The Fundamental Problem of Causal Inference

This week’s episode of the Mixtape with Scott is a little out of order. Season two’s episodes are going to be a little out of order, based on what feels like the best next episode to present at that time. So I decided after doing my interview with University of Arizona professor of economics, Hide Ichimura, that I wanted to release it because I had such a delightful time talking with him. Dr. Ichimura is an econometrician whose work I’ve gotten to know more recently because it’s been experiencing a little bit of a revival (though it’s always remained very popular over the years) within the difference-in-differences literature thanks, in part, to the Sant’anna and Zhao (2020) Journal of Econometrics on robust diff-in-diff, Callaway and Sant’anna (2021) paper on differential timing, and in many ways, other papers that conduct certain kinds of imputations and estimations that are similar in spirit like Borusyak, et al’s (2022) robust efficient imputation estimator, and even Abadie and Imbens (2011) selection bias adjustment method if you squint your eyes.

I had a wonderful experience talking with Dr. Ichimura today. This is sort of part of my broader interest, as I say in the intro, in interviewing econometricians who were active in the 1990s working on topics in causal inference, and to that end, I had in mine two Restuds by Dr. Ichimura with Heckman and Todd (1997) and a 1998 one in Restud also by Heckman and Todd (and the identical title!!), both on program evaluation. But I also just in general wanted to hear his story, and I’m so glad I did and that he would share it.

At the end of the episode, I asked him to share with me a paper that, maybe isn’t his favorite, but that has always stuck in his mind. He shared with me Stephen Nickell’s 1979 article in Econometrica entitled “Estimating the Probability of Leaving Unemployment”.

As always, opening introduction music is by Wes Cunningham (no relation).

And don’t forget to subscribe, share and maybe even support this! This podcast is subsidized by your donations and my workshops!

Scott's Substack is a reader-supported publication. To receive new posts and support my work, consider becoming a free or paid subscriber.