From Volatility Transmission to Contagion between Stock Markets : The Lighting of a Nonlinear VAR Model with Structural Breaks in Variance

We develops in this paper a nonlinear vector autoregressive model to study stock market interdependences. Among the innovations of this work, we introduce a structural break in the conditional variances-covariance's matrix of multivariate GARCH process. We consider a BEKK expand with shocks to volatility transmission across markets. The purpose of these amendments is to respond to several biases in the measurement of volatilities and correlations between markets: a primer bias is the shocks to volatility persistence over estimating; second, heteroskedasticity and omitted variables bias in market cross-correlation estimates. We use a sample of 11 markets from Europe, North America, and Asia with weekly data of market indices between 1985 and 2006. Several interesting results are obtained with this model: the reduction of shocks to volatility persistence, price and uncertainties transmission from U.S. market to European and Asian markets, regional transmission phenomenon in Europe and Asia, apart from the U.S. crash of October 1987, all crises are not always contagious. At last but not the least, it is not clear that financial liberalization isolates markets from instability and contagion, although the integration is a good tool for market efficiency. Crises and contagion phenomenon can be market equilibrating process