The tragedy at Sandy Hook Elementary is still fresh in our minds, and the debate on what should be done in response, if anything, is at full tilt. There are so many angles on the issue: gun ownership as a protection against tyranny, limiting the freedom of responsible members of society to reign in the harmful few, gun ownership as a crime deterrent or enabler, to name a few. I have talked about the first item in previous blogs (most recently concerning the Arab Spring and previously with respect to Gabrielle Giffords), arguing that guns no longer serve as any sort of protection against tyranny.
The second issue—limiting the rights of all to prevent the unstable few from committing atrocities—can be addressed rather quickly. We already do this in many areas of life, including weaponry. We do not allow people to buy bazookas, tanks, jets, missiles, etc., despite the fact that this is limiting the freedom of people who mean no harm, because the weapons are capable of causing such vast devastation. On a more mundane level, we limit the speed of cars on the road because of the safety hazard they pose, and we require people to wear seat belts (in most places) because it saves lives. In all of these cases we have chosen to forego some liberties to promote the greater good (saved lives). This is the nature of the social contract – give up some rights to gain greater benefits.
This simple acknowledgement of the way modern states function shows the absurdity of the “guns don’t kill people; people do” argument. The same could be said of speeding cars or bazookas, and yet we still place limits on such things.
That said, my primary purpose in this blog is to address the last issue in my list: does gun ownership increase or decrease violent crime, especially homicide? I am particularly interested in what gun ownership rates around the world have to do with homicide rates and the common arguments on the topic. To that end, you may have seen this graph floating around the internet recently (or others like it; here’s another one):
This looks pretty convincing. You’ve got some really high homicide rates in countries with low gun ownership, and then much lower homicide rates as you move to the right. But the problem with data-based arguments is that they tend to be accepted without any critical analysis.
Here we have several problems.
First, and most importantly, the graph implies that civilian-owned firearms cause lower homicide rates. Clever economist that he is, Mr. Davies knows he can claim no such thing with this graph and merely states that countries with higher firearm ownership rates also have lower homicide rates. Just because two variables are correlated does not mean either one causes the other. But the clear implication of causation in producing the graph at all is very misleading on the part of Mr. Davies.
Second, sometimes it does not make sense to compare every country in the world. Economists are big fans of global data, and sometimes such data can be very helpful, but other times global data actually confuses the matter by making unjust comparisons. This is one of those cases. El Salvador, with the highest homicide rate in this graph was due almost entirely to gang violence, which has raged for years in that country (here is a recent article about gangs there). The gangs there have access to heavy weaponry, including assault rifles and grenades, which they use to prosecute their conflicts. In such a case, guns for self-protection are not the point – the violence is occurring between people who already have plenty of guns. When gang violence is the key driver of violence, what is needed is stronger law enforcement and rule of law, along with an attempt to address the roots of gang conflict.
Similarly, Ivory Coast has been in and out of civil war since 2000, the main source of its violence. More guns will not solve the problem – violence from wars only stops when one side wins and sets up a legitimate and stable government. In Honduras and Jamaica, drug trafficking and gang violence contribute to the high rates of homicide, again, unrelated to the rate of civilian gun ownership. In fact, in the data I use below, which was trying to approximate the data used for the above graph, 18 of the top 20 countries by homicide rate are in the Caribbean, Central America, or major drug production areas of South America.
When drug cartels and gangs are the main source of violence in a country, normally the bulk of the violence occurs between members of those groups, who already have plenty of guns. More guns for the civilian population who live around the conflict (but are not key players) will not end the violence. That is, if Gang A and Gang B are fighting and already have plenty of guns, arming group C (the civilian population) will not reduce conflict between A and B. Group C just wants to stay out of it. Fighting will continue until the root causes of the conflicts are addressed (e.g., the lack of economic opportunities for gang members, the high price and demand for drugs, etc.).
