Safety-first portfolio selection

Das, Markowitz, Scheid, and Statman (2010) introduced portfolio optimization with mental accounts (POMA), which connects modern investment theory (MVT) and mean–variance utility (MVU) in a behavioral portfolio-style problem where the investors are concerned about downside risk. Their feasible solution is a system of implicit equations solved by numerical approximations. This article contributes several findings related to the POMA problem. First, with necessary and sufficient conditions measured by the safety-first risk management rule, we derive a closed-form solution with the threshold return relative to the orthogonal portfolio on the mean–variance frontier. Second, we show that POMA feasibility can be substantially improved through stepwise regressions in which sorting assets by the value-at-risk (VaR) constraint is equivalent to sorting the coefficients’ t-statistics. Finally, we implement mean–variance efficiency testing from the VaR perspective.