The Barclays Capital Equity Gilt Study 2004 included an analysis of commodities as an asset class for the first time. Since then, commodity investments have moved to the forefront of the alternative asset class agenda as asset managers, pension funds, insurance companies and other institutional investors seek ways to counter the modest performance of more mainstream asset classes.
Outstanding recent performance is one reason why commodity assets have garnered so much attention. Over the past five years, a passive, long-only investment in the Goldman Sachs Commodity Index (GSCI) would have generated an average annual return of around 15%, comprehensively outstripping other investment benchmarks in stocks and bonds (Figure 1).
Not surprisingly, this kind of performance has not gone unnoticed by asset managers. However, for those seeking to match their return profile to long-term liabilities, the demonstration of a robust performance over a much longer time period is necessary.
Taking the GSCI back as far as its fully audited data series allows (the GSCI has been traded since 1991, but a history of returns has been backfilled to 1970), its performance compares favourably with that of long-term equity market returns (Figure 2).
Recent academic work supports the view that investing in commodities has been as rewarding as investing in equities. Extending the analysis back to 1959 and using an equally weighted commodity futures index, Gorton and Rouwenhorst1 conclude that over the past 43 years, the average annualised return of a collateralised investment in commodity futures has been comparable to the return of the S&P 500, with both significantly outperforming corporate bonds. Analysis by Erb and Harvey2 also confirms that historically commodity futures have had excess returns similar to those of equities, though they qualify this finding by pointing out that the historical track record is not necessarily any guarantee of similar performance in the future. They also highlight the sensitivity of commodity returns to the time periods over which they are measured, making the point that measured from 1983 (when energy futures were first added to the GSCI), the excess return at 4.49% compares less favourably with that of the S&P over the same period (at 7.35%).
To this last point, it seems reasonable to respond that the return profile of any asset with a cyclical component can be made to look better or worse by varying the starting point.
Moreover, taking the starting point from 1983 ensures that the full impact on commodity futures returns of the oil price collapse, following the spike of the late-1980s, is incorporated in the results. Finally, given the well-documented negative correlation between commodity returns and those of stocks and bonds, it would be very surprising if it were not possible to pick time periods where stocks comprehensively outperform commodities and vice versa.
The historical negative correlation between commodities and other asset classes and the diversification benefit that this provides to a portfolio containing stocks and bonds is a key factor in favour of commodities as an asset class. Again, the academic literature strongly supports this characteristic. Additionally, research carried out by our rates research team found that since 1970 the GSCI exhibited a 3.2% negative correlation with the S&P 500 and a 6.3% negative correlation with a reconstructed government bond index.
Significantly, our analysis of rolling three-year correlations between commodities, stocks and bonds show that while the correlations between commodity returns and those of the other two asset classes vary between positive and negative over the past 30 years, commodity returns tend to be most strongly negatively correlated with equity and bonds when the returns from those asset classes are at their weakest. The strongly negative commodity correlations with the poor equity and bond returns associated with the 1970s oil price shock, the late-1980s share price collapse and, more recently, the negative correlation with weak equity market returns since the collapse of the 1990s tech bubble, are all good examples of this.
In order to illustrate the benefits of investing in commodities, our team of quantitative analysts has constructed and compared the performance of two portfolios: one that includes commodities and one that does not.
In this exercise, the universe of assets is aggregated into three classes: equities, bonds and commodities. The data series used for the three composite assets are monthly and span February 1970 to December 2004. The commodity series is represented by the Goldman Sachs Commodity Index (GSCI). Because there is no single representative index extending over the entire period for bonds, the series for this asset class was constructed by piecing together different proxies for the total return. In the case of bonds, the total return series for the period January 1973 to January 1981 is obtained by using the daily yields of the 5-year and 10-year US Treasury bonds in 3:4-weighted proportions. For the period January 1981 to January 1988, we use the Lehman Brothers Composite Treasury Bond Index, and lastly for January 1988 to January 2004 we use the JP Morgan US Government Bond Index. For equities, the total daily return series is the S&P 500.
