Crime way down. Who exactly knows stuff?

Share on FacebookShare on Google+Email this to someoneTweet about this on Twitter

Despite recession, crime keeps falling:

In times of recession, property crimes, in particular, are expected to rise.

They haven’t.

Overall, property crimes fell by 6.1 percent, and violent crimes by 4.4 percent, according to the six-month data collected by the FBI. Crime rates haven’t been this low since the 1960′s, and are nowhere near the peak reached in the early 1990′s.

Who expected crime to increase? Did you? I did. But I didn’t know anything about crime statistics over time so I was working off naive intuition. Did social scientists expect this? I recall a lot of worry in the media about a year ago that the crime drop which started in the 1990s would be reversed, and I shared the worry. Here’s Matt Yglesias worrying last January:

I think this is worth worrying about. One thing we know about crime is that when wages and employment levels for low-skill workers are high, crime goes down. Another is that mass incarceration works – increase the number of beds in prison and the number of sentence-years handed out and the crime rate drops. But the first of these is the reverse of what happens in a recession, and the second we’ve already pushed well past the limit of cost-effectiveness (see here) and it’s inconceivable to me that you could actually push this far enough to compensate for the declining economy in the context of declining state budgets.

It’s easy to find national uniform crime reports data back to 1960, and unemployment rates. Quick correlations between 1960-2008 are:

Violent Crime Aggregated 0.37
Murder 0.52
Rape 0.37
Robbery 0.53
Assault 0.24

Property Crime Aggregated 0.53

One seems to see a modest expectation for a rise in crime then over this time period. But poking around the ICPSR I came across Eric Monkkonen’s data sets on homicide in New York City going back to the 19th century. Below are homicides per capita by year between 1900 and 2000. The second chart is log-transformed.

It seems that there’s another “Depression Paradox” here. The economic distress of the Great Depression seems to have been associated with less crime, while the economic exuberance of the 1920s led to more crime. So if I constrained the time series from 1920-1940 the correlations might be quite different.

All things equal the recent past is a better guide to the near future than the less recent past. But it’s important to remember that history does sometimes work in cycles, and the deeper past can occasionally give us insights which the recent past can not. One could construct a tentative model whereby basal crime rates reflect cultural norms, and once norms and crime hit a particular “equilibrium” it may take a bit of a “shock” for it to shift out of the stable state.

Labels: ,


  1. I think the best policy is to do what you did and look visually at the crima data and the macroeconomic data. If recessions don’t jump out of the crime picture — and they don’t at all — then any correlation is spurious. 
    There’s a very severe recession in the early 1920s, but that doesn’t move the homicide rate noticeably up or down. There’s a boom in the ’20s associated with rising crime, and then a depression associated with lower crime — but then the post-war recovery doesn’t move the homicide rate. The Long Boom comes to a grinding halt in the mid-’70s, but you wouldn’t notice that from the homicide rate, which merely plateaus instead of being driven up or down. The early ’80s recession is associated with a fall in crime, and the subsequent boom is associated with a rise in crime. 
    Then the early ’90s recession coincides with a fall in crime again — but then the breakneck economic euphoria of the mid-’90s up through 2006 coincides with the sharpest and most prolonged drop-off in crime. Economically it was the second Roaring Twenties, but the association with crime was just the opposite. And the recent recession doesn’t show up at all in the crime rate — still heading down. 
    The reason there is very little causal relationship between violent crime and macroeconomic indicators is that violent crime levels depend mostly on how many genetically violent people there are. Those numbers will cycle for the same reason that other ecological systems cycle, as between predators and prey, hosts and parasites, etc. Those who check them grow, driving their numbers down. With few violent people, the checkers don’t have any advantage and move on to something else. With few checkers, violent people can easily get away with their plans and rise in number. That gives an advantage to violence-checkers, and the cycle repeats. 
    That has nothing to do with how prosperous or down-and-out the average person is. It is simply a decision of which margin to compete on — the margin of violence, or any other peaceful margin? Once the niche of violence-based competition becomes saturated, it pays less and those who compete along the peaceful margins will start to expand. In other words, it’s a case of frequency-dependent selection, and not facultative change.

