Measuring persistence in volatility spillovers : conference paper

This paper analyzes volatility spillovers in multivariate GARCH-type models. We show that the cross-effects between the conditional variances determine the persistence of the transmitted volatility innovations. In particular, the influence of a foreign volatility innovation on a conditional variance is even more persistent than an own innovation unless this effect is offset by an according negative variance spillover of sufficient size. Moreover, ignoring the latter causes a downward bias in the estimate of the initial impact of foreign volatility innovations. Applying the concept to portfolios of small and large firms, we find that shocks to small firm returns affect the large firm conditional variance once we allow for (negative) spillovers between the conditional variances themselves.