PROJECT ABSTRACT Modern economic theory is built on the assumption that all economic agents are completely rational, meaning that they always know what is best for them and never make mistakes. Observed behavior however, is noisy and cannot be predicted precisely. Small amounts of randomness at the individual level may have a large effect on economic outcomes when people interact. This project develops a realistic model of error-prone decision making that is used to explain a number of discrepancies between observed human behavior and the predictions of standard economic analysis. This approach is used to evaluate alternative types of auctions and other mechanisms for coordinating the activities of buyers and sellers. Other applications include charitable contributions to public causes, strategic bargaining, and intra-industry competition. In each case, data from laboratory experiments with human subjects are used to refine our approach.
