Über die Autorin bzw. den Autor

Nicholas Bardsley is senior research fellow at the National Centre for Research Methods, University of Southampton. Robin Cubitt is professor of economics and decision research at the University of Nottingham. Graham Loomes is professor of economic behavior and decision theory at the University of East Anglia. Peter Moffatt is reader in econometrics at the University of East Anglia. Chris Starmer is professor of experimental economics at the University of Nottingham. Robert Sugden is professor of economics at the University of East Anglia.

Von der hinteren Coverseite

"The authors of this book have pulled off a remarkable feat: a page-turner on experimental economics that will appeal to a wide audience. Noneconomists and nonexperimentalists will be treated to a fascinating introduction to the methods, philosophy, and controversy of this rapidly expanding area of research. Experimental economists will get a whole new perspective on their field that will make them think differently about their own work. Not to be missed!"--George Loewenstein, Carnegie Mellon University

"Experimental Economics fills an important gap in the literature. It provides a thoughtful and rigorous treatment of key methodological and conceptual issues that are frequently discussed informally by experimental economists but are rarely addressed explicitly. I would recommend it for students and current practitioners of experiments, as well as experienced economists who would like to learn more about how experiments are useful in advancing economic science."--Timothy Cason, Purdue University

"Economists increasingly run experiments. How do they justify and motivate their work? Are their methods sound? Read this book and you will be wiser."--Martin Dufwenberg, University of Arizona

"This interesting book deals with important methodological issues involved in running and evaluating experiments in economics. The authors employ numerous well-known experiments to discuss a variety of practical and philosophy of science issues as they relate to experimental economics--and the broader social science literature as well. This book will be of value to practicing social science experimenters, skeptics of experimental methods in the social sciences, and general interest readers."--John H. Kagel, coeditor of The Handbook of Experimental Economics

"With an extremely clear presentation as well as sharp and rigorous arguments, this book will become a standard introduction to the methodological foundations of experimental economics. The authors have the rare gift of effectively combining philosophical analysis with economic science, while satisfying the highest standards of scholarship in both fields."--Francesco Guala, University of Exeter

"Assessing experimental economics is an important and innovative endeavor--the authors succeed in providing a systematic and broad view of the discipline. Extremely clear and accessible, this book is full of deep methodological reflections. I gained many new perspectives and insights, and profited greatly from reading this book."--Bettina Rockenbach, University of Erfurt

"This book is a great enrichment to the experimental economic field and bridges the gap between the methodology of philosophy of science and experimental economics. Raising many excellent issues in depth, the book examines the contributions of experimental economics to economic theory, and discusses controversies between theorists and experimentalists."--Rosemarie Nagel, Pompeu Fabra University, Barcelona

"This book is important in view of the controversial nature of experiments in economics. How do experimentalists attempt to advance our understanding of economics? The authors give the reader a comprehensive view of the various approaches to experimental economics, including their strengths and weaknesses."--Peter Bossaerts, California Institute of Technology

"For this book, expertise is needed in experimental economics, but also in philosophy and psychology. These authors have it all--there is no group better suited to present this material. They explore the objections of some economists against the experimental method and the controversies that exist within the discipline."--Joep Sonnemans, University of Amsterdam

Auszug. © Genehmigter Nachdruck. Alle Rechte vorbehalten.

