John Kemp is a Reuters market analyst. The views expressed are his own.
Commodity markets are becoming more efficient at processing information about current supply/demand conditions and expectations about the future. But in the process, excess returns captured by investors for assuming price risk are vanishing.
The poor performance of first-generation passive commodity indices since 2005 is well known as the extra money allocated to the asset class has eroded roll returns. But there is some evidence returns on second- and third-generation indices with dynamic rolls and commodity allocations are also starting to evaporate as more money chases them.
“Commodity markets are coming of age,” according to a recent research note by Citigroup. “Simple strategies to make money in commodities are no longer likely to yield good returns.
“Our guess is that this is down to the massively increased investment in the asset class in recent years, which has driven returns lower,” according to the bank’s global strategy team.
Citigroup based its conclusions on a dynamic strategy of filtering a set of commodity futures for roll returns and then again for momentum before buying the five with the best roll and momentum and selling the five worst (”Global Macro Strategy - Commodity markets are becoming ever more efficient”, Jan. 3).
According to the Citi strategists, “returns in the April to December 2008 period were 40 percent, in the 2009 calendar year were 22 percent, in 2010 were 9.5 percent and in 2011 were 5.4 percent. Every year, the return seems to halve.”
Back-Testing and Achievable Returns
The Citi findings are intuitively plausible. They are also consistent with the observed changes in commodity prices and futures curves since investing in the asset class started to become popular in 2005.
Charts 1 and 2 show long-run returns on a diversified first-generation index (the GSCI Light Energy Index) and two dynamic indices that could be considered second- or third-generation products (the SummerHaven Dynamic Commodity Index and the Deutsche Bank Liquid Commodity Index Mean Reversion Enhanced strategy)
Since the 1990s, dynamic strategies have vastly outperformed passive allocations. Chart 3 shows the degree of outperformance persisted in the early years of the commodity investment boom (2005-2008). But as Chart 4 illustrates, outperformance has vanished since the beginning of 2009. Dynamic strategies have been no more successful in delivering excess returns than their passive counterparts in the last three years.
Dynamic indices are suffering the classic back-testing problem. Index creators mined data on past futures prices to select the strategy that maximizes returns. But once investors started to implement the resulting strategies, the pricing anomalies that were the source of the historic returns vanished.
No one was actually implementing dynamic index strategies in the 1990s, when the back-fitted index data show impressive gains. In the mid-2000s, only a very small amount of money was tracking these strategies; most investors were still using passive approaches.
Once returns on passive indices started to disappoint from 2008, and significant amounts of money were allocated to dynamic strategies or hedge funds, returns on dynamic strategies promptly collapsed. Once a significant number of investors tried to exploit market inefficiencies, the inefficiencies disappeared.
Risk Premiums and Market Efficiency
The academic literature is divided about the theoretical nature of expected returns to investors in commodity futures. Some treatments emphasize the concept of hedging pressure (more producers than consumers seeking to hedge their price risk) resulting in “normal backwardation”. Others focus on the idea of a risk premium investors capture from hedgers. In practice, there is little difference. All approaches suggest hedgers (on net) must pay a premium to investors to encourage them to assume unwanted price risk.
In their famous paper on “Facts and Fantasies about Commodity Futures”, published in 2004-2005, Gary Gorton and Geert Rouwenhorst characterized the excess return over bonds as a “risk premium”. They found that an investment in a fully collateralized equal-weighted basket of commodity futures between 1959 and 2004 provided about the same return as equities.
But their study seems to have suffered from its own back-testing problem (or fallen victim to its own success). For most of the period Gorton and Rouwenhorst studied, commodity markets were small and illiquid, with only a relatively small volume of investment. Returns to investors were correspondingly high. But once the asset class became popular, more investors were chasing the same source of returns, liquidity increased and premiums shrank.
Increased market efficiency is expected. But we need to be careful what we mean. Efficiency means less persistence and predictability in both outright prices and term structure (roll returns), which is what has undercut both the first-generation and second/third-generation products.
It does not mean markets are infallible at predicting the future or even that they are terribly rational in how they react to news in the short term. Markets are social communities, not mathematical models. As John Maynard Keynes understood (but many of his successors appear to have forgotten), markets are voting machines, not weighing machines.
Faster Price Dynamics
Most participants at the recent OPEC-IEA International Energy Forum in Vienna, Austria, agreed oil (and other commodity) prices could be driven by speculation in the short term, over-reacting to developments with mini-bubbles and crashes, but were firmly anchored by fundamentals in the long term.
No one was able to define what was meant by short term and long term. Presumably the short term is reflexively defined as the period when prices are driven by speculative factors, while the long term is whatever time period over which fundamentals prevail.
But if oil and other commodities do suffer from price anomalies in the short term, the short term may be getting shorter. Certainly there is statistical evidence that markets are becoming more forward-looking, such as in studies by economists at the U.S. Commodity Futures Trading Commission. They also appear to discount future changes in supply and demand more quickly than before.
As a result, it is becoming harder to make confident predictions (even dubiously accurate ones) about what will happen to prices or curve structures.
Even most big commodity trading firms, the supposed experts, deny taking “directional” views on prices, arguing they have no better than a 50:50 chance of being right, and insist they are fully hedged and focus instead on curve structure and location/grade premiums.
In oil, “few traders are venturing firm price predictions due to the large tail risks” on both sides of the market at present, according to my colleague Javier Blas at the Financial Times (”Traders fear tail risks in commodities”, Jan. 3).
While some analysts believe risks are currently tilted towards the upside, on balance, and are unusually large, the implied volatility surface for WTI and Brent options shows a more even distribution.
Oil and other commodity markets are becoming increasingly integrated with equities and other assets, as well each other, as they move faster than ever before to incorporate economic developments. Oil and gas markets are moving aggressively to price medium-term supply shifts as well as short-term political risks.
The last few years have seen a sterile (and inconclusive) debate between commentators who believe commodity markets are efficient and driven purely by fundamentals, and those who believe price moves are exaggerated and driven mostly by speculative froth. Like most divisions in economics, it mirrors views about whether markets should be more tightly regulated or not.
But recent research and Citi’s analysis suggests a middle way between these theological extremes. It suggests that commodity markets remain imperfect and inefficient, subject to strange enthusiasms and frenzies, but the massive influx of investment in recent years may actually have made them less inefficient than before.
Unfortunately, improved efficiency may actually be to the detriment of many investors, who have successfully traded away many of the anomalies on which their strategies depended.