between SPDR and every iShare series in their sample. However, their dataset included only two
and a half years of data, 18/03/1996-31/10/1998. It appears that equilibrium relations that might
have existed during 1996-1998 have not persisted in time. Our findings suggest that there may
still exist viable diversification opportunities for the US based investors over sufficiently long
time periods.
6.1.2 Bivariate Cointegration: Between G7 excluding US
From the results for the group excluding US market it follows that UK, Italy and Japan
remain segmented from other major stock markets throughout the sample period, since the null of
no cointegration cannot not rejected for these markets. Somewhat weaker evidence of
cointegration is found in the case of France - Germany (at 10 % level) and France - Canada (at
5% level). The break in the intercept occurred on 01/05/2000 for the former and on 22/05/1998
for the latter. The may 1998 period is congruent with the final notification of rates and members
for the Euro, but why only France -Canada should show a change in the relationship is puzzling.
It also comes at the time when a major Canadian purchase of French robotics technology was
reported. The 2000 period comes immediately after the resolution to a dispute between a major
French (Credit Lyonnaise) and German (Dresdner) banks regarding crossholdings. While
important, these elements however would seem to be unlikely to change market dynamics
fundamentally.
6.1.3 Multivariate Cointegration Tests’ results
To account for the influence of other markets on the returns of a single market, we turn to
the multivariate setting and employ Gregory-Hansen methodology for the group of US and
European equity markets as well as for the group of the European equity markets only. We find
18