More Fun than Truth
Say what you will about Levitt and Dubner, but they sure know how to sell a book. The co-authors (Steven D. Levitt and Stephen J. Dubner) of Freakonomics (William Morrow, 2005) apparently sent early copies to influential bloggers, they set up their own promotional blog , they just had an op-ed (based on a book chapter) in USA Today, and—honor of honors—Levitt recently appeared on Jon Stewart's The Daily Show.
These efforts have paid off handsomely, as Freakonomics is a New York Times bestseller and has soared to almost the top of Amazon's bestseller list. People are going cuckoo over the intriguing chapter titles, which include: "What Do Schoolteachers and Sumo Wrestlers Have in Common?"; "How is the Ku Klux Klan Like a Group of Real-Estate Agents?"; and "Why Do Drug Dealers Still Live with Their Moms?"
Is the book worth the hype? Yes and no, in my opinion. My quick recommendation is to definitely buy it if you're the type of person who actually enjoys it when academic economists (ghost) write books for the layperson; you'll certainly be glad that you read it. However, the book probably won't live up to your expectations. Its main virtue is a collection of statistics that will probably surprise you. In other words, there are probably certain "facts" about the world that you think you currently know, which this book will demonstrate are simply myths. However, I don't think it will fundamentally alter the way you view the world, despite the hopes of the authors (pp. 205–06).
There is plenty in the book to upset the Austrian economist. (Some of this may not be Levitt's fault; it's entirely possible that Dubner, a reporter, made broad generalizations without realizing that not all economists would endorse them.) For example, consider the following description of our field:
Economics is above all a science of measurement. It comprises an extraordinarily powerful and flexible set of tools that can reliably assess a thicket of information to determine the effect of any one factor, or even the whole effect. That's what "the economy" is, after all: a thicket of information about jobs and real estate and banking and investment. (p. 13)
What's interesting about this passage is that it implies economists have a monopoly on regression analysis; other social scientists might be surprised to hear this. For that matter, mainstream microeconomists or game theorists might be puzzled as well, for our authors aren't really describing economics but rather econometrics.
Moving on, we read the following description of an economist, which may further unsettle the Austro-libertarian reader:
Economists love incentives. They love to dream them up and enact them, study them and tinker with them. The typical economist believes the world has not yet invented a problem that he cannot fix if given a free hand to design the proper incentive scheme. His solution may not always be pretty—it may involve coercion or exorbitant penalties or the violation of civil liberties—but the original problem, rest assured, will be fixed. (p. 20)
Again, there are certainly plenty of economists who fit this description, but it is not one to which someone in the Misesian tradition would aspire. Our authors seem to be glamorizing their view of the typical economist—"Oh you! There you go again, you old dog, using coercion to get people to internalize externalities!"—and thus any reader who has serious moral objections to violations of civil liberties will be, shall we say, quite put out.
My other major complaint is that the book's style makes it very difficult to nail down exactly what the authors are claiming. This ambiguity is due to the authors' desire for controversy and comprehensibility for the layman. For example, when discussing the effects of parents on children, the authors ask "how much do parents really matter"? (At this point in the book, the reader is fairly confident that our authors are going to dispel this particular myth.) They go on to state that "bad parenting matters a great deal," because unwanted children have "worse outcomes" than children "who were eagerly welcomed by their parents." And yet our authors persist: "But how much can those eager parents actually accomplish for their children's sake?" (p. 154)
The reader is then taken on a tour of possible things that could influence a child's educational performance, such as whether the parents are divorced, how much TV he watches or how many times they read to him, or how often they take him to the museum. As it turns out, none of these things matters, and more generally, a child's peers exert a far greater influence on him than his parents' child-rearing techniques.
Now let's stop for a moment and ask what our authors mean by such a claim. Simply put, if you take children who are equal in all other respects, then it is far more significant to know that one child hangs out with gang members rather than knowing that one child's parents limit him to two hours of television per night.
Ah, but Levitt and Dubner anticipate the immediate response of the reader who wishes to cling to the myth of parental influence: "Parents must matter, you tell yourself. Besides, even if peers exert so much influence on a child, isn't it the parents who essentially choose a child's peers? Isn't that why parents agonize over the right neighborhood, the right school, the right circle of friends?" (p. 155)
Sadly, I don't think our authors ever really resolve this issue. Their final verdict (and the one they gave in the abridged USA Today treatment, linked above) is that as a parent, it matters who you are (how much money you make, your educational level, whether you like to read) versus what you do (your allowance policy, whether you send your kid to a good or bad school, whether you read to your kid in the womb). First, this still admits the "conventional wisdom" that parents matter. And second, surely this conclusion succumbs to the econometric pitfall of confusing causality with correlation.
There are obviously other variables not measured in Levitt's regressions, that drive both parental education, income, etc. and child school performance. In other words, all Levitt has really shown is that as a parent, what matters is not so much how often you read to your kid, but your other actions that demonstrate to your child that you are a person who respects and expects good grades in school. So of course parents, and their "techniques," matter—even though the typically recommended techniques might not.
The most controversial topic of the book is Levitt's (in)famous theory that the Roe v. Wade decision explains much of the dramatic and completely unexpected drop in crime rates in the early 1990s. This theory seems plausible because the drop in crime happened just when the first generation affected by Roe would've reached their prime criminal years. In other words, the theory is that a bunch of potential future criminals weren't born in the first place.
Besides the sinister implications of this thesis, critics have found other problems. For example, Steve Sailer points out that if Levitt's theory were correct, then we would expect murder rates to decline first among the younger cohorts and then to decline in the older ones. In other words, you would see the murder rate committed by people under 25 fall first and then only ten years later would you see the murder rate committed by people under 35 start to fall, because it takes time for the "pre-emptive executions" authorized by Roe to work through the entire age distribution. However, Sailer claims that we see in fact the exact opposite: Homicide rates fell first among older cohorts and only later in the youngest ones.
In any event, I feel compelled to report that Levitt does not say that abortion is justified because of the drop in crime. Indeed, in an argument that ended up totally surprising me, our authors first suppose that "1 newborn is worth 100 fetuses" (p. 144). Even under this chilling assumption, abortion would still be "terribly inefficient" at preventing homicides, because the alleged reduction in murders due to legalized abortion is only a few thousand, compared to 1.5 million aborted fetuses.
Further, I would also argue that libertarian pro-lifers needn't be so determined to disprove Levitt's case. For supposing it were true, is it really so unusual that government prohibitions on a morally repugnant but nonetheless widespread practice (abortion) could have unintended consequences? Although the arguments are not identical, is this situation not similar to narcotics, where someone could consider heroin use to be sinful, and yet could recognize that government force will only make the problem worse?
This review is already far too long, so I must bring it to an end. Despite my criticism, the book has several excellent topics, including corrupt Sumo wrestlers and cheating public school teachers. There is also an interesting discussion of the impact (or lack thereof) of a child's name on his or her success. (My wife pointed out with some satisfaction that her name is #5 on the "Most Common 'High-End' White Girl Names" while mine is #5 on the "Most Common 'Low-End' White Boy Names.") If this piques your curiosity, you should definitely pick up Freakonomics.
Later they refer to regression analysis as "the economist's favorite trick" (p. 161). Well I'm an economist, and my favorite trick has always been the old switcheroo.
 I emailed Levitt about Sailer's article, and (if I understood him correctly) Levitt replied that Sailer is being misled by the varying impact of crack. In other words, if you net out the trends in homicides attributable to crack, then the underlying effects of legalized abortion remain.