Mostly Harmless Econometrics

The new Angrist and Pischke applied econometrics text Mostly Harmless Econometrics is now published.

Here is a review from statistican Andrew Gelman, which is insightful both regarding the book itself and indirectly about differences between economics and statistics. In particular, his review highlights the relative importance of identification in economics and of getting the functional form right in statistics. Is there some reason we cannot have both? Note too the very different use of the word "model" in statistics than in economics. In some ways interdisciplinary communication might be easier if the different languages we speak did not share the same words.

I am a bit surprised by Gelman's call for more on hierarchical models; I think economists are right to treat these as a combination of useful pedagogical tool for education research design and an unnecessarily functional-form dependent way to get the standard errors right when then the unit of treatment differs from the units available in the data.

Gelman's point about treatment interactions is well-taken and their apparent omission in the Angrist and Pischke text is surprising. Also worth noting is Gelman's implicit assumption that matching is, as Gary King and his co-authors put it in a recent paper, a pre-processor that is followed by some more parametric procedure.

This book will be required for Economics 675 next year along with the Wooldridge graduate text and the Cameron and Trivedi text. Fortunately, the Angrist and Pischke book is priced very reasonably at $35 so that it does not add a huge amount to the cost of the other two budget busters.

I used the book in draft form a bit this year in preparing a couple of lectures and found it very helpful. All books reflect their authors, this one does so in a very big way, both via the nerdy humor and the choice of emphases and approach, but I think it makes a fine complement to the other books already available.

Hat tip: Chris Blattman (who is now on my "check every day" list)