A practical multivariate approach to testing volatility spillover

We propose an asymptotic N(0,1) inferential strategy to test for volatility spillover between markets consisting of multiple sectors. First, we use nonparametric kernel method to derive test statistics that assign flexible weight to each lag order and are able to check a growing number of lags as the sample size increases. Second, we propose a practical multivariate volatility modeling approach — which enjoys estimation consistency and simplicity — to facilitate higher dimensional spillover testing. Simulations show the reasonable finite sample performance of the proposed econometric strategy in a relatively large system. An empirical application highlights the merits of the proposed approach.