Long Term Trends in Homicide Rates

Here is a graph of American homicide rates, the earlier results should be taken with a grain of salt of course, but the trend is clear. N.B. These rates are per 100,000.

The American data is consistent with European data. Here is Table 2 from Manuel Eisner’s Long-Term Historical Trends in Violent Crime. Do note that some correction should be made for the fact that violence is less lethal when people are healthier and medical care is more effective.

The bottom line is that there has been a big and welcome decrease in homicide rates in Europe and America over the past several centuries. To put these numbers in perspective, however, note that the homicide rate in New Orleans today is 52 per 100,000 and in Detroit it’s 40 per 100,000 so even with a lower average there is lots of variation. Brazil today is around 22 per 100,000 not too far from America in the 19th century. The homicide rate in El Salvador is 71 per 100,000, in Jamaica (!) 60 per 100,000 and in Honduras 67 per 100,000 — all higher than fifteenth century Europe. Thus, the past was a more violent place but not so violent as to be unknown to the present.

Hat tip: Tim Harford.


Medium term features are interesting too. The hump between 1960 and now is similar in size and duration to the ones which surround the revolutionary war, and the civil war. (I assume the numbers don't actually include deaths in combat). Still, it seems there is at least one sense in which the phrase "war against crime" is correct.

I think you've made a mistake: the second chart appears to have numbers per 10,000 population (notice that all of the twentieth century averages 1.4), while the former is per 100,000. Thus your claim that modern murder rates in Jamaica/Honduras/El Salvador are comparable to those in 15th century Europe are way off the mark. I am not sure just how violent a place would have to be to be 'unknown to the present' (pre-Roman Europe was more violent still) but life was certainly cheap in medieval Europe...

A certain healthy skepicism is in order for any "stats" from before about the 18th century. How many murders escaped detection in times past? A lot of poisons were simply undetetacble after the fact, and in remote areas a body could be hidden in ways that it would never be found (or if some remains were eventually found it wouldn't be obvious that human not animals were at fault fo the death)

bbartog - I don't think so. The source of the second chart (the European historical rates) - the "Long-Term Historical Trends in Violent Crime" - indicates on page 97 that the rates are per 100,000. Alex's other statistics are also per 100k - so all the above stats are consistently per 100,000.

It obviously good that the long term trend is downwards, however America's homicide rate is still very high. According to Wikipedia the USA has a higher homicide rate than Yemen, Palestine, Afghanistan, Somalia, Syria, Iran and Libya. Now this may be due to difficulties in getting accurate data in these places meaning the true rate is under-reported. However, Canada and Western European countries which are similarly as developed as the US is also have a lot lower rates.

Source: http://en.wikipedia.org/wiki/List_of_countries_by_homicide_rate#2000s

China, India, Egypt, and Greece also have lower homicide rates than one would expect using the variables that predict the rates in the US and western countries. Tribal societies value warlike violent men and so they father more children, but in civilized societies violent men are social outcast. If violent behavior has a genetic component, after thousands of years of civilization these countries would have a smaller fraction of them. (except Afghanistan)

Assuming that violence isn't a result of environmental and cultural factors acting on some hereditary trait such as aggression that in a more peaceful place and time results in business or sporting success, which still results in more children.

The real question, is the violent crime rate for non-marginalized U.S. citizens significantly different than other countries?

Sure, the U.K. has a low rate of homicides - the negative effects of centuries of racism, imperialism, and slavery, have been externalized to Zimbabwe, South Africa, India, etc... In the U.S., because of its historically unique pattern of contiguous expansion (as opposed to acquiring far-flung overseas colonies), and of importing slaves (as opposed to enslaving people in said far-flung overseas colonies), the U.S. internalized the negative effects that Europeans where able to externalize via decolonization.

The murder rate for your average middle class person is extremely low. U.S. crime rates are concentrated in ultra-violent "Ghettos", composed of the victims and decedents of victims of slavery, segregation, and racism. What the United States has is essentially small violent third-world pockets that exist in an otherwise low-crime first-world country.

There is a missing demographic component. A graph of "murders per men aged 16-35" would show less improvement, whereas the part relating to "fraction of men aged 16-35 in the population" would show a lot more. Some of the change is not related to people being less murderous but rather to fewer people in the demographic category to which almost all murderers belong.

Does this effect show up when you compare countries with different demographics? Based on RJP's info, and the relatively low murder rates in countries with very young populations, age doesn't seem to be the primary factor involved.

