Modelling U.S. soybean and corn prices

Abstract

This paper analyses the lead–lag relationship between two closely related commodities; soybean and corn. Corn and soybeans are substitutable in terms of their end use, and therefore, these two commodities are known to compete for acres. Both these agricultural commodities have been subject to large variabilities primarily caused by the variation in the market fundamentals and changing policy conditions. We formulate a conceptual dynamic model that makes use of end-use substitutability of both commodities, demand and supply functions, and stocks. The reduced form model allows for the unpredictable shifts in demand and supply as well as the changing policy regimes, in which we make use of a Flexible Fourier estimation procedure. This procedure allows to approximate several breaks and large variability that are known to plague commodity prices. We find soybean and corn prices to be stationary around a flexible trend, supporting the commonly held belief that shocks to agricultural prices should transitory. We also find corn prices can be used to make short run predictions of soybean prices. Finally, we can establish that such allowing for flexible trends, we can generate superior forecasts than the standard benchmark model as well as the WASDE produced by the USDA.