Value of History in Social Learning: Applications to Markets for History

Abstract

In social learning environments, agents acquire information from both private signals and the observed actions of predecessors, referred to as history. We define the value of history as the gain in expected payoff from accessing both the private signal and history, compared to relying on the signal alone. We first characterize the information structures that maximize this value, showing that it is highest under a mixture of full information and no information. We then apply these insights to a model of markets for history, where a monopolistic data seller collects and sells access to history. In equilibrium, the seller's dynamic pricing becomes the value of history for each agent. This gives the seller incentives to increase the value of history by designing the information structure. The seller optimal information discloses less information than the socially optimal level.