International Market Links and Volatility Transmission

Abstract

This paper proposes a framework to gauge the degree of volatility transmission among international stock markets by deriving tests for conditional independence among daily volatility measures. We suppose that asset prices follow a multivariate jump-diffusion process, and make no parametric assumption on the functional form of the drift, diffusive, and jump components. To check for conditional independence of asset A's daily volatility given asset B's daily volatility, we consider the integrated (relative) squared difference between two nonparametric conditional density estimates. The first estimate considers only information concerning asset A's daily volatility, whereas the second estimate also includes information about asset B's daily volatility. To proxy for the unobservable daily volatility, we employ model-free realized measures, allowing for both microstructure noise and jumps. We establish the asymptotic normality of the test statistic based on realized measures as well as the first-order validity of its bootstrap analog.In addition, we investigate volatility spillovers between the stock markets in China, Japan, UK and US from January 2000 to December 2005. We find significant volatility spillovers across all markets, especially if we control for jumps and/or market microstructure effects. Apart from the expected bidirectional link between the UK and the US, we uncover significant Japan and China effects on the US stock market volatility. Their impact is particularly apparent if we further conditioning on the realized measure of the FTSE 100 index to control for the presence of global shocks.