Document Type

Working Paper

Publication Date

10-2015

College/Unit

Chambers College of Business and Economics

Document Number

15-46

Department/Program/Center

Economics

Abstract

We propose a kernel-based estimator for a partially linear regression in a triangular system where endogenous regressors appear both in the nonparametric and linear components of the regression. Compared with alternative estimators currently available in the literature (Ai and Chen 2003; Otsu 2011), our estimator has an explicit functional form, is easier to implement, and exhibits better experimental finite sample performance. The estimator is inspired by the control function approach of Newey et al. (1999) and was initially proposed by Martins-Filho and Yao (2012). It explores conditional moment restrictions that make it suitable for additive regression estimation as in Kim et al. (1999) and Manzan and Zerom (2005). We establish consistency and p n asymptotic normality of the estimator for the parameters in the linear component of the model and give a uniform convergence rate for the estimator of the nonparametric component. In addition, for statistical inference, a consistent estimator for the covariance of the limiting distribution of the parametric estimator is provided. We illustrate the empirical viability of our estimation procedure by applying it to the study of the impact of foreign aid and policy on growth of per capita gross domestic product (GDP) in developing countries.

Included in

Economics Commons

Share

COinS