Instrumental variable estimation of dynamic linear panel data models with defactored regressors and a multifactor error structure

This paper develops two instrumental variable (IV) estimators for dynamic panel data models with exogenous covariates and a multifactor error structure when both the cross-sectional time series dimensions, N T respectively, are large. The main idea is to project out common factors from of model, construct instruments based on defactored covariates. For homogeneous slope coefficients, we propose two-step IV estimator. In first step, model estimated consistently by employing as instruments. second entire extracted residuals first-step estimation, after which an regression implemented using same in step one. heterogeneous mean-group-type estimator, involves averaging estimates cross-section-specific slopes. proposed do not need seek variables outside model. Furthermore, these linear, therefore computationally robust inexpensive. Notably, they require no bias correction. We investigate finite sample performances associated statistical tests, results show that tests perform well even small T.