Journal of Economic Behavior and Organization, 51, (2003), pp. 161-193,
Abstract
We perform an experimental test of Maskin's canonical mechanism for Nash
implementation, using 3 subjects in non-repeated groups, as well as 3
outcomes, states of nature, and integer choices. We find that this
mechanism succesfully implements the desired outcome a large majority of
the time and an imbedded comprehension test indicates that subjects were
generally able to comprehend their decision tasks. The performance can
also be improved by imposing a fine on non designated dissidents. We
offer some explanations for the imperfect implementation, including risk
preferences, the possibilities that agents have for collusion, and the
mixed strategy equilibria of the game.
Other authors
Universitat Pompeu Fabra. Departament d'Economia i Empresa