Estimation for the multi-way error components model with ill-conditioned panel data

• Published in: Journal of the Korean Statistical Society Vol. 46; no. 1; pp. 28 - 44
• Main Authors: Lee, Jaejun, Cheon, Sooyoung
• Format: Journal Article
• Language: English
• Published: Singapore Elsevier B.V 01.03.2017; Springer Singapore; 한국통계학회
• Subjects: Applied Statistics; Bayesian Inference; Collinearity; Dual generalized maximum entropy; Endogeneity; Error components model; Ill-conditioned problems; Panel data; Statistical Theory and Methods; Statistics; Statistics and Computing/Statistics Programs; 통계학; secondary; Ill-conditioned problems; Panel data; Collinearity; Dual generalized maximum entropy; Endogeneity; Error components model; primary; primary 62J05; secondary 91B82
• ISSN: 1226-3192; 2005-2863
• DOI: 10.1016/j.jkss.2016.05.008

Ill-posed problems resulting from limited, partial or incomplete sample information have occurred frequently in econometric practice. The traditional methods of information recovery may cause the estimates highly unstable with low precision, known as ill-conditioned problems. In this paper, we propose a dual generalized maximum entropy estimator for the multi-way error components model with ill-conditioned panel data, based on an unconstrained dual Lagrange multiplier method. The numerical results for the panel data regression model with highly correlated and endogeneous covariates are in favor of our dual generalized maximum entropy estimation method in terms of quality of estimates.