Structural vs. Quasi-Experimental Econometrics: Does it Need to be Winner Take All?

My blog focuses on quasi-experimental econometrics because, well, that's what I do and so that's what I can (hopefully) say something useful about. Yes, I'm a huge fan of "Mostly Harmless"-style analyses, but those techniques are not the only ways we can learn about the world using data. 

In a recent blog entry, Noah Smith provides a really nice description of the structural and quasi-experimental styles of analysis. After going through the pluses and minuses of each, he concludes: "So why not do both things? Do quasi-experimental studies. Make structural models. Make sure the structural models agree with the findings of the quasi-experiments. Make policy predictions using both the complex structural models and the simple linearized models, and show how the predictions differ."

I think that's the right advice for the profession, but what if you're a first or second year PhD student? Well, then I think you do have to make a decision. Let's say, for example, you're a graduate student at UConn, and you want to write a dissertation on the impact of immigration policies on natives. Should you work mostly with me or with Hyun Lee? I think the answer depends on which techniques you'd like to learn. Regardless of the path you take, you are absolutely responsible for knowing the limitations of your analyses.