Financial Integration and China's Stock Market
Financial Integration and China's Stock Market1
R. D. Terrell
Since 1980, China's economy has gradually developed under the guidance of Deng Xiaoping's theory. Following a series of reforms which have opened up its socialist market economy, China has steadfastly increased its foreign trade and actively attracted foreign investment. China's trade value ranks fifth in the world and its annual inflow of foreign direct investment (FDI) amounting to about US$40 billion in recent years, is the highest among the developing countries. The reforms and the opening up of the country to the outside world have not only promoted the sustained, swift, and sound development of its national economy, but have also helped to restructure its economic system.
Modern China has become increasingly integrated into the global economy, especially with the United States, the United Kingdom, Japan, Taiwan, Hong Kong, and Singapore. For example, bilateral trade turnover (the sum of exports and imports) between China and the United States reached US$74.5 billion in 2000, more than thirty times that of 1979. In 2002, exports from the United Kingdom to China amounted to US$3.3 billion, while imports from China to the United Kingdom reached US$8.1 billion. Exports and imports between China and Japan were valued at US$48.44 billion and US$53.47 billion, respectively. In 2001, Taiwan's exports from, and imports to, Mainland China were US$5 billion and US$27 billion, respectively. At the same time, Hong Kong's imports from China reached HK$606 billion, or 43 percent of its total imports, while its exports to China accounted for 30 percent of its total exports, making China its second largest export destination. On the other hand, the bilateral trade volume between China and Singapore was US$14 billion in 2002.
1 Our deepest thanks to Lean Hooi and Yuan Ying for their assistance in writing this chapter. This research was partially supported by a grant from the National University of Singapore.
The stock markets in China have also developed rapidly, especially in recent years. By 2002, the total market capitalization of the Shanghai Stock Exchange (SHSE) was RMB2,536.4 billion, and the number of investors on the SHSE totalled 35,556,500. At the same time, capital raised from the SHSE surpassed RMB61.3 billion. The Shanghai Composite Index, compiled to reflect the stock price movements on the SHSE, is a weighted average stock price index, with the weighting being determined by the number of shares issued by all listed companies. The base day is December 19, 1990 and the index has been officially published since July 15, 1991 (Shanghai Stock Exchange, 2002).
Having become a member of the World Trade Organization (WTO), it is expected that China will further open its capital markets. However, it is unclear whether the degree of correlation between the returns of market indices in China and those of the major developed countries or regional economies will be comparable to the integration between these economies, and whether investment in the Chinese stock markets can diversify the risk of portfolio investment, but they will certainly draw the attention of global investors. The hypothesis here is that the Chinese stock markets may have uncorrelated relationships with those of the major developed countries and regional counterparts. Hence, investment in the Chinese stock markets can represent a feasible element of a portfolio to enhance the reward-to-volatility ratio even though China is being integrated into the global economy only gradually. The finding of a differential impact of Chinese stock markets on other stock markets can generate further insights into socioeconomic connections. Specifically, investigation into co-movement relationships will provide useful information to both domestic and foreign investors.
The benefit of international diversification is, however, limited when national equity markets are co-integrated, because the presence of common factors limits the extent of independent variation. Co-integration among national equity markets implies that there are fewer assets available to investors for portfolio diversification compared with a simple count of the stocks. Moreover, co-integration would mean Granger causality in levels and hence would be suggestive of inefficiency in the market.
This chapter contributes to the literature by investigating whether stock co- movements exist between the stock markets of China and the three world leaders—the United States, the United Kingdom, and Japan—and its three regional counterparts—Taiwan, Hong Kong, and Singapore—both before and after the Asian financial crisis of 1997–1998. The study first examines the co-movements between China's stock markets and these six stock markets by employing the Engle–Granger (1987) two-step co-integration technique. Next, the minimum final prediction error criterion is employed to determine the optimum lag structures (Hsiao, 1979, 1981). Finally, the error correction model (ECM) or vector autoregressive model (VAR) is used to find the causal relationship between the Chinese stock markets and the other six stock markets. The results provide evidence that only the stock markets of Japan and Taiwan were co-integrated with the Chinese stock market before the crisis. However, after the crisis, all six stock markets are co-integrated with the Chinese stock market. Moreover, the Chinese stock market is more co-integrated with those of its three regional counterparts than with those of the three world leaders. Their close socioeconomic, trade, and cultural relationships with China help support this important feature for global investors.
