Time Inconsistency and Learning in Bargaining Games

Abstract

The literature on time-inconsistent preferences introduced sophisticated, naive and partially naive types of agents that represent different levels of awareness of their time inconsistency. This paper incorporates time-inconsistent players in a sequential bargaining model. Under the complete information assumption, we first consider naive agents who never learn about their types and show that bargaining between such a player and a standard exponential agent ends in immediate agreement if the exponential agent offers first. A one period delay occurs if the naive player offers first. In addition, the more naive the player is, the higher share she gets. We then allow naive and partially naive agents to learn their types over time and show that there is a critical date before which there is no agreement. Hence, the existence of time-inconsistent players who can learn their types as they play the game can be a new explanation for delays in bargaining. The relationship among the degree of naivete, impatience level and bargaining delay is also characterized.

Keywords: Hyperbolic discounting, learning, bargaining, delay