Experimental Tests of Nash Refinements

点击翻译按钮获取中文摘要

A major unresolved issue in the study of noncooperative games is the appropriate refinement of the Nash equilibrium concept. A "refinement" is a theory with equilibrium outcomes that are a proper subset of the set of Nash outcomes. A sequence of successively stronger refinements has been proposed, many of which are motivated by specific examples of simple games in which the weaker refinement permitted equilibria which were thought not to be sensible. In this literature, discussions about what behavior is sensible are typically based on the author's subjective opinion about how "rational" individuals would play the game. Even though game theory is sometimes said to be a normative theory about how rational agents ought to behave, the theory is widely applied in industrial organization and other areas of economics to analyze specific issues, so the positive (empirical) content of game theory is also relevant. In addition the discussions of intuitive and unintuitive equilibria are often in the context of examples that are inspired by market applications. The contribution of this project comes from using carefully designed laboratory experiments to test competing refinements. In these experiments financial rewards are used to induce players' preferences in a manner such that competing theories give distinct predictions. The results of the experiments will provide badly needed empirical evidence on which of these refinements repeatedly fail to predict the results of controlled laboratory experiments.