Wavelet-based multi-resolution GARCH model for financial spillover effects

This study proposes a wavelet-based multi-resolution BEKK-GARCH model to investigate spillover effects across financial markets. Compared with traditional multivariate GARCH analysis, the proposed model can identify or decompose cross-market spillovers on multiple resolutions. Taking two highly correlated indices, the NASDAQ (U.S.) and TWSI (Taiwan composite stock index) for analysis, the empirical results show that the NASDAQ returns strongly predict the movements of TWSI on the raw data level, but via wavelet-based multi-resolution analysis we find that the prediction power unevenly spreads over each time scale, and the spillover patterns are totally different as that revealed on the raw data level. The direction and magnitude of return and volatility spillovers significantly vary with their time scales. Considering the fact that heterogeneous groups of investors trade on different time horizons, the results of this study help investors to uncover the complex pattern of return and volatility spillovers on their own horizon, and make a good hedge on their risk.