On Learning and Growth

We embed learning (without experimentation) in optimal growth. We extend the Mirman-Zilcha results of stochastic optimal growth to the learning case. We use recursive methods to study the effect of learning on the dynamic program by considering the case of iso-elastic utility and linear production, for general distributions of the random shocks and beliefs (i.e., there is no conjugate priors) and for any horizon. We finally address the issue of experimentation by providing a solution to an infinite-horizon optimal dynamic program.