The last decade has seen a resurgence of interest in commodity futures as an investment class, particularly from institutional investors. Their commodity allocations have grown dramatically with total commodity-linked assets under management rising to $257 billion in 2009 from $18 billion in 2003, according to Barclays Capital.
The vast majority of these investments are linked to long-only commodity indexes, with exchange-traded funds (ETF) and exchange-traded notes (ETN) based on single commodities such as crude oil, also attracting considerable interest. Much of this popularity has been fed by the notion of the commodity supercycle as propounded by Jim Rogers. The potential for inflation hedging also has been a strong draw.
However, correlations with traditional asset classes, such as stock indexes, are rising, and issues with the shape of the futures curve have meant that indexes based on futures, as well as single commodity ETFs, have failed to capture rises in spot prices. Several markets, such as crude oil, have been in contango where the spot contract is a discount to further out contracts. This acts as a drag on returns of a long commodity portfolio. For example, owners of shares of the USO crude oil ETF lost 0.7% in 2010 because of the persistent contango, although the front-month WTI contract gained 15% over this period.
Actively seeing alpha
Although the popularity of long-only indexes remains strong, there is an increasing awareness that active approaches to commodity investing might perform better.
We can construct a new class of active dynamic strategies designed to capture normal backwardation in commodity futures markets. Normal backwardation happens when the expected futures spot price is higher than the current spot price, and hence a long position in the underlying futures contract is likely to generate a positive return.
The original hedging pressure hypothesis of Keynes (1930) was that commodity futures markets always were in normal backwardation, as they served an insurance function, allowing producers to transfer price risks to speculators who would thus earn a risk premium. The Keynesian hedging pressure hypothesis provides the theoretical justification for long-only commodity indexes and single commodity ETFs. However, the empirical evidence, both long-term and current, strongly suggests that the markets are not always in backwardation and hence capturing phases of backwardation is crucial to commodity futures investment.
Our strategies are long flat weekly rebalanced strategies, in that every week either they go long an individual commodity futures contract or do not invest. We assume the underlying position is fully collateralized, but consider the performance of the timing aspect of the strategy and do not incorporate the return due to investment in Treasury bills. The return to the strategy hence may be regarded as an excess return.
Our measure of backwardation is based on the position of large hedgers and is estimated using the Commodity Futures Trading Commission’s (CFTC) Commitment of Trader’s (COT) report. We consider two kinds of backwardation — individual and aggregate. Individual backwardation is an "intrinsic" property of a commodity, which has been shown to be a determinant of future returns. Aggregate backwardation is a more recent phenomenon and refers to the increasing correlation across different commodities, which could be the result of the increasing "financialization of commodities." Our strategies endeavor to capture both of these phenomena to decide when to invest.
Data and strategy
We consider 10 of the most liquid commodities: copper, corn, crude oil, gold, live cattle, natural gas, silver, soybeans, sugar and wheat.
The basic source for hedging pressure data is the CFTC website (www.cftc.gov). It releases the COT report each Friday at 3:30 p.m. EDT. The positions refer to the Tuesday of that week, and the date reflects that. The aggregated data, which is available at a weekly frequency since 1993, is now released under the Legacy Reports (see weekly Market Pulse).
For this analysis, we need the number of long and short positions held by commercial hedgers. Commercial hedging pressure (CHP) is the ratio of long positions to the sum of long plus short positions held by commercial hedgers.
The individual commodity futures returns and the commodity index returns are based on end-of-day prices, aggregated to a weekly frequency. The futures returns are based on the front-month contract, except for the expiry month in which the next-to-front-month contract is used. The data source is Bloomberg.
We use the notion of "relative" backwardation in designing our strategies. This means we look to go long when commercial hedging pressure is low relative to the recent past. To implement this strategy, we need to decide what constitutes the recent past and what level of hedging pressure is considered to be low. The most natural time period based on harvest and storage considerations is one year, or 52 weeks. The levels for hedging pressure are based on estimated levels for "individual backwardation" and the first strategy invests when the current hedging pressure is below this level. The second strategy uses a predictive variable that measures aggregate backwardation and invests when this level is high.
Results are based on real-time out-of-sample analysis. Because strategies are executed via futures, the notional cost is zero and the basic assumption is that the investment is fully collateralized with the value of the underlying futures contract. The strategy return is reported in excess of Treasury bills and can be interpreted as an excess return.
