Thursday, June 22, 2006


See, I said I'd get around to reviewing this in the end.

Freakonomics, by Steven Levitt and Stephen Dubner, is this year's oddball non-fiction hit. It bears the curious sub-title "A Rogue Economist Explores the Hidden Side of Everything", and doesn't really explain very clearly what it's about.

Levitt is an economist. Dubner isn't. He's a journalist, who wrote an article about Levitt a while back, and ended up doing a whole book on it. Basically, the point of this book is to bring Levitt's work to a mass audience. This probably sounds extremely tiresome, because even though we all know we should pay more attention, our eyes tend to glaze over at the words "deflation" and "supply-side economics."

But that sort of thing is macroeconomics - the big picture of the economy. Levitt doesn't do macroeconomics. He does microeconomics, looking at how people behave on the small scale. Frankly, a lot of the work covered in this book wouldn't even be recognised as economics by most people. In layman's terms, he's really more of an applied statistician. He takes mounds of indigestible data and picks them apart to discover unlikely and counterintuitive cause-and-effect relationships.

Levitt acquired a measure of notoriety in 2001 for his paper The Impact of Legalized Abortion on Crime, in which he (and his co-author John Donohue) examined various possible explanations for the drop in American crime rates in the 1990s. Freakonomics helpfully explains the paper in layman's terms. In a nutshell, their analysis of the data showed that, alongside other factors, a significant part was played by Roe v Wade. Thanks to that 1973 decision, an awful lot of people who would previously have existed never got born. Those people skew disproportionately to the social groups with high crime rates, and hence a whole load of potential criminals were averted. (It's a bit more complicate than that, but you get the general idea.) They then cross-check this against the states which legalised abortion earlier on, and it confirms their findings.

Whether you actually agree with this analysis or not - and frankly, you'd have to know an awful lot of statistics to verify it for yourself - it's certainly an interesting idea, and a compelling example of what you can prove with statistics. One of the book's key strengths is its ability to explain fairly complicated statistical ideas in lay terms without completely copping out and saying "We proved it, but you'd never understand how." You won't find the detailed statistical analysis here, but you'll at least find the shape of the argument, and get some understanding of how this sort of thing works.

By the way, Levitt has never suggested that the drop in crime rate means that abortion is a good thing. On the contrary, he also points out that the effect on crime is pretty tiny compared to the number of abortions required, which means that if you place any real value on the unborn child, it's not worth it just for the social engineering consequences. He simply observes that it's an unintended benefit of a practice that goes on for completely separate reasons. Unsurprisingly, but depressingly, many people think it's distasteful to consider this sort of thing, even if it's true.

Other chapters are rather less confrontational. Levitt examines the economics of drug dealing and demonstrates why people would bother entering a profession that generally pays less than minimum wage and carries a worryingly high risk of getting shot dead. He uses statistics to prove that teachers and sumo wrestlers cheat. He argues that what you do as a parent actually makes virtually no difference to the outcome of your child; it's your qualities that affect your child (class, job, wealth) rather than anything specific that you happen to do with your kid (make them read books, take them to the museum...). Dubner, meanwhile, makes sense of all this for the layman, without losing sight of the main point - which is to explain the way Levitt thinks.

Officially, Freakonomics lacks any sort of unifying theme. It jumps around all manner of oddball subjects, partly because Levitt is recycling research that he's already published in academic papers. It's Steven Levitt's Greatest Hits, not exactly dumbed down, but certainly rendered less technical and infinitely more accessible to the non-specialist.

In reality, the main theme is Levitt's methodology itself. It's a book that tells us, with wide-eyed enthusiasm, about all the great things you can achieve through the power of critical thinking and logical analysis. It's carefully presented with a tone of intrigued fascination that reassures the reader about the reasonableness of his preconceptions before carefully dismantling them over the course of a chapter. I rather suspect that one main point of the book, which they won't point out explicitly, is to try and encourage people to think more, and to think more critically. But if you say that outright, the audience get insulted, so it's better just to try and engage their imagination.

Whether you categorise Levitt's work as economics or statistics, it's the sort of mathematics that most people find incredibly dry. The strength of this book is its ability to make it seem engaging and to give you at least the illusion that you understand what Levitt's doing. In reality, as I say, he's giving you the shape of the argument but you're really taking the results on faith. But that's inevitable in a book of this sort; if you want to go and check the small print, you can always go and read the original academic publications.

The ability to make dry academic subjects accessible and interesting without oversimplifying them into oblivion is something to be applauded. Yes, it's a bit patchwork. Even so, it sells its subject very well indeed.