A General Theory of Equilibrium Selection in Games

by John C. Harsanyi, Reinhard Selten

The authors, two of the most prominent game theorists of this generation, have devoted a number of years to the development of the theory presented here, and to its economic applications. They propose rational criteria for selecting one particular uniformly perfect equilibrium point as the solution of any noncooperative game. And, because any cooperative game can be remodelled as a noncooperative bargaining game, their theory defines a one-point solution for any cooperative game as well. By providing solutions - based on the same principles of rational behavior - for all classes of games, both cooperative and noncooperative, both those with complete and with incomplete information, Harsanyi and Selten's approach achieves a remarkable degree of theoretical unification for game theory as a whole and provides a deeper insight into the nature of game-theoretic rationality. The book applies this theory to a number of specific game classes, such as unanimity games; bargaining with transaction costs; trade involving one seller and several buyers; two-person bargaining with incomplete information on one side, and on both sides. The last chapter discusses the relationship of the authors' theory to other recently proposed solution concepts, particularly the Kohberg-Mertens stability theory.

Genres: Economics

378 pages, Paperback

First published June 29, 1988

About the author

http://en.wikipedia.org/wiki/John_Har...

Reviews

[Tyler · Rating 5 out of 5]

This book is definitely worth it if you like Game Theory. The limitations imposed by the frameworks of previous economists/game theorists is finally gotten rid of here. This is a pivotal work which explores equilibrium selection over periods of knowledge and/or decision making in games, ultimately concluding in an example of a two person bargaining game where the payoff is secured exactly 1/2 of the time. This results in different alpha (or payoff) graphs, which initially begins as an exponential function of the value of alpha and then later on completely determines the equilibrium that dominates. The book ends with two equilibrium points, which had been gotten by means of the linear tracing method, one of which is payoff dominant, and the other is risk dominant. The authors then state why payoff dominance is a necessary metric which inherently is more prior to risk dominance in regards to the choice or decision making ability of the different players in the game, which is another point that they talk about at length in the appendix. Absolutely beautiful work, everything is considered in this system, and they will use a small, finite constant, epsilon, to perform perturbations on the game in question to reveal solutions that you would not have been able to determine before. Isomorphisms are also utilized in that manner, this is a fascinating, wonderful foray into the basic framework of games.

[Don · Rating 4 out of 5]

Slightly dated given the advances in evolutionary game theory - which make equilibrium selection less necessary - nonetheless, this is a classic in game theory by two of its giants.