The article was published in the August issue of Morningstar ETFInvestor.
The holy grail of portfolio diversification: an asset class that provides long-term risk-adjusted returns similar to equities but is negatively correlated with the returns of stocks and bonds. That is what professors Gary Gorton and Geert Rouwenhorst (hereafter G&R) appeared to have identified in their seminal study, “Facts and Fantasies about Commodity Futures” 1.
The paper looked at the period from July 1959 through December 2004 and concluded that the risk premium earned from investing in a broadly diversified basket of fully collateralized commodity futures is very similar to the historical equity risk premium. Moreover, the authors found that the volatility of commodity futures (as measured by standard deviation) was less than that of stocks. According to the study, not only were these superior risk-adjusted returns negatively correlated with stocks and bonds, but they were also positively correlated with unexpected inflation.
Unfortunately, what seemed too good to be true at the time might have turned out to be exactly that. A decade has passed since the study was released and sent waves through the investment industry. This gives us fertile ground to examine the subsequent performance of commodity futures in the years following the persuasive research paper. As it turns out, since the research was published, commodity futures have behaved a bit differently from the period studied.
There are a few key points worth highlighting that could help explain why the results of the study haven’t persisted post-publication. First, commodity futures were predominantly in backwardation (futures prices were lower than spot prices) during the time period studied. Therefore, the roll-yield component of the commodity futures’ returns provided a fairly consistent tailwind. Second, using three-month T-bills as a proxy, the risk-free rate during the period studied was much higher than it has been recently. This, of course, had major implications for collateral returns. In the period from July 1959 through December 2004, the three-month T-bill rate averaged 5.6% and rose as high as 16.3% in the early 1980s. Since the study was released, the average rate on three-month T-bills has been 1.34%. For further context, today the rate is 0.02% and has averaged 0.06% over the past five years.
Remember, the returns of commodity futures are composed of three parts: (1) changes in the spot price, (2) roll yield, and (3) returns on the collateral securing the futures contract. Though many who allocate a portion of their portfolio to commodity futures may expect that changes in the spot price are the most important driver, this has not been the case historically.
In fact, collateral returns have been the largest return component, followed by roll yield. Spot prices have actually been a drag on overall returns historically. In a 2008 study2, Vanguard examined the sources of the returns of the S&P GSCI from 1983 through Sept. 30, 2008, and found that the collateral return accounted for 6.4 percentage points of the benchmark’s 7.1% annualized return during this period. The roll yield accounted for 3.3 percentage points, while spot returns actually detracted 2.6 percentage points annually (6.4% + 3.3% - 2.6% = 7.1%). Understanding the underlying components of commodity futures’ returns--and their impact on performance--is critical for making an informed decision on whether or not to make (or keep) an allocation to the asset class.
Past to Present—Fundamental Differences
Since the mid-2000s, commodity markets have predominantly been in a state of contango. But for the 45-year period that G&R researched for its study, commodity futures markets were consistently backwardated.
Based on Keynes’ theory of normal backwardation3, it was reasonable to assume that the roll yield would continue to be an important contributor to total returns. Given that foreseeable trends (such as seasonality) are taken into account in futures prices, expected movements in the spot price should not be relied upon as a source of return to an investor in futures. As G&R noted, unexpected deviations from the expected futures spot price are by definition unpredictable and should balance out to zero over time (a positive expected return here would assume consistently successful market-timing).
Therefore, G&R determined that the expected return for commodity futures boiled down to what they called the risk premium. If the current futures price is below the expected futures spot price (backwardation), the buyer of the futures contract, on average, will earn a risk premium. On the other hand, if the futures price is above the expected future spot price (contango), then the risk premium is earned by the seller of a futures contract.
The theory of normal backwardation is predicated on the assumption that producers of commodities would seek to hedge the price risk of unexpected deviations between futures and spot prices. So in a sense, a producer selling futures to lock in future prices is obtaining a form of insurance against price risk. The insurance is provided by speculators who buy the futures and, in theory, demand that the futures price be below the current spot price in order for them to receive a risk premium for providing insurance to the producer. It’s critical to point out that another implicit assumption of the normal backwardation theory is that the number of producers requiring hedging outweighs the number of speculators in the market.
In part 2 of this article, we will try to reevaluate the case for commodities as a long-term investment.
1 Gorton, G., & Rouwenhorst, K.G. 2006. “Facts and Fantasies about Commodity Futures.” Financial Analysts Journal, Vol. 62, No. 2, P. 47.
2 Phillips, C.B. 2008. “Commodities: The Building Blocks of Forward Returns.” (Valley Forge, PA.: The Vanguard Group)
3 Keynes, J.M. 1930. “A Treatise on Money,” Vol. 2. (London: Macmillan & Co.)
