Blinder and Zandi on the Stimulus

While I was getting increasingly bearish on the stimulus yesterday, everybody else was talking about a new paper by Alan Blinder and Mark Zandi claiming that we avoided a second Great Depression. Their point and my point don't have to contradict. After all, as I had said a few times in the comments, it clearly could have been much worse. The federal government did nothing other than balance out dumb state policy - but they could have added to bad state policy, the way they did during the Depression (I am not claiming Hoover didn't increase spending - I'm claiming he didn't do what Obama did). We also had better leaders at the Fed than we did in 1929, and we had a more immediate response to the banking crisis. So sure, we hopefully did avoid a second Great Depression (so far).

Blinder and Zandi put out interesting numbers on specific programs, but the analysis is a lot the same as what we've seen from them in the past, from the CBO, and from the White House. There's a good reason for that, which I've also explained here before. We don't have counterfactuals at our finger tips. If John Adams ever bothered to say anything about counterfactuals he probably would have said "counterfactuals are elusive things". So where do Blinder and Zandi get their baseline? Well, they back it out of the actual performance of the economy with a reasonable multiplier estimate. It's about all they can do. I don't personally have a problem with that, but it obviously has its limitations, and it boils down to arguments about theory and previous econometric work. I still don't know to what extent the broader public grasps what goes into these estimates, but the economics blogosphere is well aware of how these estimates are produced at this point. That's not particularly troubling for me. The general public is not aware of how we estimate how much oil is in the Gulf of Mexico right now without directly observing it - but I assume the experts are hotly debating it amongst themselves and holding each other accountable. That's the division of labor, and so far in human history it has worked out well for us.


A few links on the paper:


- Arnold Kling makes the standard macroeconometric critique but I think takes it a little too far. He goes as far as arguing that this is rejected by the profession. Considering so many professional economists are doing this right now, that seems to me to be obviously wrong. I also think its hilarious how many Austrians are chiming in on this point. If professional rejection of a modeling approach is such an important standard for them you'd think... ummmm... they wouldn't be Austrians. Kling does make one really good point: that macro models should incorporate the research on labor dynamics done by Davis, Faberman, Haltiwanger, and others. Which, of course, is exactly what I want to do in a dissertation!


- John Taylor critiques the paper. I'm loathe to even link this. Taylor's own postings on the impact of the stimulus have been completely devoid of even a discussion of a counterfactual. The idea that he would critique the way Blinder and Zandi deal with it when he doesn't even try to is a little odd. Anyway - he just offers the standard critique, which essentially only gets us to "your estimate is only as good as your model" - which is obviously no way to dissuade someone who believes the model.


- David Leonhardt talks about the paper here.


I'm sure there were many others - feel free to share them in the comment section. This paper really doesn't add anything to similar papers that have come out so far, and the critiques don't really seem to add anything to the previous critiques, so its not especially interesting to me right now.