One of the hardest parts of describing what I do at cocktail parties and other university functions is the inevitable comment disguised as a question, 'don't economists assume that everyone is selfish?' The subtext to this question is, of course, 'what's a nice girl like you doing in a field like that?' While it is true that most economic models involve actors whose utility functions are 'selfish' - that is, contain only their own consumption - this has long been recognized as a simplifying assumption (perhaps more appropriate in the world of industrial organization where firms are the actors than in the world of consumption theory, where individuals make decisions). Long ago casual empiricism and introspection led economists to realize that this assumption was not descriptively accurate. The question was, if people didn't maximize their own self-interest, then what did they maximize?

With the development of controlled economics experiments, economists have discovered a way to answer that question. This column briefly mentions some of these experiments, and points to the new, improved and (more) descriptively accurate models which have resulted. These models all incorporate other-regarding preferences into an individual's utility function. Sometimes those preferences are such that one person cares positively about another's well-being, as in the case of altruism. In other settings these preferences can be negative, as in the case of envy. In either case, models based on a broader view of preferences are more descriptively accurate than the traditional self-interested model, and allow economists do a better job of describing and predicting economic behavior in the world.

There has been an explosion of experimental and theoretical research in this area in the last five years or so. Because of space limitations, I provide just a few annotated references here. (For a more extensive list, please contact me directly. For studies that involve other psychological elements of decision-making, see Camerer's article in this issue.)

I. Altruism

Probably the first challenge to the traditional self-interested model came from research on charitable giving and altruism. Empirical evidence that individuals contribute to charities and make bequests to their children and to educational institutions suggests that others' consumption or utility matters to them. Experiments investigating this phenomenon focus on the dictator game, where an individual is given a sum of money and told to divide it between herself and another party. The money is thus divided and the game ends. If subjects were only maximizing their own payoffs, the expected outcome is for the dictator to keep all the money. In contrast, Forsythe, et al. (1994) found that subjects often offered substantial amounts of the money.

Since then, many replications and variations of this experiment have been conducted. (Camerer, 1998, provides a survey of all three types of experiments discussed here.) Generally, subjects give 20% of their endowment to the other player in the game. The result that subjects voluntarily give money away suggests that they are not solely concerned with their own payoffs, but are also concerned with the payoffs of others. Interestingly, gender differences have been found in these games with women giving more than men, although this varies with the cost and benefit of giving. (Eckel and Grossman review experiments that test for differences in the behavior of women and men.)

Andreoni (1990) extended earlier work by Gary Becker by developing a model of 'impure altruism,' where an individual's utility function includes as arguments their own consumption, the consumption of others, and a 'warm glow' factor, the extent to which the increase in others' consumption was caused by them. These models are currently in use to explain and predict observed beneficent and bequest behavior.

II. Equal Division (Fairness)

A number of psychology experiments demonstrate that subjects have a preference for equally dividing unearned gains. Economists first examined this issue using the ultimatum game. Pairs of subjects are given an amount of money to divide according to specific rules. One subject, the proposer, suggests a division of the money. The responder can then agree, in which case the money is split as suggested, or disagree, in which case neither player receives any money. If subjects are maximizing their own pay offs, the unique subgame perfect equilibrium of this game is for the proposer to offer the responder the smallest positive increment x, and for the responder to accept. In contrast, Guth et al (1982) found that subjects frequently offered large shares of money, and responders rejected small offers.

Since then, many replications and variations of this experiment have been conducted, including comparing the effect of culture and gender on behavior (e.g. Buchan, Johnson and Croson, 1998). Generally, responders in this game reject offers of less than 20% of the money about half the time. This rejection behavior is evidence that responders are not solely concerned with their own payoffs, but instead have a preference for fair division of the pie.

In response to this and other evidence, Bolton (1991) developed a model where the utility of the responders contains arguments of both absolute and relative payoffs. Bolton's results are consistent with responders rejecting low offers in the ultimatum game; responders could prefer both parties earning no money to earning some money but having the proposer earn more. Many others have developed models in which relative payoffs play a role. In a recent paper, Rabin (1998) presents a general framework of 'self-interested and fairness motivated' preferences.

Preferences for relative payoffs have important implications for economic behavior. Posted-price purchasing can be modeled as an ultimatum game, where the seller makes an offer of splitting the surplus available from a transaction to the buyer, who may purchase or not. In addition, preferences for relative payoffs have an important impact on bargaining behavior, including what offers are acceptable to each side.