We first analyze the performance of buy and hold strategies for the 10 commodities as well as an equally weighted portfolio over the 2005-10 period. This six-year period incorporates a bull phase (2005 to mid 2008), a short sharp bear phase (second half of 2008) and another possibly bull phase (2009-10). Nine of the 10 individual commodities achieve positive returns as shown in "Market returns" (below), with Sharpe ratios ranging from -0.9 (Natural Gas) to 1.15 (Gold).
The equally weighted portfolio is the best overall performer in terms of Sharpe ratio because of diversification, achieving a Sharpe ratio of 1.28. Individual commodity returns are volatile, and the maximum drawdowns for eight of the 10 commodities exceed 50%, with the equally weighted portfolio having a drawdown of 48%. Thus, even in a predominantly bull phase, commodity futures returns are volatile and buy and hold investments can incur sharp losses. The Sharpe ratios are quite high over this period and higher than equity investments in many cases, but the volatility and drawdowns are considerably higher.
Our strategy performance over the 2005-10 period is shown in "Long strategy returns" (below). Because it is a long-only strategy, the appropriate benchmark for comparison is the Sharpe ratio. Seven of the active strategy Sharpe ratios are equal to or higher than the buy and hold, and of the three that are lower, corn, soybeans and wheat, two are slightly lower. Also interesting, six of the seven strategies that outperform have higher mean returns than the buy and hold. Thus, even over a predominantly bull phase, several of our backwardation timing strategies are able to achieve higher mean returns, suggesting that timing backwardation has considerable economic benefit.
The other benefit of our long flat strategies, which is more intuitive, is the reduction in volatility and drawdown. The reduction in drawdowns is particularly interesting, being lower for all commodities, and dramatically lower for copper and crude, for which the strategy Sharpe ratios are much higher than buy and hold. The equally weighted portfolio achieves a slightly higher mean, somewhat higher Sharpe ratio and a much lower maximum drawdown of 15%. This equally weighted portfolio has the second moment (volatility and drawdown) characteristics of an equity type investment, while achieving a high mean.
We now focus on the 2008-10 period, covering the period of the financial crisis. "Crisis performance" (below) shows the performance of the buy-and-hold strategy over this period, and the decline is evident. Three of the 10 commodities have negative returns and the Sharpe ratio of the equally weighted portfolio is 0.51. The maximum drawdowns all occurred over this period, and the average volatilities are higher. Not all of the commodities performed worse, though, with sugar achieving a higher Sharpe ratio over this period.
In sharp contrast, the performance of the active strategy is not much different from that over the entire period (see "Stable under pressure"). Five of the 10 individual long flat strategies have higher Sharpe ratios over the 2008-10 period than over 2005-2010, indicating that a successful timing strategy could have made good returns even in this period. The dynamic strategies outperform their buy-and-hold counterparts both in terms of Sharpe ratio as well as mean return. The equally weighted portfolio achieves a Sharpe ratio of 1.46, slightly lower than that over the entire period, but considerably higher than the buy and hold over this period. The difference in performance over this period, which includes a short sharp bear phase, illustrates how important timing backwardation successfully could have been.
We can compare the performance of our dynamic equally weighted portfolio to that of the Barclays commodity trading advisor (CTA) index over the 2008-10 period. The geometric mean returns for the Barclays CTA index were 14.1% in 2008, -0.1% in 2009 and 7.0% in 2010, indicating that managed futures as a category weathered the financial crisis quite well.
Our equally weighted portfolio had mean returns of 32.1% in 2008, 23.8% in 2009 and 20.7% at probably similar levels of volatility (around 16%). Thus, our portfolio’s outperformance is particularly marked in 2009 when the average CTA did not perform particularly well, and overall the results seem to indicate that the "commodity supercycle" quite likely is still intact. Nonetheless, capturing backwardation seems to be quite important in any market environment.
The recent performance of commodity indexes and commodity ETFs shows that capturing the benefits of a commodity supercycle are not straightforward because of the presence of phases of backwardation. Indeed a more appropriate title for Jim Rogers’ book might be "Backwardation Now." Backwardation is present both at the level of the individual commodity and increasingly across the cross section of commodities. Timing backwardation, even over a bull phase, can provide considerable economic benefits. We have described the performance of a class of strategies that tries to capture backwardation, and our results provide some indications as to the magnitude of these benefits.
Devraj Basu, with the SKEMA Business School, can be e-mailed at email@example.com.