Sunday, February 23, 2014

Fighting the last Methodenstreit

That's a German word that means "method war" and it came to mind reading Simon Wren-Lewis' post last week, "Are New Keynesian DSGE Models a Faustian Bargain?"

The reason they might seem so is the methodological underpinnings of DSGE (Dynamic Stochastic General Equilibrium) models, which are "micro-founded" macroeconomic models derived from the optimizing behavior of individuals (or, often a "representative agent") were brought into macroeconomics by Robert Lucas, Ed Prescott and others who were seeking to overturn "Keynesian" macroeconomics (see, e.g., Lucas and Sargent, 1979, "After Keynesian Macroeconomics").

The first generation of models of this type - "Real Business Cycle" (RBC - where "real" means non-monetary) implied that economic fluctuations could be optimal, and that monetary and fiscal policy were either useless or harmful (this JEP article by Charles Plosser is a good primer).

While these models failed to convince as explanations of economic fluctuations overall (as Larry Summers explained, though they can still be a useful part of the macro toolkit, as Chris House argues), the methods introduced by the RBC theorists have become nearly universal in macroeconomic modelling under the broader moniker "DSGE".  The last couple of decades have shown us that a number of "Keynesian" features, such as "sticky" prices can be incorporated into such models, which then go by the name "New Keynesian."

So the "New Keynesians" are using methods that were introduced by a cohort of macroeconomists that were explicitly anti-Keynesian.  That is, Lucas et al. won the methodological war about how to build macroeconomic models, but their anti-Keynesian view of the economy itself did not prevail.

Wren-Lewis' answer to the question posed in the title of his post is "no." Paul Krugman summarizes and responds:
Wren-Lewis’s answer is no, because New Keynesians were only doing what they would have wanted to do even if there hadn’t been a de facto blockade of the journals against anything without rational-actor microfoundations. He has a point: long before anyone imagined doing anything like real business cycle theory, there had been a steady trend in macro toward grounding ideas in more or less rational behavior. The life-cycle model of consumption, for example, was clearly a step away from the Keynesian ad hoc consumption function toward modeling consumption choices as the result of rational, forward-looking behavior. 

But I think we need to be careful about defining what, exactly, the bargain was. I would agree that being willing to use models with hyperrational, forward-looking agents was a natural step even for Keynesians. The Faustian bargain, however, was the willingness to accept the proposition that only models that were microfounded in that particular sense would be considered acceptable. It’s one thing to accept that models with an Euler condition at their core can sometimes be useful; it’s quite different to restrict your discourse to models with that characteristic, while ruling out everything else.
A couple of things to note here:

Politics: Both the academic sort in terms of who gets hired and what gets published - as Krugman alludes to, some of it was pretty vicious (at least that's my sense - this was all well before my time) and some are still holding grudges - and the political implications of the theory.  In its purest form, RBC theory has some pretty right-wing policy implications (though RBC macroeconomists are not necessarily Republicans), so some view RBC (and, by extension, DSGE) models as cover for a conservative political agenda.

How academia works: Publishing papers requires at least some incremental degree of novelty (i.e., a journal article must make a "contribution to the literature").  While the events of the last six years have underscored the usefulness of the standard textbook Keynesian approach I (and many others) teach our intermediate-level macroeconomics students, as far as publishing it, well, it was done 77 years ago.  While it is useful for policymakers and the economists working in policy institutions, academic economists are going to focus on developing new theory - which hopefully leads to better policy-making, in the long-run at least.  That is, the divide between "scientists" and "engineers" described by Greg Mankiw applies.

For more interesting thoughts on this see: Brad DeLong, Roger Farmer, Steve Williamson's response to the Krugman post quoted above, another post by Krugman.

No comments: