An important movement in econometrics in the l980's has been the development of estimation techniques that do not rely heavily on restrictive functional form assumptions of traditional econometric models. The aim of such techniques is to permit the economic effects of variables to be measured in ways that do not require ad hoc assumptions on how individual agents (or separate data observations) vary in unobserved dimensions. Parametric models explicitly or implicitly utilize such assumptions, which can lead to unknown and unmeasurable biases in parameter values and precision measures. The purpose of the proposed research is to create and study fully implementable nonparametric estimators of average derivatives. The statistical properties of the proposed estimators will be established, and their practical performance will be thoroughly analyzed via Monte Carlo simulation. The aim is to create a new empirical tool for the study of economic relationships, which does not rely on restrictive modelling assumptions, and to establish its practical value by examining its performance.
