Commodity indexing - A complex benchmark problem

The benchmarks for equities and fixed income are generally very well established, but determining the appropriate benchmark is still one of the most important strategic issues for a pension manager. Most investors will employ the S&P 500 for equities and use the Lehman Aggregate for domestic fixed income. These benchmark standards are well accepted and deviations are undertaken for specific reasons to exploit market features such as a focus on small capitalization or international. Customization has increased, but still has often been described in terms of its relative merits versus the core benchmarks.



There has been less research on analyzing the appropriate benchmarks for commodities and with more benchmarks there has been increased positioning to create a better alternative. There have been good comparisons of the major benchmarks to show their characteristics and performance, but there has been less discussion on what should constitute a good benchmark. For example, should the driver for allocation weights be liquidity, diversification, or determined by production? How should the total return be measured for the index? What should be the roll convention for futures trading?All current benchmarks have valid reasons for their construction, but a single standard has not been clearly identified or agreed upon by the market even though the GSCI is the market leader in volume of trading.

These construction differences can lead to significant difference in the return performance between these benchmarks. Look at the difference between the major commodity indices as listed by Bloomberg. For the last year, there is over a 19% difference between the Goldman Sachs Commodity Index and the Dow-Jones AIG Commodity Index. Over a five year period, the difference is only slightly less than 200 bps per year, but still relatively large. Consequently, the strategic choice of the benchmark will have a large impact on performance. This strategic choice may be greater than the alpha added by a active manager if there are tight restrictions on the level of deviation from the benchmark.


The large performance differences are caused by three major factors:

• low correlation across the major markets within the index;
• higher volatility in many markets than found in equities;
• the limited number of commodities traded in the index.

The low correlation means that smaller changes in the allocation of a benchmark will have a larger impact on the total return of the index. Each commodity has a higher degree of uniqueness than what would be found with a stock or bond index. While there is relatively high correlation in the energy markets where all of the futures markets either represent the core commodity crude oil or refined products of crude. The greatest differences in correlation in the energy sector occur between natural gas and crude oil.

In other sector groupings there are also large differences in correlation. For example, the base metals sector have seen significant differences in performance with copper rising significantly over the last year and aluminum only increasing slightly. In some case, the correlation of between seemingly close substitutes is not high. For example, the price correlation between London (robusta)and New York (arabica) coffee, albeit different beans, is only .58 over the last year. The price correlation between London (white) and New York (#11 world) sugar was .69 over the last year. The price correlation between Chicago (soft red winter)and Kansas city (hard red winter) wheat is higher at .81 for the last year.

This low correlation is coupled with the high volatility for many markets to create more distortions across indices. There will be significant differences in the total return across many of these markets over a short time period. For example, with coffee or any agricultural product that is subject to a supply failure, there is the opportunity for large price changes. The lack of substitutability means that supply shocks could create large price moves. Nevertheless, there is a tendency for mean reversion in the longer run for commodities. This can explain why the one year total returns may differ so substantially from the five year performance numbers.

While volatility for many equities may be higher than what is found in many commodities markets, the impact of volatility is greater when there is less correlation across the markets. Currently, there is over a 50% spread between the best performing commodity market, nickel which is up just under 38% for the year and zinc which is down over 26%. Both are in the base metals sector of the market.

Finally, the lack of markets means that the allocation to any one market will be greater than what is seen in either equity or fixed income indices. For example, the large allocation to energy markets is difficult to reduce beyond a specific level because the index would then face liquidity constraints.

There has to be greater awareness and sensitivity to benchmark issues which may suggest the need for customization in the commodity space. Additionally, there is greater opportunity for active management to offset some of the problems with benchmark construction. Slight tilts or adjustment in the indices will provide opportunities for return enhancement.