Commodity return harvesting – Blend the factors to get a better return to risk mix

Research over the last few years has found that a number of factors can explain return performance in commodity futures but there has not been any exhaustive analysis of how to blend all these factors within a portfolio. This is the critical issue for many commodity investors; improving the return profile through mixing styles along with markets.

In the recent paper “Harvesting Commodity Styles: An Integrated Framework", the authors Adrian Fernandez-Perez, Ana-Maria Faeroes, Joelle Miffre has looked at a diversified portfolio of 28 markets, 11 different factor models, and 6 different portfolio weighing scheme. This is important investment management work for anyone who wants to blend commodity risk factors within a portfolio.

Their conclusion is simple and straight-forward, equally weight all factors and you will get a more attractive return to risk portfolio than looking at any one or two factors.

There are a large number of factor models that can be used to explain returns in commodity markets but few researchers have tried to integrate them in a single portfolio structure.  Some of these are classics across all asset classes like carry, value and momentum but there are also a number that are more unique to commodities. The issue for this paper, which may be true for any asset class, is determining how to blend factors. The authors offer a set of integration approaches that can be applied to both long-short and long-only portfolios. The risk factor models tested include:

• Term structure (backwardation/contango)
• Hedgers’ hedging pressure
• Speculators’ hedging pressure
• Momentum
• Value
• Volatility
• Open Interest
• Liquidity
• Dollar beta
• Inflation Beta
• Skewness 

The return to risk characteristics of these factors are dynamic and with some going negative during the large commodity down cycle. There is deterioration when they are blended but the general view is that blends will provide diversification benefit.

Each of the models can be formulated as separate long-short portfolios but the integration questions really focuses on how these factors are blended in a portfolio. The integration of factors models is one of the key portfolio building issues facing investment managers. The authors test six different blending mechanisms;

• Equally Weighted Integration (EWI)
• Optimal Integration (OI)
• Rotation of Styles Integration (RSI)
• Volatility of Styles Integration (VTI)
• Cross-Sectional Pricing Integration (CSI)
• Principal-Components Integration (PCI)

The numbers show declines over the down portion of the super-cycle but the equal weighted outperforms the other combinations. 

The research finds the equal weighted approach does better than those dynamic approaches that may be subject to estimation risk. The individual risk factors all have high volatility and variable performance and the combinations do not show stability. The lesson is clear, use the combination of factors available but keep it simple when it comes to portfolio integration. The value-added even exists after accounting for transaction costs. 

Unfortunately, the equally weighted portfolio did not always outperform the GSCI index. The Sharpe ratio between 1992-2016 was negative while the index was slightly positive. While it is able to better control risk, the factor model cannot generate positive returns in all sub-periods. More work on how to properly manage a commodity portfolio needs to be done but this research provides an improved roadmap.

.