Asymmetry in the Transmission of Return Shocks; The Skewed-t Connectedness Framework

In this report we study return spillovers among U.S. asset classes and global stock markets based on nancial connectedness measures as introduced by Diebold and Yilmaz (2012). We show that the assumption of normally distributed residuals in the underlying VAR models is poorly substantiated. Therefore, we introduce new connectedness measures based on skewed-t distributed innovations which account for excess kurtosis and skewness and enable us to make distinctions between the transmission of negative and positive extreme shocks. We nd that the system's total connectedness is little aected by this new assumption, but that there are large dierences in the transmission of extreme positive and negative shocks among individual markets. Since the original model of Diebold and Yilmaz (2012) is nested in our new model, we conclude that it is better to use our newly proposed skewed-t connectedness measures.