The Federal Reserve Board published a research paper setting out an estimation and inference framework for time-varying impulse response functions in structural vector autoregressions identified using external instruments. The approach extends kernel-based estimators to allow for nonparametric time variation and derives asymptotic distributions for key quantities, with methods designed to remain usable even when instruments may be weak. The paper describes the resulting estimators as simple and computationally trivial, and reports simulation evidence of satisfactory empirical coverage in relatively small samples when parameter instabilities evolve smoothly. An empirical illustration applies the methods to assess how the effects of global oil supply news shocks on US industrial production change over time.
Federal Reserve Board 2025-01-01
Federal Reserve Board research develops nonparametric methods to estimate time-varying impulse responses in IV-SVARs
The Federal Reserve Board released a research paper introducing a framework for estimating time-varying impulse response functions in structural vector autoregressions using external instruments. The approach enhances kernel-based estimators for nonparametric time variation and provides asymptotic distributions, remaining usable with weak instruments. An empirical example evaluates the changing impact of global oil supply news shocks on US industrial production.