Wednesday, September 21, 2011

Is looking at it from a distance supposed to make it look prettier?

Gene has a typology of business cycles up, the last of which is:

"Those that resemble microeconomic models from a distance"


The typology is supposed to be funny, but this is one type of business cycle theory that for some reason everybody goes weak in the knees over. I think I'm cynical because in class today - going over properties of binary relations - we ended up talking about a lot of assumptions of neoclassical consumer theory. It all sounds reasonable of course at first glance, but by the time the professor was done with it you wondered how this stuff survived. The reason, of course, is that abstracted consumers in neoclassical microeconomic theory provide an approximation that is useful for explaining observed behavior.

I happen to think that's a pretty good answer. I've never been one to throw out the neoclassical baby with the bathwater.

But I'm still suspicious enough of it all to really scratch my head at the obsession that many people have with "microfoundations of macroeconomics". When I told a lot of my colleagues at the Urban Institute that I was hoping to do macroeconomics, many were surprised. My supervisor said she could never get into macro because "it seemed so made up", that "micro seemed real". Another senior researcher said the same thing - that micro was "about real stuff". I really don't understand how people can think like this. Microeconomics seems like the faker to me. I know what GDP is. It's the market value of all final products. Easy. I know what the money supply is. It may be hard to count and there may be multiple definitions. But you don't have to make weird mathematical assumptions to talk about what it is. We know what it is - we just have a hard time measuring it. All the fundamental building blocks of macroeconomics are very real and you don't have to make any crazy assumptions to get a sense of them. You start making abstractions and assumptions when you begin to model their relations, of course, but you have a very firm base.

Microeconomics isn't like that. Even the building blocks of microeconomics are abstract and unrealistic, to say nothing of the theorized relations of one building block to another.

I think microfoundations will be looked back on as probably a good thing to think about some in the sixties, but something that was blown way out of proportion to its importance over the ensuing decades.


  1. You've written a number of posts on microfoundations, but I still don't know what you are trying to say. Sure, if we had stable relationships (not just negative sign, but in terms of the elasticities) between macroeconomic variables that we care about modelling, we probably could get away without microfoundations. But the problem is that we don't. And even worse, many of the correlations are contaminated the decisions that are affected by how other variables move. Because we typically don't have good exogenous variation, it forces us to model the way agents respond to movements in our aggregates. That's microfoundations.

    Marginal propensity to consume, or money demand curves don't make much sense if you don't have a model which tells you where they come from, and how they should change in response to policy.

    Could you specify what other world you have in mind?

  2. A couple things -
    First, I should compartmentalize better. My real motivation here was simply that almost all micro theory seems like it's built on a less realistic foundation than even the most naive macro. I think I've tried to stay clear in all my posts of saying "we shouldn't do microfounded macro", because I don't think that. But the idea that we should microfound macro models because realistic microeconomic assumptions are necessary for realistic macro seems bassackwards to me. There are a lot of nice things about neoclassical micro - including its provision of a strong logical foundation for macro. But realism is not among the qualities of micro I would jump to praise.

    The fact that our justification for macroeconomic relations are not ideal also does not imply that microfoundations is the obvious direction to go in. Take gravity. We don't have a great "microfounded" (i.e. - quantum mechanical) theory of gravity right now. I don't think it's unreasonable to stick with relativity but stay optimistic about string theory when it comes to gravity. The incompleteness of a macro approach does not mean that the micro approach is ideal. This is kind of my orientation towards microfounded models - cautiously optimistic.

    This point: "And even worse, many of the correlations are contaminated the decisions that are affected by how other variables move. Because we typically don't have good exogenous variation, it forces us to model the way agents respond to movements in our aggregates."

    Is good, but the other option of course is to find exogenous variation, which people try to do all the time.

    So perhaps I came across too strong. My point here is really just that I seem to be one of the only ones out there that thinks micro is a lot whackier on a fundamental level than macro. I'm comfortable modeling processes and finding which models are plausible. I don't feel the need for a grand theory. We're not physics and they're not even sure they have a grand theory waiting for them, so I'm not holding my breath for one in economics. If we can find important macroeconomic consequences of microeconomic insights, that's great. But I don't think we should feel obligated to pin down everything any more than an anatomist feels obligated to build up everything he or she does from the molecular level. When it's useful, use it. Often it will be useful. But don't think moving forward without microfoundations is obviously the wrong answer.


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