A1694
Title: Inference in linear models with structural changes and mixed identification strength
Authors: Bertille Antoine - Simon Fraser University (Canada) [presenting]
Otilia Boldea - Tilburg University (Netherlands)
Niccolo Zaccaria - Tilburg University (Netherlands)
Abstract: The estimation and inference in a linear IV model are considered in the presence of parameter instability. When the reduced form is stable but the structural form exhibits structural change, new GMM estimators are proposed and is proven that they are more efficient than the standard subsample GMM estimators, even in the presence of weaker identification patterns. For detecting change points in the structural form, two test statistics are proposed: when the reduced form is stable and when the reduced form exhibits structural change. The limiting distribution of these test statistics is derived and is shown that they have correct asymptotic size and non-trivial power even under weaker identification patterns. The finite sample properties of the proposed estimators and testing procedures are illustrated in a series of Monte-Carlo experiments, and in an application to the NKPC.
