Last time, I showed that gun-related deaths (both suicide and homicide) were staggeringly high in the United States. This time, I would like to get a little more rigorous, by looking at cross-country patterns of gun deaths and gun ownership across a sample of OECD
countries (the sample is limited by data availability– the data source can be found here
). It has been well known for quite some time that there is a clear association between gun deaths and gun ownership rates, going back to an influential study by Martin Killias
. Here are what the picture look like:
The pattern is clear. Of course, this is a simple bivariate relationship. Might gun deaths be related to a bevy of socio-economic factors that might have little to do with gun ownership per se? To test this, I decided to run some simple regressions for the sample of 19 countries: Australia, Austria, Belgium, Canada, Denmark, Finland, France, Germany, Italy, Japan, Netherlands, New Zealand, Norway, Singapore, Spain, Sweden, Switzerland, UK (technically England and Wales), and the US. To match the gun data availability, these data are also taken from the early 1990s.
Here are the data I choose:
(1) The log of GDP per capita, measuring living standards
(2) Ethnolinguistic fractionalization, measuring the ethnic and racial divisions in society
(3) Age dependency ratio, or the relative young and old population
(5) Gini coefficient, measuring inequality
The idea being tested is that violent gun deaths could reflect wealth, social divisions, the presence of a lot of young people in the population, urbanization rates, and inequality. Of course, let’s not forget gun ownership! Here is the result:
Gun deaths = 9.7 GDP per capita + 1.63 ethnolinguistic fractionalization + 7.51 age dependency -0.01 urbanization + 0.05 Gini + 0.19 Guns***
The three stars (***) signals statistical significance at the 1 percent level. What does this mean? It means that controlling for the availability of guns, none of the other stuff affects gun deaths in a statistically significant manner. The results are similar when breaking it down between homicide and suicide, with only one difference: for suicides, the age dependency ratio is also significant at the 5 percent level. (This shows, I think, that suicide is relatively more prevalent among the young.) It is also worth noting that these countries are all similar to a certain extent, as they are the richest countries in the world. The results merely show that living standards, for example, do not affect the differences in gun deaths across this narrow sample of countries. The results could be quite different if we included some of the more violent countries in the world. But, based on this limited data availability, I think we can conclude that what causes gun deaths is the availability of guns. Occam’s razor triumphs.
There is one more thing I would like to do. It is quite possible that the availability of guns acts not only on its own, but through its affect on some of the other socio-economic variables. For example, if ethnic divisions affect violence in society, the availability of guns may enhance this level of violence. Statistically speaking, this means adding non-linear terms to the regression, interacting gun availability with some of the other variables. I found that such a result can be found for both ethnolinguistic fractionalization and the Gini coefficient:
Gun deaths = 1.48 – 6.95 Ethnolinguistic fractionalization* + 0.08 Guns** + 0.44 (Guns x Ethnolinguistic fractionalization)**
Gun deaths = 2.25 -0.08 Gini – 0.16 Guns + 0.02 (Guns x Gini)***
(Again, *** means significant at the 1 percent and 5 percent significance levels, respectively).
This tells us that the presence of guns feeds on the underlying tensions brought on by ethnic, racial, and class divisions in society. Once again, the results differ little for homicides or suicides (the Gini coeffient
, though, matters far more for murder than suicide).
Guns matter. Notice that I look only at deaths caused by guns, either homicide or suicide. An interesting point to note, following the research of David Hemenway (from Harvard’s School of Public Health) is that the US is actually not that exceptionally violent, at least among other high-income, industrialized nations like the ones in the present study. Crimes like assault, car theft, burglary, robbery, and sexual incidents are not particularly high by OECD standards. What differs about the US is “lethal violence”. So while guns don’t induce people to commit crimes, they make crimes lethal. This matters for both homicide and suicide.
So, can people still hold onto an antiquated notion about the “right to bear arms” and not admit that this is very much a culture of life issue in the US?