Figure 3 illustrates that the inclusion of commodities within this simple three asset portfolio results in a significant improvement in the overall risk/reward profile, raising return for a given level of risk at all but the very highest risk tolerance levels. Our portfolio optimisation program selected commodities at a consistent 12% share of the optimal portfolio, with that share only beginning to fall at a risk tolerance level of over 13.6% for the annualised standard deviation of returns.
To further test the optimal weighting for commodities in a portfolio for a given target return rate, we assume a target return of 8% by the investor and then vary portfolio weightings in order to minimise risk. Figure 3 shows the results of this exercise. The analysis shows that during the period under examination, the optimal allocation of commodities within the portfolio ranged between 0% and 43%, averaging 16%.
The diversification benefits that commodities bring to a mix of assets justify a significant portfolio share. We would repeat the recommendation made in the 2004 Equity Gilt Study that as a general statement, a 10% allocation to commodities is consistent with the historical risk/reward characteristics of the asset class.
A combination of the strong recent return performance of commodity index investments, the positive experience of a number of high profile institutional investors that initially made commodity investments in the late-1990s, plus a clearer understanding of the historical diversification benefits of holding commodities as part of a portfolio have helped spur a significant increase in commodity investments by institutions. Increasingly prominent media coverage, due primarily to soaring oil prices, has also helped.
We estimate that the total amount of institutional investor fundstracking commodity indices worldwide has risen from less than US$10bn in 2000 to around US$55bn by May 2005, with most of that growth occurring since 2002. This total includes investments by a broad spread of asset managers, pension funds and endowment funds in North America and Europe. US mutual fund exposure to commodities has risen particularly rapidly from less than US$300mn at the end of 2002 to nearly US$10bn by the end of March.
Moreover, there is growing interest in commodities as an asset class, evidenced by the impressive attendance (more than 150 existing and potential clients, including pension funds, private banks and hedge funds) at Barclays Capital's inaugural commodity investor conference held in Barcelona on Friday, 4 February 2005.
In a survey taken of the attendees, 67% of them had no exposure to commodities in their portfolios in 2004, but only 11% felt that their exposure would remain at zero over the next three years (Fig. 4), with 35% expecting commodities to become more than 10% of their portfolio. Portfolio diversification was the primary driver for investing in commodities for 64% of attendees (Fig. 5).
Tactical investors in commodity markets in the form of specialised hedge funds, such as CTAs, have been around for many years but they are also getting more active. The big difference between hedge fund activity and that of most institutional commodity investment is the more active approach taken to trading different commodity markets and different points on the futures curves, in addition to going long or short, depending on market view. Since mid-2004, hedge funds have been increasingly playing commodity markets from the short side and their gross short positions across 23 of the major US commodity futures markets are near the largest on record.
Following the big increases in commodity prices, plus the flow of new funds into the commodity sector, the two key questions being asked by current and potential investors in commodity index products are:
Further gains in store for commodity prices
Many commodity prices have surged towards multi-year highs over the past year and the extent of these moves has forced all industry participants (producers, consumers, traders, investors and bankers) to consider whether we are simply approaching the peak of a bull market or are in the midst of a secular shift or "super-cycle".
We are strong advocates of the view that the underlying level of most commodity prices will be higher over the rest of this decade. Reflecting this, Barclays Capital's Commodity Research Team recently made big increases to its long-term average price forecasts across a broad range of different commodities. The rationale for this is based on three key assumptions:
In the short term, we see considerable upside price potential in 2005 in a number of commodity sectors that are important for index investors, most notably in oil and US natural gas. Given the dominance of energy as a share of the GSCI, rising oil and gas prices would go a long way to ensuring further gains in the index in 2005.
Turning to the debate over the capacity of commodities to absorb fresh long-only investment flows, we do not believe in the thesis that investment in commodity markets is creating a bubble pushing commodity price levels above those that are justified by underlying market fundamentals. However, there is some evidence of an impact on the term structure of prices at the front end of the commodity price curves which could have important implications for returns to index investors.
The first point to make is that institutional investor involvement in relation to overall commodity market size is still small. Using the aforementioned estimate of around $55bn for the current level of funds tracking commodity index products, institutions account for less than 5% of combined monthly average turnover in the commodity markets that typically constitute the main traded commodity index products.