The chapter next provides a review of the literature and then presents the data and methodology. The following section discusses the empirical findings and interpretation of the results, and this is followed by a summary.
A number of studies have examined the co-movements of international stock markets. Considerable work has been done on the interrelationships among world equity markets, especially those of the major developed countries such as the United States and Japan. The presence of co-movements among national stock markets limits the benefit of international diversification.
The performance of the developed markets was the focus of world attention before and after the stock market crash of October 1987. The crash made people realize that most national equity markets are closely integrated. The developed markets, notably the United States, exert a strong influence on other smaller markets. Eun and Shim (1989) use vector auto-regression to study the interdependence among world equity markets and find evidence of co-movements among these markets with the United States. By using a single equation model, Cheung and Mak (1992) examine the causal relationships between the Asian and the developed markets. They also find that the U.S. market is an important global factor. Lee and Kim (1994), following a correlation approach, examine the effect of the October 1987 crash on the co-movements among the national stock markets. They find that the national stock markets became more interrelated after the crash, and the strengthening co-movements among the national stock markets continued for a longer period after the crash. In addition, the co-movements among the national stock markets were stronger when the stock markets in the United States became more volatile.
There is also a literature on price discovery in world markets. Naturally, co-integration and error correction modeling provides a useful framework for analysing price adjustments in internationally linked markets. Mclnish and Wood (1992) have reported that regional exchanges contain information that is relevant for traders on primary markets (Garbade and Silber, 1979).
In recent years, new capital markets have emerged in many parts of the world, and foreign capital controls have also been relaxed to some extent. With this relaxation of capital controls there has been an increase in investors’ interest in international diversification, as it gives investors a larger basket of foreign securities to choose from and add to their portfolio assets in order to diversify investment risk. A number of studies have examined co-movements in stock returns with reference to the expected returns and diversification benefits of emerging-market investments (Harvey, 1991; and Wilcox, 1992).
Asian capital markets are new stars among the emerging markets. Many studies have been done in the 1990s and thereafter to study the co-movements between Asian markets and the stock markets in the developed countries. Chan et al. (1992) conducted unit root and pairwise co-integration tests to examine the relationship among the Asia-Pacific markets,
and concluded that these markets were not co-integrated. Chowdhury (1994) studied the relationships among four newly industrialized economies (NIEs), Japan, and the United States. He found that the U.S. market led the four NIEs (Hong Kong, South Korea, Singapore, and Taiwan) and there were significant links between the stock markets of Hong Kong and Singapore, Japan, and the United States. Phylaktis (1995) found that there was increased integration of capital markets in the Pacific Basin region with those of the United States and Japan. Kwan et al. (1995) studied the stock markets of Australia, Hong Kong, Japan, Singapore, South Korea, Taiwan, the United Kingdom, the United States, and Germany, and suggest that these markets were not weak form efficient as they found significant lead-lag relationships between these equity markets. Cashin et al. (1995) used co-integration tests to assess the extent to which equity prices moved together across countries and regions. They reported increased integration of emerging equity markets since the beginning of 1990 as a result of greater regionalization of national stock markets. Moreover, if national stock markets experience a global shock that causes them to deviate from their long-run equilibrium relationship, it would take several months for the long-run relationship to reassert itself.
