Asset Allocation Under Distribution Uncertainty

This paper shows how uncertainty about the type of return distribution (distribution uncertainty) can be incorporated in asset allocation decisions by using a novel, Bayesian semiparametric approach. To evaluate the economic importance of distribution uncertainty, the extent of changes in ex-ante optimal asset allocations of investors who factor in distribution uncertainty into their portfolio model is examined. The key findings are: (a) distribution uncertainty is highly time varying; (b) compared to investors facing parameter uncertainty, investors under distribution uncertainty, on average, allocate less money to risky assets; their allocations are less variable; and their certainty-equivalent losses from ignoring distribution uncertainty can be economically significant; (c) portfolio strategies of such investors generate statistically higher returns, even after controlling for common factors