We examine the properties of equilibrium stock returns in an incomplete information economy in which agents need to learn the hidden state of the endowment process. Both Bayesian and suboptimal learning rules are considered, including near-rational learning, conservatism, representativeness,...