Is there an important distinction to be made between the type of microfoundations used in DSGE modelling and the type of microfoundations used in the incomplete markets/contracting literature? or are they effectively the same?(I am aware that this is not actually a True or False question...)
Tuesday, April 12, 2011
True or False...
The following question was inspired by Larry Summers claim at the recent INET conference that microfounded, DSGE-style models played no role in informing policy making during the recent financial crisis.
Subscribe to:
Post Comments (Atom)
There is no distinction, in that one could construct DSGE models with asymmetric information and contracts - Bernanke Gertler Gilchrist (1999) is such a model as are the slew of recent attempts to incorporate financial frictions into macro models. But it's true that most DSGEs, in particular (as far as I know) all incarnations of the canonical New Keynesian model, assume symmetric information and complete markets.
ReplyDeleteThe pedant in me thinks `DSGE' is a bad name for this class of models. Many so-called DSGEs are not general equilibrium models in the strict Arrow-Debreu sense, since they have monopolistic price setting, search frictions, nominal rigidities, etc. All that these models have in common is that a) someone maximizes something, subject to constraints which depend on aggregate variables (prices, market tightness) and b) there are some conditions imposing consistency between agents' actions (market clearing, rational expectations). This is an unhelpfully large class of models; it's not obvious that all such models share the same virtues or flaws. So maybe we should stop talking about DSGEs, and talk instead about particular *assumptions* - optimizing behavior, RatEx, complete markets, a representative agent, symmetric information, market clearing (as vs. the alternatives: learning, incomplete markets, asymmetric information, nonclearing markets) - and particular *tools* for solving models - mainly perturbation methods/first-order approximation around a steady state, as vs. alternatives e.g. agent-based modelling.
Approaches with significantly different 'microfoundations' are Marxian/Kaleckian models, based on class division (eg Goodwin's class struggle business cycle model or Kalecki's income determination equation). Its not really clear if these should actually be called microfoundations, however - as there aren't really any corresponding microeconomic theories that justify them.
ReplyDeleteInasmuch as DSGE models are concerned, I'd say that if you construct a model that has a representative agent, you don't really need, or have, microfoundations - its patently clear that all aggregate outcomes are the basis of choice-theoretic behaviour! In addition, I believe there is a proof that single-agent equilibria are identical to multiple-agent equilibria, so there are 'microfoundations' in that sense for representative agents.