Opal Financial Group FX & Commodity Summit for Institutional Investors Chicago. Term Structure Properties of Commodity Investments

Transcription

1 Opal Financial Group FX & Commodity Summit for Institutional Investors Chicago Term Structure Properties of Commodity Investments March 20, 2007 Ms. Hilary Till Co-editor, Intelligent Commodity Investing, Research Associate, EDHEC Risk and Asset Management Research Centre, and Principal, Premia Capital Management, LLC, 1

2 Term Structure Properties of Commodity Investments This presentation is based on chapters from: The Risk Books publication, Intelligent Commodity Investing, edited by Hilary Till and Joseph Eagleeye, Spring

3 Term Structure Properties of Commodity Investments A. The Statistical Behavior of Commodity Futures Contracts B. The Importance of Yields as a Long-Term Driver of Returns C. Recent Research on Term Structure as the Primary Driver of Returns Icon above is based on the statue in the Chicago Board of Trade plaza. 3

4 A. The Statistical Behavior of Commodity Futures Contracts The empirical behavior of commodity prices can be described as follows: Commodity prices are extremely volatile; There exist rare but violent explosions in prices; In normal times, there is a high degree of autocorrelation; Source: Deaton and Laroque (1992). 4

5 A. The Statistical Behavior of Commodity Futures Contracts In spite of volatility, prices tend to revert to their mean or to a trend level; There is substantial positive skewness in the price distributions; and There is substantial kurtosis with tails much thicker than those of a normal distribution. Source: Deaton and Laroque (1992). 5

6 A. The Statistical Behavior of Commodity Futures Contracts These observations are illustrated with a long-term chart of sugar prices: Source: Deaton and Laroque (1992). 6

7 A. The Statistical Behavior of Commodity Futures Contracts Commodity prices exhibit positive skewness for the following reason. When there are ample supplies, more inventories can be held and the price can decrease in order to balance supply and demand. 7

8 A. The Statistical Behavior of Commodity Futures Contracts In the short run, new supplies of commodities cannot be instantly mined, grown, or drilled. So when there is not enough of a commodity, only the price can adjust to balance supply and demand. 8

9 A. The Statistical Behavior of Commodity Futures Contracts Sorensen (2002) analyzes agricultural price behavior from 1972 to He formally models the prices of soybean, corn, and wheat as the sum of following factors: a. permanent trend shifts, b. seasonality, and c. mean reversion. 9

10 A. The Statistical Behavior of Commodity Futures Contracts Permanent trend shifts can be the result of either inflation or permanent changes in supply or demand. Sorensen estimates mean-reversion parameters for each commodity: their half-lives range from 0.9 years for wheat to 0.7 years for soybeans. The half-lives express the time it takes before a given shock to [a price] process is expected to have leveled off by half of the shock. 10

11 A. The Statistical Behavior of Commodity Futures Contracts The shocks to agricultural prices subsided within 5 years (on average.) Graphical Representation of Sorensen's Fitted Mean-Reversion Parameters for Wheat, Corn, and Soybeans Prices, Which Were Inferred from 1972-to-1997 Data 100% 90% Level of Shock (Normalized to 100% at Time of Initial Shock) 80% 70% 60% 50% 40% 30% 20% 10% 0% Years Since Initial Shock to Prices Fitted Mean-Reversion for Wheat (w ith half-life of shock to prices at 0.9 years) Fitted Mean-Reversion for Soybeans (w ith half-life of shock to prices at 0.7 years) Fitted Mean-Reversion for Corn (w ith half-life of shock to prices at 0.9 years) Source: Till (2007). 11

12 A. The Statistical Behavior of Commodity Futures Contracts Geman (2005) states: commodity prices neither grow nor decline on average; they tend to mean-revert to a level which may be viewed as the marginal cost of production. This has been evidenced a number of times in the literature [for both] agricultural and energy commodities. Hence, mean-reversion is one of the main properties that has been systematically incorporated in the literature on commodity price modeling. 12

13 B. The Importance of Yields as a Long-Term Driver of Returns Because of the mean-reverting nature of spot commodity prices, it should naturally follow that across time, roll yields (and therefore, backwardation) have to be the dominant explanatory variable for individual futures contract returns over long enough time horizons. The yield comes from the benefit of physical possession, which provides all sorts of optionality, mainly in being able to take advantage of the possibility of ensuing spot shortages. 13

14 B. The Importance of Yields as a Long-Term Driver of Returns Annualized Return Vs. Average Annual Backwardation ( ) 25.0% 20.0% Gasoil (since 6/86) Brent (since 8/89) Crude Oil Gasoline (since 1/85) 15.0% Annualized Return Natural Gas (since 4/90) Lean Hogs 5.0% Soy Beans Cotton Bean Oil Sugar Platinum Soy Meal Copper Live Cattle Heating Oil Gold 0.0% % % % -5.00% 0.00% 5.00% 10.00% 15.00% Wheat Coffee Corn Silver -5.0% Cocoa 10.0% -10.0% Average Annual Backwardation (since 4/83, as a % of price) Source: Nash and Shrayer (2005). 14

15 B. The Importance of Yields as a Long-Term Driver of Returns The finding that a yield variable is the dominant driver of long-run returns is not unique to commodities. Cochrane (1999a, 1999b) discusses how the evolution of yield variables is key to the long-run returns in equities, bonds, and foreign exchange. 15

16 B. The Importance of Yields as a Long-Term Driver of Returns Example of a Long-Horizon Effect Horizon b sigma(b) R-squared 1 year (0.33) years (0.66) years (0.88) years (1.24) 0.59 Ordinary least squares regressions of excess returns (value-weighted New York Stock Exchange - Treasury Bill rate) on valueweighted price-dividend ratio over different time horizons. b is the coefficient on the price-dividend ratio, and standard errors in parenthesis (sigma(b)) use Generalized Methods of Moments to correct for heteroskedasticity and serial correlation. Sample Source: Cochrane (1999a). 16

