Monday, March 5, 2012
Can You Say Sonnenschein?......I knew you could
What bothers me about the current discussion on the relative merits of micro-founded versus non-micro-founded macro is the crazy idea that RBC/DSGE models are micro-founded. These are representative agent models, people! They are micro-founded models of a macroeconomy consisting of one agent, and so we should by all means use them whenever we come across such an economy, and good luck with that. Unfortunately, the macroeconomy we'd like to understand has many agents. And any attempt to apply general equilibrium analysis to such an economy is faced with an insurmountable problem. Hugo Sonnenshein showed conclusively that aggregate excess demand functions, aggregated from "well-behaved" individual excess demand functions, can behave any way you like - arbitrarily. Just sayin'.
Mr. Market as Gosplan superhero!
ReplyDeleteI can remember being in class with George Von Furstenburg at IU and making the same point. I got a withering glare and got told to shut up. As he put it "I don't care whether the assumptions of the model are correct. I want mathematical elegance"
ReplyDeleteI remember my first year in my Ph.D. program when we first learned of aggregation problems (things like the Gorman Form) and we confronted our macro professor about his use of a function we'd just learned doesn't aggregate properly. After a few awkward moments of stumbling through bad logic concerning why it didn't matter, he eventually just shrugged his shoulders and said, "Well, it's better than not having a model at all."
ReplyDeleteOf course you are right. But I jsust the other day (well, a couple months ago) heard Sam Bowles, of all people (hi Peter!) acknowledge this problem and then immediately brush it away, saying (my paraphrase), Oh, people have made too much of this critique, we know now that the real economy behaves more or less like a representative agent model so it's not worth worrying about.
ReplyDeleteThis is your problem. Criticize the mainstream all you want, until you can offer a positive alternative -- until you can say , look at this important problem that we can solve and you cannot -- nobody is going to listen.
"we know now that the real economy behaves more or less like a representative agent model"
ReplyDeleteHe means "less." When you consider that the Representative Agent is skewed to (let's be nice) the 80-85th percentile, it's difficult to argue the model actually covers the "real economy"--unless you already assume that at least 50% of people are, by definition, not part of the real economy.
JW: Say it isn't so: Bowles of all people! But Sam is calling for and doing a post-Walrasian economics: he thinks, and I agree, that Walrasian GE belongs on the scrap-heap - an unnecessary detour as he and Herb put it.
ReplyDelete