Multivariate Stochastic Volatility with Large and Moderate Shocks

Abstract

Summary The paper proposes a multivariate stochastic volatility model where shifts in volatility are endogenously driven by large return shocks. The model proposed generalizes the univariate stochastic volatility model of Dendramis and colleagues to a multivariate context. Allowing for multivariate dependence permits the volatility of common return factors to affect individual stock returns volatility jointly. The model is further extended to allow for endogenous thresholds that depend on covariates. Model selection priors are introduced and the new techniques are applied by using data from the FTSE100-index.