A0175
Title: Large vector autoregressions with stochastic volatility in mean
Authors: Jamie Cross - Melbourne Business School (Australia) [presenting]
Aubrey Poon - University of Strathclyde (United Kingdom)
Gary Koop - University of Strathclyde (United Kingdom)
Chenghan Hou - Hunan University (China)
Abstract: Recent research has shown that incorporating large datasets and volatility feedback effects into vector autoregression (VAR) models is useful for structural analysis. However, computational complexities have made it challenging to estimate VARs with both of these features. We propose an efficient Bayesian posterior simulator to estimate large VARs with stochastic volatility in mean (SVM) dynamics. The algorithm is more efficient (both computationally and statistically) compared to conventional particle-filter based algorithms. In two empirical applications on the US economy, the large VAR-SVM model provides (1) novel macroeconomic insights about multi-sectored spillovers of uncertainty and (2) competitive out-of-sample forecasts to conventional large VAR models.
