Bayesian Cross Hedging: An Example From the Soybean Market

Following Lence and Hayes (1994a), we study the problem faced by an Iowa farmer who wishes to hedge a soybean harvest using Chicago futures contracts. A time-series model for spot and futures prices is postulated, and numerical Bayesian procedures are used to calculate predictive densities and optimal hedges. The numerical procedures extend earlier analytical work, and easily accommodate alternative views about specification (levels vs. logarithms, trends vs. no trends, etc.), uncertainty about parameterisations (estimation risk), as well as other non-sample information (the likely size of the difference between spot prices in Iowa and Chicago, the tendency of the basis to be large in the spring, the shrinking of the basis as expiration of the future looms, etc.)