Efficiency profiles of MM estimators in dynamic panel data models

In dynamic panel data models with unobserved individual effects least-squares estimators are inconsistent when the number of cross-section units N gets large while the number of time-series observations T remains finite. For that situation an abundance of method of moments (MM) estimators is available, which differ in the way unobserved heterogeneity is dealt with and regarding the number and nature of instruments that is being exploited. For some stylized models we derive and compare characteristics concerning instrument weakness (or fitness) and the resulting effectiveness with respect to estimator efficiency for T small and N infinite. We make extensive use of graphical methods to show the characteristic qualities of and differences between estimation methods over relevant areas of the parameter space.