Commodity futures investing is arguably the most misunderstood asset class in the financial marketplace.We want to change that state of affairs.
Commodity futures strategies are fascinating and can be beneficial to investors. However, commodity futures are often complex, involve leverage and significant volatility, and have suffered periods of poor performance for decade long periods (the last 10 years is a good example).
Nonetheless, because of the unique properties of the asset class, sophisticated investors should at least consider commodity futures as part of their investment toolkit, and at the very least, consider the exploration into commodity futures as part of a dedication to education and becoming a better investor.
We begin our exploration with two simple questions that investors should consider regarding all investments:
- What are they buying?
- Why are they buying it?
This post helps investors check both of those boxes if they choose to invest in commodity futures based investment strategies.
An Introduction to Commodity Futures
Commodity futures investing can appear to be deceptively simple, and investors can have basic misunderstandings about the asset class. For example, many investors (to include the great investors at GMO) believe that investing in commodity-producing equities allows them to replicate the results of investing in similar commodity futures. If this were true it would be great on many levels — more transparent, more liquid, more tax-efficient, and less complex. But there is a problem — one can’t replicate commodity future trading strategies via equities because there are 2 elements of commodity futures: spot returns and roll yield return.
We’ve posted here and here about this reality. And while commodity equities and commodity futures can be similar, there are profound differences when one digs into the weeds. What commodity futures can provide for a diversified portfolio is also very different from what commodity equities provides for a diversified portfolio (i.e., more equity risk).
In many respects, commodity futures offer variations of commodity-related exposure, offerings premiums that are only loosely or partially related to commodity prices themselves. For example, we posted here about a Fuertes et al. paper that describes a strategy that combines momentum and term structure (e.g., roll yield) signals.
There have been a number of papers over the past decade that have explored commodity futures returns from different perspectives. By examining the literature, some competing views emerge, but collectively they contribute to a better understanding of the asset class.
Back in 2006, Gorton and Rouwenhorst (GR) wrote a seminal paper that showed that an equally-weighted index of commodity futures from 1959 to 2004 had equity-like returns, performed better in periods of unexpected inflation, and had a negative correlation with traditional asset classes. In short, there seemed to be a lot to like about the asset class, especially as a diversifier.
From the paper:
Fully-collateralized commodity futures have historically offered the same return and Sharpe ratio as equities. While the risk premium on commodity futures is essentially the same as equities, commodity futures returns are negatively correlated with equity returns and bond returns.
Similar performance, but with negative correlation to equity and bonds? This is an asset allocator’s dream come true!
Backtests Work…Until they Don’t!
Of course, for the decade following the results from the GR paper, some researchers identified that the returns to commodity futures stunk it up.
For example, Erb and Harvey (2015) observe that long-only commodity returns for the Bloomberg Commodity Index and the S&P GSCI commodity index over the period were -4.6% per year, which was a much lower return than for stocks and bonds. The chart below highlights the poor performance over the most recent decade. Yuck!
Graph from Erb and Harvey Highlighting Poor BCOM/GSCI Commodity Index Returns
Erb and Harvey note that a key driver of these returns was a poor “income return” over the recent period (roll yield plus collateral return). Erb and Harvey leverage this fact to make the reasonable argument that investors expecting to earn a “safe” carry return from investing in commodity futures may be disappointed. Periods of poor income returns from commodity investments are certainly possible and can be expected to last for fairly long (10 year) time frames. In short, commodity futures investing is definitely risky!
Wait…Maybe the Backtests Were Correct?
Erb and Harvey paint a picture that perhaps investor’s should question the historical performance of commodity futures. Perhaps the historical returns were an artifact of data-mining and/or a lucky streak in commodities during the original GR paper time period?
To combat the arguments in Erb and Harvey (2015), Bhardwaj, Gorton and Rouwenhorst wrote a 2015 paper that found that 10-years of additional data (2005-2014) did not change their basic conclusions from Gorton and Rouwenhorst (2006): The average risk premium to an equal weight commodity futures index during this out-of-sample period was 3.7% per annum, which is comparable to its long-term in-sample average of 5.23% from 1959-2004. The authors attributed much of the lower out-of-sample returns to lower collateral returns. The chart below highlights their core result that commodity futures haven’t been wonderful the past decade, but they haven’t been a tragedy either:
Graph from Bhardwaj, Gorton and Rouwenhorst Highlighting Equal-Weight Commodity Index Returns
Two Research Groups with Two Different Conclusions: What gives?
Erb and Harvey find that commodities “don’t work” post GR, whereas, Bhardwaj, Gorton and Rouwenhorst (2015) highlight that the performance of commodity futures is arguably consistent with their historical estimates.
Why are the conclusions so different? The different results are driven by different methodologies.
In the Erb and Harvey paper, the authors look at indices that are weighted by trading volume or macroeconomic factors. Indirectly, this portfolio construction skews the results to overweight the experience of energy futures (oil in particular).
A heavy exposure to energy futures would also explain why Erb and Harvey calculate such a large negative “roll yield” estimate relative to the results in Bhardwaj, Gorton and Rouwenhorst. Oil futures structurally moved to a strong “contango” state, following the 2008 crisis. The intense negative roll yield prevented oil futures from capturing the gains in the spot returns (red line in the chart below versus the blue line).
The results are hypothetical results and are NOT an indicator of future results and do NOT represent returns that any investor actually attained. Indexes are unmanaged, do not reflect management or trading fees, and one cannot invest directly in an index. Additional information regarding the construction of these results