Causal identification is dead. Long live causal identification.

I have been misinterpreted, for which I only have my brevity and haste to blame.

Yesterday I wrote:

more and more I wonder whether the causal fetish club in labor and development economics (of which I am a card-carrying member) is asking too many old and tired questions with clever causation rather than new and important questions. I think the young economic historians, political economy people, and comparative politics crowd are way out in front on this.

A commenter asked:

What do you mean by the new and important questions that rely less on estimating causal effects? Can you give an example from the fields you mention?

And Duncan Green tweeted

Is ‘causal fetish club in labor and development economics’ tired and obsolete asks @cblatts. Yup.

To be clear, I don’t think causation is less important than before. Of course it’s not the only thing empirical scholars should focus on (I have a prediction paper coming out soon). But some of the best work does it well.

What I get a little tired of is the 300th paper on the returns to education in the US. Or the 44th paper that revives the Catholic schools and performance question. Or the 3000th paper that uses information to increase voter turnout.

A big part of the reason scholars work on them is not because there is something new to learn, but because other people worked on them before. The schooling literature is one used to teach us economists the lessons and evolution of causal identification. People of my generation like cute new schooling papers just like we like new Star Wars movies, or a re-release of The Goonies. We love them and despise them.

The new and exciting stuff is still causally identified, but it’s asking new and interesting questions. In the sphere of political economy of development, if you want to see examples look up Nathan Nunn, Jake Shapiro, David Atkin, Suresh Naidu, Fotini Christia, Macartan Humphreys, Nancy Qian, Ted Miguel, or Leonard Wantchekon, to name just a few. See my syllabus for more.

4 Responses

  2. Phenomenal syllabus. I’m curious how you think about merging the causal ID readings (i.e. the authors you name in the post) with more traditional observational empirics (like Frye, Boix, Nielson et al., Tsai, Querubin, Arriola, Scheve and Stasavage, Acemoglu and Robinson ) with big-think qualitative work (like Thelen, Tilly, Weingast et al., the POP pieces by Levitsky/Way and Staniland). Just curious – it seems like an interesting intellectual/pedagogical challenge in terms of making them complementary, or setting them up in opposition to one another, or something else. Maybe that’s fodder for another post. . . .

  3. Reminds me of an academic reading club I used to attend on animal cognition. The prof’s seemed really hung up on tiny issues like is it associative learning or “real” learning (and hw would we know? and what does that mean?)

    In some ways it seemed overly narrow. But then again, isn’t this what academics are *supposed* to do? These are the people who write monographs. They are supposed to dig-dig-dig at a laser-thin question until they truly, finally, oncea-and-for-all put it to bed. Then the human race has figured out one thing for sure.