Questioning Alternative Microfoundations

There’s been a bit of discussion on a Rajiv Sethi post about Duncan Foley’s attempts to bring strong microfoundations into macro. Mark Thoma first had comment and then Leigh Caldwell chimed in. If you have time, start with Rajiv’s post. Thoma sums up a key problem well:

That is, you need to make the representative agent assumption in order to aggregate individuals up to the macroeconomic level and still be able to guarantee uniqueness, stability, or many other properties we need to have a reasonable model.

Caldwell goes a little bit deeper into smoothing out the problem. He argues that we should abstract models of the economy into systems of agents and goods, mediated by beliefs and values. I think the problem is that Caldwell over-abstracts, though. He seems most excited that, “the model has the potential to be simple enough to be tractable.” Even though he acknowledges that, “the actual mathematics of this theory will look like is not yet obvious,” I don’t think Caldwell moves the ball forward much. What he wants boils down to:

But a simple model of choice arising from values, mediated by beliefs, under constraints on attention, accuracy and myopia provides a parsimonious and expressive description of reality. By implementing a realistic theory of decision-making into the model, we will have a closer match to the real world than current theories.

However, he eschews agent-based models which require too many assumptions. However, I think we have two choices if we want to formally model the economy. We either observe behavior on measurables and fit the model, or we make abstract assumptions about how agents behaves on squishier concepts like beliefs and values and then apply them throughout. I don’t think, though, we can get by based on observing these hard-to-measure concepts. Even if we can measure beliefs and values in some dimensions, I’m not sure how we implement them in these models without making further assumptions. And finally, if we want micro-foundations, they have to be agent-based, with a full slew of interactions. For now I side with Thoma’s third option:

give up the idea of providing microeconomic foundations for macroeconomic models and begin modeling the aggregate level directly (e.g. see Kirman’s discussion of network models)

And of course, in modeling this aggregated activity, we should be mindful that there are a variety of questions we can ask besides “what will GDP growth look like?”

About these ads