| A standard assumption in game theory is that in equilibrium players have correct beliefs about both the equilibrium strategies of other players and the distribution over the fundamentals in the economy. The notion of a self-confirming equilibrium (SCE) relaxes such an assumption by requiring beliefs to only be consistent with whatever information feedback players obtain while repeatedly playing the game, but not necessarily correct. In this dissertation I first show how the idea of a SCE can be applied to study how information policy affects market outcomes by determining what players know about their strategic environment and the strategies being followed by other players. In particular, I study how the amount of feedback about past bids that is provided by an auctioneer affects equilibrium bidding in first price auctions. I then extend the notion of a SCE by making explicit that players' beliefs are not only restricted by information feedback but also by how players use their cognitive skills to interpret this information as well as by players' knowledge of their strategic environment. This extension is useful for two reasons. First, it leads to an extension of the idea of equilibrium to cases where players suffer from cognitive biases. Second, it provides a framework to study how the equilibrium set changes with different assumptions about what players know about the fundamentals, what they know about the rationality of other players, what they know that other players know, and so on. In particular, the proposed framework provides a way to relax the standard common knowledge assumption. |
