Recent market volatility has revived the debate on the role of commodities in strategic and tactical asset allocation and led to an increasing recognition of the relevance of long/short strategies.

The rise in commodity prices over the last 10 years and their recent volatility has generated considerable interest on the part of investors, regulators and policy-makers. Attracted by the prospect of robust returns, diversification benefits and potential for hedging inflation and macroeconomic risks, investors have increased their allocations to commodities over the period, primarily via passive investment into commodity futures indexes.

While the potential for alpha generation through long/short dynamic trading in commodity futures markets has been recognised, the focus of the attention has been on passive portfolios of long positions in futures owing to the long-only nature of traditional market indexes. Recent market gyrations have contributed to reviving the debate on the role of commodities in strategic and tactical asset allocation and led to an increasing recognition of the relevance of long/short strategies.

Some market participants and policymakers have been quick to associate the strong inflows into commodity investments with the recent commodity spikes of 2007-08 and 2009-11. The increased participation of financial investors in commodity markets has also led to concerns about their possible increased integration with traditional financial markets, which could have resulted in the weakening of the diversification benefits from commodity investments.

Recent research by Edhec-Risk Institute with the support of CME Group1 has shed new academic evidence on these concerns with particular emphasis on long/short commodities futures investing. The first contribution is a study of the performance and risk characteristics of long-only commodity portfolios and of long/short commodity strategies implemented by managers with a focus on commodities, such as commodity trading advisers (CTAs) and commodity pool operators (CPOs). The research investigates the conditional volatility of commodity futures investments and their conditional correlations with traditional assets.

The strategic decision to include commodity futures in a well-diversified portfolio does not solely depend on the risk premium of commodity futures viewed as an asset class but is also driven by a desire for risk diversification and thus depends on how the returns of commodity investments correlate with the rest of the investor’s portfolio over time.

The research analyses the conditional volatility of long-only and long/short commodity portfolios and the conditional correlations of their returns with those of traditional assets in periods of high volatility in traditional asset markets. The research extends the literature, which focused on the conditional correlations between long-only commodity portfolios and traditional assets.

The second contribution of the research is to look at the possible impact long/short trades may have had on the volatility of commodity prices and on their conditional correlation with traditional assets. This topic is of interest given the recent debate among politicians and policy makers surrounding the financialisation of commodity futures ­markets.

The 2009 staff report by the US Senate permanent subcommittee on investigation argues that commodity index traders were disruptive forces, driving prices away from fundamentals. If established, this would support calls for an increase in transparency, position limits and margins to curb excessive speculation and, it is hoped, volatility.

Since then the claim that the financialisation of commodity markets is responsible for the observed volatility in commodity prices has been the subject of an intense academic debate. The overwhelming conclusion has been that it is not possible to empirically link investments in commodity futures and commodity futures prices.

However, the debate has so far focused mainly on the potentially destabilising role of long-only as opposed to long/short, investors. Our research fills the gap by studying the potentially unsettling role of long/short investors on price volatility and cross-market correlation.

Mimicking patterns
To address these objectives, we first start by mimicking the trading behaviour of long/short participants in commodity futures markets over the period 1992-2011. This is done by implementing a battery of long/short strategies where these strategies are based on a momentum signal, the slope of the term structure, a double-sort that combines momentum and term structure signals or on the positions of commercial traders (also often termed “hedgers”) and non-commercial traders (also often referred to as ­‘speculators’).

The hedger/speculator distinction is not clear since commercial traders can hold views on the market and take speculative positions accordingly. Non-commercial traders are generally investors seeking exposure to commodities without holding the underlying. Hedging motives may nonetheless motivate them.

The rule-based strategies we implement specify minimum liquidity requirements for commodity contracts and then include in a long/short portfolio the commodities with the strongest momentum, the highest absolute roll-returns or the most extreme hedging pressures.

The dataset spans October 2, 1992 to March 25, 2011 and includes Friday settlement prices for 27 commodity futures as obtained from DataStream International. The frequency, time series and cross-section are chosen based on the availability of the positions of commercial and non-commercial traders in the Commodity Futures Trading Commission (CFTC) commitment of traders report (COT).

The cross-section includes 12 agricultural commodities: cocoa, coffee C, corn, cotton number 2, frozen concentrated orange juice, oats, rough rice, soybean meal, soybean oil, soybeans, sugar number 11, wheat; five energy commodities: blendstock RBOB gasoline, electricity, heating oil number 2, light sweet crude oil, natural gas; four livestock commodities: feeder cattle, frozen pork bellies, lean hogs, live cattle; five metal commodities: copper, gold, palladium, platinum, silver; and random length lumber. The positions of commercial and non-commercial traders are collected every Tuesday and made available to the public the following Friday.

Futures returns are calculated by assuming that investors hold the nearest contract up to one month before maturity and then roll their position to the second nearest contract where the rolling takes place to avoid physical delivery of the underlying commodity.

