Title: Estimating volatility spillovers: A large t-Vector AutoRegressive approach
Authors: Luca Barbaglia - KU Leuven (Belgium) [presenting]
Christophe Croux - Edhec Business School (France)
Ines Wilms - Maastricht University (Netherlands)
Abstract: Commodity markets have recently experienced large price fluctuations. Volatileness and fat-tailedness have become typical features of commodity prices. We develop a general framework for the analysis of volatility spillovers accounting for fat-tailed error distribution. We propose a penalized estimator for the Vector AutoRegressive model with Student $t$-distributed errors that incorporates the estimation of the degrees of freedom of the $t$-distribution. Moreover, we provide a network analysis tool based on forecast error variance decompositions accounting for $t$-distributed errors to show the volatility spillovers between a large number of commodities. We study the dynamics of volatility spillovers between energy, biofuel and agricultural commodities. Our results highlight that the overall level of volatility spillovers increases in periods of low energy prices. Moreover, we find evidence of bidirectional volatility spillovers between energy and agriculture, which cannot be completely explained by presence of biofuels.