Shadow assets, such as human capital, are nonfinancial assets owned by investors that are nontradable and exogenous to the investor’s asset allocation decision. Traditional mean–variance portfolio optimization typically excludes shadow assets. The author expands the traditional mean–variance portfolio optimization framework to include shadow assets and concludes that optimal allocations can be quite different if shadow assets are introduced.

The author shows that the introduction of shadow assets in the standard allocation model increases total wealth and traditional mean–variance speculative demand for speculative assets. Examples of shadow assets include human capital and corporate pension promises for individual investors, underground oil reserves for sovereign wealth funds, and future alumni contributions for university endowments. In a mean–variance framework, the attractiveness of a traditional financial asset depends on the covariance between it and the investor’s shadow asset(s). Financial assets that covary strongly with shadow assets are less attractive to investors because the volatility of aggregate wealth increases for a nondiversifying financial asset.

Using the asset liability framework with shadow assets, the author shows that asset demand is a weighted combination of traditional mean–variance speculative demand, shadow asset hedging demand, and liability hedging demand. Financial assets are held if they show attractive (standalone) risk–reward trade-offs, if they help reduce fluctuations in shadow assets (negative covariance with shadow assets), or if they help hedge liability-related risks (positive covariance with liabilities). Paradoxically, the existence of positive (negative) shadow assets increases (decreases) the aggressiveness of an investor regardless of their covariance with financial assets.

This research is useful to wealth managers and sovereign wealth funds, among other fiduciaries that, for the client’s best interests, need to incorporate financial assets, shadow assets, and liabilities in their recommendations.

The author generalizes the traditional mean–variance optimization approach to include shadow asset and liability betas in a one-period nonstochastic framework.

More than 50 years after the introduction of utility theory and the mean–variance framework, it seems that a dynamic and stochastic multistage approach to individual financial planning is long overdue. The last few years have produced very interesting research on the subject, through the use of simulation, optimization, and dynamic stochastic programming. Incorporating shadow assets in these models, as suggested by the author, is one more step in the right direction.