AN INQUIRY INTO THE EFFECTS OF LEARNING AND DIFFERENTIAL INFORMATION ON RESOURCE ALLOCATION (RATIONAL EXPECTATIONS, GENERAL EQUILIBRIUM)

MICHAEL R FRIERMAN, Purdue University

Abstract

It has long been recognized that agents' expectations, in many instances, have a major impact on their behavior. The solution concepts developed, rational expectations and Bayes-Nash equilibrium, have led to a deeper understanding of the market's operation. However, in most cases this has been done at the expense of very strong informational requirements on the part of the agents involved. Agents are assumed to know either the relevant model or the class of distributions up to some parameter. In the latter case it has been shown that Bayesian players can learn the "truth" when they know the correct likelihood function generating their observations. Recognizing that the feedback between agents and the market environment implies that away from a rational expectations equilibrium market realizations are not "drawn" from a stationary distribution, this study pays particular attention to the short run nature of information acquisition and learning on the part of agents when the structure of their environment is unknown. In a general equilibrium setting where agents predict utility relevant information from market prices by means of a collection of models, Blume and Easley have shown that under a Bayesian type learning process, non-rational expectations equilibrium may be locally stable. Using a simple example of their environment, Chapter II shows that under classical conditions on aggregate excess demand, not only will the rational expectations equilibrium be unique, but it will also be globally stable. This result is illustrated by means of a technique developed here, which allows convenient diagramatic treatment of the issues surrounding rational expectations. In a simple labor market where workers must learn the monetary authority's true policy, Chapter III studies the convergence of real economic variables to their long run natural rates. Finally, Chapter IV investigates the incentives among duopolists to share private information. It is found that under certain conditions firms prefer to share information while under others they would keep their information private. Further, it is shown that conditions may exist which induce one firm to choose to remain uninformed while its opponent chooses to become informed.

Degree

Ph.D.

Subject Area

Economic theory

COinS