An Overview of Experimental Economics

Economic experiments are of two types. Laboratory experiments are generally conducted with undergraduate students who interact with one another to play “games” by computer in a laboratory. Field experiments are conducted with more experienced people in real environments. The key element of experimentation is that the experimenter has substantial control over the economic environment faced by the participants, and this makes it possible to determine the consequences of altering the environment, for example, through economic policy.

Both laboratory and field experiments have a long history in economics. Laboratory experimentation originated in the pioneering work of Vernon Smith and Charles Plott dating back to the early 1960s. Field experiments also have a long tradition in economics. Experiments on the negative income tax, for example, were conducted beginning in the late 1960s. In psychology, experimentation has a much longer history, dating back to the late 19th century. Laboratory experimentation in economics and allied disciplines has grown explosively in recent decades, and field experimentation has taken off in the last five years as well. Very recently a third type of experimentation has become popular in which physiological measurements are taken, often using fMRI (functional Magnetic Resonance Imaging) machines to directly measure brain activity. In addition to economics, “economic style” experiments have become widespread in business disciplines such as marketing, in political science, law, and anthropology.

Laboratory experimentation in economics has much in common with experimentation by social psychologists. There are, however, some notorious methodological disputes between the two groups, notably over deception (frowned upon in economics) and monetary incentives (viewed as essential by economists). One other controversy is perhaps deserving of comment. Despite its innocuous nature, experimentation in the United States is highly regulated by the Federal Government. The regulation is enforced by an offshoot of the National Institute of Health, which is charged with supervising a set of regulations designed to protect subjects in high-risk medical experiments. The enforcement of these often meaningless rules in the context of harmless and legal social science and social psychology experiments is a matter of significant friction within academia.

Experimental studies, especially laboratory experiments, are not without controversy within economics. Many economists argue that the stakes are too small to make much difference to the participants, and that the circumstances in the laboratory are not terribly familiar and do not reflect the types of decisions of greatest interest to economists which involve day to day ordinary behavior. Proponents of experiments point out that there is extensive data on the size of stakes, and that while it is true that participants play more carefully when more money is on the table, many central findings are quite robust, and are observed in the field as well as the laboratory. However, the validation of experimental results outside the laboratory is still in its infancy. Finally, proponents of experimental research point out that the type of broad general theory that is the ideal of economists ought to be able to predict behavior, even when incentives are weak.

Experimental studies in economics serve two major purposes. First, they are a useful complement to theory, serving both to validate and invalidate theories, as well as to suggest new theories. They are used to answer questions of interest to economists and social scientists such as: How do individual decisions result in particular outcomes for the entire community or group? How do fluctuations in external factors influence these outcomes?  Second, experiments serve an important practical purpose by testing policy proposals and other “mechanisms.” For example, laboratory experiments were widely used to test auction procedures designed for large government auctions such as those used to allocate electromagnetic spectrum. Similarly, the negative income tax experiments were designed to analyze the efficacy of a negative income tax as an alternative to welfare payments. Experimental knowledge has found broad application in such diverse topics as the analysis of capital markets, political and voting trends, crime and punishment, and family dynamics.

Some important centers of experimental economic research are studies of prisoner’s dilemma type situations where private incentives conflict with the common good, bargaining, studies of auctions and markets, and studies of individual decision making, especially under uncertainty.

A brief reading of the experimental literature might lead the casual observer to conclude that economic theories are not of much use in the laboratory. This, however, reflects the fact that anomalies are generally of greater interest than confirmations of accepted theory. In the environments of greatest interest to economists – competitive environments where the individual participants are not essential to the whole – the theory explains laboratory behavior exceptionally well. The central anomalies in interactive behavior that have been discovered are that people are mildly altruistic and spiteful – some people being willing to contribute more to the common good than theory suggests, and some willing to accept losses to punish an opponent who is perceived as uncooperative. The central anomalies in decision making are the exceptionally high degree of risk aversion in the laboratory and the failure of the widely used expected utility theory under a variety of not terribly common circumstances. Much current research focuses not on the existence of these anomalies, which is well established, but on their quantitative importance, and on resolving conflicting theoretical explanations.
Economic “style” experiments have spread beyond the profession as a method of measuring human behavior. One remarkable example is the cross-cultural ultimatum bargaining experiment conducted by Henrich et al [2001]. In the ultimatum bargaining game a fixed “pie” must be divided between two players. The first mover proposes a division rule, which the second mover may accept or reject. If the second mover accepts, the pie is divided as agreed upon; if the second mover rejects, nobody gets any pie (hence the “ultimatum”). A naïve application of economic theory that assumes players to be completely selfish suggests that the first mover should get almost all the pie. In fact second movers are quite willing to reject ungenerous offers, and first movers quite sensibly offer a somewhat “fair” division of the pie. What is most striking is that this outcome is quite robust across many cultures both modern and primitive. There are two interesting exceptions. One Amazonian tribe plays in an almost completely selfish way, with the first mover getting almost all the pie. In the opposite direction, in one competitive gift-giving culture offers are very generous – and offers that are too generous are rejected. This shows on the one hand that while culture matters, at least with respect to ultimatum bargaining, cultures are more similar than not.

References:
Henrich, Joseph, Robert Boyd, Samuel Bowles, Colin Camerer, Ernst Fehr, Herbert Gintis, Richard McElreath [2001]: “In Search of Homo Economicus: Behavioral Experiments in 15 Small-Scale Societies,”  The American Economic Review, Vol. 91, No. 2, Papers and Proceedings of the Hundred Thirteenth Annual Meeting of the American Economic Association (May), pp. 73-78.
Kagel, John H. and Alvin E. Roth (editors), The Handbook of Experimental Economics, Princeton University Press, 1997.

______________________________
David K. Levine is Professor of Economics at Washington University in St. Louis, USA. He is co-editor of Econometrica, co-editor of NAJ Economicss and President of the Society for Economic Dyamics.