correlations. As suggested by the results of the correlation analysis, WEBS offer better
diversification opportunities after the Asian crisis than CECF.
Phengpis and Swanson (2004), on the other hand, discuss construction of optimal portfolios
and in this context, use results from cointegration analysis to investigate whether or not
incorporating information about long run relationships can help in improving diversification
gains than relying exclusively on short-term information. The authors conclude that relying on
national indices (as opposed to iShares) to evaluate diversification gains may overstate the actual
benefits and moreover, including long term information as an additional input to portfolio
construction can improve diversification benefits.
Olienyk et al. (1999) is the starting point of the present study. We begin by considering
pairwise, bivariate cointegrating relation between various combinations of G7 markets using a
more recent cointegration technique developed by Gregory and Hansen (1996) that allows for
endogenously determined structural break and hence reflects significant improvement over prior
cointegration techniques. Next, we extend Olienyk et al. (1999) study to investigate the existence
of multivariate cointegrating relationship between G7 markets, since the absence of bivariate
relation cannot rule out the existence of multivariate relationship between markets. Subsequently,
we focus on the time varying dynamics of both long-term and short-term cointegrating
relationship by using recursive cointegration procedure of Hansen and Johansen (1999), residual-
based cointegration test of Gregory and Hansen (1996) and Dynamic Conditional Correlation
GARCH (DCC-GARCH) of Engle (2000) respectively. The resulting graphs present a detailed
picture of the time varying nature of the equity market linkages
11