Improved patching-up of the grievously injured may have an influence too.

I think that's what he was referring to with the "correction should be made for the fact that violence is less lethal when people are healthier and medical care is more effective" bit, Einstein.

Here is Richard Whately (1831) on change in media exposure misleading our impressions about crime trends:

"It should be observed also, that in large towns, and in populous districts intersected by roads which furnish a rapid conveyance of intelligence from place to place, and where newspapers are in common use, much more in proportion is known of every enormity that is perpetrated, than in remote country-districts, thinly peopled, where there is less facility of mutual communication, and where the natural appetite for news is compelled to limit itself to the gossip of the nearest hamlet. Much apparent increase of crime (I will not undertake to say how much) consists, I am convinced, in the increase of newspapers. For crimes, especially (be it observed) such as are the most remote from the experience of each individual, and therefore strike him as something strange, always furnish interesting articles of intelligence. I have no doubt that a single murder in Great Britain has often furnished matter for discourse to more than twenty times as many persons as any twenty such murders would in Turkey. We should remember, that there are not more particles of dust in the sunbeam than in any other part of the room; though we see them more where the light is stronger."

That is from Introductory Lecturers on Poltical Economy, which contains many remarkable passages. Available at the Online Library of Liberty.

This completely ignores institutionalized homicides though wars, genocides, and other state-sponsored killings. Admittedly, these may have also declined over the past few decades. Regardless, claiming that there's been a decline in the amount of lethal interpersonal violence while ignoring the large, and growing, sector of the society that specializes in committing acts of violence doesn't make much sense. This is similar to the claim that was made here last week that crime in general has fallen during the recession; just as long as you ignore the institutionalized criminality of the rapidly expanding state. In both cases, it might easily be true that crimes committed by private individuals have simply been supplanted by state-sponsored aggression.

Actually, state violence and interpersonal violence have both tended to decline over time. What you are looking at is what historical criminologists refer to as the "civilization curve," and it holds for both homicide rates and (say) capital punishment statistics. I'm really not sure what you mean by the "large, and growing, sector of society that specializes in committing acts of violence," but it seems to allude to the fact that as societies modernize, the state monopoly on violence tends to be put into the hands of "violence specialists" (i.e., taken away from the mob, and given to bureaucrats). However, the specialization of violence actually tends to inhibit its use; you seem to be suggesting the opposite here.

Wouldn't this theory (we can call it the "Criminal Crowding Out" theory) be undercut by the notable "humps" in the graph around the Civil and Revolutionary war periods? Or the rise in crime during the 60's, which saw both considerable expansion of the welfare state, as well as significant exports of state-sponsored aggression to Southeast Asia?

Yes, if there was a causal mechanism suggested. I’m actually not referring to any theory, here; just the empirical pattern that, in terms of long-term trends across modernity, violence has tended to decline—interpersonal and state violence, alike, and that the latter has generally declined as bureaucratization of justice system has advanced.

Personally, I tend to think Elias’ theory of the civilization process (where the term “civilization curve” originated) has the most going for it, here. In essence, he argues that the rationalization of social life through the proliferation of codes of manners from one generation to the other have effectively built up layer and layers of psychical inhibitions on our baser instincts. (Think Freud, but historicized.) In that case, disruptions to institutional life (as seen during periods of war, the disestablishmentarian movements and counter-cultural conflicts of the 1960s, etc.) should predict a reversal in the trend. Of course, the theory is difficult to directly test. However, I think it is one of the more persuasive ones out there.

Also, you shouldn't gloss over the medical access and effectiveness angle. How does it look when plotted against other measures of violent crime, e.g., assault w/ a deadly weapon?

I think Gawande covered this in Better.

To help clear up the influence of better medical care, something like "attempted" homicide may be more interesting. There may not be as good of records on that though. Plus, it may include anyone merely accused of attempted homicide. Still, the fact that you can now (hopefully) survive a gun shot to the gut probably skews these numbers.

We don't have any statistics nearly as reliable for attempted homicide. Lots of people don't report crimes committed against them, but with homicide you've always got a body or at least a missing person.

Considering that the US has gained wealth every year (not the least of which was via technological advances), the poverty-induces-crime model correlates very well with this data.

