Thread by @JWMason1

@themountaingoa1 @robertwaldmann This isn't really a conversation for twitter, but:

"Microfoundations" can mean a lot of things. I'd distinguish 4 senses:

1. aggregate model that is motivated/justified by explicit views of behavior of units at less aggregated level

@themountaingoa1 @robertwaldmann 2. aggregate model that rests on/incorporates formal model of units at less aggregated level

3. aggregate model derived from formal model of behavior of "agents", atomic units of economy who are consumers of final output, suppliers of labor, owners of wealthy etc.

@themountaingoa1 @robertwaldmann 4. aggregate model derived from formal model of behavior of agents, which must be described as the solution to an explicit intertemporal maximization problem.

@themountaingoa1 @robertwaldmann From my point of view, heterodox macro models are already better micro founded than mainstream models in sense 1, since they describe an economy that includes firms, usually banks, etc.

@themountaingoa1 @robertwaldmann On the other side, there is a lot of criticism of microfoundations in sense 4, and willingness to allow various departures from it, but that still are microfoundations in sense 3, i.e. we still have atomic agents but some of them may fail to optimize for in various ways.

@themountaingoa1 @robertwaldmann I would argue however that it is sense 3 that is the real problem -- it is simply not possible to describe an actual economy in terms of an explicit model of the behavior of atomic agents, however many types you include and however flexible you are about the constraints on them.

@themountaingoa1 @robertwaldmann That leaves sense 2. Once we abandon idea of "agents" as fundamental constituents of economy, there is always another layer of disaggregation possible. So it's a question of how far you want to go. Sometimes explicit models of firms are helpful, but often they're not needed, imo.