Title: SVARs with breaks: Identification and inference
Authors: Emanuele Bacchiocchi - University of Milan (Italy)
Toru Kitagawa - University College London (United Kingdom) [presenting]
Abstract: The aim is to study identification of structural vector autoregressions (SVARs) with structural breaks in structural error variances and/or structural coefficients. We first provide the point-identifying conditions for the structural parameters in the presence of structural breaks, which generalizes the existing results in the literature. We then investigate scenarios where the point-identifying assumptions are relaxed and the structural parameters are only partially identified. We provide analytical characterizations of the identified set and discuss how to compute them given knowledge of reduced-form parameters. A inferential analysis is proposed based on robust Bayesian techniques.