Friday, March 30, 2012

The Lucas Critique is a theory/empirics interface point, not a theoretical point

We're into the second week discussing monetary policy in macro, and the Lucas Critique came up last night - so I wanted to share some brief thoughts.

One of the things that bothers me is the way some people use the Lucas Critique as a bat to swing at Old Keynesian theoretical orientations. A good example is Williamson's recent criticism of the DeLong-Summers paper as "lazy" and a "piece of trash" and the authors plus Krugman as "arrogant loud-mouths" (ye who swoon over Krugman's rudeness and expect me to take you seriously - I eagerly await your posts criticizing Williamson and would be happy to repost them here). I don't think this is entirely appropriate (and to be honest I think Krugman could have been a little more specific when he talked about "decadent Lucasian doctrine" that dominates journals).

The Lucas Critique, in my mind, pointed out that lots of structural models can produce the same reduced form model. The moral of the story is you can't take an empirical relationship and infer a structural model from it. You have to do more work to support your argument.

This, in my mind, is very different from saying that all non-structural models are rubbish. Ever since Lucas, there's been a lot of work on microfoundations, and there's been a lot of empirical work trying to get at price formation and that sort of thing. And a lot of people have converged on a New Keynesian Phillips Curve that is A-OK with the Lucas Critique. We also have microfounded IS-LMs that seem to make good sense.

Given that leg-work, I don't see anything wrong with going back to an aggregate model and using it to talk about policy or teach students. Any questions about the structural model can be referred to the literature.

The point is, it's almost a historical accident that Lucas butted heads with neoclassical Keynesianism. That just happened to be the sort of economics people did at the time that this sharp guy named Lucas came along. There's nothing inherent in Keynesianism that is averse to thinking more deeply about structural parameters and models.

So in that sense, we should be careful not to mistake the Lucas Critique as a critique of the actual positions of the Old Keynesians. Instead, it was a critique of the way that the Old Keynesians defended their positions. The New Old Keynesians are fully on board with the Lucas Critique, the leg-work has been done (although it certainly ought to continue), and we now have aggregate models we talk about casually that come with implicit microfounded justifications. I think people ignore the implicit defense at their peril.


  1. I'd really like to see Steve Keen address the Lucas Critique in his models. Or come up with a good reason he doesn't have to.

    I'm guessing Lucas-esque objections are your main problem with Keen's models, Daniel?

    1. I honestly don't know Keen well enough to have an opinion one way or another.


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