March 28, 2005
Notes toward a review of Steven D. Levitt and Stephen J. Dubner's forthcoming economics book, Freakonomics. I thank Julia Bannon, Online Marketing Manager for HarperCollins Publishers, for sending me a proof copy.
For the most part, the book succeeds. It is interesting, useful, and, in spots, entertaining. But it is also frustrating. (I concede that the book is not intended for professional economists.)
First, the good news.
1. Economics is based on a handful of simple, even obvious, ideas. What distinguishes economists is the single-mindedness and depth with which they pursue those ideas to their logical conclusions. Consider, for example, one of the most basic ideas: economists believe that people tend to respond to incentives. This seems reasonable, but it doesn’t seem surprising or terribly useful.
But Levitt and Dubner push this assumption to its limits and in so doing produce some impressive insights. We find out that, contrary to what at least some criminologists once believed, if criminals are punished more severely, they definitely commit fewer crimes (p. 121). We learn whatever merits high-stakes educational tests have, they also clearly have at least one cost: teachers affected by them will cheat. Cheating was occurring in at least 5% of Chicago’s public school classrooms (p. 32). Not only did economics suggest to Levitt that teachers would cheat, it provided the strategy for detecting it. Levitt also found evidence that sumo wrestlers in Japan cheat (pp. 36-42).
2. In an especially nice touch, Levitt and Dubner remind would-be social planners that while people respond to incentives, they don’t necessarily respond in simple, predictable ways. Paul Feldman sold bagels to white-collar workers on the honor system, but found that an overwhelming majority of the workers paid anyway (pp. 45-49). Ten Israeli day-care centers wanted parents to pick their children up on time. When some parents didn’t, the centers imposed a $3 fine per child. Result? The number of late pickups more than doubled (pp. 17-18).
3. Levitt and Dubner stress a truth that even most good books on economics do not: how hard it is to know—really know—something in the social sciences. First, you need good data, which as chapter three emphasizes, often requires long, tedious, unglamorous work. But even after good data is in hand, the investigator usually has to confront a difficult issue, the issue of causality. For instance, crime rates tend to be positively correlated with the amount of money spent on police (p. 124). Do the police induce more crime? Or does higher crime produce a higher demand for police? Seem obvious? Be careful. Levitt and Dubner report that Atlanta police underreported crime before the city won the right to host the 1996 Olympics (p. 90). If reported crime rates can be manipulated downward, they presumably can be manipulated upward (perhaps to support a request for more manpower and funding). Even if the investigator can demonstrate that yes, causality runs from more crime to more police, how can he then determine whether more police also deters some criminals and incapacitates others, thereby lowering the crime rate?
For another example, consider “distinctively black” names. Does such a name cause a student to do less well in school and later in life because teachers and employers discriminate? Or is such a name merely a reflection of factors—having a mother who is young and poorly educated—that tend to make it harder for someone to do well (pp. 180-187)?
Such questions are difficult. A big reason why Professor Levitt’s work is so respected by other economists is that he finds exceptionally clever ways to address them.
4. Scattered through the book are tidbits that should interest non-economists and economists alike. In 1987 when the IRS required filers claiming dependents to provide the dependents’ Social Security numbers, one in ten dependent children—7 million—vanished (p. 23). The internal organization of a drug-selling gang is an example of the economic theory of tournaments (pp. 102-03). The renowned sociologist William Julius Wilson assigned a graduate student to administer a multiple-choice survey to residents of Chicago’s poorest neighborhoods. The first question was “How do you feel about being black and poor?” and the five possible answers were “very bad,” “bad,” “neither bad nor good,” “somewhat good,” and “good”. After a day of trying to get answers to this question, the graduate student realized a sixth answer was needed: “fuck you” (pp. 92-93). Projected to be among the most popular girls’ names in 2015 (at least in California): Annika, Clementine, Eleanor, Grace, Maeve, Philippa, Quinn, and Waverly (p. 201).
And now, the book’s frustrations.
