Economics Seminar: An Asymptotically Smoothed Two-Stage Nonlinear Least Squares Estimator for Threshold Regression Models with Endogenous Variables

When?

Wednesday 26 January 2011, 16:00 to 17:00

Where?

04AD00

Open to:

Public, Staff, Students

Dr Danielle Massacci (University of Surrey)

"An Asymptotically Smoothed Two-Stage Nonlinear Least Squares Estimator for Threshold Regression Models with Endogenous Variables"

Abstract

This paper proposes a two-stage estimator for the parameters of threshold regression models with endogenous right-hand-side variables (including the threshold variable). The objective function is constructed in terms of squared deviations of the dependent variable from its expected value conditional upon the available instruments, and the bias correcting terms are estimated by their sample counterparts. The proposed estimator is shown to be consistent. Under mild regularity conditions on the distribution of the error terms, the original threshold effect is transformed into an asymptotically smooth transition effect: in this way, a √T asymptotically normally distributed estimator for all model's parameters (including the threshold parameter) is obtained. An extensive Monte Carlo study is undertaken to support the validity of the proposed estimator. The issue of statistical inference is also addressed, and a solution based on the distance metric statistic is discussed.