All portfolios shortlist the most liquid contracts out of the universe of commodity futures (75%) that are available at the time of portfolio construction. Buying the remaining 20% with the highest past performance and shorting the remaining 20% with the lowest past performance over a chosen observation period compose the momentum portfolio.

Out of the 75% of contracts that are the most liquid, the term structure portfolio buys the 20% with the highest roll returns (for example, the most downward-sloping term structure and sells the 20% with the lowest roll-returns or the most upward-sloping term structure.

Finally, the hedging pressure strategies use as signals for asset allocation the positions of commercial traders and non-commercial traders, as reported by the CFTC. The strategies take long positions in liquid backwardated commodities for which commercial traders were net short and/or non-commercial traders were net long as well as short positions in liquid contangoed commodities for which commercial traders were net long and/or non-commercial traders were net short.

Either the single or double-sort strategies we end up with aim at taking long positions systematically in the 15% of commodities whose prices are expected to appreciate and short positions in the 15% of commodities whose prices are expected to depreciate.

Having mimicked the returns that long/short commodity speculators earned over the period 1992-2011, we then move on to comparing the performance and risk characteristics of these long/short commodity portfolios with those of long-only index positions using both an equally weighted portfolio of the aforementioned 27 commodity futures and the Standard and Poor’s Goldman Sachs Commodity Index (S&P-GSCI) as benchmarks.

We also analyse the conditional correlations of these long/short and long-only commodity portfolios with traditional asset classes, using the Standard & Poor’s 500 Index (S&P 500) for equities and the Barclays Capital US Aggregate Bond Index for fixed income products.

Commodity domination
In terms of standalone performance, long/short commodity portfolios are found to dominate long-only commodity indexes. The average mean excess return of the single sort long/short portfolios equals 7.99% a year and that of the double-sort portfolios equals 9.03% a year.

Over the same period the mean excess return of the S&P-GSCI equals 0.64% a year and that of the long-only equally weighted portfolio of the 27 commodities included in this study is at 4.28% a year. The S&P 500 and the Barclays Capital US Aggregate Bond Index returned 4.21% and 2.95% in excess return form, respectively.

The conclusion is similar once we adjust for risk. The Sharpe ratios of the long/short portfolios average out at 0.5093 with a range from 0.2711 for the single-sort momentum strategy) to 0.6302 (for the single-sort hedger-based strategy.

These Sharpe ratios always substantially exceed those of long-only benchmarks – 0.0529 for the long-only equally weighted portfolio and 0.1965 for the S&P-GSCI. The S&P 500 and the Barclays Capital US Aggregate Bond Index had Sharpe ratios of 0.2401 and 0.6631, respectively.

In the context of multi-asset class investment, the relationship between the conditional volatility of commodity investments and that of traditional investments as well as the conditional correlations between commodity investments and traditional asset classes have important implications for risk management and ­diversification.

The averages of the conditional volatilities of the long/short commodity strategies studied here range from 15.09% to 17.61% a year and are less than the average of the conditional volatilities of the S&P-GSCI, which stands at 20.81%. These differences are statistically significant. The equally weighted portfolio of the 27 commodities is found to have a conditional volatility of 11.28%.

The conditional volatilities of the long/short commodity strategies are also found to rise by less than that of the S&P-GSCI or the equally weighted benchmark in periods of increased volatility in equity markets. Furthermore, they fall in periods of increased volatility in fixed income markets at a time when the conditional volatilities of the long-only commodity indexes rise.

Other things being equal, this is welcome news to long/short speculators as it indicates they can reduce the total risk of their multi-class portfolios in a more effective way by being long/short commodities as opposed to being long-only.

Over the period the conditional correlations of the S&P 500 with the commodity investments studied are low, confirming their strategic role as risk diversifiers. The conditional correlations modelled relative to the long/short commodity portfolios, averaging 0.00 and statistically not different from zero, are lower than those modelled relative to the long-only commodity indexes, averaging 0.19, positive and significant.

This suggests the risk diversification benefits of commodity futures are stronger within long/short portfolios.

Focusing on the period following the demise of Lehman Brothers, we observe a sharp rise in the conditional correlations between the S&P 500 and long-only commodity indexes to an average of 0.54 while those between the S&P 500 and long/short commodity strategies remain very low at an average of 0.01. This documents the much stronger diversification benefits associated with long/short strategies versus long-only indexes, in the recent past.

The conditional correlations between the long/short and long-only commodity portfolios and the Barclays Capital US Aggregate Bond Index are also found to be low over the period.

As previously reported in the literature, the correlations with the long-only commodity indexes are particularly low: -0.0737 for the equally weighted portfolio and -0.0267 for the S&P-GSCI. They are also low for the long/short portfolios ranging from -0.0324 to 0.0320 and ­averaging 0.0006.

