Collective Sampling: An Ex Ante Perspective

I study collective dynamic information acquisition. Players determine when to end sequential sampling via a collective choice rule. My analysis focuses on the case of two players, but extends to many players. With two players, collective stopping is determined either unilaterally or unanimously. I develop a methodology to characterize equilibrium outcomes using an ex ante perspective on posterior distributions. Under unilateral stopping, each player chooses a mean-preserving contraction of the other's posterior distribution; under unanimous stopping, they choose meanpreserving spreads. Equilibrium outcomes can be determined via concavification. Players learn Pareto inefficiently: too little under unilateral stopping, while too much under unanimous stopping; these learning inefficiencies are amplified when players' preferences become less aligned. I demonstrate the value of my methodological approach in three applications: committee search, dynamic persuasion, and competition in persuasion.