Keywords
multivariate garch, asymmetry, volatility, spillovers, risk, sensitivity
Start Date
1-7-2004 12:00 AM
Abstract
Although market interdependence would seem to be conceptually straightforward, being based on international fundamentals, there are no generally accepted testing strategies. This paper tests for the sensitivity of the empirical results reported in Veiga and McAleer (2004), who use the vector autoregressive moving average asymmetric generalised autoregressive conditional heteroskedasticity (VARMA-AGARCH) model of Chan, Hoti and McAleer (2002) to test for the existence of volatility spillovers among FTSE 100, SandP 500 and Nikkei 225. The existing literature is extended to analyse the robustness of the empirical results reported in Veiga and McAleer (2004) to: (1) the choice of currency used to denominate asset prices, where it is found that the results are not affected by the choice of currency; (2) the inclusion of another asset in testing for volatility spillovers, where it is found that the results can be changed substantially following the inclusion of another asset; (3) the choice of the conditional mean specification, where it is found that the results are sensitive to the choice of conditional mean specification; and (4) the stability of the conditional correlation matrix over time through the use of rolling windows, where it is found that the conditional correlations tend to be time-varying.
Testing the Sensitivity of Spillover Effects Across Financial Markets
Although market interdependence would seem to be conceptually straightforward, being based on international fundamentals, there are no generally accepted testing strategies. This paper tests for the sensitivity of the empirical results reported in Veiga and McAleer (2004), who use the vector autoregressive moving average asymmetric generalised autoregressive conditional heteroskedasticity (VARMA-AGARCH) model of Chan, Hoti and McAleer (2002) to test for the existence of volatility spillovers among FTSE 100, SandP 500 and Nikkei 225. The existing literature is extended to analyse the robustness of the empirical results reported in Veiga and McAleer (2004) to: (1) the choice of currency used to denominate asset prices, where it is found that the results are not affected by the choice of currency; (2) the inclusion of another asset in testing for volatility spillovers, where it is found that the results can be changed substantially following the inclusion of another asset; (3) the choice of the conditional mean specification, where it is found that the results are sensitive to the choice of conditional mean specification; and (4) the stability of the conditional correlation matrix over time through the use of rolling windows, where it is found that the conditional correlations tend to be time-varying.
