Optimal sequential grain marketing decisions under risk aversion and price uncertainty

A method is developed to find sequences of expected utility maximizing decisions under risk aversion when random elements are time-dependent and additive separable utility of income is implausible. A Taylor-series approximation to expected utility is used. In an application to marketing stored wheat, expected seasonal sales patterns, early fractional sales of total inventory for risk reduction, and negative skewness in resulting income distributions are noted. Sensitivity to the number of income distribution moments used to approximate expected utility is examined. Six moments produce a good approximation. Use of only mean and variance can give doubtful results.