17 B. The Importance of Yields as a Long-Term Driver of Returns As the R-squared values [for the previous slide s regression] show, these are long-horizon effects: annual returns are only slightly predictable and month-to-month returns are still strikingly unpredictable, but returns at five-year horizons seem very predictable. The results at different time horizons are reflections of a single underlying phenomenon. If daily returns are very slightly predictable by a slow-moving variable, that predictability adds up over long horizons. 17

18 C. Recent Research on Term Structure as the Primary Driver of Returns One issue with the long-term findings that an individual commodity futures contract s returns are (or have been) linearly related to its roll yield is that the roll yield for a commodity is not constant. Schneeweis and Spurgin (1997) note that there is a lack of evidence of a consistent convenience yield for oil. This is similar to equities where the dividend yield over time is not stable either. 18

19 C. Recent Research on Term Structure as the Primary Driver of Returns We examined the agricultural futures markets since these markets have continuous data since the late 1940 s. We also examined how long the time horizon needs to be before roll yields (and backwardation) are the dominant explanatory variable for investment returns in soybean, corn, and wheat futures contracts. 19

20 C. Recent Research on Term Structure as the Primary Driver of Returns We first examine the historical evolution of front-month futures prices for soybeans, corn, and wheat. These graphs are through June 2006, and start in January of

21 C. Recent Research on Term Structure as the Primary Driver of Returns 21

22 C. Recent Research on Term Structure as the Primary Driver of Returns 22

23 C. Recent Research on Term Structure as the Primary Driver of Returns In viewing the crop futures price charts, we see the emergence of substantial instability in prices during the early 1970 s. This is largely coincident with the collapse of the post-world-war II Bretton Woods framework of fixedcurrency regimes. In viewing the post-1970 s crop price patterns, it may be hard to argue that weather got any more volatile in disrupting crops. 23

24 C. Recent Research on Term Structure as the Primary Driver of Returns We provide two illustrative graphs, which show how well roll yield (and backwardation) explain excess returns when analyzed over 5-year time horizons from 1950 to

25 C. Recent Research on Term Structure as the Primary Driver of Returns This figure shows average excess returns versus roll yield for all three agricultural futures contracts. Five-Year Annualized Excess Return as a Function of Five-Year Annualized Roll Yield 1950 to % Annualized Excess Return 30% 20% 10% 0% -10% -20% Soybeans Corn Wheat -30% -20% -15% -10% -5% 0% 5% 10% 15% 20% 25% 30% Annualized Roll Yield 25

26 C. Recent Research on Term Structure as the Primary Driver of Returns This figure shows annualized returns as a function of averagetime-in-backwardation. Five-Year Annualized Excess Return as a Function of Five-Year Time-in-Backwardation 1950 to 2004 Annualized Excess Return 40% 30% Soybeans 20% Corn 10% Wheat 0% -10% -20% -30% 0% 10% 20% 30% 40% 50% 60% 70% Average Time-in-Backw ardation 26

27 C. Recent Research on Term Structure as the Primary Driver of Returns In each graph, the outlier observations of extreme returns relative to backwardation and roll yield are from the 1970-to period, which incorporates the collapse of the Bretton Woods system and the consequent decline in the value of the U.S. dollar. But other than the 1970-to-1974 period, it is apparent that the driver of 5-year returns has been roll yield. 27

28 C. Recent Research on Term Structure as the Primary Driver of Returns If spot prices generally mean-revert over long horizons, then spot prices cannot (usually) be the driver of return over longterm horizons. An examination of the 1970-to-1974 period shows that this is a proper caveat. A rare trend shift upward in spot prices can also be a meaningful source of returns, separate from the term structure for a futures contract. 28

29 References Cochrane, J., 1999a, New Facts in Finance, Economic Perspectives, Federal Reserve Board of Chicago, Third Quarter, pp Cochrane, J., 1999b, Portfolio Advice for a Multifactor World, Economic Perspectives, Federal Reserve Board of Chicago, Third Quarter, pp Deaton, A. and G. Laroque, 1992, On the Behavior of Commodity Prices, Review of Economic Studies, Vol. 59, pp Feldman, B. and H. Till, 2006, Separating the Wheat from the Chaff: Backwardation as the Long-Term Driver of Commodity Futures Performance; Evidence from Soy, Corn and Wheat Futures from 1950 to 2004, EDHEC-Risk Publication ( and Premia Capital & Prism Analytics White Paper; a version of this article is in the Journal of Alternative Investments, Winter 2006, as Backwardation and Commodity Futures Performance: Evidence from Evolving Agricultural Markets, pp Geman, H., 2005, Commodities and Commodity Derivatives, (Chichester: John Wiley & Sons). Nash, D., and B. Shrayer, 2005, Investing in Commodities, Morgan Stanley Presentation, IQPC Conference on Portfolio Diversification with Commodities, London, 24 May. Schneeweis, T. and R. Spurgin, 1997, Energy Based Investment Products and Investor Asset Allocation, CISDM Working Paper Series, 29 May. Degas, Edgar, The Cotton Exchange at New Orleans, 1873, Musée Municipal, Pau, France. 29

30 References (Continued) Sorensen C., 2002, Modeling Seasonality In Agricultural Commodity Futures, Journal of Futures Markets, Vol. 22, No. 5, pp Till, H., 2007, A Long-Term Perspective on Commodity Futures Returns, in H. Till and J. Eagleeye (ed) Intelligent Commodity Investing (London: Risk Books). Presentation Prepared By Katherine Farren, Premia Capital Management, LLC. 30