A0416
Title: Singular vector autoregressions
Authors: Rodney Strachan - The University of Queensland (Australia) [presenting]
Eric Eisenstat - The University of Queensland (Australia)
Abstract: The purpose is to develop methods for the empirical analysis of singular processes. A strong rationale, a well-developed theoretical framework, and, as shown, empirical support exist for multivariate time series with a singular spectral density. A singular spectral density is consistent with the economic theory underlying, for example, DSGE models, in which the number of variables is greater than the number of structural shocks. This assumption guarantees the existence of a finite order VAR representation, but a unique probability density function does not exist with respect to the Lebesgue measure. A density on a compact submanifold is therefore defined with respect to the Hausdorff measure, and in a Bayesian framework, an HMC algorithm is developed that jointly samples coefficients, lag length, and the number of shocks. The proposed framework is used to carry out a structural analysis of the US macroeconomy with COVID-19 shocks.
