Bias corrected generalized method of moments and generalized quasi-likelihood inferences in linear models for panel data with measurement error

For the estimation of the parameters involved in a linear dynamic mixed model for panel data, it has been shown recently by Rao, Sutradhar, and Pandit (2012, Brazilian J. of Probability and Statistics, Vol. 26, 167–177) that the so-called generalized quasi-likelihood (GQL) approach produces more efficient regression estimates as compared to the generalized method of moments (GMM) approach. These standard approaches, however, produce biased and hence inconsistent regression estimates when the time dependent covariates in such panel data models are subject to measurement errors. In this paper, we develop a bias corrected GQL (BCGQL) estimation approach and demonstrate that the BCGQL approach produces more efficient regression estimates than the BCGMM approach.