Neoclassical economic theory provides an important conceptual framework for the analysis of agricultural production. Theory provides little guidance, however, in the actual specification of empirical models. This paper applies an integrated approach for choosing between price expectation mechanisms in a multiple-equation model when the alternatives are non-nested. Nine alternative specifications of market price and policy information are developed. Price forecasting accuracy, non-nested tests of hypotheses, and out-of-sample predictive accuracy are examined for agricultural production in Iowa. The results call into question the reliability of using forecasting accuracy as the sole guide to selecting a price expectation proxy.