Although they offer the important benefits of inflation protection and diversification, commodities do not figure meaningfully in most investors’ portfolios. The authors focus on performance as one of the factors that contributes to this underweighting. They explain commodities’ performance drivers and explore ways to improve risk-adjusted returns. They then describe a new approach to constructing a long-only index that identifies attractive futures contracts and allocates more to commodities that are experiencing higher roll yields and price momentum. They demonstrate in simulation that the combination of contract selection and dynamic weighting can improve long-term performance without sacrificing the main benefits investors expect from the asset class.
Supply and demand forces in the physical market may create upward or downward pricing pressure on futures contracts that are further from expiration relative to those with earlier expirations. Commodity futures with later expirations can thus trade at a premium or a discount relative to near-dated futures contracts. Known as “contango” and “backwardation,” respectively, these conditions determine the roll yield or returns from selling expiring contracts and buying later-dated contracts. Backwardation is profitable, and contango unprofitable, to long-only index investors in the front months.
Both commodity prices and the term structure of commodity futures prices have been volatile. In the 2012–2014 period, commodities were in backwardation more often than in the 2005–2011 period, when contango was nearly constant. To achieve superior performance in volatile environments, commodity index investors need a strategy that is smarter about weighting, rebalancing, and rolling.
Holding contracts further out the curve offers the most advantageous strategy. First, front-month contracts need to be rolled often, creating higher turnover, and tend to experience higher volatility and more-negative roll yields when the market is in contango. Second, a strategy can change its allocations to different commodities over time, favoring those in backwardation and reducing exposure to those in contango. Given that roll yields are dynamic and vary significantly across commodities, this approach has the potential to significantly improve the strategy’s overall roll yield.
Existing commodity indices contain most of the key commodity futures that are liquid and representative of global production. The Standard & Poor’s GSCI uses a weight based on world production to reflect the general economic significance of commodities. The Bloomberg BCOM Index uses a 2-to-1 ratio of liquidity to world production weighting, then further limits weights by commodity, commodities derived from each other, and group. The Dow Jones Commodity Index (DJCI) is a modern version of BCOM that simply gets to the point of diversification by dropping world production from its weighting scheme; instead, it liquidity-weights commodities and equal-weights sectors. These indices invest generally in front-futures contracts.
In contrast, the authors’ approach to constructing a long-only commodity index moves away from the front contracts for commodities that are in contango and dynamically raises the weights of commodities that are more attractive due to their higher roll yields and greater price momentum. In designing the new index, the authors give special attention to maintaining high capacity, liquidity, diversification, and economic representation.
The universe at the base of the new index contains exactly the same number of commodities as the DJCI: 23 for the year 2014. Diversification is explicitly achieved by dividing the commodity universe into three groups—energy, metals, and agriculture plus livestock—that are then equally weighted. Using fixed weights in the three groups controls risk by reining in undue concentrations in any group, and improves long-term performance through what is known in the literature as the “rebalance premium” or “diversification return.” The rebalance premium is particularly strong in this asset class, because, first, commodities in different sectors tend to have very low, or even negative, correlations, and second, commodity prices mean revert over long horizons. Within each group, the commodities are weighted by a five-year average of dollar-volume traded, ensuring high liquidity and capacity. Overall, the stable-base weighting scheme has the further benefit of reducing costly turnover.
The authors calculate two characteristics that are widely targeted in practice by commodity trading advisors (CTAs): roll yield and momentum. In this methodology, roll yield is measured on the basis of the slope between the nearest contract and its next-year counterpart, rather than the nearest two contracts. The fixed one-year difference gives a more homogeneous measure, reduces the turnover occasioned by roll yield volatility, and eliminates misleading signals due to seasonality. Similarly, the authors calculate momentum as the ratio of spot prices for the nearest contracts today and 12 months ago.
The authors then rank each commodity on the basis of these measures and adjust the base weights accordingly to form separate roll yield and momentum portfolios. The final index weights are calculated as the simple arithmetic average between these two portfolios. Because the two return components, roll yield and momentum, provide information about time-varying market conditions of each individual commodity, it is important to measure and update them frequently. Thus, while the base weights are recalculated only once a year, the index is rebalanced once a month. Given this dynamic approach, the new commodity index could be viewed as a low-cost, high-transparency, large-capacity strategy providing an alternative to active CTA offerings that trade commodities.
In addition to weighting commodity sectors on the basis of roll yields and momentum, the method the authors use for index construction looks across the curve each month and selects contracts most likely to produce favorable results. The first step is to ensure that the index has high capacity. This is accomplished by applying three criteria to screen the universe: eligible contracts have a tenor that is less than or equal to 24 months, and average open interest that is at least US$100 million and 5% of the front contract’s average open interest.
The second step is to rank the screened contracts by implied roll yield, a measure that is specific to the term structure of the sector. If the opportunity set includes 6 or fewer eligible contracts, the rule is to select the one with the highest implied roll yield. If between 7 and 12, then the contract with the best implied roll yield is selected, unless the current contract is among the top two candidates (in which case, it is retained). Finally, if the number of liquid-candidate contracts is between 13 and 24, then the contract with the best implied roll yield is selected, unless the current contract is among the top three candidates. Formulated with the practicalities of index implementation in view, these rules are intended to select attractive contracts without causing undue turnover. Using simple arithmetic averages across sectors as a rough indicator of the improvement, there is only about half as much turnover under the contract selection methodology described in this paper (34.2%) as there is when the front contract is routinely chosen (68%).
The authors tested the performance of the new index over the period from January 1999 to July 2014. The index produced a simulated Sharpe ratio of 0.77, compared with 0.16 for the S&P GSCI and 0.19 for the BCOM Index in the same measurement period. Analysis reveals that the new index hypothetically generated competitive spot returns, but achieved a positive roll return of 2.2% compared to the two benchmark indices that produced roll returns of −6.7% and −7.1%, respectively. Over the simulation period, the new index outperformed the benchmarks in high-inflation regimes, and by eliminating negative roll returns, also fared well when inflation was low.
The hypothetical results validate the authors’ argument that performance can be improved without impairing inflation protection, all the while retaining other desirable characteristics such as large capacity, high liquidity, effective diversification, and broad economic representation. Dynamically selecting futures contracts on the forward curve, rather than simply using the most liquid nearby contracts, makes it possible to reduce exposure to contracts in contango and increase exposure to contracts in backwardation. In addition, the authors establish that the successful CTA practice of taking momentum as well as roll yield into account can be built into a transparent, rules-based process for selecting and weighting futures contracts.
Summarized by Philip Lawton, CFA