Friday, October 12, 2012

A quick thought on Duflo/Banjeree vs. Card/Angrist/Krueger

Tyler Cowen writes:

"I believe Duflo and Banerjee (and possibly Michael Kremer too, maybe even Robert Townsend) will get it sooner than people are expecting, though not this year as they just presented in Stockholm. Next year I think."

Indeed.

But I can't decide if I'd prefer to see Card/Angrist/Krueger before or after Duflo/Banjeree [btw - these are obviously Nobel ideal-types. There are several possible combinations that could get at what these refer to].

My concern is that justifiable excitement over experiments in economics will overshadow non-experimental empirical work. I ultimately can't see how economics will ever be a primarily empirical science. Experiments work well for behavioral/micro stuff, and they work well for policy analysis (Duflo's strength), but there are a lot of economic questions where not only is experimentation hard to imagine - we also have lots of great non-experimental data to work with already.

None of these non-experimental methods are perfect. Indeed, I've been fairly hostile to a lot of IV work in the past myself. But we can still learn a lot from a large body of competing non-experimental studies. I don't want that to be lost in all the progress in experimental methods.

No comments:

Post a Comment

All anonymous comments will be deleted. Consistent pseudonyms are fine.