查看文章

The paper explores the issue of endogeneity in random effects models, focusing on the case where the random effects are correlated with a level 1 covariate in a simple linear random intercept model. This type of endogeneity arises from a wrong equality restriction on the between-cluster and within-cluster slopes. We show that estimation of the between-cluster slope is affected by measurement error and we propose a correction to obtain consistent estimates. Then we review the proposed tests for endogeneity and compare alternative estimators. A a small simulation illustrates the main findings.