Sankhya: The Indian Journal of Statistics

2004, Volume 66, Pt. 2, 253--262

Generalized and Feasible Generalized Median Estimators for the Linear Regression with AR(1)  Error Model

Yi-Hsuan Lai, Lin-An Chen , National Chiao Tung University, Hsinchu, Taiwan

Wen-Yaw Chan , University of Texas-Houston, Houston, Texas

SUMMARY. The generalized least squares estimator (GLSE) and the feasible generalized least squares estimator (FGLSE) are, separately, extended to the generalized and the feasible generalized median estimators for the linear regression with AR(1) error model. The large sample theory for these estimators is developed. Furthermore results of Monte Carlo studies and an example of real data analysis are provided for the feasible generalized median estimator.


AMS (1991) subject classification. 62E20, 62G05, 62G20, 62G35.

Key words and phrases. Generalized least squares estimator, median, regression quantile.

 Full paper (PDF)