III. Reciprocity

Other experimental evidence suggests that individuals reciprocate the treatment they receive at the hands of others. Berg et al. (1995) examine behavior in a 2-player trust game. The sender is given an endowment, any portion of which he can send to the receiver. Any money sent is tripled. The receiver then acts as a dictator, and can return any portion of the money to the sender. Pure self-interest implies no money will be returned, and thus none sent (the unique subgame-perfect equilibrium). Instead money is both sent and returned, and the proportion returned is positively related to the amount sent. This experiment has been replicated and extended, including an examination of gender differences, with women returning a significantly higher percentage of their earnings than men (Croson and Buchan, 1999).

Other experimenters too numerous to mention have tackled the subject of reciprocity. An interesting example is the work of Ernst Fehr and his co-authors (e.g. Fehr, Gachter and Kirchsteiger, 1997), where subjects in a simulated labor market reciprocate efficiency wages with super-optimal effort.

New models incorporating reciprocity have been developed in response to this evidence. In Rabin (1993), players take the intentions of other players into account when making their decisions. If others treat them well, they reciprocate positively; if others treat them badly, they reciprocate that treatment as well. The general framework proposed by Rabin (1998) can be used to capture reciprocity as well as simple inequality aversion. These types of models of reciprocity have been used to explain economic phenomena as diverse as gift exchange and contracts in labor markets.

IV. Conclusion

The interaction of theory and data form the basis of most empirical sciences, including physics, chemistry, and even psychology. Economics has suffered somewhat because of the lack of data available for theory testing, particularly in areas like game theory and individual decision-making where the context of actual decisions bear only weak resemblance to those assumed in the theory. The rise of experiments in economics has ameliorated this problem by providing a controlled setting in which to test theoretical predictions. The underlying assumptions of the theory can be implemented in the laboratory, providing a clean test of both baseline and comparative-statics predictions.

Results from a number of experiments (very sketchily reviewed here) demonstrates a richer picture of behavior than theorists had hypothesized. This behavior involves preferences for altruism, fairness and reciprocity, as well as self-interest. The natural response -- to build theories consistent with our observations -- has begun in earnest. These theories accommodate observed behavior by incorporating other-regarding preferences into individuals' decision-making process. The next step in theoretical development is a generalization and unification of these theories which will allow us to predict and explain behavior across different games and settings. New experiments will be designed to test the implications of new theories, and the dialectic will continue.

Experiments like these and their resulting theoretical work provide an opportunity to change the field of economics by addressing the assumptions of individual preferences at its very foundation. And they also give me something to say at cocktail parties!

References
Andreoni, James, 1990, Impure Altruism and Donations to Public Goods; A Theory of Warm Glow Giving, Economic Journal 100, 464-477.

Bolton, Gary, 1991, A Comparative Model of Bargaining: Theory and Evidence, American Economic Review 81, 1096-1136.

Buchan, Nancy; Johnson, Eric; Croson, Rachel, 1998, Culture, Fairness and Power: Contrasting Influences on Repeated Negotiation Behavior in Japan and the United States, Working Paper, The Wharton School, University of Pennsylvania.

Camerer, Colin, 1998, Social Preferences in Dictator, Ultimatum and Trust Games, Working Paper, California Institute of Technology.

Croson, Rachel; Buchan, Nancy, 1999, Gender and Culture: International Experimental Evidence from Trust Games. American Economic Review 89(2), Papers and Proceedings, May.

Eckel, Catherine; Grossman, Philip, 1999, 'Differences in the Economic Decisions of Men and Women: Experimental Evidence.' Forthcoming, Handbook of Experimental Results, edited by C. Plott and V. Smith. 1999.

Forsythe, Robert; Horowitz, Joel; Savin, N.; Sefton, Martin, 1994, Fairness in Simple Bargaining Experiments, Games and Economic Behavior 6, 347-369.

Fehr, Ernst; Gachter, Simon; Kirchsteiger, Georg, 1997, Reciprocity as a Contract Enforcement Device: Experimental Evidence, Econometrica 65, 833-860.

Guth, Werner; Schmittberger, Rolf; Schwarze, Bernd, 1982, An Experimental Analysis of Ultimatum Bargaining, Journal of Economic Behavior and Organization 3, 367-388.

Rabin, Matthew, 1993, Incorporating Fairness into Game Theory and Economics, American Economic Review 83, 1281-1302.

Rabin Matthew, 1998, Bargaining Structure, Fairness and Efficiency, Working Paper, University of California, Berkeley.