Source on Kahneman's "magical" behavior

Revision as of 21:39, 13 December 2011

I had trouble finding this on the internet, so here is a scan of the hard copy that came in.

pdf

Its important because Camerer and others have cited Kahneman's surprise at convergence in the market entry game, or what he calls the N* game. In his words "Observing the regularity of behavior in these markets was a bewildering experience -- to a psychologist, it looked almost like magic."

Besides summarizing the market entry results---in which psychologists are subverted by economics---he describes "reluctance to trade" and other cases of economists being subverted by psychology. The paper is short and sweet; full of lots of other good quotes. For example, he succinctly asserts a claim that Gode and Sunder fleshed out a decade later: "Robots programmed to obey simple dominance would establish the optimal market price."

Kahneman, D., (1986) Experimental economics: A psychological perspective presented in Bounded Rational Behavior in experimental games and markets: Proceedings of the Fourth Conference on Experimental Econoimcs, Bielefeld, West Germany, Spetember 21-25, 1986. eds. R. Tietz, W. Albers, R. Selten published in Lecture notes in Economics and Mathematical Systems, Eds M. Beckmann and W. Krelle

Abstract:

In a recent review Bruno Frey (1986) listed three assumptions of standard economic analyses: economic agents are supposed to be rational, to be selfish, and to have unchanging tastes. These assumptions contradict common sense knowledge of human nature as well as the conclusions of other social sciences. Economists know this, of course. Their position of deliberate o unconcern to for the validity of assumptions about economic agents was brilliantly stated in Friedman's (1953) classic essay, and additional arguments in its defense are added from time to time. For social scientists reared in other sultures, the intellectual position of economics presents a dual challenge. First, the position must be understood -- which is not easy to do across the cultural gap. Second, the limits of its validity must be established. The economic predictions sometimes succeed in ways that are surprising to other social scientists, and sometime fail in ways that are surprising to economics --- or would be surprising if the failure were admitted, which is not always the case. In the following comments I briefly describe two sets of recent experimental observations that bear on these issues. The first project is concerned with a situation in t which an economic prediction is upheld, in a manner that is quite surprising for a psychologist. The second documents a failure of an important assumption in economic analyses of exchanges.

extended quote:

Observing the regularity of behavior in these markets was a bewildering experience --- to a psychologist, it looked almost like magic.  The bewilderment was not eased by the debriefing conversations in which we engaged participants after the experiment.  They described a large variety of strategies and expectations as guiding their behavior.  Most of the strategies were completely unfounded (as they must have been, since the equilibrium effectively precluded any successful strategy).  Furthermore, in at least one case there was evidence that the strategy which the individual described did not in fact account for his choices.  The equilibrium outcome (which would be generated by the optimal policies of rational players) was produced in this case by a group of excited and confused people, who simply did not seem to know what they were doing.
Psychologists are trained to believe that aggregate phenomena can be explained by finding relevant regularity in individual behavior.  The N* game provided me with  first -hand experience of a clear failure of this belief.  The only solid explanation of the results of the N* game belongs to a type that is quite familiar to economists, but not to other social scientists.  As J. Brander  pointed out, E had to be close to N* because there a was no stable alternative.  Any systematic deviation from near-equality would have been obvious to the participants, and their attempt to take advantage of it would have ce cancelled it.  The achievement of global equilibrium in such a situation requires  very little  intelligence from the participants: just enough to recognize apparent regularities in the group response to variations f of N*, and to move to counter them.  It is very likely that all the participants had the intelligence to detect arbitrage opportunities -- and as a result there were none.