Believe those who are seeking the truth. Doubt those who find it. Andre Gide


Sunday, July 29, 2012

The Microfoundations Windmill

Well done, Don Paul!
I can't help it. I just have to say something about  Paul Krugman's latest complaint (in a series of seemingly never-ending laments) concerning yet another "problem with the economics profession." See here: Making Ourselves Useless.

A well-aimed critique constitutes an important step in helping us understand things better. In this case, however, I think he is largely making things up--methinks our fair knight is chasing windmills.

Krugman begins by quoting Simon Wren-Lewis (who I happen to find quite sensible most of the time, just not in this case) in reference to the profession's alleged obsession with "microfoundations:"
If you think that only ‘modelling what you can microfound’ is so obviously wrong that it cannot possibly be defended, you obviously have never had a referee’s report which rejected your paper because one of your modelling choices had ‘no clear microfoundations’. One of the most depressing conversations I have is with bright young macroeconomists who say they would love to explore some interesting real world phenomenon, but will not do so because its microfoundations are unclear.
Oh, please. Papers are rejected all the time and for all sorts of reasons. That goes even for papers with microfoundations. And as for those "bright" young economists, they sound truly misguided. Not sure who's to blame for that, however. 

It is true that "microfoundations" are valued in the profession (and Wren-Lewis has several excellent pieces explaining why). But just what are these pesky "microfoundations," anyway?
  
A narrow view of  "microfoundations" is reflected in the idea that the methodology of microeconomic theory (specifying individual preferences, information sets, endowments, constraints, together with an equilibrium concept) can and should be brought to bear on macroeconomic questions. This is in contrast to an earlier methodology that specified and estimated behavioral relations at the aggregate level. (One can legitimately weigh the pros and cons of these (and other) methodologies.)

Not many macro models are "microfounded" in a pure sense. Almost all models make at least some assumptions that may be viewed as ad hoc and provisional (subject to further investigation). I think of an ad hoc assumption as a restriction on behavior that is inconsistent with other aspects of the model, like maximizing behavior.

To give an example, in most "microfounded" models of money there are ad hoc restrictions placed on the set of assets that might serve as exchange media. Consider a model with money and bonds. The modeler typically assumes that money is used to buy things and that bonds are not. While a "cash-in-advance constraint" of this sort may be descriptively accurate, it does not explain why bonds cannot be used to buy things. In short, liquidity is assumed and not derived. It is generally understood that shortcuts of this sort may matter for some questions and not for others. Understanding where and how these assumptions matter for the particular question at hand is part of the skill set that defines a good economist.

Sticky nominal wages is another popular example. Actually, in this case, I think it's rather worse. As I explain here, sticky nominal wages are likely only relevant if one adopts the questionable assumption that the labor market operates as a sequence of  anonymous spot markets.

Anonymity is a very bold assumption. In particular, it rules out the formation of relationships--something that most of us would recognize as being an important element of most labor market transactions. If the labor market works more like a marriage market, then spot wages (whether real or nominal) are inconsequential for resource allocation. What matters is the manner in which surplus is divided. And generally, there are many wages paths (real and nominal) that are equivalent to dividing the surplus in a particular manner. (This is something Barro (1977) pointed out long ago.)

So why do I mention this? Well, let's see what Krugman has to say:
And this [Lucas Critique] is fair enough. But what if you have an observed fact about the world — say, downward wage rigidity — that you can’t easily derive from first principles, but seems to be robust in practice? You might think that the right response is to operate on the provisional assumption that this relationship will continue to hold, rather than simply assume it away because it isn’t properly microfounded — and you’d be right, in my view. But the profession, at least in its academic wing, has largely chosen to take the opposite tack, insisting that if it isn’t microfounded — and with all i’s dotted and t’s crossed, no less — then it’s not publishable or, in the end, thinkable.
Can you spot what's wrong in that passage? No, it's not the first sentence--it's everything that follows.

First, I see a lot of other facts in the labor market that I might like to model, like the coexistence of large gross flows of workers into and out of employment--something, sticky nominal wage models frequently ignore. So maybe I want to ignore sticky nominal wages because I'd rather model worker flows--not because I can't "microfound" the phenomenon.

Second, and more important, it is clear that he is just making things up here. Why do I say this? Well, just take a look at one of the dominant paradigms in macroeconomic theory--the New Keynesian framework. As anyone who is familiar with the paradigm knows, it is built around models that embed ad hoc assumptions reflecting the alleged costs associated with nominal wage and price adjustments in auction-like settings. It seems to me, on the basis of this (and plenty other) evidence, that the profession cannot be obsessed with microfoundations in the way that Krugman suggests. On the whole, the profession is much more pragmatic than he makes it out to be.

By the way, 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. (My colleague, Arthur Robson, for example, is exploring the "microfoundations" of preference formation.) I certainly hope that this search for deeper understanding is not the "obsession" that Paul Krugman is concerned with.

What I find puzzling is that I'm pretty sure that the K-man knows all this. But if so, then what motivates his insatiable desire to tar-and-feather the profession as a whole in this manner? I find the following passage illuminating:
Now we’re having a crisis that makes perfect sense if you’re willing to accept some real-world behavior that doesn’t arise from intertemporal maximization, but none at all if you aren’t — and to a large extent the academic macroeconomics profession has absented itself from useful discussion.
Well, maybe not that illuminating. I mean, he can't literally believe this given that he has a paper with Gauti Eggertsson that makes use of use of intertemporal maximization that purports to explain recent events.

At root, I think the source of the man's bitterness toward the profession is that in his view, we are doing this stuff called "research" into questions for which we already know the answers (the answer is to increase G, something I partially agree with here). We are fiddling like Nero while the economy burns.

I would like to ask Krugman whether he believes there is anything left to learn about how an economy functions in the aftermath of a financial crisis. Is the profession wrong in devoting a good part of its time searching for a deeper understanding ("microfoundations") of how monetary and fiscal policies work using its best available tools? How would he rather we spend our professional time?

Or is the science now settled?