1. Two parts of the marketing are ill-conceived. One is the title, Freakonomics. The authors thank someone for the title so they evidently approve. They might not have teenaged children yet, but is there no one at William Morrow who does, who could inform them about the unsavory connotations of “freak”? I do, so I’ll help: http://www.urbandictionary.com/define.php?term=freak.
Second, the subtitle is “A Rogue Economist Explores the Hidden Side of Everything”. “Rogue” is also a poor choice. Whatever Professor Levitt may be, he is not a “rogue” economist; see, for example, http://www.bartleby.com/61/62/R0286200.html.
These observations are more than just nitpicks. “Freak” and “rogue” are meant to suggest that Professor Levitt is unorthodox, that economics never asks the kinds of questions he asks. To the contrary, Professor Levitt’s work is very mainstream. He publishes in the leading, most established journals of economics: American Economic Review, Quarterly Journal of Economics, Journal of Political Economy, Journal of Law and Economics, etc. He is working in a well-established tradition, a tradition led by honored economists such as Gary Becker, Milton Friedman, George Stigler, Ronald Coase, and James Heckman. These two terms will suggest to the average reader something completely false about economics, and that’s too bad.
(Also, a reader who expects that he is going to learn about the hidden side of “everything” in this book of 205 generously-spaced pages is going to be disappointed. But this is just normal marketing hype, so perhaps it’s more excusable.)
2. There are a number of places where the book wanders and that seem to me like the authors were simply padding the book. A prime example is nearly a half page of completely gratuitous, ad hominem attack on John Lott (p. 131). I know there are hard feelings between Levitt and Lott, but must the authors stoop to this? (Note also that their claim that Lott's work has "not been replicated" by other researchers is utterly false. Some other researchers disagree with his findings, but none, so far as I know, have claimed they can't replicate his work. This apparently small difference in terminology is important, at least in academia.) Other discussions that should have been trimmed: almost a half-page of slang terms for crack (p. 215); six pages on the invention of crack (pp. 107-112); eleven-plus pages on the history of the Ku Klux Klan and the person who allegedly caused its fall (pp. 53-64); more than a dozen lists of first names (chapter 6); and, finally, a contrast between a black economist at Harvard, Roland Fryer, and the Unabomber, Ted Kaczynski. Professor Fryer grew up in very tough circumstances; Mr. Kaczynski had a nurturing family. The contrast makes the point that however important heredity and early environment are, they don’t completely dictate life’s outcomes.
But what person over the age of, say, fifteen doesn’t know this?
3. There are a number of issues that merit further discussion and that would be better use of the space. I, for one, would like to read more on how Professor Levitt develops his ideas. Chapter four discusses one of his best-known ideas (in joint work with John J. Donohue III), the claim that legalized abortion is responsible for about half the steep drop in crime rates in the 1990s. What prompted them to explore this? The book mentions that someone running for office suggested this in a 1990 book; did that inspire the work, or was it something else? Similarly, what gave Levitt the idea to use electoral cycles in police hiring to study the effect of policing on crime? Why did he look at sumo wrestling for evidence of cheating? Even if the ideas come to him as bolts out of the blue, it would be interesting to know more about his thought process.
Some further issues that warrant more discussion are as follows. The legalized abortion-crime drop idea has been criticized and not much space is devoted to considering the criticisms. (A good summary of the debate appears in a paper by Jonathan Klick, “Econometric Analysis of U.S. Abortion Policy: A Critical Review.”) Levitt’s results in his police effectiveness paper have been weakened by subsequent research, and that should at least have been noted (Justin McCrary, "Using Electoral Cycles in Police Hiring to Estimate the Effect of Police on Crime: Correction,” American Economic Review, September 2002). The statement (p. 66) “. . . information asymmetries everywhere have in fact been mortally wounded by the Internet" is questionable. In fact, the authors seem to question it on the next page: “The Internet, powerful as it is, has hardly slain the beast that is information asymmetry.” The statement (p. 173) “. . . it isn’t so much a matter of what you do as a parent; it’s who you are” is questionable and claims more than the evidence presented seems to support.