Value Oriented Equilibria in Repeated Games of Complete Information

Two refinements, called value oriented equilibria, of the Nash equilibrium concept are proposed for repeated games of complete information. Value sufficient equilibria make each player's strategic response to the another player's previous actions depend only on the value of those actions to the responding player. In value monotonic equilibria no player punishes another for taking actions which increase the first player's payoff. The use of value oriented equilibria enables the set of outcomes consistent with equilibrium to be reduced. Outcomes which are unilaterally inefficient (i.e., can be Pareto dominated by the unilateral action of one player) are never part of the equilibrium path.