Kevin Quinn wrote: > I think that RATEX is important precisely because it is a > potent generator of multiple equilibria, given that the world > we're trying to model is evidently a world characterized by such > multiplicity.
If I may interject myself here, I'd like to make a methodological point. Or insist on it, since I've written similar things before on PEN-L. The world is hugely complex. There's no way that any given model of the economy (e.g. EM) will ever be capable of completely capturing its full complexity. There are hard limits. At some point, economies behave as self-referential dynamic systems. That limits the power of our analytical tools. Those systems exhibit messy dynamics. Our tools to make sense of those phenomena are (and may necessarily be) very limited. That said, in a given model, it's *not* assuming rational expectations which generates multiple equilibria. It cannot be. It's the other equations of the model, as they interact with rational expectations. And one can *always* assume conditions that rule out multiple equilibria. Whether or not to force a unique equilibrium by assumption is really up to you. It depends on what you want the model to do for you. You can always shock and tweak your model every which way. The difficulty is to get straight results out of that tweaking. In my understanding, what Azariadis and Guesnerie did in their 1985 sunspots paper was to anchor people's expectations not on economic fundamentals, but on collective beliefs unrelated to fundamentals -- e.g. beliefs that sunspots affect the economy. That's their clever way to generate bubbles and all that. Azariadis and Guesnerie set out specifically to model bubbles and phenomena of the type. That's *not* what Samuelson or Fama intended to do. Fama was concerned with whether or not price changes followed, basically, an i.i.d. kind of process. And just so that people know that they are not naively pushing the idea that their models are perfect images of the real world, off the bat, Fama admits in his 1965 and 1969 papers that *actual* asset price data don't have to fit the ideal i.i.d. process mold. Fama's is a very different goal. If the hypothesis was correct, then people would be able to extract profitable information out of past prices only by chance, not systematically. This is a more long-run type of concern than bubbles. That bubbles exist in asset markets is obvious. But who claims to be able to identify and time the bubbles well and extract systematic profits? Moreover, suppose that one can calibrate an A-G model, time bubbles, and make a buck. How long will it take for others to mimic the strategy and take away those rents? I'm commenting on Fama's EM papers, but something similar could be said of other rational expectations models, like Lucas'. They were on to something else. One can fairly blast a rational expectations Ramsey type model like that in Prescott and McGrattan's 2003 paper, not only because the testable predictions were so wrong, but mainly because it was the wrong model to predict stock prices. It was, if I may exaggerate a little, like using a model of the big-bang to time the rain in northern New Jersey. IMO, one cannot say that of some of Lucas' or even Fama's models. So Azariadis and Guesnerie are not "refuting" EM or Lucas' or some Minnesota type model. EM, Lucas, etc. are points of reference. A and G are into something related but different. Moreover, just to show how apples-to-oranges the comparison may be without getting down to the last detail, A and G use overlapping generations models (as opposed to the regular Ramsey type models used by, say, Lucas or Prescott). Since Allais, Samuelson, and Cass proposed them, overlapping generations models need not have a unique equilibrium, and their equilibrium or equilibria need not be Pareto optimal. Although the welfare analysis gets complicated, that's a reasonable cost to pay when you want to focus on certain macro issues -- money, inter-generational transfers of wealth like social security, etc. Personally, I don't find the models of messy dynamics to be particularly useful. Once upon a time, I spent some time looking at them. At some point, I realized that there's something of a blind-alley quality to the approach. It can only help you so much. In my experience, even very simple mathematical models are capable of generating a tremendous amount of complexity, sometimes as a result of tweaking seemingly innocuous parameters or assumptions. At that point, grasping why the world is messy becomes a truism. Modern capitalist societies, economies, human societies, life are extremely fragile little thingies. Yet, we're here. If even highly hyper-simplified mental replicas of the world have such volatile properties, then it's kind of a miracle that human societies exist at all. And that they have some measure of permanence. Some people may think that a good economic theory of capitalism is one showing that the system is inherently unstable. Among some Marxists, the emphasis Marx placed in Capital on central tendencies, centers of gravity (in a word, equilibrium) appears disproportionate only because Marx couldn't complete his entire project. At some point, he was going to get into the messy aspects of capitalism. I don't mean as side historical references, but as the subjects of direct analysis. I'm not so sure about that. In any case, I that showing the inherent instability of capitalism is not very hard or very illuminating. There are n ways in which the system can explode. The real mystery IMHO is why capitalism has the bit of historical robustness exhibited. IMO, if we're in the business of building a better society, alternative to capitalism, understanding what makes the system stable has the highest payoff. _______________________________________________ pen-l mailing list [email protected] https://lists.csuchico.edu/mailman/listinfo/pen-l
