Kevin McCabe is a Professor of Law at George Mason University and holds appointments at George Mason’s Interdisciplinary Center for Economic Science, the Mercatus Center, and Krasnow Institute.
Having started my career as an experimental economist I probably have a little different, but I hope complimentary, perspective on behavioral economics and other experimental programs in general.
I view the difference between experimental and behavioral economics in terms of (1) what is studied, and (2) how it is studied. Experimental economists are interested in institutional and organizational rules and how these rules affect both, the joint behavior of participants, and the outcome generating, or process, performance of the institutional rules in question. To study this the experimental economist induces preferences and implements a microeconomic system. One major problem for this approach is that ‘risk preferences’ are very noisy, when induced, due either to, the added complexity imposed on subjects of having to work with induced preferences, or that the induced preferences conflict with a subject’s actual preferences. A second major problem with this approach is that institutional rules that are isolated in the lab often depend on on additional rules that are not being studied, or social and cultural norms that are not present in the lab. Experimental economists have learned to manage these problems and many interesting research papers have been produced.
Behavioral economists are interested in individual behavior, whether it be individual choices, strategic decision making, or competitive strategies in markets. The behavioral economist does not in general induce preferences, but does often use salient rewards in a well defined decision theoretic problem defined by decision theory, game theory, or price theory. As a consequence of not inducing the behavioral economist is interested in the nature of preferences, and the nature of decision making. One major problem for this approach is that preferences and decisions interact, and it is often not clear whether one is studying the former, the later, or a combination of both. A second major problem with this approach is that behavior observed in the lab may not capture the full computations that people are capable of making when augmented by technology and institutions. But again, behavioral economists have learned to manage these problems and many interesting research papers have been produced.
When I refer to experimental economists, or behavioral economists, I am referring to a researcher employing a specific methodology to explore a specific class of problems. So, in my experience, there are many researchers who employ more than one methodology, and this has proven to be very useful. But now I can be more specific, thus narrower, and I’m sure subject to more debate. Lets hypothesize that experiments are all about exploring the computations that humans make. Under this hypothesis both experimental economics and behavioral economics are methods for exploring computational mechanisms. In the former case institutions are mechanisms than make computations, and in the later case individuals are mechanisms that make computations, but in the end we will want a computational theory of economics that includes both. I think this is where we are heading and when I look at some of the most promising experimental programs including economic systems design, which seeks to engineer better institutions, and neuroeconomics, which seeks to understand the computations occurring in embodied brains, it seems that the computational hypothesis is one that will best integrate the different experimental methodologies and best serve to move experimentation forward.
This raises the question, should we use experiments to study the law? By my hypothesis anything computational can be studied experimentally, and in legal institutions and in legal decision making many interesting computations are made. This suggests that we could use experiments to study the law. The downside of course is that our experiments could mislead us, but any source of data could mislead us. In its favor experiments invite a form of structured debate that is almost impossible to have without them. In particular, if I don’t like your experiment, then I’m free to run my own counter experiment, and as long as both our experiments replicate, a good theory should be able to explain both results and lead us to a better understanding of the mechanism in question. If we agree to the theory but are still hesitant to apply our knowledge to the field we are now in a better position to design, and run, a field experiment that can help us decide.