Essays on Coordination Games

This thesis consists of three chapters on coordination games. The first chapter studies how the existence of Imitators in population affects the success of the threshold public good production. Threshold models require at least m out of n players to form a team to produce a public good. We model the dynamics of contribution decisions when the population consists of Best-responders and Imitators. We show that the existence of Imitators improves the feasibility of production only when Imitators are willing to join the team in the early stages. The production even when achieved is not necessarily optimal. When the number of Imitators is high, contributions exceed the efficient level. In the second chapter, we test how subjects in lab divide attention between public and private information when they need to coordinate (or anti-coordinate) their actions on the unobserved state. Subjects are provided with one public and one private piece of information about the state in the form of noisy signals and asked to assign a weight to each information type. The public signal is perfectly correlated, whereas private signals are independent or partially correlated. We find that subjects weigh private signal less when they need to coordinate. However, they do not increase the weight in the anti-coordination game. Even though subjects do not play the optimal strategy in the anti-coordination game, they react to correlation in private signals as predicted by theory. Correlated signals attract higher weight in coordination, and lower weight in the anti-coordination game, relative to the independent private signals. The third chapter is a model of information supply to players who want to coordinate their actions on the unobserved state. A profit-maximizing monopolist designs a discriminating mechanism to extract higher surplus from the buyers with different valuations for the information. We show that the buyer with the highest valuation always receives a fully informative signal. The quality of the signal offered to the low valuation buyer is a continuous and increasing function of the frequency of low types. Correlation in actions causes externalities in prices; each type is also willing to pay for the precision of the signal received by their partner. When there is a high probability of meeting the low type, accuracy of low type’s decision increases the high type’s payoffs. Hence a low type receives a positive amount of information.