some exceptions for sports or hunting licences - those can be issued to younger people under some conditions).

You also have to pass a knowledge and practice test like you would for a driving licence, but you don't have to go to any classes beforehand, you can learn it at home and the total cost of the licence (including these exams) is roughly USD 100 (the average monthly wage is currently about USD 1500).

There are no "assault weapons" laws, anything is ok as long as it is not automatic. If you want an automatic weapon or if you want to buy a suppressor, you need an approval of your police department and that is no longer shall-issue.

The right to concealed carry comes automatically with the defence licence (essentially that is what the licence is, otherwise you can still transport unloaded firearms with just the sports or hunting licences). In contrast to the US, open carry is completely illegal (save for the police, the army is not allowed to operate inland unless an emergency state is declared). The reason is probably not to cause panic.

Cold weapons are not limited in any way and you don't need any licence for them (but you need to be at least 21 to buy modern hunting crossbows...those things can be much more powerful than the heaviest medieval hand-held crossbows). The only exception are cane swords and similar concealed weapons that mask themselves as everyday objects. I guess that is mostly to try to protect the policemen in case of a violent arrest.

The homicide rate is 0.75 homicides per 10 000 inhabitants (as of 2015), lower than in the UK or Germany, both of which are significantly more restrictive in their gun laws.

It is true though, that the Czech population density is only about 55% the population density of Germany and it can be argued that the higher concentration of people, the more trouble. Still, it is still much more densely populated than the US.