A generalised stochastic volatility in mean VAR

This paper introduces a VAR with stochastic volatility in mean where the residuals of the volatility equations and the observation equations are allowed to be correlated. This implies that exogeneity of shocks to volatility is not assumed apriori and structural shocks can be identified ex-post by applying standard SVAR techniques. The paper provides a Gibbs algorithm to approximate the posterior distribution and demonstrates the proposed methods by estimating the impact of financial uncertainty shocks on the US economy.