Shagufta Mubarik (smc7824@yahoo.com)
College of Statistical and Actuarial Sciences, University of the Punjab, Lahore, Pakistan
Dr. Maryam Ilyas (maryamilyas@hotmail.com)
College of Statistical and Actuarial Sciences, University of the Punjab, Lahore, Pakistan
ABSTRACT
Regression estimators are not robust in the presence of unusual observations. Heteroscedasticity and unusual observations cannot proceed together therefore to address these problems robust versions of weighted least square are used. Yet, weighted least square estimators are also affected in the presence of unusual observations. Therefore, an estimator is required which perform well. A modified estimator is proposed in this study which is more outlier resistant. The comparative performance of modified estimator is evaluated by conducting two studies. The performance is investigated by Monte Carlo simulation and confirmed by real data. The results showed that the modified estimator outperformed.
Key Words: Heteroscedasticity; High leverage points; influential observations; outliers; weighted least squares.
JEL Classification: C01, C15
