ABSTRACT OF PAPER

Title: On the role and limitations of experimental/behavoural economics
Author: Boland Lawrence


Whether one wishes to conduct laboratory experiments to falsify aspects of traditional economic theories and behavioural propositions or one wishes to build replicable experimental designs to accumulate credible ‘exhibits’, most experimental model builders today recognize some unavoidable criticisms and the inherent logical problems of model building whether the model is just an experimental design or the pre-1980s’ notion of testing a theory by building a model of it and seeing whether or not it is consistent with available data. There are two main criticisms – questionable ‘external validity’ of the test or of exhibits produced by an experimental design and the inherent ambiguity of the results that one produces whenever the results depend on the additional assumptions or conditions needed to conduct the experiments. These two criticisms are just two aspects of what today’s experimental economists know as the Duhem-Quine problem. As Vernon Smith explains the Duhem-Quine problem, ‘experimental results always present a joint test of the theory (however well articulated, formally) that motivated the test, and all the things you had to do to implement the test’ [2002, p. 98]. The problem is a matter of logic and simply is that we do not know whether any observed refuting evidence is due to the theoretical behavioural assumptions we were trying test themselves or one of the additional assumptions we made to construct the model. In the case of seeking credible exhibits, we cannot know for sure whether the apparent behavioural regularity we observe in the application of a particular experimental design was due to those additional assumptions we needed to add to make the experiment practical. This logically ambiguous situation is the Duhem-Quine problem that Smith was explaining. Almost all experimental economists seem to consider the Duhem-Quine problem to be insurmountable. And so, rather than making any claims about the truth status of falsifying test or positive exhibits, experimental economists have for the most part given up and turned to looking to experiments to provide ‘positive evidence of a particular regularity in behaviour’ [Sugden 2008, p. 625]. However, the results of laboratory experiments can present puzzles that theorists need to address. After all, economic theory is supposed to apply to real economic situations and even the limited nature of laboratory experiments still may involve the choice behaviour purportedly explainable with traditional micro economic theory and models. And if an experimental design presents exhibits or regularities that appear to be anomalies, it can indicate that there is some work to be done perhaps to fix some aspect of traditional theory. This paper will examine whether the many attempts to address these problems have been successful or at least useful.

Registred web users only can download this paper - Go back


Please note that files available for download have not been checked for viruses. These files have been submitted by authors of the conference to this web site. Conference organisers can't accept any responsibility for damages caused to users by downloading such files.