Sankhya: The Indian Journal of Statistics
Efficiency of GMM Estimation in Panel Data Models with Measurement Error
Zhiguo Xiao, Jun Shao, Ruifeng Xu and Mari Palta, University of Wisconsin-Madison, USA
SUMMARY. Wansbeek (2001) proposed a generalized method of moments (GMM) framework for analysing static panel data with measurement error in one regressor, without requiring external information such as a validation or replicate data set. We follow Wansbeek (2001) and provide a corrected formula for the asymptotically optimal second-stage GMM estimator. Furthermore, we show that, in some important special cases, the first-stage unweighted GMM estimator is actually asymptotically optimal. Some Monte Carlo results are presented to examine the finite sample performance of the first-stage and second-stage GMM estimators.
AMS (2000) subject classification. 62J05, 62J99.
Key words and phrases. Panel data, measurement error, efficient GMM, first-stage GMM.