| A conventional method of testing for financial contagion examines if cross-market correlation coefficients increase significantly after a crisis. Since the conventional test assumes that asset return dynamics are homoscedastic and cross-market correlations are time-invariant, it fails to take into account the well-documented positive relationship between time-varying correlation and volatility. To control for the volatility effects on return correlation, this dissertation proposes an alternative test based on a modified dynamic conditional correlation (DCC) model. We apply the proposed test to the 1997 Hong Kong stock market crisis. We compare our results to those of the conventional test. We find that when the volatility effects are controlled for, the evidence of contagion under the conventional test substantially weakens. We also find that contagion effects, which are related to investors' behavioral change, exhibit persistence. |
