A0897
Title: Bayesian modeling of TVP-VARs using regression trees
Authors: Niko Hauzenberger - University of Strathclyde (United Kingdom) [presenting]
Florian Huber - University of Salzburg (Austria)
Gary Koop - University of Strathclyde (United Kingdom)
James Mitchell - Federal Reserve Bank of Cleveland (United States)
Abstract: In light of widespread evidence of parameter instability in macroeconomic models, many time-varying parameter (TVP) models have been proposed. The purpose is to propose a nonparametric TVP-VAR model using Bayesian additive regression trees (BART) that models the TVPs as an unknown function of effect modifiers. The novelty of the model arises from the fact that the law of motion driving the parameters is treated nonparametrically. This leads to great flexibility in the nature and extent of parameter change, both in the conditional mean and in the conditional variance. Parsimony is achieved through adopting nonparametric factor structures and the use of shrinkage priors. In an application to US macroeconomic data, the use of the model is illustrated in tracking both the evolving nature of Phillip's curve and how the effects of business cycle shocks on inflation measures vary nonlinearly with changes in the effect modifiers.