Furthermore, commodity futures market volumes are getting larger as their use by a wide range of participants – from producers and consumers to traders, banks and investors – continues to grow. Since mid-2003, volumes in the 23 futures markets included in the most commonly traded index products are up by almost 50%. It is true that the growth in commodity index investment has risen much more quickly than overall market growth, rising almost three-fold over the same period, but from a very low base.
But is this increase now stretching the capacity of commodity markets to absorb investor interest and still generate competitive returns? To answer this question, some discussion of what generates those returns is necessary.
Index investment is very specifically concentrated at the front end of commodity market price curves, where futures contracts that form part of an index are typically rolled on a regular basis from front to second month. In a market in backwardation (i.e., a downward sloping futures curve), this generates a return (for a fuller explanation of this "roll yield" see the 2004 Equity Gilt Study). This roll yield is extremely important in generating overall commodity market returns. For example, Erb and Harvey find that between 1982 and 2004, the returns generated from rolling forward contracts accounted for more than half of the excess return level of the GSCI (2.59% of the 4.49% excess return).
In the past year, there is some evidence that roll yields are contracting, though it would be very difficult to prove that this is the result of the flow of investment money into commodity markets. This contraction is most evident in the term structure at the front-end of the underlying futures price curves where backwardations, notably in crude oil, have now shifted to contango and where on average the level of backwardation during roll periods in 2004 tended to be lower than outside them.
A common explanation for the source of return to long only investors in commodity markets, first put forward by Keynes, is that it arises from being paid to take price risk away from market participants that want to avoid it and that the normal state of a commodity market is to be in backwardation, so long as the demand for insurance is greater than supply.
If this is the case, then investors in index products are being paid for insuring a very specific type of risk, that of price volatility at the very front end of the price curve, where market activity tendsto be dominated by hedgers of short-term trading risk. One explanation for the contraction in roll yield in recent months could therefore be that the sellers of commodity price insurance are beginning to outnumber buyers at this point on the forward curve.
However, this type of explanation of the basis for returns to commodity investors does not take enough account of the importance of underlying physical fundamentals in determining futures market price structure and hence the roll yield gained by investors.
Commodity futures markets are inextricably linked to the underlying physical markets by the need to deliver the commodity against a short position still open on contract expiry or to receive delivery against an open long position. Front end time structure is therefore very closely driven by the balance between a commodity's supply and demand.
This can best be demonstrated by examining the relationship between reported inventory levels (a useful proxy for underlying supply and demand balance) and nearby price spreads. In the examples presented in Figure 6 and Figure 7, we show how movements from contango to backwardation are closely linked to inventory levels. The figures show that in both markets, as inventory levels fall below a certain range, the nearby futures market is likely to move into backwardation, providing a positive roll yield to investors holding long positions.
Viewed from this perspective a viable alternative explanation for the move from backwardation to contango in oil markets in recent months is the rise in US crude oil inventory levels that commenced in October. A mild northern hemisphere winter has contributed to easier availability of crude oil and products in most regions. In the US, commercial inventory levels are now significantly above average for the time of year. However, for the reasons given earlier in this report, the likelihood is for fundamental tightness to return to the oil market later in 2005, for inventory levels to fall back, returning the futures curve to backwardation and restoring a positive crude oil roll yield.
We would conclude by stating that the prevailing high level of commodity prices does raise the question of whether passive investment in a commodity index is the appropriate structure for all investors at this point in time. There is further upside to commodity prices but future gains are unlikely to match those of the past two years. In addition, there is evidence of some contraction in roll yield over recent months and these two factors suggest that the upward path for commodity total return indices seen over the past four years may now be moderating.
This does not mean the opportunity for institutional investors to invest in commodities has passed. Investment choices for investors to obtain commodities exposure are growing rapidly and are far from being merely restricted to passive index investing at the front end of the commodities curve. The choices range from modest changes in the structuring of commodity indices – such as different weightings than those offered in the standard indices or changing the rolling dates – to specialised structured notes that are individually tailored to the investor's requirements, be that increased leverage to particular commodities, or taking advantage of opportunities further out along the forward price curve.
The investment case still holds
Kevin Norrish and Kamal Naqvi, Directors, Commodities Research, Barclays
Originally published in the June/July 2005 issue