Colin F. Camerer
1200 E California Blvd
MC 228-77
Pasadena, CA 91125
(626) 395-4054

Email me

Until relatively recently (the 1970's), economists thought of economic systems as being like astronomical systems of planets and stars--they could only be observed from afar, and not touched or created experimentally. But economic systems can be created in artificial laboratory environments and studied experimentally, as in most older sciences (physics, chemistry, and biology). In an economics experiment, the experimenter specifies "endowments"--what agents start out with--, the messages they send and choices they make, and how the messages and choices agents pick determines their economic outcomes. (Usually they are paid substantial sums according to their experimental performance, to be sure they are thinking carefully and acting like agents do in naturally-occurring economic interactions.) The experimenter does not determine how the experimental participants actually behave, because seeing what people do is the whole point of the experiment. (Usually we have one more competing theories about what is likely to happen; an experiment can tell us which of these various theories, which may all sound intuitive, is just wrong.) A good place to learn how to do experiments is from the books by Friedman and Sunder and Davis and Holt. A good place to learn what we found out from economics experiments through 1995 is the Handbook of Experimental Economics, and since then, from the Handbook of Experimental Economics Results (in press).

Some of my early experiments concerned how people weigh the chance of winning (probability) and the amount they can win when they choose among risky gambles. (Economists use gambles over money as simple metaphors for risky economic activity like investing in education or assets, starting up a business, buying a house, and so on.) In the 1970's and 1980's many theories were proposed about how people weight probabilities and value outcomes differently than is proposed in "expected utility theory". My experiments and analyses found that of the new theories that were proposed, prospect theory seems best able to explain patterns in experimental data. Those papers are:

  • "An Experimental Test of Several Generalized Utility Theories," Journal of Risk and Uncertainty, 2, 1988, 61-104. (Reprinted in J. Hey and G. Loomes (Eds.), Recent Developments in Experimental Economics, Edward Elgar Publishing, Ltd.)
  • "The Predictive Utility of Generalized Expected Utility Theories," with David Harless, Econometrica, 62, 1994, 1251-1290. (Reprinted in J.D. Hey (Ed.), The Economics of Uncertainty, Edward Elgar Publishing Ltd., 2000.)
  • "Violations of the Betweenness Axiom and Nonlinearity in Probabilities," with Teck Ho, Journal of Risk and Uncertainty, 8, 1994, 167-196.
It is often noted that most data evaluating theories of risky choice have been collected by offering subjects simple choices between simple monetary gambles in the lab. But, in fact, many of these models can also be used to understand labor supply, asset pricing, consumer choice, and gambling in field settings which matter for everyday life. Using prospect theory elements to explain interesting patterns in field data is discussed in my paper "Prospect theory in the wild: Evidence from the field," (pp 288-300) in D. Kahneman and A. Tversky (Eds.), Choices, Values, and Frames, 2001. Cambridge: Cambridge University Press. An exciting new development in experimentation is studying unusual, important special populations. A clever undergraduate, Stephanie Kovalchik, with a little coaching from me, John Allman, Dave Grether, and Charlie Plott, studied an amazing sample of 80-year olds and compared them to 20-year old students on a variety of judgment, bargaining, and game theory tasks. The older and younger folks are remarkably similar, except on how much they know about the world and how good their self-knowledge is (i.e., whether they know how much they know): The 80-year olds know more, and know when they don't know and when they do. (Is that a definition of wisdom?) Our paper is here; it's forthcoming (2003-4) in the Journal of Economic Behavior and Organization.

Keith Weigelt and I did some early studies on experimental asset markets. In these experiments participants get a valuable asset, which will pay a cash dividend if they hold it at the end of a trading period. We study the prices at which people buy and sell the asset. In 1991 we published the first study on "information cascades" (which we called "mirages")--namely, is it possible for nobody in a market to have "inside information" about what an asset is worth, but for some traders to think that price movements mean other people have inside information, which creates a self-fulfilling kind of cascade or "herd behavior"? The answer is Yes, cascades do occur. But they only occur early in the experiments when traders are inexperienced. After participants trade for a while, they learn to figure out whether other traders have inside information by whether the market is lively or quiet (if it's quiet, nobody wants to trade because nobody has inside information) and the cascades stop. That paper is: "Information Mirages in Experimental Asset Markets", with K. Weigelt, Journal of Business, 64, October 1991, 463-493.

Keith and I also studied "price bubbles". If an asset lives a long time, like a share of stock or a house, prices can go up simply because people think they can sell at a higher price in the future. Financial economists have known about the theoretical possibility of such bubbles for decades, and there are many famous examples like the Dutch tulip bulb bubble in the 1600's. But it is hard to establish when prices are really in a bubble because we never know the true value of a naturally-occurring asset. In the experiments we create the value of the asset and so we know what it should be worth (and we know that subjects know its "fundamental value" too). In one experiment we observed many bubbles--in some cases the asset traded for five times its intrinsic value. That paper is "Convergence in Experimental Double Auctions for Stochastically Lived Assets," with K. Weigelt, in D. Friedman & J. Rust (Eds.), The Double Auction Market: Theories, Institutions and Experimental Evaluations, Redwood City, CA: Addison-Wesley, 1993, 355-396.

An early survey of ideas about bubbles and fads (well before the "tech stock" and Japanese stock bubbles!) is: "Bubbles and Fads in Asset Markets: A Review of Theory and Evidence," Journal of Economic Surveys, 3, 1989, 3-38. (Reprinted in Italian in G. Vaciago and G. Verga (Eds.), La Teoria dei Mercati Finanziari, Italy, Societa Editrice II Mulino.)

One of my current research projects, with Roberto Weber is about organizational culture. A culture is a set of values, rules for behaving ("institutions"), and symbols and language. In our experiments subjects get a set of pictures which they must describe to each other, under time pressure, by creating a slang or code. Their code is an element of culture which we can create artificially, and quickly, in the lab and study. We wrote one paper (''Cultural conflict and merger failure: An experimental approach'') on this project and are doing more in 2003.

Charlie Hornberger and John Lin developed nice "CultureX" software for studying code development using the kind of picture-naming Rob Weber and I used to study "corporate culture", which we are happy to share. Documentation is here. If you use it please give us feedback.

Besides Caltech, there are many universities with labs in experimental economics. This brief list omits many important centers (please email me to correct omissions), but among the most active labs are: Amsterdam; Arizona; Harvard; University College London; Nottingham; Ohio State; Oxford; Technion; Texas; Texas A&M; Trento; UCLA; Wisconsin; and Zurich.

Division of the Humanities and Social Sciences
Mail Code 228-77
California Institute of Technology
Pasadena, California 91125

Office: Room 101
Phone: (626) 395-4054
Fax: (626) 432-1726
email icon  E-mail: camerer@hss.caltech.edu
Caltech Home Page