Three and a half years in the past, again in early 2022, I began a podcast referred to as The Mixtape with Scott. The unique thought was easy: hint out the historical past of causal inference by means of its connection to Orley Ashenfelter, and every other aspect quests I wished to pursue. I had begun realizing, vaguely at first then a bit extra clearly, throughout the writing of my guide that Princeton’s Industrial Relations Part had been the fountain head of causal inference inside economics, however later I noticed it extra squarely part of Orley’s legacy within the career. So I pursued the podcast as a manner of making an oral historical past of causal inference in economics. I wished to hint these connections, hear these tales, perceive how concepts moved from individual to individual.
However like most issues in my life, it sprawled. The podcast grew to become “College students of Gary Becker,” then “College students of Jim Heckman,” then “College students of Orley.” Then it grew to become departments I used to be interested by—Stanford, as an illustration. Fields I cherished—labor economics, market design, econometrics. Textbook writers. Economists who left academia for tech. On and on. Round 130 interviews later, right here we’re.
And now I’m asking myself, “How do I land this airplane?”
I’m not achieved with the podcast, not precisely. However it might must evolve. I’m speaking with a few individuals a few totally different format. I’ve acquired two extra interviews already recorded—one with Alan Auerbach, one with Andrew Gelman—sitting within the can. However one thing else has been pulling at me.
I need to write a guide concerning the historical past of causal inference in economics instructed not merely by means of concepts, but in addition the sociology and biography of the concepts. I don’t actually imagine “concepts” do something; fairly I believe individuals do issues, and since individuals have concepts and speak, concepts do issues. They simply do these issues as mediated by means of individuals. Actually, I’m actually undecided if “concepts” can do something ever and so I’m actually not within the camp that may ever say “Concepts Have Penalties”. Not with out individuals anyway. It’s folks that have penalties. Persons are the confounders. Folks and locations and functions.
The guide can be concerning the evolution of causal inference in economics, transferring primarily by means of Orley Ashenfelter and his college students, and the way that stream ultimately linked up with Don Rubin’s potential outcomes framework by means of the work of Guido Imbens and Josh Angrist. It’s concerning the three main causal inference traditions within the trendy period of economics: Princeton (pure experimental design), Harvard (experimental design) and Chicago (Heckman and structural approaches). It’s concerning the position that quantitative labor economics performed in all of this. The Industrial Relations Part at Princeton was a labor group. Orley is arguably a first-generation quantitative labor economist—Card and Farber referred to as him precisely that in a festschrift. The position that labor economics performed, and the position that Princeton performed, is one thing I preserve turning over in my thoughts. I simply don’t assume I’ll relaxation till I get it out of me, too.
So what I’m doing now could be constructing out a database I began years in the past however by no means completed. And I’m calling this database the Orley Family tree Undertaking.
I’ve gotten private lists of pupil advisees from a number of individuals—Card, Katz, Angrist, Currie. I scraped CVs for others, like Heckman. I don’t even bear in mind how I acquired Krueger and Imbens’s lists, although I believe Guido despatched me his immediately. There’s knowledge you’ll be able to pull from ProQuest, and I’ve dabbled there, however getting it at scale requires a subscription and a few infrastructure I haven’t constructed but. It’s on my to-do record. However truthfully? I get pleasure from doing it this manner. Whenever you scrutinize every node and every edge your self, you be taught the story. You’re feeling it.
The newest additions got here from scraping the Arithmetic Family tree Undertaking, which helped me fill out extra of the third and fourth generations—what I’m calling the grandchildren and great-grandchildren. The construction seems to be one thing like this (simply displaying a number of examples):
Technology 1: Orley Ashenfelter (a ton of missingness)
Technology 2: Jim Heckman, David Card, Joshua Angrist (each Card and Orley suggested Angrist), Robert LaLonde, Janet Currie, …
Technology 3: Alberto Abadie, Esther Duflo, …
Technology 4: College students of Abadie, college students of Duflo…
Alongside I put Don Rubin, however as this isn’t the story of causal inference a lot as it’s the story of causal inference inside economics, I’ve solely selectively pulled Don in, by way of his connection to economists like Josh and Guido, in addition to a number of others (Dehejia and Wahba, as an illustration).
It followers out quick. And it’s an advanced story, so becoming it onto a graph is its personal problem. I’m at round 1,100 economists tracked and need to preserve going—there are nonetheless a whole lot of lacking nodes. Getting Orley’s full record, as an illustration, is difficult as a result of he didn’t preserve information the way in which Katz did. Katz saved meticulous information. Orley… didn’t. Anyway, try what I’ve to date. Fairly proper? That is with the brand new Math Family tree Undertaking scrape I did right this moment.
I’ve additionally acquired Claudia Goldin in right here. She was at Princeton, I believe within the Part high, earlier than she left and located her method to Harvard. She’s related to the cliometrics custom by means of Robert Fogel—a distinct stream, however one which flows into the identical river as a result of she is a colleague of Orley’s.
The best way I’ve been occupied with that is by means of metaphor. It’s a community. It’s a tree. And it’s two rivers converging. The Princeton quasi-experimental custom on one aspect, the Rubin potential outcomes framework on the opposite, and someplace within the center, Angrist and Imbens linking instrumental variables to the Rubin causal mannequin. And by way of Heckman, the structural department. These metaphors assist me assume. For no matter motive, I appear solely capable of perceive issues by means of story, emotion, and picture.
Which brings me to you.
I’ve arrange a kind on my web site referred to as The Orley Family tree Undertaking. It’ll let me crowdsource this undertaking. And so in the event you’re studying this, I might use your assist.
I’m searching for the kids, grandchildren, and great-grandchildren of the causal inference motion inside economics. This implies Orley’s descendants—together with colleagues like Alan Krueger. It means Heckman’s college students. (Heckman is positioned in Orley’s direct lineage, though I’ve gotten three separate sources—together with Heckman’s personal CV and his early acknowledgments—that every one identify totally different major advisors. I requested Orley as soon as who the “true advisor” was, and he principally stated he thought he was most likely one of many major ones. So for now, Heckman sits as a descendant of Orley, although clearly Chicago, Princeton, and Cambridge every signify their very own distinct streams.)
It additionally means the Rubin aspect—Paul Rosenbaum, Andrew Gelman, Elizabeth Stuart. I place collaborators on right here which is how I linked Guido in — by way of his coauthorship with Josh after which Don. As soon as Imbens and Angrist linked IV to the Rubin causal mannequin, these grew to become one household. Heckman and Robb linked causal inference to Rubin earlier (1985) too, however that’s one other story for one more time.
Right here’s what I want:
If you understand somebody who belongs on this family tree—an advisor-student relationship, a key collaboration, a lacking node—please submit it by means of the shape. However examine first. I’ve posted a searchable record of everybody already within the database so you’ll be able to see in the event that they’re there earlier than you submit. The shape asks for a supply—a CV, a webpage, a dissertation acknowledgment—so I can confirm.
This can be a residing household tree. It’ll preserve altering as I be taught extra and as you assist me fill within the gaps.
Once more, the web site is right here.
Thanks for studying. Thanks for serving to. I’ll preserve you posted as this factor grows. Within the meantime, right here’s some nice photos.
