Mostly Harmless Econometrics
by Joshua D. Angrist & Jörn-Steffen Pischke
Notes
Mostly Harmless Econometrics is an admittedly narrow, but very unique and fun to read textbook on modern (reduced form) micro-econometrics. If one wants to understand the credibility revolution1 in modern economics, and where it's led us, they should start here.
The title of this paper, by the same authors of the book, is a nod to a famous article by the late Ed Leamer, which set off a slow turn away from model-based inference towards design-based inference. The former represents a primary concern for correctly specified economic models as a pretext for unbiased effect estimates; the latter is instead concerned with independence/exogeneity and randomization. Angrist and Pischke clearly take inspiration from Leamer, first introducing the study of observational data with regression as an analogue of controlled, randomized experiments. Leamer's article has aged very well, and is well worth reading today.