Regression Is Matching (one more way; with discussion!)

Pat Kline has a nifty new interpretation of the old Blinder-Oaxaca regression estimator for two-group comparisons (in this case, applied to treatment effects in a selection-on-observables setup) . . .  It’s a matching estimator, of course!

Here’s the version we saw at ASSA in Denver, niftier than ever!

and my discussion, which is pretty good too . . . JA