Daniel McFadden Department of Economics University of California, Berkeley August 1996 (revised July 1997, September 1998)
ABSTRACT: Rationality is a complex behavioral theory that can be parsed into statements about preferences, perceptions, and process. This paper looks at the evidence on rationality that is provided by behavioral experiments, and argues thatmost cognitive anomalies operate through errors in perception that arise from the way information is stored, retrieved, and processed, or through errors in process that lead to formulation of choice problems as cognitive tasks that are inconsistent at least with rationality narrowly defined. The paper discusses how these cognitive anomalies influence economic behavior and measurement, and theirimplications for economic analysis.
Forthcoming, Journal of Risk and Uncertainty, Special Issue on Preference Elicitation
RATIONALITY FOR ECONOMISTS? Daniel McFadden1
1. INTRODUCTION Economics has always been concerned with the motivations and behavior of consumers. Rational behavior, in the broad meaning of sensible, planned, and consistent, is believed to govern most conduct in economicmarkets, because of self- interest and because of the tendency of markets to punish foolish behavior. However, rationality has been given a much more specific meaning in the classical theory of consumer demand perfected by Hicks and Samuelson that forms the cornerstone of courses in economic theory. In Herb Simon's words, "The rational man of economics is a maximizer, who will settle for nothingless than the best." While this model of consumer behavior dominates contemporary economic analysis, there is a long history among economists of questioning its behavioral validity and seeking alternatives. What has come to be known as Behavioral Decision Theory had its origins in the von Neumann & Morgenstern (1947) treatise on choice under uncertainty and game theory.2 This work had two majorimpacts beyond its direct effect of providing a prescriptive framework for analyzing risky behavior: It made formal, axiomatic analysis fashionable in economics and psychology, and it invited laboratory experimentation to test the descriptive validity of the axioms. Most of this work
1. This paper is dedicated to the memory of Amos Tversky, whose brilliant life profoundly influenced psychology andeconomics. In the subject known as Behavioral Decision Theory, Tversky's hand appears everywhere, through his papers, and through his ingenious and definitive experiments that have made clear the importance of heuristics and judgment in human cognition. He will be counted among the great minds of the 20th Century. It was a delight and an education to have been his friend. Early versions of thispaper were presented at the European Meetings of the Econometric Society, Istanbul, 1996, and at the NSF Symposium on Eliciting Preferences, University of California, Berkeley, July 1997. I have benefitted from discussions and comments from Moshe Ben-Akiva, Baruch Fischhoff, Tommy Garling, Danny Kahneman, Mark Machina, Charles Manski, John Payne, and Drazen Prelec. Research support from the E.Morris Cox Fund is gratefully acknowledged.
2. There is an early history of economic thought on risk-taking behavior, in the work of Bernoulli (1736), Fisher (1930), Keynes (1921), Menger (1934), Knight (1921), and Ramsey (1931), as well as important developments by Friedman & Savage (1948), Marschak (1950), and Arrow (1951) that parallel the von Neumann-Morganstern contribution.
1concentrated on choice among lotteries, but the ideas spread to other decision-making situations. In the following two decades, behavioral science and cognitive psychology came of age, with the participation of notable economists such as Allais (1953), Chipman (1960), Marschak (1950), Papandreau (1960), and Simon (1959). Figure 1. The Decision Process
Information Attitudes Affect