Experimental Economics

Princeton University Press

Contents

Preface...................................................................................................................vii1 Introduction............................................................................................................11.1 Experiments in Economics.............................................................................................11.2 Does Economics Need Experiments?.....................................................................................41.3 The Practice of Experimental Economics...............................................................................111.4 The Illustrations and the Structure of the Book......................................................................231.5 Methods, Methodology, and Philosophy of Science......................................................................362 Theory Testing and the Domain of Economic Theory........................................................................462.1 Domain Restrictions: Economic Theory and the Laboratory..............................................................462.2 Generality and External Validity.....................................................................................492.3 The Blame-the-Theory Argument........................................................................................542.4 The Concept of Domain................................................................................................562.5 The Laboratory and the Three Senses of Domain........................................................................642.6 Application to Experimental Tests of Choice Theory...................................................................712.7 Application to Experimental Tests of Equilibrium Predictions.........................................................852.8 Conclusions..........................................................................................................923 Experimental Testing in Practice........................................................................................953.1 Preliminaries........................................................................................................953.2 Experimental Testing and the Duhem-Quine Thesis......................................................................953.3 On the Significance of the DQT for Testing...........................................................................1063.4 On Testing Game Theory...............................................................................................1143.5 Hard Cores, Progress, and Experiments................................................................................1283.6 Conclusion...........................................................................................................1384 Experiments and Inductive Generalization................................................................................1414.1 Preliminaries........................................................................................................1414.2 Deduction versus Induction...........................................................................................1424.3 How Inductive Investigation Works....................................................................................1454.4 Experiments as Tests.................................................................................................1504.5 Exhibits.............................................................................................................1564.6 Why Be Interested in Exhibits?.......................................................................................1604.7 Do Exhibits Need to Be Explained?....................................................................................1664.8 Multiple Causation...................................................................................................1694.9 Explaining Exhibits Inductively......................................................................................1724.10 Investigating Exhibits without Trying to Explain Them................................................................1844.11 Experiments as Models................................................................................................1894.12 Conclusion...........................................................................................................1945 External Validity.......................................................................................................1965.1 Introduction.........................................................................................................1965.2 Are Economics Experiments Models?....................................................................................1975.3 Tests of Applied Economics Theories..................................................................................2045.4 Types of Artificiality Criticism.....................................................................................2145.5 Alteration Contrasted with Omission and Contamination................................................................2285.6 Evaluating Alteration Criticisms.....................................................................................2335.7 Field Experiments....................................................................................................2375.8 Conclusions..........................................................................................................2426 Incentives in Experiments...............................................................................................2446.1 Preliminaries........................................................................................................2446.2 Incentives, Design, and Control......................................................................................2466.3 Incentives in Experimental Economics: Convention in Action...........................................................2486.4 Three Perspectives on the Effect of Incentives on Behavior...........................................................2506.5 Incentive Mechanisms.................................................................................................2646.6 Conclusion...........................................................................................................2847 Noise and Variability in Experimental Data..............................................................................2867.1 "Noise" in Economics and in Experimental Economics...................................................................2867.2 "Noise" in Individual Decision Experiments...........................................................................2897.3 "Noise" in Experimental Games........................................................................................3067.4 Exploring Different Stochastic Specifications........................................................................3227.5 Concluding Remarks...................................................................................................3298 Conclusion..............................................................................................................3318.1 How Successful Has Experimental Economics Been in Developing a Sound Methodology?....................................3328.2 How Successful Has Experimental Economics Been in Increasing Understanding of Economic Behavior?.....................3388.3 Has Experimental Economics Had a Positive Impact on Wider Economics?.................................................342References................................................................................................................347Index.....................................................................................................................369

Chapter One

1.1 Experiments in Economics

Over the last thirty years, there has been a revolutionary change in the methods of economics. For most of the twentieth century, reports of experiments were almost unknown in the literature. Economics-as viewed by economists, and as viewed by professional methodologists-was generally taken to be a nonexperimental science. This understanding of economics is encapsulated in an incidental remark in Milton Friedman's famous essay on the methodology of positive economics-an essay that deeply influenced economists' methodological self-perceptions for at least three decades. Friedman says:

Unfortunately, we can seldom test particular predictions in the social sciences by experiments explicitly designed to eliminate what are judged to be the most important disturbing influences. Generally, we must rely on evidence cast up by the "experiments" that happen to occur. Friedman (1953, p. 10)

The implication is that the methods of economics, like those of astronomy (in the philosophy of science, the traditional example of a nonexperimental science), are an adaptation to the practical impossibility of controlled experiments. But, from the 1980s onwards, there has been an explosive growth in the use of experimental methods in economics. In terms of most obvious signals, these methods are now accepted as part of the discipline. Experimental research is carried out by many economists around the world. Its results are routinely reported in the major journals. In 2002, Daniel Kahneman and Vernon Smith were awarded the Nobel memorial prize in recognition of their work as pioneers of experimental economics.

