Modeling optimal grain-marketing decisions when prices are generated autoregressively

A model is developed to find decision rules that maximise expected utility of income from selling stored wheat when future price expectations are updated weekly using newly observed prices. Expected utility is approximated in terms of six moments of the income distribution, and efficient procedures are developed to calculate moments. Marketing losses that arise when the autoregressive structure of wheat prices is ignored are estimated. Losses were found to be equivalent to reductions of 1.9-14.5% in expected income, depending on initial prices and risk aversion.