We study the market for a risky asset with heterogeneous valuations. Agents seek to learn about their own valuation by acquiring private information and making inferences from the equilibrium price. As agents of one type gather more information, they pull the equilibrium price closer to their valuation and further away from the valuations of other types. Thus they exert a negative learning externality on other types. This, in turn, implies that a lower cost of information for one type induces all agents to produce more information. When evaluating agents’ welfare, the learning externality has to be offset against a gains from trade externality, since agents who learn less because their valuation is further away from the price also stand to profit more from trading. In equilibrium, agents’ information acquisition decisions are clustered together more than is socially optimal.
