Critiquing Micro-Foundations, a Response to Mark Thoma

There has been a lot of talk on the blogosphere about so-called "micro-foundations" for years. It has especially picked up since David Levine's post on Keynesian economics.

So it should not surprise me to wake up and read two posts on the topic: one from Paul Krugman and another from Mark Thoma. After thinking about it more by myself and on Twitter, I am still left confused about what the critique is of micro-foundations. Thoma's article is more substantive and interesting, so I'll focus on his.

Macro, by definition, deals with aggregates, such as unemployment, GDP, and inflation. While individual actions determine GDP, economists study them as aggregates statistics. So first off, all macro is aggregation.

According to Thoma, there are two main ways to do macro. The newer version uses micro-foundations. These models start with standard micro-style consumers and producers. The other approach, older in origin, assumes some steady macro relationship, such as a constant relationship between aggregate consumption and aggregate output. This is the style of Keynes and Fisher.

Since the 1970's, macroeconomists have tended to use the micro-founded approach. But it's still an aggregate. Both are aggregates. Even Thoma recognizes this when he describes the modern approach which must ultimately "aggregate across household and firms to determine macroeconomic relationships." The most common way to do this is using a "representative agent." When Thoma draws a distinction between micro and aggregate models, I'm left confused. I might not be reading it right.

Leaving that aside, Thoma levels a reasonable critique of the representative household models:

Unfortunately, the representative agent approach is unsuited for studying behavior in financial markets. The problem is that there is no way for a single representative household to trade stocks and bonds with itself based upon different forecasts of future economic conditions (e.g. a person who thinks the price of a stock will fall in the future sells the stock to someone who believes the price will rise).

I agree. The representative agent model is not suited for financial markets. But the old Keynesian model of aggregates is suited for financial markets? How does the aggregate Y and aggregate C or IS/LM or whatever aspect of the Keynesian model explain financial markets? I don't see it. To explain financial markets, there must be two people who value a financial asset, e.g. a stock, differently.  I just don't see how the representative agent or the old Keynesian model can explain. Any critique of the micro-founded DSGE model in this respect also is a critique of the old models. So the score is 0-0.

There is one difference that I see. DSGE models have developed beyond the representative agent models and are working to model financial markets. Are the models perfect? No. Should the models have been more common before the Great Recession? Probably.

The only critique I understand of Thoma's,which he makes in a linked-to article, is that DSGE doesn't give clear policy advice the old Keynesian model does. Sure, but the model wasn't designed to manipulate the economy like an engineer. That's not the purpose of the DSGE model, at least not the purpose that I learned in class. It may have been the purpose of the Keynesian model. But as Thoma says in the linked-to article "when a model is applied to situations it was not designed to address, it is not the model that failed. The model has been misused."

I am left wondering what are the "simple, fast, and accurate answers to our questions" that the old Keynesian model provides. I would love to hear in the comments.

Thoma ends with

Thus, the approach to take depends critically on the question the researcher is asking. For some questions, the aggregate approach (sic) is best despite the criticism it has received in recent years from those using modern models, and we shouldn’t think of it as going backwards if we adopt this approach when it provides simple, fast, and accurate answers to our questions. The “correct” model to use is not an either/or decision, and macroeconomists should be open to both approaches as we try to improve our ability to understand the macro-economy, and provide policy advice when the economy experiences problems.

Here I am absolutely in agreement. As micro teaches us, there are trade-offs everywhere. Just as consumers don't buy only one type of good, economists should probably not use only one type of model. But Thoma has left me unconvinced that the old aggregate approach is the way to focus efforts on the margin.

I can't believe I wrote a post defending DSGE models. But if the critique is not compelling, it's not compelling. This is true regardless of what it is critiquing.