A STATE-SPACE FORECASTING APPROACH TO OPTIMAL INTERTEMPORAL CROSS-HEDGING

Cross-commodity hedging between fishmeal and soybean meal is investiga ted. The approach uses successively updated out-of-sample forecasts to approximate subjective price expectations, and forecast error varianc e-covariance matrices to measure risk. Forecasts are generated by stat e-space models of vector-valued time series. In a stationary environme nt, uncertainty reduces to the difference between the historical autoc ovariance of the random process and the variance-covariance of out-of- sample forecasts. Results indicate that weakly risk-averse agents can increase average marketing returns within acceptable risk levels by co mbining information from price forecasting models with an appropriate hedging strategy.