Palac-McMiken (1997) used monthly ASEAN market indices (Indonesia, Malaysia, Philippines, Singapore, and Thailand) between 1987 and 1995, and found that, with the exception of Indonesia, all the markets were linked with each other and were not collectively efficient. He suggested that there was still room for diversification across these markets despite the evidence of interdependence among the ASEAN stock markets. Masih and Masih (1999) found a high level of interdependence among markets in Thailand, Malaysia, the United States, Japan, Hong Kong, and Singapore from 1992 to 1997. Johnson and Soenen (2002) studied equity market integration between the Japanese stock market and twelve other markets in Asia. They concluded that the equity markets of Australia, Hong Kong, Malaysia, New Zealand, and Singapore were highly integrated with the stock market in Japan. The results from Yang et al. (2003) show that both long-run co-integration and short-run causal linkages among the United States, Japan, and ten Asian emerging markets have become more integrated since the Asian financial crisis. It appears that past empirical studies on the relationship between world stock markets do not provide consistent results. The reasons for this are numerous, including the choice of markets, the different sample periods, the frequency of observations, and the various methodologies employed. Because of the increasing importance and integration of the Chinese economy in the world market, this chapter takes China into account where it had not been examined previously. The goal of assessing co-movements between China and the six stock markets is unique to this study.
The weekly stock indices of the major stock exchanges in China (Shanghai Composite), the United States (S&P 500), the United Kingdom (FTSE 100), Japan (Nikkei 225 Stock Average), Taiwan (Taiwan SE Weighted), Hong Kong (Hang Seng), and Singapore
(Straits Times Index) are used in this study. All the indices are expressed in terms of local currencies and obtained from DataStream. The sample covers the period from January 1, 1991 through June 30, 2003. Weekly indices are used to avoid representation bias from some thinly traded stocks, that is, the problem of non-synchronous trading. In addition, the Wednesday indices are used to avoid the “day-of-the-week” effect of stock returns (Lo and MacKinlay, 1988). The sample was divided into two periods to look at the effects of the Asian financial crisis: January 1, 1991 to December 31, 1996 (before the crisis), and January 1, 1997 to June 30, 2003 (after the crisis).
To examine the co-movements between the stock indices in China and the six markets with the three developed world leaders, the United States, the United Kingdom, Japan, and the three regional counterparts, Taiwan, Hong Kong, and Singapore, the following relationship is used:
where Y t denotes the Chinese stock index; X t denotes the index of any of the six other countries, and e t denotes the error term. As the stock indices are likely to be non- stationary, co-integration plays a major role in determining the validity and reliability of the relationship.
Co-integration tests, which are important in determining the presence and nature of equilibrium economic relationships, was first introduced by Granger (1981) and later developed by Granger (1987). A detailed description of co-integration can be found in Hamilton (1994), Manzur et al. (1999), Tiku and Wong (1998), Tiku et al. (2000), Penm et al. (2003), and Wong et al. (2004). Before testing for co-integration, a unit root test has to be performed to determine non-stationarity for both endogenous and exogenous variables.
Co-integration tests in this study consist of two steps. The first step is to examine the stationary properties of the various stock indices. If a series, yt, has a stationary, invertible, and stochastic ARMA representation after differencing d times, it is said to be integrated of order d, and is denoted as yt = I(d). To test the null hypothesis H0:yt=I(1) versus the alternative hypothesis H1:yt=I(o), the Dickey–Fuller (1979, 1981) unit root test procedure is applied, based on the OLS regression
or the augmented Dickey–Fuller (ADF) test, based on the OLS regression
where ∇yt =yt -yt-1 and yt can be Y t or X t as defined in (1). The regressions in (2) and (3) allow for a drift term (b0) and a deterministic trend (a0). The regression in (3) allows a stochastic structure in the error term, t, while p is chosen in equation (3) to achieve white noise residuals. Testing the null hypothesis of the presence of a unit root in y t is equivalent to testing the hypothesis that a1=0 in equations (2) and (3). If a1 is significantly less than zero, the null hypothesis of a unit root is rejected. When p = 0, the test is known as the Dickey–Fuller (DF) test. This test assumes that the residuals t are independently
and identically distributed. If serial correlation exists in the residuals, then p. >0 and the augmented Dickey–Fuller (ADF) test must be applied.