  2. tx jon. fixed it.

  3. This doesn’t apply as much to your post, which just discusses NYC, but it’s claimed <a href=”“>here</a> that violence in other parts of the U.S was greatly undercounted in the early 20th century. The official stats show a big increase a few years into the 20th, but if Douglas Eckberg is right that just shows the numbers coming into alignment with a reality that was already there. Kind of like Alan Reynolds on inequality. 
    agnostic, I think your genetic theory has much to be said for it (in line with “Farewell to Alms” and “10,000 Year Explosion) but we’re looking at some very abrubt changes over short time periods. Could you explain more about “checkers”? 
    Roy Baumeister blames high self-esteem for violence, particularly shifts in inter-racial violence. 
    I thought Randall Collins’ book on violence was really good, but he focused on micro-situations rather than explanations for changes over time. Have any of you read it, and if so, what did you think?

  4. Agnostic - 
    Genetics is only a small part of the story.  Even among whites, crime rates have been an order of magnitude in the U.S. than in Europe.  For instance, in the 1910′s L.A., Buffalo, Minneapolis, were all 98% white yet had 5-25X the murder rate of London, Paris, and Vienna.  I cannot think of any plausible genetic explanation that would account for the difference.   
    The real factor affecting crime rates is the obvious one – how effective the criminal justice is at finding and convicting criminals.  The United States has always had an incredibly ineffective justice system.   In the 1910′s Raymond Fosdick spent several years studying the police systems in the U.S and Europe.  He wrote: 
    There is no part of its work in which American law fails so absolutely and so ludicrously as in the conviction and punishment of criminals. ” It is not too much to say,” said President Taft in 1909, ” that the administration of criminal law in this country is a disgrace to our civilization, and that the prevalence of crime 

  5. (continued from above) 

    Fosdick then compares that to London: 


    In England the situation is far different.   In the whole of England and Wales for 1916, 85 murders were committed and 59 people arrested in connection therewith were committed for trial. Fifty-three trials resulted during the year. Twelve of the accused were found insane on arraignment and were confined; sixteen were found guilty but were adjudged insane and confined; ten were acquitted, and fifteen were sentenced to death.

  6. The mere fact that both American and European cities were white doesn’t suffice to show genetic similarity. Some Europeans have claimed that the more cautious among them staying home while risk-takers immigrated to America, hence their aversion to “cowboy capitalism”! 
    Good point about the virtuous circle. I think Mark Kleiman makes a similar argument (particularly in the case of Highpoint NJ). That also helps to explain behaviour during riots: the activity of others makes it less likely that any given individual will be punished for criminal activity. I haven’t read the whole thing, but that’s part of the argument behind Alex Tabarrok’s simple model of crime waves, riots and revolutions (which I suspect owes much to Gordon Tullock & Timur Kuran).

  7. Agree with Devin Finbarr about the prohibition effect, which is surely also operating with the rise in homicide in the 1960s: the “war on drugs” having the same effect as the “war on alcohol”.  
    Also agree about the enforcing of crime. In effect, it is about enforcing “property rights”. Where the state is effective at enforcing property rights in one’s body and property, there are fewer transgressions. Where the state tries to deny people “property rights” in their own body (i.e. attempt to stop them consuming alcohol and drugs: or gamble or purchase sex, for that matter) it attempts something it does not have the power to do and, worse, undermines its capacity to protect “ordinary” “property rights” in one’s body and property since it creates social domains where private violence will be used to defend property rights it refuses to acknowledge or protect.

  8. You seem to say that if it’s not economic then it’s cultural. No argument was offered in defense of the statement that violence is simply genetic and no cultural factors were addressed at all. 
    As for the “checkers”, how exactly do checkers impede breeding and how does fewer checkers lead to greater breeding? Do yo have data to offer as to the number of police (or prison population, deceny leagues, religiousity, or other “checkers”) in relation to breeding of the violent? 
    In other words, Citation Needed!