Stochastic growth with a likelihood-increasing estimation process (Q2490714)

The paper deals with the following problem of stochastic growth. While preferences and technology are subject to random shocks, the true probability distribution is unknown to the planner, who can only observe the sample realization generated by a stationary ergodic stochastic process. Basing on this information, he/she must estimate the true probability distribution in deciding on an optimal program. The estimation process is based on the realized sample path, and updated by taking the likelihood of the sample path into consideration. For each period, an estimator of the true probability distribution that assigns a higher likelihood to the given sample path than did the estimator in the previous period, is adopted. In such a framework, the author formulates an optimal estimation process. He considers a general reduced model of capital accumulation with an infinite horizon, and introduces a learning process in the stochastic dynamic programming. The main result of the paper is that the optimal estimation process based on likelihood-increasing behavior converges to the true probability measure and the likelihood increasing estimator defines a transition function on the sample space. Two opened questions are pointed out by the author. The first is whether or not the sequential limit of an optimal program has the turnpike property, and the second concerns the existence of a support price that characterizes an optimal program.