Economic theory is frequently criticized for assuming that economic agents are always rational. But economic theories based on more realistic assumptions about the way individuals decide have not been very productive. The results from these "realistic" theories are sensitive to specific assumptions about the decision rules used by agents. This project promises to bridge the gap between traditional rational choice economic models and models based on more realistic decision rules. In this project adaptive, heuristic decision rules emerge from individually rational behavior under imperfect information. Behavior predicted by the models of dynamic choice developed in this project conform to observed behavior, but the results are rigorously derived from general assumptions and yield useful insights. There are two parts to this project. In the first, models are studied in which individuals play a game repeatedly, learning and experimenting as they go. A wide class of behavioral rules are shown to yield convergence to Nash equilibria. Restrictions on individual behavior are related to various Nash equilibrium refinements. This is important because Nash equilibria and Nash refinements are mathematically tractable. This part of the project could provide economic theorists with powerful mathematical tools for studying dynamic models of bounded rationality. In the second part, models are developed in which individuals have a general preference for flexibility because they know there are future contingencies they cannot foresee. Dynamic game theory and contract theory are used to study individual choice where the desire for flexibility changes with circumstances. This project should be supported because of its contributions to economic theory and decision theory. Methodologically the project should provide economists with new approaches to modelling dynamic choice in a changing economy. Substantively, the applications of the investigator's approaches to economic problems, especially his work in the first part of the project on models of experimentation and learning, should provide important new insights into a wide range of economic issues.

Agency
National Science Foundation (NSF)
Institute
Division of Social and Economic Sciences (SES)
Application #
8908402
Program Officer
Daniel H. Newlon
Project Start
Project End
Budget Start
1989-07-15
Budget End
1992-12-31
Support Year
Fiscal Year
1989
Total Cost
$109,469
Indirect Cost
Name
Stanford University
Department
Type
DUNS #
City
Palo Alto
State
CA
Country
United States
Zip Code
94304