On Term Structure Models of Commodity Futures Prices and the Kaldor-Working Hypothesis by

Both prices and the volatility of storable agricultural commodity futures contracts have been rising since 2005 and particularly since 2007. This paper aims to answer two principal questions: (i) How has the behavior of these futures prices over time and across maturities changed with the rise of biofuels and their demand-side pressure on corn and related crops?, and (ii) Is there now stronger or weaker evidence of the Kaldor-Working convenience yield-storage hypothesis, whereby futures price backwardation can be explained by the high value of remaining inventory stocks when these are near stockouts? The empirical application is to Chicago Board of Trade corn, wheat and soybeans futures. To make use of all available futures data rather than only the nearby, this paper adopts a recently developed affine term structure model approach and conducts estimation in state-space form using the Kalman filter. A novel aspect of the research is that it allows an arbitrary number N of state variables, where more variables provide further precision and curvature but at a higher computational cost. It is found that a three-state variable model containing both random walk and mean reversion components provides the most parsimonious fit during 1988-2004, but that a simple one-state variable model is optimal for the period 20052007. The main implication is that futures prices since 2005 behave much more like a “random walk” than before. Also, the model allows us to estimate the term structure of volatility and it is found that distant maturity futures should be expected to be much more volatile than historically normal. Two practical but only tentative implications are: (a) hedgers should use significantly lower hedge ratios than before, and (b) for traders, the classic Black-Scholes option pricing solution should perform better now than it has historically. Lastly, the paper finds partial empirical support for the convenience yield relationship with relative inventory stocks, especially for soybeans and wheat. JEL Classification Codes: C52, C53, G12, G13, Q13, Q14.