There was a time when the conventional wisdom was that, because economics is a science concerned with complex, naturally occurring systems, laboratory experiments had little to offer economists. But experimental economics has now become a well-established tool of economic research. The initial impetus for this transformation came from studies of individual choice behaviour. As economists focused on microeconomic theories which depend on individuals� preferences, the fact that these are difficult to observe in natural environments made it increasingly attractive to look to the laboratory to see if the assumptions made about individuals were in fact descriptive of their behaviour. The publication in 1944 of von Neumann and Morgenstern�s Theory of Games and Economic Behavior accelerated the interest in experimentation. The expected utility theory they presented gave a new focus to experiments concerned with individual choice, while the predictions of game theory - and how these depend on the rules of the game - sparked a wave of experimental tests of interactive behaviour. This has blossomed into large literatures on topics as diverse as bargaining behaviour, the provision of public goods, co-ordination and equilibration, auction theory, learning, and the effects of different rules of market organization.

Formal tests of economic theories of individual choice go back at least as far as L.L. Thurstone (1931), who used experimental techniques common in psychology to investigate whether the indifference curve representation of preferences could coherently describe individuals� choices (he concluded that it could). Expected utility theory made more pointed predictions, which allowed more powerful tests, and while some early experiments (e.g. Mosteller and Nogee 1951) supported the conclusion that utility theory could serve as an adequate approximation of behaviour, others, such as Allais (1953) identified systematic violations of expected utility theory. There have since been hundreds of experiments designed to explore further systematic violations of utility theory, and of the alternative choice theories that have been proposed to account for various parts of the experimental data.

In the 1970�s, the psychologists Daniel Kahneman and Amos Tversky systematically explored how decision-making heuristics introduce a number of biases in human behavior, adding considerable richness to our understanding of how the assumption of idealized rationality may or may not be a useful approximation. One influential part of their work, Prospect Theory, summarized their results in a form that could be viewed as an alternative to expected utility theory (Kahneman and Tversky 1979). Kahneman shared the 2002 Nobel prize in Economics (Amos Tversky passed away in 1996). More recently, studies of individual choice behavior have focused not only on choices involving risky alternatives, but also choices made over time, with studies focusing on robust but �irrational� behaviors that lead to problems of self control, or procrastination, such as arise in consumption and savings behavior (see e.g. Laibson (1997), Lowenstein and Elster (1992), Thaler(1987) for three strands of this work). Much of this work is surveyed in Camerer (1995).

An early game theoretic experiment, conducted by Melvin Dresher and Merrill Flood in 1950 (Flood, 1958), first introduced the much studied game that subsequently came to be called the Prisoner�s Dilemma. They observed that, even in a game with a unique equilibrium, the observed behaviour may deviate from the game theoretic prediction. But these experiments also confirm the game theoretic prediction that the incentives for individuals to act in their own interest may in some circumstances make it difficult to achieve the gains available from co-operative action. One body of experimental research which pursues this investigation concerns the provision of public goods; this literature is surveyed by Ledyard (1995).

A different kind of problem emerges in games having multiple equilibria, in which the players must co-ordinate their expectations and actions. Early experiments on co-ordination were reported by Thomas Schelling (1960). Bargaining presents an important class of problems in which co-ordination of expectations is of the essence, and it is a subject in which there has been considerable interplay between theory and experiment. Self-interest is not the only driving force shaping behavior in such games. Experimental results suggest that considerations of fairness often play an important role: Subjects are willing to forgo some monetary gains in order to avoid being treated unfairly. For various experiments and models of fairness, see Fehr and Schmidt (1999) and Bolton and Ockenfels (2000). Surveys of experiments concerning co-ordination and bargaining may be found in Ochs (1995) and Roth (1995a).

Experiments are particularly useful for isolating the effects of the rules of the game by which markets are organized. (For example, cattle and flowers are auctioned by different rules, but are also very different commodities, so one cannot isolate the effect of the different rules by relying only on field studies of cattle and flower auctions.) Chamberlin (1948) introduced a design, now widely used by experimenters, to create markets for artificial commodities in which the reservation prices of buyers and sellers can be controlled by the experimenter. Chamberlin�s design permits experiments in which different rules of market organization (e.g. different forms of auction) can be compared while holding all else constant.

Vernon Smith and Charles Plott expanded on Chamberlin�s 1948 design to investigate different rules of market organization. Smith (1962) famously showed that in a double oral auction (in which buyers and sellers can both propose prices to the market) there is a strong tendency for prices to converge to a competitive equilibrium. Smith received the other half of the 2002 Nobel Prize in economics. Kagel (1995) and Kagel and Levin (2002) survey the large modern literature on auctions, Sunder (1995) considers markets for commodities (such as financial securities) in which information plays a dominant role, and Holt (1995) surveys experiments in industrial organization generally.

Much modern work has gone into understanding when equilibrium predictions will be descriptive, and when they will not. Experimental evidence has helped spark an interest in theories of learning and adaptation (see e.g. Roth and Erev 1995), in contrast to theories of static equilibrium. There has also been growing use of experiments as an engineering tool, to help test new market designs (see e.g. Roth, 2002).

A general overview of experimental economics, including details of its early history, can be found in Roth [1995b].
[See also