Factor-Based Asset Allocation vs. Asset-Class-Based Asset Allocation

This article addresses the issue of the alleged superiority of risk-factor-based asset allocations over the more traditional asset-class-based asset allocation. The authors used both an idealized model, capable of precise mathematical treatment, and optimizations based on different periods of historical data to show that neither approach is inherently superior to the other. Although the authors appreciate the role of risk models in portfolio management, they urge caution with respect to unwarranted claims of their dominance.The article addresses the issue of the alleged relative superiority of risk-factor-based asset allocations versus the more traditional, asset-class-based asset allocations. The topic has been the subject of several recent articles that implicitly or explicitly claim such superiority. After a brief synopsis of the history of risk factors as such, the authors examine the logic of these claims and find it generally defective: In most cases, the alleged superiority is simply a result of mis-specified “apples to oranges” comparisons, where a robustly diversified set of risk factors is held up as a winner against a very basic asset allocation model that consists of just stocks and bonds. The authors then introduce a model in which there is a one-to-one mapping between asset classes and risk factors and show that in such an idealized setting, one can mathematically prove that neither approach is superior; the same mean–variance-optimal portfolio is returned by both approaches. (This finding is not new—it goes by the name “rotational indeterminacy of the factors”—but it seems that theoretically minded investors or researchers may have lost sight of it.) Finally, the authors use real-world historical data from different time periods to perform optimizations. It turns out that which approach would have yielded a dominant portfolio depends on the time period and the risk level selected. The authors, who appreciate the role of risk-factor models in portfolio management—asset classes are, after all, an example of risk factors themselves—conclude by urging caution against hasty, uncritical acceptance of these new claims of inherent superiority. The real advantage in portfolio management results from an ability to better model risk and reward assumptions, whether of asset classes or risk factors, rather than from simply rearranging the format of the inputs.