Leo Krippner, Singapore Management University - Specifying and estimating vector autoregressions using their eigensystem representation

Title:   Specifying and estimating vector autoregressions using their eigensystem representation

Abstract: In this seminar, based on two related papers, Leo will introduce the eigensystem vector autoregression (EVAR) framework, where VARs are explicitly specified and applied directly via their eigenvalue and unique eigenvector parameters. To demonstrate the types of benefits the EVAR offers, the applications include: (1) transforming VAR data and forecasts to AR1 component form for assessing persistent/important and transitory/negligible dynamic contributions; (2) obtaining closed-form forecasts and forecast error variances for arbitrary horizons; (3) imposing eigensystem constraints in estimations to control a VAR’s permissible dynamics, including non-explosiveness in time-varying VARs; (4) imposing zero, unit, and/or repeated eigenvalues in VAR estimations for parsimony and interpretability.

Presenter: Leo Krippner, Singapore Management University

If you would like to subscribe to the Melbourne Institute Seminar Series email list, please contact us.