ESTIMATION OF SOME PARTIALLY SPECIFIED NONLINEAR MODELS

This paper presents a procedure for analyzing a partially specified no nlinear regression model in which the nuisance parameter is an unrestr icted function of a subset of regressors. The procedure does not requi re parametric modeling of the nuisance parameter but assumes that the model can be transformed into a partially specified linear equation by inverting some nonlinear functions. The model parameters are estimate d by applying Robinson's (1988a) procedure and the estimator is shown to be root N-consistent and asymptotically normal. One attraction of t he estimator is that it is computationally simple, requiring no more t han least squares regressions. A simulation study indicates that the e stimator has practical values.