Adjusted robust mean-value-at-risk model: less conservative robust portfolios

Abstract

We examine the robust mean-VaR portfolio optimization problem when a parametric approach is used for estimating VaR. A robust optimization formulation is used to accommodate estimation risk, and we obtain an analytic solution when there is a risk-free asset and short-selling is allowed. This renders the model computationally tractable. Further, to avoid the conservatism of robust optimal portfolios, we suggest an adjusted robust optimization approach. Empirically, we evaluate the out-of-sample performance of the new approach, the robustness of obtained solutions and level of conservatism of the resulting portfolios. The empirical results highlight some benefits of our approach.