Michele De Benedictis has for some time taken an interest in behavioral models for agricultural entrepreneurs. Often, such models are specified by means of a production function and a cost function with the associated derived demands for inputs. In order to make such models operational in a given empirical setting, the approach in the traditional literature has been to estimate either the production function and the associated first order conditions or the system of derived demand functions for inputs. This paper proposes an encompassing approach which consists in the joint estimation of the production function and the associated first order conditions and the system of derived demand functions. Empirical verification lends support to the hypothesis that full utilization of the available information requires a primal-dual approach, as presented in this paper.