A1448
Title: Unveiling large dynamic systems via simultaneous impulse response inference
Authors: Mikihito Nishi - University of Tokyo (Japan) [presenting]
Yoshimasa Uematsu - Hitotsubashi University (Japan)
Takashi Yamagata - University of York (United Kingdom)
Abstract: An inferential procedure is proposed to investigate the shock propagation mechanism in large vector autoregressions. The goal is to uncover dynamic links between shock variables and outcomes, both of which may be high-dimensional. Given the high-dimensionality, the dynamic relationships are measured via the generalized impulse response functions, which are invariant to the ordering of variables. Formulated as multiple hypotheses testing, the attempt to discover potential dynamic relationships within a large system is designed to control the false discovery rate. The use of asymptotic e-variables based on post-lasso OLS estimators is proposed. Theoretical justifications and numerical experiments are presented.
