Games with Unrestricted Information Acquisition (Job Market Paper)
Current Version: November 21, 2015
Abstract: In a game, when there is uncertainty about the state of fundamentals, the players' behavior depends not only on their information about the state, but also crucially on their information about each other's information. Before taking actions, what do the players choose to know about what the others know? In this paper, I propose a tractable model of information acquisition to address this question in a systematic way. To unmask the primitive incentives to acquire information, the model does not impose restrictions on the information choice technology: the players can acquire information not only about the state, but also about each other's information in a flexible way. In coordination games, I show that the players have a strong incentive to know what the others know, even if it is unrelated to the state. In investment games, this leads to risk-dominance as the unique solution. In linear-quadratic games, this generates nonfundamental volatility. I further show that this incentive weakens as the game gets large and players small. In large investment games, multiple equilibria arise where the players focus on information about the state. In linear-quadratic games, nonfundamental volatility vanishes if no player is central in the game.
Rationally Inattentive Preferences and Hidden Information Costs (Joint with H. de Oliveira, M. Mihm, K. Ozbek)
Theoretical Economics, Conditionally Accepted
Abstract: We show how information acquisition costs can be identified using observable choice data. Identifying information costs from behavior is especially relevant when these costs depend on factors–such as time, effort and cognitive resources–that are difficult to observe directly, as in models of rational inattention. Using willingness-to-pay data for opportunity sets–which require more or less information to make choices–we establish a set of canonical properties that are necessary and sufficient to identify information costs. We also provide an axiomatic characterization of the induced rationally inattentive preferences, and show how they reveal the amount of information a decision maker acquires.
Endogenous Informational Smallness
Current Version: November 14, 2015
Abstract: An agent is informationally small if, given the information of the other agents, her information is not pivotal to determine the state of fundamentals. When all the agents are informationally small, the inefficiencies due to asymmetric information are small and can be disregarded. Informational smallness is often assumed in settings with many agents. In this paper, I study whether informational smallness emerges as an endogenous outcome of large games with unrestricted information acquisition. I show that, in equilibrium, informational smallness does endogenously emerge, provided that a mild requirement of non-triviality is satisfied.