In addition, a test is done on the hypothesis that y t is a random walk with drift, that is (b0,a0,a1) =(b0, 0, 0), and yt is a random walk without drift, (b0,a0, a1) =(0, 0, 0), using regression (2). The test statistics use the likelihood ratios Φ3 or Φ2 found in Dickey and Fuller (1981). If the hypotheses that a=0, (b0,a0,a1) = (b0, 0, 0) or (b0,a0,a1) = (0, 0, 0) are accepted, it can be concluded that y t is I(1). If the hypotheses that y t is I(1) is rejected, then the null hypothesis H 0:y t = I(2) needs to be tested against the alternative hypothesis H1:yt = I(1). Note that most non-stationary series are integrated of order one.
If both Y t and X t are of the same order, say I(d), with d. >0, the co-integrating parameter of (1) is estimated using OLS regression. If the residuals of equation (1) are stationary, the two series, Y t and X t are called co-integrated. Otherwise, Y t and X t are not co-integrated.
The most common tests for the stationarity of estimated residuals are Dickey–Fuller (CRDF), and Augmented Dickey–Fuller (CRADF) tests based on the OLS regression:
where êt are residuals from the co-integrating regression (1) and p is chosen to achieve white noise residuals.
Engle and Granger (1987) pointed out that when a set of variables is co-integrated, a vector auto-regression in first differences will be misspecified. The first differencing of all the non-stationary variables means that any potentially important long-term relationships between the variables will be unclear. Thus, inferences based on vector auto-regressions in first differences may lead to incorrect conclusions (Granger, 1981, 1988; and Sims et al., 1990). However, there exists an alternative error correction representation of such variables, which takes into account both the short- and long-term equilibrium relationships between these variables.
Once the co-integration relationship between the Chinese stock market and those of other countries has been decided, a bivariate VAR model can be adopted to test for Granger causality. If there is no co-integration between the two markets, following Granger et al. (2000), the following is used:
where Y t and X t represent the index of the Chinese stock market and any of the other six stock markets, respectively, n and m are the optimum lags, and u1t and u2t are the error terms. A test of the null hypothesis, H0:a21 = a22 = … = a2m = 0 implies that none of these six stock markets Granger cause the Chinese stock market. Similarly, another test H0: b21 = b22 = … = b22m = 0 is done to confirm that the Chinese stock market does not Granger cause any of these six stock markets.
If the series are co-integrated, there is a long-term, or equilibrium, relationship among the series. Their dynamic structure can be exploited for further investigation. An error correction model (ECM) abstracts the short- and long-run information in the modeling process. The ECM, first used by Sargan (1984) and later popularized by Engle and Granger (1987), corrects for disequilibrium in the short run. Engle and Granger (1987) show that co-integration is implied by the existence of an error correction representation of the indices involved. An important theorem, known as the Granger representation theorem, states that if two variables Y and X are co-integrated, then their relationship can be expressed as an ECM (Gujarati, 2003). In this situation, an error correction term (et-1 = Yt-1− δ Xt-1) is added to the equation to test for Granger causality.
The existence of co-integration implies causality among the set of variables as manifested by |a| + |b|> 0, where a and b denote speeds of adjustment (Engle and Granger, 1987). If H0: a21 = a=22 = … = a2m = 0 and a = 0 is not rejected, it means that none of the six stock markets Granger cause the Chinese stock market. Similarly, if H0: b21 = b22 = … = b2m = 0 and b = 0 cannot be rejected then the Chinese stock market does not Granger cause the other six stock markets individually (Granger et al., 2000).
To test the hypothesis H0: a21 = a22 = … = a2m = 0, the sum of the squared residuals is found for both the full regression, SSE F, and the restricted regression, SSER, in (6) and the F test is applied:
where N is the number of observations, and n and m are defined in (5) or (6). If H 0 is true, F is distributed as F(m, N − m − n − 2). Hence, the hypothesis H 0 is rejected at αlevel of significance if F. > F(>;m, N − m − n − 2), and the reduced model is accepted if H0 is not rejected. Similarly, the hypothesis: H0: b21 = b22 = … = b2m = 0 can be tested and a decision made on the causality. The usual simple t statistics are applied to test H 0: a = 0 and H 0: b = 0.
