Covariate Measurement Error in Endogenous Stratified Samples

摘要: In this paper we propose a general framework to deal with the presence of covariate mea-surement error in endogenous stratifield samples. Using Chesher?s (2000) methodology, develop approximately consistent estimators for parameters structural model, sense that their inconsistency is smaller order than conventional which ignore existence measurement error. The approximate bias corrected are obtained by applying generalized method moments (GMM) modifeld version moment indicators suggested Imbens and Lancaster (1996) stratified Only specification conditional distribution response vari-able given latent covariates classical additive model assumption required, availability information on both marginal probability strata population variance not being essential. A score test detect arises as by-product approach. Monte Carlo evidence presented suggests that, samples moderate sizes, modified GMM perform well.