Except that crime dropped like a rock during the Great Depression and has also been falling during the recent recession, though it was on the rise in the 1920s and 1960s. Also, looking internationally many of the poorest countries (poorer than Latin America) do not have the highest crime rates. A more plausible theory is that there are some other factors correlated with both long-term wealth and low crime rates. Exogenous shocks to wealth thus fail to affect the crime rate. Greg Clark might says it's conscientiousness.

The same societies you refer to have all been advanced by technology over the last 100 years in far greater proportion than their inter-national crime rates and historical variances in GDP. Crime rates are tied to the prevalent legal funding and structure - it seems just as likely that in recessions, legal structures are defunded or underfunded which cuts the rate of conviction, and a criminal offense in one legal framework is often perfectly acceptable in other frameworks.

The relevant data are violent crime and long term, aggregate economic value (possibly regardless of currency valuation). Per the data above, peaceful societies are not dirt-poor, but quite advanced.

Crime dropped like a rock in the 30s because of the repeal of prohibition, which is what caused the spike you see around 1920.

Anyone know why the big drop took place between the 17th and 18th centuries? I'd have expected it to take place in the 19th century when they invented practically everything I'd associate with lowering crime: police, speedy transportation, widespread news, accurate government statistics, etc.

Even if you want to say that, painted in broad brush strokes, European institutions were similar in the 17th and 18th centuries, those similar institutions were working better in the 18th century than the 17th. Widespread warfare and its accompanying population displacements, infrastructure destruction, and production of a generation of underemployed young men with experience in lethal violence, together do nothing to the aid the effectiveness of one's civil institutions.

"Data" is plural.

Only if you're an irredeemable geek.

"Data" is short for "Data set".

Maybe we are running out of the people who need killin'.

Well that would certainly tie in with evolutionary theory.

Strange it's taken so long though.

It's pretty simple. People who aren't hungry aren't as likely to kill each other.

The average income of the 1950s is about where today's poverty line lies.

That is, in my opinion, too simple to be true.

For starters, it is hard to kill someone if your physical strength is undercut with frequent hunger.

But it vastly increases your willingness to try. Given a knife or gun, killing someone strikes me as being 99% motivation, 1% ability.

Anyways, every species that hunts must have a mechanism that focuses their hunting instinct and conserves energy when hungry, else they'd quickly fall below the minimum necessary energy to kill something and die.

Outside of Donner Party situations where killing someone gives you flesh to eat, what's the point of doing so?
It would be interesting though to correlate instances of homocide with major famines. Did murder go up in Ireland in the 1840s? In Ethiopia in the 1980s?

Maybe it's harder to survive an assault if your are weak with hunger?

really interesting, what you mean by "some correction should be made for the fact that violence is less lethal when people are healthier and medical care is more effective."?

that people survives an attack just cause medical attention is available? perhaps the murder instinct is still with us, but it is not that efficient anymore. maybe we can find the answer comparing with the "attack with murder purpose rate".

I like to think that the principal reason for the decline in crime is that it has gotten so much easier to get caught.

I think we're just getting better at hiding the bodies.

That could also explain the rise in crop yields...

Very interesting post. I, for one, suspect that cultural factors played a major role in the reduced level of interpersonal violence. Steve Pinkker had an interesting article in the New Republic a few years ago that focused on the reduction in violence since the 16th Century. It can be found here (sadly suspription required).

Every time the homicide rate goes up or down, we all cast about for causes. The usual suspects, the economy, policing, and number of prisoners, do not work out. The changes are usually national, while policing and prison policies differ over the country. Crime rates were low in the Depression, are low now, in our deep recession and were high during the prosperous 80's.

The historian David Hackett Fischer, in his book "The Great Wave" using over 700 years of British records shows that the homicide rate are inflation are closely correlated. High inflation, high crime, low inflation low crime. It certainly holds for the examples above. Fisher himself concedes that correlation is not causation, but it rules out the usual explanations.

Crime has gone down in both states with high rates of incarceration and low ones too.
In cities with new policing techniques and in cites with still use the old techniques

Might I suggest:


The latter results should also be taken with a grain of salt.

Isn't Venezuela's homicide rate off the charts?

Congratulations, Progressives. You've managed to keep us on roughly the same trend line we were on before a literally unimaginable increase in media exposure (more people know about crimes faster,) forensic technology, and medical science occurred. Now, if over the NEXT hundred years you don't continue to respond to improved pumping technology by knocking bigger holes in the boat, we might get somewhere.

This shows which they last very much lengthier and thus saving you income which could otherwise are actually utilized to purchase new ones.'/\/

Comments for this post are closed