RELEVANT MOMENT SELECTION UNDER MIXED IDENTIFICATION STRENGTH

This paper proposes a robust moment selection method aiming to pick the best model even if this is condition with mixed identification strength, that is, conditions including functions are local zero uniformly over parameter set. We show relevant procedure of Hall et al. (2007, Journal Econometrics 138, 488–512) inconsistent in setting as it does not explicitly account for rate convergence estimation candidate models which may vary. introduce new based on criterion automatically accounts both model’s estimate and entropy estimator’s asymptotic distribution. The benchmark estimator we consider two-step efficient generalized moments estimator, known be framework well. A family penalization introduced guarantees consistency procedure. finite-sample performance proposed assessed through Monte Carlo simulations.