Finite Rationality and Interpersonal Complexity in Repeated Games.

Finite complexity strategies suffice for approximating all subgame perfect equ ilibrium payoffs of repeated games. Generically, at such equilibria, no player's complexity exceeds the product of his opponents' complexi ties. Also, no player's memory exceeds the maximal memory of his oppo nents. The complexity of a strategy is defined here to equal the numb er of distinct strategies it induces in the various subgames. It equa ls the size (number of states) of the smallest automaton describing i t and also the number of states of the smallest information system ne eded for the implementation of the strategy. Copyright 1988 by The Econometric Society.