Do commodities improve traditional equity-bond asset allocations?

Master Thesis
Author
Pantatosakis, Pavlos
Παντατοσάκης, Παύλος
Date
2026-03View/ Open
Keywords
Asset allocation ; Portfolio choice ; CommoditiesAbstract
This thesis investigates whether the inclusion of commodities can improve the performance of traditional equity-bond portfolios in an environment of shifting macroeconomic regimes, using data from 1980 to 2021. While the 60/40 stock-bond allocation has historically provided diversification, the tendency of stock-bond correlation to increase during periods of inflation uncertainty and market stress suggests a need for alternative asset classes. We create a dynamic, out-of-sample investment strategy using Time-Weighted Least Squares (TWLS) and an Exponentially Weighted Moving Covariance (EWMC) matrix. This approach allows portfolio weights to adapt to potential structural breaks by prioritizing recent data. Four commonly used macroeconomic predictors are employed to estimate conditional expected excess returns, subject to theoretically motivated sign constraints on the slope coefficients. We evaluate three generations of commodity indices to determine the impact of index construction on portfolio efficiency. The results indicate that while first-generation indices provide limited value, third-generation indices offer diversification benefits, which stay robust after various tests. Specifically, the Certainty Equivalent Return (CER) test confirms that the addition of commodities to a stock-bond portfolio provides statistically significant (p-value < 10%) benefits for risk-averse investors. Ultimately, this study constructs a practical allocation strategy in real time and demonstrates that commodities serve as a tool for enhancing portfolio resilience and capital preservation.


