Research

Abstract: In most real-life interactions players are free to act and react at whim: actions may incite others to react (which may then incite further reactions and so forth), and inaction leaves the floor for others to act. I propose a solution concept where players correctly predict the long-term consequences that their actions or inaction might bring, but know nothing about the process determining who acts when.

Abstract: We conduct a lab experiment to test the predictions of various myopic and farsighted solution concepts in dynamic multi-player games, as well as the way the dynamics of play are affected by the composition of players in terms of their cognitive abilities and the disclosure of this information. The experiment consists of an initial part measuring participants' cognitive ability, and a subsequent part where participants strategically interact within groups. Results lend empirical support to farsighted solution concepts allowing for preemptive moves. In addition, we find that high-ability individuals tend to insist harder on achieving Pareto-efficient outcomes. This effect is further amplified by the disclosure of information on other group members' cognitive abilities.

Abstract: This paper explores the existence of externalities from network architecture (so-called network externalities) in undirected link formation games. It develops a structural estimation method for static games of incomplete information where link proposals are only partially observable. We provide existence, consistency and asymptotic normality results for the proposed two-step estimator, and we document its performance through a simulation exercise. When the estimation method is applied to data on risk-sharing arrangements in a Tanzanian village, results indicate that indirect connections (i.e.  friends of friends) matter. Assuming that link formation follows a bilateral process, the estimated probability of proposing a link to a potential partner increases by 9% for any additional indirect connection provided.