Even so, it would be a mistake to think that experimental methods are no longer controversial in economics. Most economists do not conduct experiments and many remain unconvinced of their usefulness, as experimentalists still often discover when invited to present research papers to general economics audiences. Perhaps more significantly, the apparent consensus that experiments have a legitimate role in economics hides major disagreements about what that role is. Experimental economics is not a unified research program. Indeed, the two Nobel memorial prize winners represent two very different lines of research: Smith is an economist who has developed novel experimental techniques to investigate traditional economic questions about the workings of markets; Kahneman is a psychologist who has used the well-established experimental methods of his discipline to challenge economists' conventional assumptions about the rationality of economic agents. Some commentators have seen these two styles of research as so different that they have reserved the term "experimental economics" for Smith's program, in distinction to the behavioral economics of Kahneman's program. We find it more natural to define all forms of experimental research in economics as "experimental economics" and to use the term "behavioral economics" to refer to work, whether experimental or not, that uses psychological hypotheses to explain economic behavior. But whatever terminology one uses, it is undeniable that the research programs pursued by Smith and Kahneman began with different presuppositions and methodologies.

Economists can and do use experimental methods in their own work while rejecting the different methods used by other experimenters. They can and do recognize the value of some programs of experimental research while expressing skepticism about, or even hostility toward, others. There are ongoing disputes about what economics should learn from experimental results, about whether (or in what sense) economic theory can be tested in laboratory experiments, and about how far traditional theory needs to be adapted in the light of experimental results.

Given the speed with which experimental methods have been taken up, and the absence of a tradition of experimental research in economics, the existence of such controversies is hardly surprising. Perhaps for the same reasons, it is not always easy to discern exactly what the disputants are arguing about. In part, these controversies can be seen as normal scientific disagreements about how to interpret new findings. In part, they reflect disagreements about particular features of experimental method. In some cases, however, opponents may be arguing at cross purposes, failing to appreciate that different types of experiments have different purposes and potentially different methodologies. In other cases, apparent disagreements about experimental method may be the surface indications of much deeper differences between rival understandings of what economics is and how its claims to knowledge are grounded. Because widespread use of experimental methods is so new to the discipline, professional methodologists have only just begun to revise their accounts of the methods of economics to take account of the change. Among the profession generally, there is no recognized set of general methodological principles that can be used to structure these controversies.

This book is the result of our sense that economics needs a methodological assessment of the claims to knowledge that can be derived from the various kinds of experiments that are now being used. Our aim is to offer such an assessment-to describe, appraise, and, where possible, adjudicate between different positions on how experiments do or do not help us to understand the real economic world. In doing so, we hope to enrich the practice and understanding of experimental economics.

We hope to interest at least three kinds of reader: practicing experimental economists engaged in these controversies at first hand; nonexperimental economists trying to decide how to interpret (or whether to take any notice of) experimental results and the claims that experimentalists make about them; and philosophers of science who want to examine the status of the knowledge claims made in economics, or are curious about how a scientific community that once disclaimed experimental methods adapts to their introduction. Clearly, these groups of readers will come to the book with different background knowledge. In the rest of this chapter we provide some basic orientation for our varied readers. In section 1.2, we take a brief look at the history of experiments in economics and ask why economics saw itself for so long as a nonexperimental science. This leads into a discussion of some of the reservations that economists continue to express about experiments, and that feature in ongoing methodological controversies. In section 1.3, we provide outline descriptions of eight experiments, chosen from across the range of experimental economics, broadly interpreted. Our aim here is to give readers who are not familiar with experimental economics a preliminary sense of what this form of research is, and the kinds of claims that its practitioners make. In section 1.4, we use these examples to illustrate the main issues that will be addressed in the rest of the book. Finally, in section 1.5, we explain the stance that we take as authors, as practicing experimental economists writing about the methodology of our own branch of our discipline.

1.2 Does Economics Need Experiments?

Perhaps surprisingly, given the general perceptions among economists, the idea that controlled experiments can contribute to economics has a long history.

In an account of the history of experimental economics, Alvin Roth (1995a) uses as his earliest example the work of Daniel and Nicholas Bernoulli on the "St. Petersburg paradox" Bernoulli (1738). The St. Petersburg paradox is a hypothetical problem of decision under risk, in which most people's ideas about reasonable choice contravene the principle of maximizing expected monetary value. In an informal use of experimental methods, Nicholas Bernoulli tried out this decision problem on a famous mathematician to check his own intuitions about it.

