Collective Sampling: An Ex Ante Perspective

Yangfan Zhou

DOI: https://doi.org/10.48550/arXiv.2311.05758

2024-06-24

Abstract:I study collective dynamic information acquisition. Players determine when to end sequential sampling via a collective stopping rule. With two players, collective stopping is determined either unilaterally or unanimously. I develop a methodology to characterize equilibria 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 mean-preserving spreads. Equilibrium sampling regions are determined via concavification. Preference misalignment generates Pareto inefficiency: there is too little learning under unilateral stopping and possibly too much under unanimous stopping. Learning inefficiencies are amplified when players' preferences become less aligned. I demonstrate the value of my methodology in three applications: committee search, dynamic persuasion, and competition in persuasion.

Theoretical Economics