Neoclassical econometries: The kernel (Q1098537)

The paper considers the problem of determining the parameters in static equality-constrained neoclassical models frequently encountered in economic analysis. These include the utility-maximizing household, subject to income and time constraints; and the profit-maximizing firm, subject to a given technology. The approach applies equally to the traditional (regression-based) stochastic econometric model (with additive normally distributed errors) or to a deterministic analog, where the ``distance'' between the actual data and model solutions for the agents' decisions is minimized. Conventional econometric practice is extended in that closed-form, analytic solutions for the decision functions are not required; and essentially arbitrary configurations of available data are permitted. The computer is employed to jointly: (1) solve the agents' optimization problems, (2) impute missing data, and (3) determine whether there is sufficient information in the available data set to uniquely calibrate or estimate the model's parameters.