The rogue conventional wisdom of Freakonomics

In the introduction of Freakonomics – A rogue economist explores the hidden side of everything, we are told that Steven D. Levitt is a young, successful, and serious–if controversial–economist. We should believe what Levitt writes because he has the right credentials: he is a professor of economics at the University of Chicago. But soon enough we find out that Levitt doesn’t believe in “expert” opinions:

[The typical expert] is prone to sound exceedingly sure of himself. An expert doesn’t so much argue the various sides of an issues as plants his flag firmly on one side. That’s because an expert whose argument reeks of restraint or nuance often doesn’t get much attention. An expert must be bold if he hopes to alchemize his homespun theory into conventional wisdom. His best chances of doing so is to engage the public’s emotions, for emotion is the enemy of rational argument.

Experts, the authors Steven D. Levitt and Stephen J. Dubner write in Freakonomics, create and then nourish what John Kenneth Galbraith calls “conventional wisdom:” a set of unproven facts, hypotheses we like to believe true because they make us feel comfortable and give us some sense of control over what happens around us. Experts have their own agenda and their own interest in mind; and expertise is nothing more than a clever but often ruthless way to make money or to gain power from the monopoly of information.

The premise of Freakonomics is intriguing. The execution is lazy and disappointing. In the end, the rogue Levitt and Dubner behave exactly as the “experts” they criticize in their book.

One – I read Freakonomics in two days and I am a slow reader. My brother-in-law got it from Christmas, read it in one day, and handed it to me. So, Freakonomics is a very fast read, even after controlling for the increased amount of free time available during the Christmas holidays and the rainy weather we had on December 26. Some critics suggest that one of the problems with Freakonomics is that the book is no more than a 200-page stretched version of the New York Times magazine article about Levitt that Dubner wrote in August 2003. (The last chapter, 33 pages including notes, is a quite pointless list of babies’ names by socio-economic status, race, and mother’s years of education plus lengthy description of regression analysis. Talk about fillers.)

Two – Freakonomics is an uneven, half-baked book; it must have been sent to the presses before a good editor had a chance to work on it. Content aside, the presentation of the reasoning is too often approximate and the writing sloppy; the writer has a predilection for liberal use of exclamation points. (As we learn from the book, one should be suspicious of exclamation points: “And an exclamation point (…) is bad news for sure, a bid to paper over real shortcomings with false enthusiasm.”)
Freakonomics could have been a good book, had the authors taken the time and made the effort to write a real book. Not that they didn’t warn us: “In New York City, the publishers were telling Levitt he should write a book. ‘Write a book?’ he said. ‘I don’t want to write a book.'”

Freakonomics has some interesting insights, provocative hypotheses, intriguing case-studies of “conventional wisdom” deconstruction, the notion that truth is not always as politically correct or reassuring as we would like, and the statistics-101 notion that correlation is not causation. All good stuff. However, from a market economy point of view, the books is not worth the $25.95 cover price (but wait, you can find it at Amazon for $15.11; is that a sign?). If you want to read it because all the cool kids have already read it and are talking about it, borrow it from a friend or from you local library.

Three – Dubner mentions several times the lack of a “unifying theme” in Levitt’s work; Freakonomics, he writes, is a book about “everything.” After reading it, I feel that this is just a weak justification for the book’s mediocrity. It’s of little interest to me whether Levitt should pursue a unifying theme in his research. But I think that the lack of unifying theme is an outcome of Levitt’s methodology.

Levitt was able to access some large databases, mostly from Chicago, the city where he works. He did not collect data. He did not hunt down the answer to big burning questions. In a likely–but unproven–scenario, some graduate students ran a bunch of regression analyses and Levitt looked at the results. He noticed some interesting and unpredicted correlations and took note of them. He talked about these results with his buddy Stephen Dubner. Dubner took notes, read a few other articles, and wrote the book. An interesting start, but not nearly good enough. (By the way, this doesn’t mean that Levitt is not a gifted young economist; hopefully, his peer-reviewed papers are more serious and rigorous than this book.)

Four – The book shows us that you can get some wacky and unexpected results when you play with data analysis techniques and large enough data sets. But the authors make clear that they don’t have any interest in finding solutions to high-impact complex social problems.

Morality (…) represents the way that people would like the world to work–whereas economics represents how it actually does work.

I passionately agree that before we look for solutions we need to understand the problems, even if the truth refuses to be comfortable, reassuring, or pretty. But the definition of “understanding how the world actually works” implicit in Freakonomics bothers me. The authors seem to have a lot of fun creating fragments of understanding and then happily moving to the next topics. They don’t put all the pieces together to create insights on complex and dramatically relevant problems. They don’t provide a convincing explanation of their methods; they don’t present enough data to convince us that what they are describing are really causes, not just correlations. They don’t point in the direction of possible solutions or at least additional factors to explore. Judging from this book, the job of an economist is to come up with a bunch of regression analysis results that are provocative enough to create controversy. To me, this looks very much like conventional wisdom created by experts for their own personal gain.

Five – There is a deeper and more fundamental problem with Freakonomics. Levitt analyizes past situations and looks at what variables might have affected the observed outcome. He admits these are not controlled experiments, so the investigation is conducted after the fact. This means that there is no guarantee that future situations will behave the same way. Past events are constrained by the context in which they occurred.
As in the complex and volatile world of financial markets, we should be warned that “past performance does not guarantee future results.”

I deal with this problem every day. Let’s assume that one of our products has problems. People don’t use it, or maybe they use it but they are not successful. We examine people’s current behavior with the product and we rapidly realize that what people do is constrained by how the product works. When we are examining an interactive, living system, rather than a mechanic juxtaposition of additive parts, looking at the current behavior is not enough to predict the effect of changes to the system.

Six – Let’s examine the most controversial issue discussed in the book, the analysis of the unexpected decrease in violent crime during the ’90s. Levitt mentions, among others, the following factors that according to his analysis contributed to this outcome:

1. Increased rate of imprisonment
2. Increased number of police
3. Legalization of abortion

There are critics who doubt that Levitt’s analysis is correct, but let’s assume for a moment he is right and that in the ’90s violent crime decreased because of these three factors. What does this mean for our current situation? Would the same factors work in the future? Should we continue to put more people in jail, to encourage abortion in lower income black young women, and to put more police in the street?

How sad if this was all we could do. Unless, rather than analyzing readily available data set, we dare to ask big questions, and then look for the data that can help us answer them. Of course, this requires a lot more effort.

For a moment, let’s follow the hypothesis that a high percentage of unwanted poor children are going to grow up to be criminals. So, what makes a child poor and/or unwanted? Maybe we should have a better birth control campaign rather than using abortion as a way to control unwanted births (let’s examine violent crime trends in countries that have just started a serious program on contraception education; or let’s start our own controlled experiment). Or maybe family policies have something to do with it (let’s examine countries who have changed their laws to ensure more protection for mothers and children). Perhaps excessive economic inequality in this society creates large pockets of uneducated poor mothers (let’s examine historical data on violent crime trends as a function of wealth distribution and access to education). And so on.

The problem is not that Levitt thinks in an unconventional way and does not bend to liberal or conservative ideologies. The problem is that Levitt according to Dubner thinks too narrowly. He looks at the past, not at the future. He looks at the world as a giant clock to disassemble. He doesn’t have a vision. In the end, what is presented as the strength of Freakonomics becomes its greatest limit: economics without ethics is not more than a vain exercise in regression analysis.