Communication and Information Acquisition in Committees (Job Market Paper)
Abstract: Many collective decisions rely on costly information acquisition by group members, giving rise to a free-riding problem. This paper examines how pre-vote communication affects this problem. I present a model in which voters with common interests decide whether to acquire private information before voting, and compare institutions with and without an intermediate communication stage. The model identifies two opposing effects: communication raises the incentive to acquire information by allowing a member's signal to influence others' votes (persuasion effect), but lowers it by making others' information a closer substitute for one's own (substitution effect). In equilibrium, communication increases information acquisition and improves group accuracy and welfare when information is relatively expensive, but reduces them when information is relatively cheap. I test these predictions in a pre-registered laboratory experiment with three-member committees. Communication increases acquisition when information is expensive, as predicted, but does not reduce it when information is cheap, and it improves group accuracy and welfare at both cost levels. Additional evidence helps explain the missing downside: subjects recognize the persuasion effect but insufficiently account for the substitution effect, leading them to overvalue information acquisition under communication.
Talk Helps, but Excess Talk Can Hurt: Pre-vote Communication and the Swing Voter's Curse (under review)
Abstract: This paper investigates the effect of pre-vote communication on collective decision-making when abstention is allowed. We present a model of majority-rule voting with common-interest and asymmetrically informed voters, with and without a communication stage. Communication does not alter the set of equilibrium outcomes. Still, we hypothesize that communication decreases efficiency, based on the conjecture that messages from poorly informed voters dilute those from better-informed voters. Our experimental implementation of the model shows that communication instead improves the efficiency of collective decisions, primarily through an unanticipated mechanism: it curbs irrational voting by less-informed voters. Nevertheless, consistent with our message-dilution conjecture, efficiency is lower when everyone speaks than when the less-informed voters stay silent, because after communication less-informed voters rarely abstain and instead largely vote with the majority message.
Persuasion with Two-sided Messages (with Jan Potters)
Sequential Deliberation in Committees (with Marina Agranov and Thomas R. Palfrey)