A Nonparametric Least Square Estimation of the Slope Parameters in the Truncated Regression Model

Abstract:

      In this paper, we consider the truncated regression model. A new method of estimating the regression parameters based on truncted data is given. The residual distri-bution is allowed to be unspecified. Using the conclusions on nonparametric estimation of error distributions in our work early, we obtain consistency of the estimation under some regular conditions. An example is used to show that our results is an improvement of Heckman's work (1979) essentially.