The Handbook of Experimental Economics. Edited by John H. Kagel and Alvin E. Roth. New Jersey: Princeton University Press, 1995. 721p. $55.00 cloth.

Kenneth C. Williams, Michigan State University

This volume, edited by John H. Kagel and Alvin E. Roth, contains an impressive set of essays that summarizes the present state of experimental economics. In light of the current controversies in political science concerning the empirical testability of formal models, this volume illustrates the merits of using laboratory experiments to test theories when there is a lack of naturally occurring data or when the available data is tenuous at best. Although the use of experimental methods among political scientists to test formal models has increased in recent years, it still has not attained the nearly "mainstream" status it has in economics. This volume is remarkable in demonstrating the progression of laboratory experiments from a fringe methodology to a reasonable way to not only test but explore economic theory. It provides accounts of how laboratory experiments have confirmed existing theoretical predictions, helped decipher puzzles that have baffled researchers for years, and have lead researchers to investigate different aspects of a theory that would have gone unnoticed without experimentation.

Each chapter focuses on a series of experiments within a given topic area. This is particularly useful since it allows the reader to trace the progression of the experimental process revealing how hypotheses are formulated and reformulated into new experimental designs. As a result, the authors not only provide compelling evidence for the usefulness of laboratory experiments, but also do not mask their inadequacies. The authors make it clear that laboratory experiments are not a substitute for naturally occurring data, rather when there is a sparsity of relevant data experiments provide a reasonably good mechanism to test parts of a theory under special conditions. Each author is cautious about not over generalizing results obtained from the laboratory. They realize that experiments test a particular hypothesis. Consequently, the environment that is created often has no real world counterpart, or if a counterpart exists the experimental environment is a gross simplification of it. Also experimental data may be biased since the rewards offered to subjects may not be sufficient to control their incentives. A more fundamental problem is that experiments are often conducted in different environments making it difficult to make comparisons across sets of experiments. Given these shortcomings however, it is clear from this volume that there are research questions that cannot be answered without the use of experimental data. We learn that carefully designed laboratory experiments can yield fruitful results.

While this volume concentrates primarily on issues relevant to economists, most of the topics should be of interest to political scientists. Of particular interest, including the introduction, are the chapters by John O. Ledyard on public goods, Jack Ochs on coordination problems, Alvin E. Roth on bargaining, and Colin Camerer on individual decision making. While the chapters by Charles A. Holt on industrial organizations, Shyam Sunder on experimental asset markets, and John H. Kagel on experimental auctions, are excellent, they may be beyond the interest of some political scientists. However, Sunder's chapter is a must read for researchers interested in how various environments help to aggregate information.

The introduction to this volume is splendidly done. It gives the reader an adequate history of experimental economics, tracing its roots from the 1930's with individual choice experiments, to the present. The introduction also provides an excellent overview of the major methodological issues that are discussed in subsequent chapters.

Experiments that seek to study behavior in voluntary contribution mechanisms are surveyed by Ledyard. Of interest are variables that lead subjects to contribute to a public good when it is in their self interest not to contribute. Experiments conducted by sociologists, social psychologists, political scientists, and economists are discussed. Data reveals that even though economic theory suggests that subjects should not contribute, they generally do in the laboratory. Even in environments conducive to noncooperative behavior the rates of contributions cannot be forced below 10 percent. Experimenters seek to determine which factors alter the relative rates of cooperation. There are strong findings that show that cooperative behavior improves when communication is permitted and when the marginal payoff for contributing increases. Repetition of play generally causes contributions to decline. Other factors may be relevant, but the evidence is inconclusive.

Ochs' chapter focuses on macroeconomic experiments that examine how subjects coordinate strategies when there are multiple equilibria to a game. Experiments are conducted to understand how different mechanisms (e.g., the existence of dominated strategies), and environments (e.g., overlapping generations and decentralized matching) allow subjects to coordinate actions. This series of experiments illustrates an important characteristic of economic experiments -- repetition. Experimenters in this series are not concerned about whether in the early rounds subjects identify and select an equilibrium. Rather, they examine behavior over time to determine if subjects learn to select strategies that converge to a predicted equilibrium. Convergence occurs in most cases in experiments that specify an unique equilibrium, when multiple equilibria are specified, only experienced subjects tend to coordinate on one of them. Results suggest that different theories of "strategic" learning may better explain behavior.

Inefficiencies in bargaining experiments are the focus of Roth's chapter. Data from sequential bargaining experiments, in which players make alternating offers, suggest inefficient outcomes may be a by-product of uncontrollable effects. For example, some deviations from the predicted outcome may result because bargainers often have nonmonetary incentives not specified by the theory. Additional studies have attempted to control for incentives such as concerns over relative shares and notions of fairness. Other experiments also reported by Roth suggested that experience and learning are important determinants of bargaining outcomes. Similarly, as with coordination, further learning theories need to be studied. Also, as revealed in the public goods experiments, bargaining in the laboratory is also affected by controls over communication.

Camerer surveys individual decision making experiments and in the process provides an excellent contrast between psychological and economic experiments. Experiments of this nature attempt to determine whether subjects in the laboratory emulate the assumptions of rationality. Methodological differences between the two disciplines seem to account for varying results. Economic experiments which rely on repeated play, monetary incentives and anonymous interactions, tend to provide evidence that subjects' actions are generally consistent with the assumptions of rationality. By contrast psychological experiments, which often do not provide monetary incentives, do not allow repeated choices, and sometimes deceive subjects, find evidence to the contrary. Camerer points out however that different results between the two disciplines are directly related to disparate research questions.

Overall, this volume gives the reader a greater appreciation for laboratory experiments as an integral part of theory confirmation and development. While the laboratory is not flawless at the present, with a proliferation of formal models, it is an invaluable research tool. This volume is excellent in displaying this methodology's strong points and its shortcomings. Each chapter leaves the reader with more questions than answers, providing material for a new volume in the near future.