A little less freak, a little more -nomics, please...

I just finished reading Freakonomics, the acclaimed bestseller that explains many of modern life's little quirks with basic economic theories. Its most controversial claim, which was involved in the recent Bill Bennett controversy, is that abortion has cut crime rates in the U.S. by preventing likely criminals (unwanted children of poor single moms) from being born.

When I first read this claim, I was skeptical. I live in Japan, where the abortion rate steading climbed over the 60s, 70s, and 80s, and where the youth crime rate has climbed since the 80s - meaning that kids of Japan's "abortion generation" commit more crimes.

Looking over the web, I found this scathing review of Freakonomics. The writer points out that homicide committed by Americans born in the high-abortion years of 1974-1979 was actually much higher in the late 80s and early 90s than crime committed by the same age groups in the early 80s (who had been born before Roe v. Wade). It also noticed a few other places where the statistics don't support - or even contradict - the Freakonomics hypothesis.

So if it mostly wasn't abortion that dropped crime in the late 90s (homicide has now leveled off), what was it? One possibility is that the other factors Freakonomics identifies as contributing to the crime drop - more police, more incarceration, and a drop in the price of crack - might not be linear in their effects on crime. For example, potential criminals, seeing their chances of making money go down and their chances of going to prison go up, might have collectively crossed some "incentive threshold" where a life of crime just didn't seem like a good bet anymore. When Freakonomics claims that a certain factor "accounts for only [X] percent of the drop in crime", it's assuming a linear model (Crime = Ax + By +...). But if crime is really nonlinear (eg. Crime = Ax + By +Cxy +...), then the entire drop in crime might just be due to the combined effect of incarceration, police presence, and drug markets.

Another possibility is that welfare might have had something to do with it. Check out this graph of welfare rolls in the U.S. and compare it to this graph of homicide rates. The graphs are almost identical. Now that could be a coincidence - welfare and murder could have both dropped due to the same outside factor(s).

But maybe not. In fact, Freakonomics notes in its third chapter (right before the chapter on abortion) that most drug dealers live with their (poor, single) mothers, selling drugs for chump change while they wait for the big break that will make them a rich gang leader. It certainly seems possible to me that, with welfare no longer coming in, those wannabe drug dealers - who had been the ones committing the most murders - were simply forced to ditch the gangsta lifestyle and go work at Wal-Mart.

Anyway, I wouldn't be making this fuss if Freakonomics hadn't presented this claim with just as much self-assurance as it presented much less speculative ones (for example, that sumo wrestlers cheat); as one consequence of that, we've got guys like Bennett running around saying that aborting black fetuses would reduce crime.