The minimum final prediction error criterion (FPE) from Hsiao (1979, 1981) is used to determine the optimum lag structures in regressions (5) and (6), where n and m are the maximum lags of the corresponding variables to be used on the right-hand side of equations (5) and (6); and u1t and u2t are disturbance terms obeying the assumptions of the classical linear regression model. The FPE statistic of ∇Yt with n lags of ∇Yt and m lags of ∇X t is
where N is the number of observations. The FPE statistic for ∇ Xt is found in the same way. To determine the minimum FPE∇ Y, the first step is to run the first regression in equation (5) excluding ∇ Xt and including only lags of ∇Yt. Starting from m = 0 and n = 1, a calculation is made for FPE∇ Y (1,0). The same steps are repeated until n = n* where FPE is minimized for m = 0. Then, by holding n = n*, m is systematically lagged until m = m* minimizes the FPE. The same procedure is repeated with the second regression in equation (5) where n = n** and m = m** minimizes FPE∇X1(n, m).
Table 9.1 shows the results for testing the order of integration of the seven series, both before and after the Asian financial crisis. All seven stock indices are I(1)2 at the 5 percent significance level for both periods. Some researchers believe there is a lead-lag effect
|Table 9.1 Unit root test results for the weekly indices in the Chinese stock market and six others|
1. DF is the Dickey–Fuller t–statistic; ADF is the augmented Dickey–Fuller statistic.
|2. Φ2 and Φ3 are the Dickey–Fuller likelihood ratios.|
|3. *denotes p < 0.05; **denotes p < 0.01.|
|Shanghai SE Composite||1997–2003||–1.89||–1.89||0.99||2.80|
|S&P 500 Composite||1997–2003||–2.00||–2.00||1.35||3.74|
|Nikkei 225 Stock Average||1997–2003||–2.00||–2.00||0.93||2.07|
|Taiwan SE Weighted||1997–2003||–2.68||–2.68||0.16||3.71|
among different stock markets (Kwan et al., 1995). Cross–correlation is employed to account for any lead–lag effects between the SHSE daily index and the other stock markets’ daily indices.3 The findings4 show that there are no strong lead–lag effects and thus the results originally obtained are a good measure for testing co–movement of stock indices between China and other markets.
Having established that the stock indices in this study are all I(1), the co–integrating equation (1) is then estimated and unit root tests are conducted on the residuals from equation (4) to determine co–integration. Table 9.2 shows that in the period before the Asian financial crisis, only Japan and Taiwan were co–integrated with the Chinese stock market at the 5 percent significance level, and Japan was also co–integrated with China at the 1 percent significance level. However, in the period after the crisis, Table 9.2 shows that all six stock markets were co–integrated with the Chinese stock market at the 5 percent significance level. Furthermore, the three regional counterparts, Taiwan, Hong Kong, and Singapore, were co–integrated with China at the 1 percent significance level. This suggests that after the crisis, the Chinese stock market has become more integrated into the global economy, especially with its regional counterparts. These relationships could be related to geographical proximity, partnerships in trade, and cultural and historical similarities. Yang et al. (2003) point out that the Asian crisis altered the degree of market integration in the region over time. Although China was not included in their study, this study extends their approach to include China.
