Author(s):
1. Gargi Tyagi:
Department of Mathematics & Statistics, Banasthali Vidyapith, Banasthali, Rajasthan, India
2. Shalini Chandra:
Department of Mathematics & Statistics, Banasthali Vidyapith, Banasthali, Rajasthan, India
Abstract:
In this paper, an alternative estimator has been developed for the estimation of unknown regression coefficient vector in linear regression model to combat the problem of multicollinearity when additional stochastic restrictions are available. The proposed estimator is a generalization of the mixed regression (MR) estimator (Theil and Goldberger (1961)) and the principal component two parameter (PCTP) estimator (Huang and Yang (2015)), named as two parameter stochastic restricted principal component (TPSRPC) estimator. Necessary and sufficient conditions for superiority of the TPSRPC estimator over some other competing stochastic restricted estimators have been derived under the MSE matrix and scalar mean squared error (MSE) criteria. Further, tests to verify the conditions under MSE matrix have been derived. A Monte Carlo simulation and a numerical example have been given to evaluate the performance of the estimators.
Page(s):
127-154
DOI:
DOI not available
Published:
Journal: Pakistan Journal of Statistics, Volume: 35, Issue: 2, Year: 2019
Keywords:
Principal Component Two Parameter estimator
,
Stochastic linear restrictions
,
Mean squared error
References:
References are not available for this document.
Citations
Citations are not available for this document.