Looking at the evidence
Statistical tests of differences between these correlations suggest that, other things being equal, bond investors are better off from a risk management perspective holding the S&P-GSCI than commodity portfolios based on past performance or on past roll-returns. The evidence is less clear cut for strategies based on the positions of commercial and non-commercial traders.

We then focus on the behaviour of conditional correlations between traditional asset classes and commodity investments when the former are under stress.

In periods of high volatility in equity markets, the conditional correlations between the S&P 500 index and the long/short commodity portfolios based on the positions of commercial and non-commercial traders are found to decrease. This is good news for equity investors as it is precisely when the volatility of equity markets is high that the benefits of diversification are most appreciated.

In contrast, the conditional correlations between long-only commodity indexes and equity indexes rise with the volatility of the S&P 500. This suggests the risk reduction that comes from diversification prevails less when needed most.

The four weeks that followed the demise of Lehman Brothers provide an acid test of these relationships. As depicted by the chart, the S&P 500 lost 8.18% a week, the long-only commodity portfolios retreated sharply dropped 4.73% for the equally weighted portfolio and fell 6.42% for the S&P-GSCI. However, the long/short commodity strategies studied here earned between a negative 1.72% and plus 2.02% a week.

These results suggest that, unlike long-only commodity portfolios, long/short commodity strategies can serve as partial hedge against extreme equity risk.

In periods of high volatility in fixed income markets, the conditional correlations between the Barclays Capital US Aggregate Bond Index and the long/short commodity portfolios based on the positions of commercial and non-commercial traders are found to remain constant, whereas the conditional correlations measured relative to long-only commodity portfolios are found to rise sharply.

This suggests that, other things being equal, long/short commodity portfolios based on the positions of hedgers and speculators can serve as better diversifiers than long-only commodity portfolios in periods of extreme risk in fixed income markets.

Our research confirms the relevance of commodity futures investment and presents reasons as to why long/short strategies should prevail over long-only investing. They provide superior risk-adjusted performance as reflected by higher Sharpe ratios and have lower conditional volatility than the leading long-only benchmark index. In addition they offer more effective diversification qualities for equity portfolios, especially in the recent period.

Furthermore, long/short commodity portfolios based on the positions of commercial and non-commercial traders are found to have partial hedging characteristics providing protection against extreme-risk in the equity markets and to offer stable diversification properties in times of turbulence on the fixed income markets.

The second purpose of our research is to present evidence on the financialisation of commodity futures markets and investigate the possible impact that long/short trades may have had on the volatility of commodity prices and on their conditional correlation with ­traditional assets.

The financialisation of the commodity futures market is portrayed in the graph, which plots the long and short open interests of non-commercial traders across the cross-section of commodities studied (the number of contracts is shown on the left side) and the level of the S&P-GSCI (the index level on the right side).

Two points are worth noting. First, the dramatic changes in the long open interests of non-commercial traders seem to parallel the dramatic ups and downs of the S&P-GSCI over the period 1992-2011. This gives credibility to the claim that changes in the long open interests of investors could have increased the volatility of the S&P-GSCI.

Second, both the long and short positions of investors have risen sharply over the period 1992-2011, suggesting an increase in the financialisation of commodity futures markets.

We investigate whether the increased role of financial investors has been a disruptive force in commodity futures markets. This is done by testing whether the increase in the long, as well as short, interests of speculators caused a change in the conditional volatility of long/short commodity portfolios or a change in the conditional correlation between their returns and those of traditional assets.

Our Granger-causality tests are tests of the null hypothesis that changes in the positions of investors (whether long or short) do not cause changes in the volatility of the long, short and long/short commodity portfolio returns or have no impact on the conditional return correlation between the long/short commodity portfolios and traditional assets. A failure to reject the null hypothesis indicates a lack of causality.

We find no support for the hypothesis that long/short investors have destabilised commodity prices by increasing volatility or co-movements between commodity prices and those of traditional assets. Interestingly, this conclusion holds irrespective of whether investors are labelled as ‘non-commercial’ in the CFTC commitment of traders report or ‘professional money managers’ (for example, CTAs, CPOs and hedge funds) in the CFTC disaggregated Commitment of Traders (DCOT) report.

The analysis presented here does not call for a change in the regulation in relation with the increased participation of professional money managers in commodity futures markets.

The research from which this article was drawn was produced as part of the research project on “Exploring the Commodity Futures Risk Premium: Implications for Asset Allocation and Regulation” at Edhec-Risk Institute sponsored by CME Group.

Joëlle Miffre, professor of finance at Edhec Business School is the author of this article.


Footnote
1 Miffre, J, “Long-Short Commodity Investing: Implications for Portfolio Risk and Market Regulation,” EDHEC-Risk Publication, August 2011. This research was conducted with the support of CME Group.