26.8 PERFORMANCE ENHANCEMENTS OF ENHANCED COMMODITY INDICES

A variety of enhanced commodity indices are available, which claim to provide return improvements over the first generation of commodity indices without increasing risk. As discussed earlier, enhanced commodity indices may be able to enhance returns by increasing weights to commodities that have recently performed well and underweighting commodities that have recently performed poorly, to capture short-term momentum. They may also seek to maximize roll return by increasing weights in commodities that are in steep backwardation and by underweighting those in steep contango. Furthermore, they could gradually roll their futures contracts or utilize longer-maturity contracts to minimize the price impact of their roll activities or reduce volatility. Unfortunately, validating the performance improvement due to the methodology enhancements of these indices is far from trivial. As Erb and Harvey (2006) point out, commodity futures markets tend to exhibit high volatility and very low correlations with one another. As a result, the returns of commodity indices have been driven by a few commodities that historically performed extremely well. This makes ex-post comparisons between index methodologies difficult, since a high weight in a commodity that happens to perform extremely well can have a large impact on returns.

To analyze the potential improvements of enhanced commodity indices while addressing this issue, Fuertes, Miffre, and Rallis (2010a) compare two first-generation indices (S&P GSCI and DJUBSCI) with enhanced versions of the indices. To create the enhanced versions of the indices, Fuertes et al. utilize the weighting and rolling methodology of the first-generation indices as a baseline, and adjust the weighting or time to maturity of the futures contracts. The weighting is adjusted based on a momentum signal (overweight the winners), term structure (overweight commodities with higher roll return), or a combined momentum and term structure signal to create reweighted enhanced indices. The time to maturity is also altered to create enhanced indices that retain the baseline weights. Fuertes et al. examine the performance of the base indices and the enhanced versions of the indices over the period from October 1988 to November 2008 for the S&P GSCI, and the period from January 1991 to November 2008 for the DJUBSCI. The authors find that all of their hypothetical enhanced versions of the S&P GSCI and DJUBSCI improve the returns of the first-generation indices, providing significant positive alphas. The largest improvement comes from the time to maturity enhancement, with the longer-maturity versions exhibiting the highest performance. As important, the authors find that the enhanced indices provide effective risk diversification and inflation hedging comparable to the underlying first-generation indices.

26.8.1 Second-Generation Enhanced Commodity Indices

The second-generation commodity indices attempt to enhance returns through forward curve positioning. In contrast to first-generation indices, second-generation indices tend to spread the roll period or target different segments of the forward curve. For example, the Bache Commodity Index (BCI) spreads its roll over its entire holding period by rolling a small portion of its positions every day, gradually rolling the entire position of each commodity from one contract month to the next further out.

The Merrill Lynch Commodity Index eXtra (MLCX) differs from the S&P GSCI and DJUBSCI in its longer average maturity. The MLCX rolls over a period of 15 days, from the first to the fifteenth business day of the rolling month, and rolls from next to second next contract (i.e., from the second contract to the third contract) instead of the more conventional front to next contract (i.e., from the first contract to the second contract). Rolling one month ahead of the S&P GSCI and DJUBSCI gives the MLCX an average maturity of about one month longer than the DJUBSCI and six weeks longer than the S&P GSCI.

The Deutsche Bank Liquid Commodity Index (DBLCI) Optimum Yield Index uses fixed commodity weights similar to first-generation indices, but employs a variable curve positioning strategy. The particular contract month selected for each commodity is determined by selecting the contract with the highest implied roll yield from all contracts expiring in the following 13-month period (Dunsby and Nelson 2010).

26.8.2 Third-Generation Enhanced Commodity Indices

The third generation of commodity indices adds yet another enhancement: active commodity selection. Active commodity selection may be algorithmic (such as momentum, inventory levels, term structure signals, etc.) or discretionary. For example, the UBS Bloomberg CMCI Active Index invests in the same commodities as the standard UBS Bloomberg CMCI Index, but varies the weights and tenors of individual commodities based on UBS research analysts' performance expectations for each contract.3