Information aggregation and learning in a dynamic asset pricing model

This paper analyses a dynamic framework where an unobservable fundamental can be learned over time through two signals: one exogenous and private and the other, prices, endogenous and public. As information cumulates over time through Bayesian learning, prices become fully revealing and agents disregard their private information, suggesting a possible route through which fundamental values and prices can become misaligned. The analysis is then extended to a setting where agents need to infer the statistical properties of the signals they receive, merging Bayesian with adaptive learning. By introducing uncertainty about the moments of the relevant distributions used for Bayesian learning, adaptive learning can improve the ability of prices to track changes in fundamentals and thus their efficiency.