Microfoundations
Matthew Martin
8/10/2012 08:52:00 AM
Tweetable
I like to take a broader view of "microfoundations;" or, rather, the search for microfoundations. Microfoundations does not, in my mind, mean stopping at preferences and technology, or anywhere else, for that matter. It simply means seeking a deeper understanding.This is a point that I think has been missing from a lot of the debate over microfoundations in the economics blogosphere (the econosphere?). There are a number of so-called "heterodox" economists who legitimately loath microfoundations, but for the most part the "orthodox" economists who critiqued microfoundations--I have Krugman and Wren-Lewis in mind--are in fact major proponents of microfoundations in their own research. So I think that to a certain extent they did not communicate their views very effectively: they came off a opposing microfoundations, when I think their actual view is that economists definitely should pursue microfoundations, as long as they that there are a lot of relationships in the real world that could be highly relevant to their model but that we haven't yet found a way to microfound. In other words, their view is actually a lot like Andolfatto's.
As an aside I would just like to make a broad critique of some of the heterodox economists. Like Krugman and Wren-Lewis, I do not have an intellectual objection to their approach, per se. But if you look at, for example, this paper by Steve Keen, you can see that that kind of modeling involves a subjective judgment on the part of the researchers. He uses words like "caused" but I don't feel like after reading the paper I have any sense of what "caused" the asset bubbles--all he really shows are correlations. Indeed, he seems to overlook the fact that he is saying something very strange--investors, apparently, chose not to maximize profits. Now, if there were microfoundations, he might be able to prove me wrong--maybe there were microeconomic constraints and firms actually were maximizing profits. But we don't know, because Keen doesn't model it.
As a final aside, I just want to vent some frustration at those, like Keen, who complain about "orthodox" macro. Their complaint often goes like this: all of DSGE modeling is based on an underlying neo-classical model that is fundamentally incorrect, and therefore all of DSGE is fundamentally incorrect. I have some problems with this. First, New Keynesian economics is not neo-classical, but whatever, this is semantics. More importantly, the objection that Keen and others have with so-called "neo-classical" economics is entirely based on a much more specific version of the neo-classical model that the learned as an undergrad. In reality, the neo-classical model is far more general than the version they were taught--the model was massively simplified and restricted in order to reduce the mathematical complexity down to a level that was easy enough to teach to a crop of economics undergrads who may not yet have completed their calculus courses. So yes, the version of the neo-classical model they were taught is rife with inaccuracies. But what basis do they have for striking down the entire class of generalized neo-classical models based on the failure of a single, highly specific and restricted version of the model?
I think that part of the confusion here is that they have it backwards--they are acting as if the specific version of the model they learned was the "general case" and that all the other models are little tweaks of the generalized model. But this is inaccurate: all of these models are examples of specific versions of the generalized neo-classical model. The generalized neo-classical model makes assumptions like this: consumers have preferences that are complete, transitive, and continuous. Almost all the behaviors we could possibly find data on will approximately fit that description. By contrast the model that Keen objects to makes assumptions like this: markets are perfectly competitive, and investors have perfect information. Yeah, I can find hundreds of cases where that undergrad-economics assumption leads to inaccurate results. So if you don't like that restriction on the neo-classical model, don't use it! But don't complain that the models of monopolistic competition (a central feature of New Keynesian models) are inaccurate because the models of perfect competition are false.