Since hard and inconclusive thought is needed to transfer the results learned from randomized experiments into other domains, there must therefore remain uncertainty and ambiguity about the breadth of application of any findings from randomized experiments… (You and I know that truly consistent estimators are imagined, not real.)
Angrist and Pischke understand this. But their students and their students’ students may come to think that it is enough to wave a clove of garlic and chant “randomization” to solve all our problems just as an earlier cohort of econometricians have acted as if it were enough to chant “instrumental variable.”
That is Ed Leamer responding to Angrist and Pischke’s article on the causality revolution econometrics. Here is Leamer interviewed by EconTalk.
Those and other articles in the latest issue of JEP. Well worth reading. Sadly gated. Please post any ungated links in comments if you know them.