Stochastic dynamic optimization is performed for a representative consumer and producer in the soybean market. Each is assumed to maximize the expected value of the infinite sum of the present value of his (her) income stream. Optimum rules for each side are solved as a function of, among other things, expected future prices. Three price expectations are applied--rational, adaptive, and cash-futures. Estimation, based upon the aggregate U.S. soybean market. Time series analysis and Granger-causality test are utilized at the first stage of estimation in order to obtain information which help to forecast prices. The derived decision rules under rational expectations hypothesis are nonlinear function of parameters appearing in agents' objective functions. All variables that are in the information set which help to predict future values of prices are in the decision rules. The response functions of all decision rules under rational expectations hypothesis depend upon the values of all structural parameters. Dynamic simulation of Quasi-Rational expectations model is performed with relatively good results. The estimation under the other two hypotheses--adaptive and cash-futures price expectations--are relatively inferior.