Collaborative Research: Bargaining and Behavioral Pertubations in Repeated Games

Princeton University, Princeton NJ

Investigators

Abstract

By arriving at self-enforcing agreements, agents in an ongoing strategic situation create surplus that benefits them both. Little is known about how that surplus will be divided. This project concerns the role of reputation formation in such environments. It studies a model in which players entertain the slight possibility that their opponents may be subject to a variety of behavioral biases. Happily, it seems that these informational complications make it easier to predict what will usually happen in the repeated game. There emerges a theory of bargaining in repeated games that strengthens some links between cooperative games and noncooperative game theory. The proposal also addresses dynamic games, in which the different periods of play are con-nected by "state variables" such as capital stock or inventories. Here one can examine some of the most important problems in industrial organization, including collusion in the face of potential entry. One of the goals is to derive a generalization of the "Nash bargaining solution with threats" that can be applied in dynamic games.