國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
1. INTRODUCTION
The degree of correlations among different international stock markets is an
im-portant issue for both theoretical and empirical research in international finance.
By estimating the intensity of the interdependency between national stock
mar-kets, we can measure the benefits of international portfolio diversification. While
the earlier analysis has mainly focused on major developed markets, recent research
has been extended to the linkages between emerging and developed markets. One
reason doing this is that benefits of international diversification rely increasingly
on investment in emerging markets (Goetzmann, Li, and Rouwenhorst, 2001). In
this paper we investigate the fundamental factors that affect American and
Chi-nese stock return correlations.
Given China’s increasing global importance, it is not only of academic interest
but also of significance from the perspectives of policy makers and financial analysts
to investigate if and how the Chinese stock market is correlated to global markets,
and if China’s liberalization policies on cross-country investment strengthen the
international linkages. Figure 1 shows the impact of China’s liberalization policies,
‧
captured by the net inflows of portfolio equity from 1997 to 2010. The net inflows
are clearly on the upward trend, and the growth rate is around 454% within 13
years. However the net inflows of portfolio equity are volatile, in response to
chang-ing economic conditions. This has implication for asset diversifications. If China’s
stock market becomes more integrated with overseas markets, it could decrease
the benefits of international diversification. Hence, it is interesting to investigate
the benefits of international diversification by examining to what degrees Chinese
stock market integrates with overseas markets.
[INSERT FIGURE 1 HERE]
There have been some analyses of interactions between the stock exchanges
in China (Chui and Kwok, 1998);(Kang, Liu, and Ni, 2002). Also, a few studies
have investigated interdependence between the Chinese stock market and overseas
markets. Huang, Yang, and Hu (2000), for example, using Granger causality and
cointegration test, find that stock markets’ variations of Hong Kong and Taiwan
are affected by the previous trading day’s American stock market variations. Shih,
Hsiao, and Chen (2007) examine if there is a long-time relationship among China,
the U.S., and Japan. They conclude that there is no co-integration relationship
among these markers. However, Shanghai B Shares are found to be ahead of other
three stock markets, and the U.S. stock market ahead of Shenzhen B Shares and
Japan’s. They all show that there is a lead-lag effect among China and overseas
‧
In this paper, instead, we focus on whether there are co-movements between
American and Chinese stock markets, with particular attention to decomposing
the underlying shocks into global and competitive ones. Global shocks are those
that affect the value of all firms in the same direction. Competitive shocks
in-crease the market value of firms in the same direction relative to firms in another
country. On the basis of the hypothesis that stock linkages are related to trade
links (Bekaert and Harvey, 2003), we select the U.S. stock market, China’s most
important trade partners, in this study. By sorting the nature of different shocks,
we could analyze the covariances of underlying portfolio more precisely. Most of
the past studies calculated covariance by looking at return surprises (deviations
from expected return) in each scenario as the indicator to adjust portfolio
combi-nations.
Karolyi and Stulz (1996) examine the U.S.-Japan stock return comovements
by decomposing shocks into global and competitive ones. They find that the U.S.
macroeconomic announcements, shocks to the Yen/Dollar foreign exchange rate
and Treasury bill returns have little influence on the U.S. and Japanese return
correlations. However, large shocks to aggregated market indices positively
im-pact the magnitude and persistence of the return correlations from 1988 to 1992.
Motivated by this study, we investigate if the correlations between Chinese and
American stock markets would differ from market shocks. Our analysis is enabled
‧
by applying the two-stage latent variable regression provided by Pindyck and
Rotemberg (1990) to distinguish global shocks from competitive ones.
We discuss the impact of each shock on the co-movements between two
port-folios. The shocks we choose are composed of competive and global ones, such as
returns of S&P 500, Shanghai A shares, and Hong Kong Hang Seng index, trading
volumes of each markets, exchange rate (Yuan/Dollar), and treasury bill rate.
It is important to justify the choice of these shocks. Hong Kong Hang Seng
index plays a key role in international stock markets. Chinese stock markets are
closely related to its variations recently. On the other hand, trading volumes of
each markets reflect the inflows and outflows of capitals, and they are key
in-dicators to estimate stock markets’ trend. Yuan/Dollar exchange rate impacts
economic activities on both the countries.
We use daily transactions data from 2005 to 2010 to construct overnight and
daytime returns for a portfolio of Chinese cross-listed stocks using their
NYSE-traded American Depository Receipts (ADRs) and a matched-sample portfolio of
the U.S. stocks. The portfolio of China’s ADR is composed of 12 companies which
are distributed over 11 different industries. The total market values of these stocks
play crucial roles in Chinese stock market. For each ADR, we select three matching
American firms of comparable size within the Chinese firm’s industry. Industries
are defined using four-digit Standard Industrial Classification (SIC) codes. The
reason for adopting Chinese cross-listed stocks using their NYSE-traded ADRs is
‧
non-synchronous trading periods need to be reduced in order to reveal the impact
of information on returns simultaneously. This is different from the past studies
where the lead-lag effect is emphasized. This is because the studies look at
differ-ent time intervals in differdiffer-ent stock markets.
An alternative approach to analyzing market co-movements is using
time-varying second-order methodologies such as ARCH or GARCH framework to
esti-mate the variance-covariance transmission mechanisms between countries. Hamao,
Masulis, and Ng (1990) employ ARCH model to examine the 1987 crisis in the
U.S. stock market, and find that there are price volatility spillovers from New
York to Tokyo, London to Tokyo, and New York to London. Becker, Finnerty,
and Tucker (1992), adopting GARCH-M model, on the other hand, uncover that
an hour before the opening of Japan’s stock market, the variances of prices would
reflect the latest trading conditions in the U.S. stock markets. Theodossiou and
Lee (1993) reveal that statistically significant mean spillovers radiate from stock
markets of the United States to those of the U.K., Canada, and Germany, and then
from the stock market of Japan to that of Germany. To summarize, each of these
tests reaches a similar indication: there are spillover mechanisms working among
the countries under investigation. However, this approach is unable to answer the
question of what the underlying shocks drive the co-movements.
Other series of papers examining international transmission mechanisms
at-tempt to directly measure how different factors affect co-movements by looking
‧
at return correlation coefficients. Applying the factor analysis, Ripley (1973)
investigates the systematic covariation between stock prices of developed
coun-tries. Results show that the major pattern of covariance between national indices
can offer interesting economic interpretations, for instance, interdependence on
in-ternational trade, geographic proximity, policies on monetary system, or cultural
similarities. Johnson and Soenen (2002) examine to what degrees twelve equity
markets in Asia are integrated with Japan’s equity market. Evidence suggests
that Asian markets become more integrated overtime, and factors triggering stock
markets co-movements include import shares, inflation rates, real interest rates,
and gross domestic product growth rates. However, these analyses have mainly
focused on major developed markets and are unable to understand benefits of
in-ternational diversification on investment in emerging markets.
The remainder of the study is organized as follows. Section 2 describes our
sim-ple framework of the two-stage latent regression procedure proposed by Karolyi
and Stulz (1996). Section 3 presents the sample data and reports basic statistical
results. In Section 4, we present the empirical results based on the latent variable
regression for returns of Chinese and a matching American portfolios. Also we
discuss the primary application in international portfolio diversification. Section
5 concludes.