The Experiment in the History of Economics. Edited by PHILIPPE FONTAINE and ROBERT LEONARD

Routledge , London . xii+158 pp . £65 .

This book, the conference volume of the Fifth Annual Conference on the History of Economics in 1999, consists of six papers loosely structured around the theme of the history of experiments.

The first paper, by Robert Diamond, is a straight history, from Chamberlin's early experiments up until the 1970s. Diamond's emphasis is on the link between game theory and the rise of experiments. This link was close, because game theory is more amenable to testing than other types of economic theory. The paper traces the influence of individuals on the development of experiments, and concludes that it was the death of key individuals and the breaking up of key research groups that prevented experimentation from emerging as a key part of economic research before the 1980s.

As it stands, this hypothesis is not particularly satisfying, as no reason is given as to why game theorists should have been interested in experiments while other economists virtually ignored them. Rizvi, in the second paper, tries to answer this by looking at the theoretical structures involved. According to this author, the axiomatic approach of the general equilibrium approach blocked the use of experiments because of the emphasis on introspection to validate its axioms. The axioms of general equilibrium were a part of ‘everyday experience’; so, given valid deduction, there was no need for testing until the model was complete. The model would then be so complex that it could be tested only by using econometrics. Rizvi (like Diamond) also links experimentation to game theory, but points out that game theory became a major part of economics only when general equilibrium theory suffered a theoretical breakdown in the 1970s as a result of the Sonnenschein et al. results. As a result, general equilibrium theory was largely replaced by rational choice game theory.

Although this is undoubtedly true, I cannot agree with Rizvi's next move—which attributes the switch from rational choice to evolutionary game theory to experimental results on the ultimatum game. As Sugden (Journal of Economic Methodology8, 2001) has commented, this does not accord with the mainly theoretical concerns of game theorists, who were more concerned with the problem of multiple equilibria and the corresponding failure of the equilibrium refinement project. This led theorists to look for alternative foundations for equilibrium, which were found in evolutionary game theory. Furthermore, it is doubtful that evolutionary game theory is particularly responsive to experiments. As Camerer (Behavioral Game Theory, Princeton University Press, 2003) points out, in spite of its popularity among game theorists, evolutionary game theory has not performed well in the experimental lab.

The paper by Jallais and Pradier revolves around the events of the 1952 Paris conference and the introduction of the Allais paradox. It has some fascinating detail on the intellectual background of Allais's objections, and also corrects some myths about the introduction of the paradox; for example, the paradox itself was only shown to Savage over a lunch at the conference, and there was no paper presented on it.

The chapter by Innocenti and Zappia compares Morgenstern and Hayek. Both theorists started off as deductive theorists in the Austrian school. However, disagreements about the possibility of equilibrium led to Morgenstern becoming an inductionist. The contention is that Morgenstern helped create game theory as a step towards making economics an empirical science. Hayek, by contrast, doubted the possibility of such empirical work within economics, describing it as ‘scientistic’. He remained a deductionist until the end of his life.

The fifth paper, by Boianovsky and Erreygers, concerns the monetary precursors of the current New Monetary School. The paper examines the idea that a pure credit economy would allow easier control over the price level (or would give price stability). This is an interesting work in the history of economics, but I did not feel that it had any real relevance in this collection. The attempts to sell it by referring to Solvay's experiments in the natural sciences or Wicksell's ideas on experiments were the weakest parts of the paper.

The final paper, by Nancy Cartwright, is not historical, but it is a good contribution to the methodology of experiments. The author uses an analogy between experiments and economic theory to examine the simplifying assumptions that can be made in models (or ‘analogue economies’) for them to identify real tendencies in the economy. This was done by analogy with ‘Galilean experiments’, whereby controls are placed on all factors except those that are needed to study the effects of a given cause.

In a model, ‘simplifying’ assumptions either act as controls or contribute to the effect. If assumptions contribute to the effect then they cannot be seen as merely ‘simplifying’, so the model may not identify a tendency in the real economy. However, if assumptions are controls, the model has general application. A good model therefore can be transformed into a Galilean experiment. This point is of crucial importance when thinking about theory design, as it classifies those simplifying assumptions that are valid and those that are invalid.

In general, this is an interesting collection of papers. There are some worthwhile papers, and those by Rizvi and Cartwright are particularly interesting.