Strategic asset allocation with distorted beliefs

Abstract

We propose an approximate solution for the asset allocation model in which the dynamics of the investment opportunity set are subject to regime shifts and the regime updating/predicting procedures are contaminated by psychological biases, including optimism, pessimism, conservatism, representativeness, momentum, and reversal. We employ monthly U.S. data to study the model and find that the optimistic investor earns higher returns and outperforms the rational (Bayesian) investor. We find that the outperformance of optimism results from the return predictability and that learning errors of optimism are offset by the information contained in the data sample. Psychological biases that can generate a large variability in beliefs or state uncertainty induce the investor to tilt toward cash and bonds.