We suggest that David Hume is another candidate for the experimental economists' Hall of Fame. Hume's A Treatise of Human Nature (1739-40) is now generally regarded as one of the canonical texts of philosophy, but it can also be read as a pioneering work in experimental psychology and decision and game theory. Significantly, the subtitle of Hume's book is: Being an Attempt to Introduce the Experimental Method of Reasoning Into Moral Subjects. In the preface, Hume describes his work as a study of "the extent and force of human understanding, ... the nature of the ideas we employ, and of the operations we perform in our reasonings." He undertakes to use the methodology of the natural sciences, investigating the workings of the human mind by "careful and exact experiments, and the observation of those particular effects, which result from its different circumstances and situations" (Hume 1739-40 pp. xv-xvii). In the course of the book, he describes the designs of a series of psychological experiments, and invites his readers to try these out on themselves. Among the results he finds are phenomena that were rediscovered (as so-called anomalies of decision-making behavior) by experimental psychologists and experimental economists in the late twentieth century.

It is particularly significant that neoclassical economics-the orthodox approach to the subject for most of the twentieth century-was, in the first years of its existence, based on experimental research. The pioneers of neoclassical economics were strongly influenced by what were then recent findings of experimental psychology. In launching the "marginal revolution" in economic theory, Stanley Jevons (1871) and Francis Edgeworth (1881) based their analyses of diminishing marginal utility on psychological findings about the relationship between stimuli and sensations. These authors were well aware of the work of psychophysicists such as Gustav Fechner and Wilhelm Wundt, which they saw as providing the scientific underpinning for the theory of demand. It was only from the beginning of the twentieth century that neoclassical economics separated itself off from experimental psychology, in a self-conscious process initiated by Vilfredo Pareto (1906). Intriguingly, Jevons (1870) may have been the first person to report the results of a controlled economic experiment in a scientific journal. This report, in the second volume of Nature, is of a series of experiments carried out by Jevons himself, investigating the relationship between fatigue and the effectiveness of human muscular effort. This was a matter of real economic importance at a time when major civil engineering works were being constructed by men with spades and wheelbarrows. Jevons (1871, pp. 213-16) tells us that he ran these experiments to illustrate "the mode in which some of the laws forming the physical basis of economics might be ascertained." As one might expect of a pioneer of neoclassical economics, Jevons was interested in such maximization problems as determining the optimal size of spade for shifting different materials, and the optimal rate of marching for an army.

Nevertheless, for much of the twentieth century, experimentation was a marginal activity in economics, barely impinging on the consciousness of most economists. With hindsight, it is possible to pick out landmark contributions to experimental economics, some even published in major economics journals; but it is striking that, for many years, very little was done to build on these isolated pieces of work. It seems that they were seen as having curiosity value, rather than as being part of the real business of economics.

For example, an experiment by Louis Thurstone (1931) is now seen as a classic. Thurstone, who was based at the University of Chicago, was one of the leading psychophysicists of his time. Through conversations with his colleague Henry Schultz, an economist doing pathbreaking work on the statistical estimation of demand functions, Thurstone had become aware that the concept of an indifference curve in economic theory had no direct empirical grounding. His experiment attempted to elicit individuals' indifference curves from responses to binary choice problems. Over the following three decades, the project of investigating whether the preferences postulated in theory can be elicited from actual choice behavior was pursued by only a tiny number of economists and decision theorists (see, for example, Mosteller and Nogee 1951; Allais 1953; Davidson et al. 1957; Davidson and Marschak 1959). Maurice Allais's discovery, in the early 1950s, of a systematic divergence between theory and behavior (that we describe in chapter 2) did not much trouble economists for another twenty years.

