We use a parsimonious approach, modeling data are as a DCC-GARCH (1,1) process,
within bivariable system of each iShare return versus the U.S. An asymmetric GARCH process
of Glosten, Jaganathan and Runkle (1993) with t-distribution is assumed. The extraction of the
conditional time varying correlations allows us to examine the short-run dynamics of the series. It
also allows us to monitor the effects attributed to the sequence of crisis events that took place
throughout the sample.
6. Empirical Findings
6.1. Gregory-Hansen Residual Based Cointegration Test Results
6.1.1 Bivariate cointegration: US SPDRS vs. G7 iShares
To analyze long-term relationships between SPDR and the G7 iShares return series, we
perform Gregory-Hansen (1996) test that allows for a single break of unknown timing in the
coefficients of the cointegration vector as given by equations (1)-(3) above. Since Gregory-
Hansen tests are sensitive to the direction of causality, we apply the test using U.S. SPDR series
first as dependent and then as independent variable. Since the results of both specifications turn
out to be virtually identical, we choose to report only the results with SPDR as the dependent
variable in Table 3 below.
[Table 3 around here]
Results of the bivariate Gregory-Hansen tests fail to reject the null hypothesis of no
cointegration for all three alternative models as applied to the six pairs of markets. At a first sight,
such a finding seems to be at odds with that of Olienyk, Zumwalt and Schwebach (1999). These
authors, using conventional Engle-Granger testing methodology, found cointegration relations
17