The problem with all of this is that murder in some countries has very different causes than in others. In wealthy countries like the US, murder is not due to civil war, country-wide gang violence, or massive drug cartel armies battling it out. So comparing such countries with the US and other wealthier countries is completely absurd and tells us nothing significant about gun ownership.
So I set out to look at the data from a different perspective. First, I tried to recreate the data from Mr. Davies’ graph above, using this data, so that I could compare the results with his. The data is not precisely the same, as Davies did not link to his precise sources, but I used similar time frames (2007). The graph I created is below:
It looks very similar. The US is still way to the right. Honduras is up top. My main point here is to show that the data is similar to Mr. Davies’ (I think the other differences are due to fluctuation year to year in some of the worst conflict zones).
Then I decided to add GNI per capita (similar to GDP per capita, a simple measure of average income; data from the World Bank) and graph that against homicide. The results are below:
The thing to notice here is that the graph looks very similar to the plots of firearm ownership and homicide. My point is that it is not hard to create a graph that shows murder rates going down as some other variable goes up. I’m sure we could produce similar graphs with variables like years of education, lifespan, and many others. Except in the case of income and homicide, there is some theoretical reason to think the relationship may be causal. After all, drug cartels and gang violence do not thrive in rich countries because they have stronger law enforcement, better prospects for their citizens (so they don’t turn to gangs), and diversified economies that provide many legal sources of income for their citizens. Many problems improve as countries get richer. Nonetheless, the relationship looks weak. So now let’s consider something else.
Now this is a much different picture, isn’t it? This graph shows firearm ownership and homicide rates in OECD countries (a club of rich countries) that also have a GNI per capita of more than $25,000. As you can see, among wealthy nations similar to the US, as firearm ownership increases, so does homicide. This is completely different than the depiction of the data when looking at the whole world, comparing incomparable countries. And of course, that nation on the upper right, with very high firearm ownership and murder rates, is the US. The implication is obvious: even if guns are sometimes used to protect the innocent, any deterrence of homicide is outweighed by an even larger increase in homicides due to the availability of guns.
So if anything, we should expect that more guns will lead to more violence, not the other way around.
But, as I have discussed, correlation is not causation. So, in this new case, are the higher gun ownership rates causing higher homicide rates?
This possibility is supported by a number of studies, particularly by David Hemenway, Matthew Miller, and Deborah Azrael. A good summary of some of their main findings (on this topic and related ones) and a few of their articles are cited here: Harvard Injury Control Research Center. There are some fairly obvious candidates for why this relationship may exist. For instance, the availability of a gun may lead to fatal escalation in moments of anger and passion, when otherwise fists or less lethal knives would have been the weapons of choice. Guns may be taken by other members of a household (besides the gun owner) and used contrary to the owner’s will, simply because they are available. Attempts at deterrence by a gun-owner untrained to actually fight in such situations may lead to an even greater use of lethal force by criminals. These are just some of my own hypotheses, but I think the first one is probably the most likely.
My main point in this article is to show that the oft-cited data about guns and crime internationally is often misread. However, I also acknowledge that there are opposing views, most especially by John Lott and his collaborators (one of whom was my former professor, for whom I have immense respect). Lott and Mustard’s original article that started this debate can be found here, but it only deals with the right to carry concealed weapons, not overall ownership, and not in comparison to other countries. Lott’s later book that fleshes out his argument is called More Guns, Less Crime, and he wrote another book later again responding to critics. A good critique of his second book is here. Finally, another good summary of info on gun violence in the US is here. Overall, I think that research in recent years has largely shown that more guns does NOT equal less crime, contrary to Lott.
I do not believe all guns should be banned, nor do I think that such an outcome is at all possible in the US any time soon. But strict controls on who can buy guns and the types of weapons and accessories available for purchase would reduce gun violence in this country (if also combined with efforts to reduce the vast quantity of guns already floating around the US).
As many have pointed out, the same day of the Newtown shootings, a man attacked young students at an elementary school with a knife in China. While it was also horrible, without the gun the outcome was very different. The man in China wounded 23, but none died. If only that had been the outcome in Connecticut.