Similarly, Edward Chamberlin's (1948) investigation of price-determination in an experimental market would appear on any present-day list of great experiments in economics. Chamberlin was a leading industrial economist, famous for his theory of monopolistic competition. His experiment (described in chapter 4) was motivated by his awareness that price theory, despite its formal sophistication, provided no real explanation of how equilibrium is reached in real markets. His results seemed to confirm his hunch that equilibrium would not be reached under conditions typical of real-world markets. His paper was published in the Journal of Political Economy, but little further work was done for more than a decade. Systematic research on experimental markets was getting under way from the end of the 1950s (see, for example, Sauermann and Selten 1959; Siegel and Fouraker 1960; Smith 1962), but it remained very much a minority taste. It seems that most economists did not think that price theory was in need of experimental support.

Notwithstanding the existence of a few studies now seen as landmarks, it is probable that the large majority of economists saw their subject as fundamentally nonexperimental at least until the last two decades of the twentieth century. For many trained in the third quarter of the century, Friedman's 1953 essay would be their sole excursion into the subject's methodological literature; and echoes of its incidental remark on experiments could also be found in introductory textbooks of the time. For example, consider the following quotation from the 1979 edition of Richard Lipsey's classic textbook:

Experimental sciences, such as chemistry and some branches of psychology, have an advantage because it is possible to produce relevant evidence through controlled laboratory experiments. Other sciences, such as astronomy and economics, cannot do this. Lipsey (1979, p. 8)

Even now, one occasionally finds serious writers who echo Friedman's remark. For example, in the abstract of a paper on the methodology of economics published in a recent issue of Philosophy of Science, Marcel Boumans (2003, p. 308) asserts: "In the social sciences we hardly can create laboratory conditions, we only can try to find out which kinds of experiments Nature has carried out." Boumans's paper is an extended discussion of the question of how, given the supposed infeasibility of controlled experiments, economics can discover lawlike relationships within its domain of investigation.

Why did economists accept for so long the idea that their discipline was nonexperimental? It is sometimes suggested that the widespread use of experimental methods in economics has become possible only as a result of developments in information technology. It is certainly true that many experimental designs that are now used routinely would have been simply infeasible a few decades ago. The availability of generic software for economics experiments, such as the widely used z-Tree package designed by Urs Fischbacher (2007), has greatly reduced the investment in skills necessary to run computerized experiments. But, as our historical sketch has illustrated, there was no shortage of feasible and potentially informative experimental designs in the first three quarters of the twentieth century-just very little interest in using them. Even in the 1980s-the decade in which experimental methods began to be accepted in economics-many of the most significant experiments used pencil-and-paper technology. What has to be explained is why economists believed for so long that the information that such experiments would produce would not be useful.

Recall that Friedman's comment was that social scientists can seldom test particular predictions in controlled experiments. Since controlled experiments with human subjects are clearly possible, it seems that Friedman must be interpreted as saying that the kinds of experiments that are possible cannot be used to test the kinds of predictions that economics makes. Lipsey's use of the qualifier "relevant" suggests a similar view. Such claims should be understood in relation to two features of mid-twentieth-century economics. First, the domain in which economics was expected to make predictions was, by modern standards, narrow. As is suggested by the then-common use of the term "price theory" as a synonym for "microeconomics," the main focus of microeconomics was on explaining and predicting the values of statistics of aggregate market behavior-in particular, prices and total quantities traded. Macroeconomics worked at an even higher level of aggregation. Thus, the useful predictions of economics operated at a level at which, it was thought, direct experimental tests would be enormously costly and perhaps even unethical. The second feature was a prevailing conviction-a conviction for which Friedman (1953) argued strongly-that the "assumptions" of a theory are not claims about how the world is, but merely "as-if" propositions that happen to be useful in deriving predictions. Although price theory was derived from apparently restrictive assumptions about individuals' preferences, those assumptions were not to be interpreted as empirical hypotheses to which the theory was committed. Thus, experiments that purported to "test" the assumptions would be pointless.

(Continues...)

Excerpted from Experimental Economicsby Nicholas Bardsley Robin Cubitt Graham Loomes Peter Moffatt Chris Starmer Robert Sugden Copyright © 2010 by Princeton University Press. Excerpted by permission.
All rights reserved. No part of this excerpt may be reproduced or reprinted without permission in writing from the publisher.
Excerpts are provided by Dial-A-Book Inc. solely for the personal use of visitors to this web site.