Bootstrapping volatility spillover index

Abstract

Concentrating on confidence interval, a bootstrapping method is developed for volatility spillover index proposed by Diebold and Yilmaz via a vector autoregressive (VAR) model. The method is based on residual based bootstrapping with normal pivot which is a winner of a Monte-Carlo comparison of several possibly competitive methods. The Monte-Carlo simulation demonstrates finite sample validity of the bootstrap confidence interval. The proposed method is illustrated by standard errors and confidence interval estimation for the volatility spillover indexes of various financial real data sets.