Multiperiod mean-variance portfolio optimization in Markovian markets under imperfect information

Abstract

ABSTRACT In this thesis, we analyze multiperiod portfolio optimization problems in stochastic mar kets where periodic returns are serially correlated and information flow is imperfect. Serial correlation and imperfect information flow are constructed by using two processes, one of which is observable and the other is hidden. Both processes are assumed to be Markov chains. The market consists of a riskless asset and several risky assets whose returns directly depend on state of the unobserved market process. The state of the unobserved stochastic market in a certain period depends on the prevailing economic, social and other relevant fac tors. We consider two different models that describe the imperfect information flow between the unobserved stochastic market and the observed process. Considering such a stochastic market modulated by a hidden Markov chain, the multiperiod mean- variance formulation is solved by using dynamic programming. The explicit optimal solution is obtained, and some illustrative cases which demonstrate the application of the solution procedure are given. Keywords: Multiperiod Portfolio Optimization, Mean- Variance Models, Dynamic Pro gramming, Hidden Markov Chain, Imperfect Information IV