A REVISIT OF INTERNATIONAL STOCK MARKET LINKAGES: NEW EVIDENCE FROM RANK TESTS FOR NONLINEAR COINTEGRATION
| Author | Xiao‐Ming Li |
| Date | 01 May 2006 |
| DOI | http://doi.org/10.1111/j.1467-9485.2006.00375.x |
| Published date | 01 May 2006 |
A REVISIT OF INTERNATIONAL STOCK
MARKET LINKAGES: NEW EVIDENCE
FROM RANK TESTS FOR NONLINEAR
COINTEGRATION
Xiao-Ming Li
Abstract
Based on a theory proposed for the possible link between financial market
integration and nonlinear cointegration, this study reinvestigates international
stock market linkages by performing both conventional linear cointegration tests
and newly developed rank tests for nonlinear cointegration. The stock price indexes
of Australia, Japan, New Zealand, the United Kingdom and the United States are
used, with daily data spanning from 29 May 1992 to 10 April 2001. Much more
evidence of market integration emerges from nonlinear than linear cointegration
analysis, suggesting that comovements among various national stock markets may
well take nonlinear forms. Our findings challenge the conclusion of market
segmentation reached in some previous studies that only conducted linear
cointegration analysis.
I Intro ductio n
As a result of financial globalisation, extensive literature has been devoted to
examining short- and long-run relationships among stock markets in different
countries. Studies of short-run phenomena mainly focus on the return
correlations, while studies of long-run relationships investigate the possible
existence of comovements in stock prices or the cross-country efficient market
hypothesis (EMH). As far as the latter is concerned, the presence (or absence) of
comovements implies collective market inefficiency (or efficiency) in the sense
that international investors are able (or unable) to explore arbitrage profits. This
provides them with guidance for diversifying portfolios by buying stocks from
noncointegrated markets. To this end, cointegration analysis offers an effective
method to test empirically the EMH in an international context.
The present paper falls into the aforementioned second strand of the
literature that is interested in examining the long-run relationship among
different national stock indexes. International stock market integration may be
implied by one or more of the following factors: less cross-country restrictions
n
Massey University (Albany)
Scottish Journal of Political Economy, Vol. 53, No. 2, May 2006
rScottish Economic Society 2006, Published by Blackwell Publishing, 9600 Garsington Road, Oxford OX4 2DQ, UK
and 350 Main Street, Malden, MA 02148, USA
174
on stock investment and foreign ownership (Ng et al., 1991; Lam and Pak,
1993); contagion effect (Roll, 1988; King and Wadhwani, 1990); strong
economic ties and policy coordination among countries that are in the same
continent or within the same time zone (Engle and Susmel, 1993); and deliberate
regional and global cooperation under technological and financial innovation
(Chen et al., 2002). A natural question then arises here: what kind of
cointegration, linear or nonlinear, would these factors result in? So far, all of
the studies on international stock market integration that adopt cointegration
analysis have taken the former for granted if cointegration is indeed present,
while completely ignoring the latter. See, for example, Manning (2002), Chen
et al. (2002), Ghosh et al. (1999), Chan et al. (1997), Arshanapalli and Doukas
(1993) and Chan et al. (1992). Such an assumption may well lead to incorrect or
incomplete conclusions.
To motivate the nonlinear relationship among stock indexes from different
countries, let us consider a simple theoretical model that involves only two
national stock markets: domestic and foreign. Denote the domestic and foreign
indexes by Pand Pand the corresponding rates of return by rand r,
respectively. Recall that financial integration is defined as a situation where assets
in different countries display the same risk-adjusted expected returns. That is,
when domestic and foreign stock markets are perfectlyintegrated so that investors
can instantly alter the composition of their cross-board investments, perceived
differences in risks between the two markets lead to the relationship between the
expected domestic and foreign rates of return as
r¼
rþj(assuming static
exchange-rate expectations), with jbeing the risk premium. Generally speaking,
it may be assumed that the risk premium depends positively on the risk (standard
deviation) of domestic stocks and negatively on the risk (standard deviation) of
foreign stocks: j¼jðs;sÞwith @j/@s40and@j=@so0. jðs;sÞmay take
a linear form such that jðs;sÞ¼as bs(in the special case of a5b,
jðs;sÞ¼aðssÞ, that is, the risk premium depends positively on the excess
risk of domestic stocks over foreign stocks). Using the capital market line, it can
readily be shown that jðs;sÞ¼jð
r;
rÞ¼ a=KðÞð
rrfÞ b=K
ðÞð
rr
fÞor
r¼
rþa
Kð
rrfÞ b
Kð
rr
fÞwith K
rMrf
sM
and
K
rMr
f
sM
:ð1Þ
In this equation, r
f
and r
fare, respectively, the risk-free rates on domestic and
foreign treasury bonds (taken as exogenously given),
rMand s
M
are,
respectively, the expected value and the standard deviation of the world-stock-
market rate of return (taken as exogenously given) and Kand Kare,
respectively, the prices of domestic and foreign risks (taken as exogenously
given). As
r¼dln
Pand
r¼dln
P, a differential equation may be derived from
equation (1) as follows:
1a
K
dln
P1b
K
dln
P¼b
Kr
fa
Krf:ð2Þ
REVISIT OF INTERNATIONAL STOCK MARKET LINKAGES 175
rScottish Economic Society 2006
To continue reading
Request your trial