He doesn’t like it. (HT: Paul Solman) I share Robb’s skepticism about drawing conclusions from empirical work that either ignores other factors or purports to control for them in the name of creating a “natural experiment.” Having been in the kitchen, I know how this dish gets cooked. Those who have never been in the kitchen assume it must be good food.
Robb on Superfreakonomics
Share
Previous post: Apologies to Paul Krugman
Next post: Financial reform
{ 8 comments }
I haven’t read Superfreakonomics yet, but that was exactly my take on Freakonomics. It was a very interesting and intriguing read, but I wouldn’t commit to believing any of it. It was clever, innovative, well argued, thought-provoking, and of course fun to read. That’s a contribution in and of itself. Doesn’t always make it accurate.
There’s another downside to the natural experiment/IV mentality. Not only do few people usually believe it outside of the authors and a small group of followers – but it also drives research in directions that are amenable to natural experiments or IV models. Research should be dictated by what knowledge is important for us to have. It shouldn’t be dictated by how it allows us to cleverly identify a model.
Having read the book and the Solman article, I can only conclude that Mr. Solman hates Levitt, and tries to discredit or ridicule him as much as possible. What a humourless idiot.
The article is by Robb, not Solman. Solman was just my source for the article.
Sorry to Mr. Solman. Seems I can’t edit my post though.
I’ve admired the idea behind “Freakonomics” of making more people think economically, but I have also been concerned about their easy use of econometrics to prove a prior opinion held by the researcher. Prior opinion: Unwanted children lead to unloved children lead to criminal behaviour. Then do a regression that appears to prove it and write a paper that makes headlines in all the leading newspapers because it confirms journalists’ prior opinions. However, numerous research papers have examined the flaws in that original paper and found no connection. A recent paper from NBER summarizes those findings:
“Abortion and Crime: A Review”, by Theodore J. Joyce, NBER Working Paper No. 15098, June 2009.
The argument that Freakonomics is not good economic/econometric analysis is not new. There is an article in The New Republic, by Noam Scheiber, critcially evaluating Freakonomics. It is here:
http://www.tnr.com/article/freaks-and-geeks-how-freakonomics-ruining-the-dismal-science
Manfred
There is also an article by Ariel Rubinstein in the Economists’ Voice, Volume 3, Issue 9, Article 7 (year 2006) commenting on Freakonomics.
Of course, Ariel Rubinstein is a die-hard game theorist, thus his comment need to be taken with a grain of salt. But still….
Empiricism is very useful. But disentangling the line of causality requires more than raw data. It requires UNDERSTANDING. That means context and, yes, some level of theory against which to test that understanding.
Love the analogy of cooking, Russ. Recipe in a book vs. Execution is a brilliant metaphor for the hubris of distance experts vs. the experiential local knowledge of practitioners.