evidence of cointegration between the US and European equity markets group at the 5 % level
with the break in both trend and intercept occurring on 31/12/1999. This of course reflects the
final transition to the euro, with the introduction of notes and coins and the withdrawals and
demonetization of the national currencies. Somewhat weaker evidence of cointegration, at 10 %
level, can also be found for the group of the three European markets (Germany, France and Italy)
with the break in the entire vector occurring on 12/07/2000.
Summing up our findings, there is little evidence in favor of bivariate cointegration
relationships between the G7 markets using SPDR and iShares returns. Our results in this respect
lend support to the existing cointegration literature for G7 countries based on analysis of broad
market indices. While Kasa (1992) found the strongest evidence in favour of cointegration
between the world largest stock markets (US, UK, Germany, Japan and Canada) using both
monthly and quarterly prices of market series over the period 1974-1990, his findings were later
questioned by Richards (1995) who pointed out the need to apply small sample critical values to
avoid over-rejection of the null hypothesis of no cointegration. Using adjusted critical values,
Richards found very weak evidence of bivariate cointegration for the dataset of sixteen developed
stock markets during 1969-1994 and argued that substantial risk-reduction benefits may arise
from investing abroad.
Our findings with respect to presence of multivariate cointegration suggest somewhat
limited diversification benefits available to US investors investing in iShares of other developed
equity markets. It is worth noting that using broader market indexes for eighteen developed and
emerging markets during 1961-1992, Chan, Benton and Pan (1997) also found somewhat weaker
evidence of cointegration in case of the four European markets (UK, Germany, France, and
Italy).
19