The rates of learning with public and private signals

This paper presents a dynamic market-based learning model, in which agents trade a risky asset repeatedly while observing public and private endogenous signals about market prices and demands. We explicitly characterize the changes in precisions of agents’ beliefs about the market, and demonstrate that complete learning is achieved in the limit. Moreover, we show that the asymptotic learning rates of both public and private signals attain the highest possible rate in the model, which is linear, thereby establishing the asymptotic learning efficiency. Factors that influence the learning speed and asymptotic rates are also identified. We also prove that the asymptotic learning efficiency applies to other learning situations.