|Table 9.2 Co–integration results for the Chinese stock market and six others|
|Before crisis (1991–1996)|
|1. CRDF is the co–integration regression Dickey–Fuller statistic for stationarity of the estimated residuals.|
|2. CRADF is the comparable test statistic for the augmented Dickey–Fuller test.|
|3. *p < 0.05 and **p < 0.01.|
|United States||Y(China) = - 4.6332 + 1.7756Y (US) (- 4.63) (10.94)||0.2780||–1.74||–1.74|
|United Kingdom||Y(China) = -11.3909 + 12.2106Y (UK) (- 7.39) (11.49)||0.2980||–1.76||–1.76|
|Japan||Y(China) 5 38.959 =- 3.2994Y (Japan) (18.94) (-15.87)||0.4475||–2.61 **||–2.61 **|
|Taiwan||Y(China) 5 6.1897 + 0.0147Y (Taiwan) (3.87) (0.08)||0.0000||–2.13*||–2.13*|
|Hong Kong||Y(China) 5 =- 4.4922 + 1.2157Y (Hong Kong) (-7.70) (18.55)||0.5252||–1.91||–1.91|
|Singapore||Y(China) 5 =- 4.3212 + 1.45Y (Singapore) (=- 4.23) (10.41)||0.2582||–1.74||–1.74|
|After crisis (1997–2003)|
|United States||Y(China) 5 3.4549 + 0.5497Y (US) (9.18) (10.25)||0.2377||–2.06*||–2.06*|
|United Kingdom||Y(China) 5 4.6906 + 0.3053Y (UK) (8.40) (4.69)||0.0613||–2.16*||–2.16*|
|Japan||Y(China) 5 9.3015 =- 0.2086Y (Japan) (23.83) (-5.10)||0.0717||–2.19*||–2.42*|
|Taiwan||Y(China) 5 10.1419 =- 0.3224Y (Taiwan) (28.69) (-8.01)||0.1600||–2.55*||–2.71 **|
|Hong Kong||Y(China) 5 4.1844 + 0.333Y (Hong Kong) (8.92) (6.67)||0.1165||–2.69**||–2.69**|
|Singapore||Y(China) 5 5.4315 + 0.2536Y (Singapore) (14.44) (5.00)||0.0690||–2.67**||–2.67**|
One possible explanation for the existence of co–integration between the Chinese stock market and other markets is that it is the outcome of economic reforms and the opening up of China. Since the 1980s, the Chinese economy has been gradually integrated into the global economy, especially with world leaders such as the United States, the United Kingdom, and Japan, and with its regional counterparts, such as Taiwan, Hong Kong, and Singapore. China's comprehensive national purchasing power has been remarkably strengthened, and a high gross domestic product (GDP) built up. China is now the seventh largest trading nation in the world. The Chinese government has reduced tariff rates and opened up the country to international trade in goods and services. China is progressively liberalizing its service sectors, such as finance, insurance, telecommunications, transportation, and tourism. With greater economic integration with the world, the Chinese stock market cannot remain isolated. Thus, all fundamental economic factors are reflected in the performance of the Chinese stock market.
Since all six markets are co–integrated with that of China after the crisis, an ECM is employed to test for Granger causality. The ECM is only applied to Japan and Taiwan before the crisis where there is evidence of co–integration, and the other markets are tested by VAR. The significance of a 0 in Table 9.3 rejects the null hypothesis that any other stock market does not Granger cause the Chinese stock market. However,
|Table 9.3 Granger causality results for the Chinese stock market and six others|
|Before crisis (1991–1996)|
|1. NA means ECM not applicable in the model because of the absence of co–integration between the two variables.|
|2. ® Implies Granger causes.|
|3. a = p–values of F test on H0: a 21 = a 22 = · · · = a2m = 0 or H 0: b 21 = b 22 = · · · = b 2m = 0;|
|4. b = p–values of t test on H0:a = 0 or H 0:b = 0 in ECM model.|
|5= *p < 0.05 and **p < 0.01.|
|United States (us)||us → cn||1||1||0.3777||NA|
|cn → us||6||1||0.4077||NA|
|United Kingdom (uk)||uk → cn||1||2||0.2165||NA|
|cn → uk||2||6||0.0326*||NA|
|Japan (jp)||jp → cn||1||1||0.1381||0.0385*|
|cn → jp||1||3||0.0259*||0.4765|
|Taiwan (tw)||tw → cn||1||1||0.4489||0.0376*|
|cn → tw||6||1||0.9294||0.3985|
|Hong Kong (hk)||hk → cn||1||2||0.0355*||NA|
|cn → hk||1||1||0.4569||NA|
|Singapore (sg)||sg → cn||1||3||0.0361 *||NA|
|cn → sg||5||1||0.6111||NA|
|After crisis (1997–2003)|
|United States (us)||us → cn||6||1||0.9444||0.0086**|
|cn → us||1||1||0.3155||0.7970|
|United Kingdom (uk)||uk → cn||6||1||0.6545||0.0070**|
|cn → uk||4||1||0.3337||0.2238|
|Japan (jp)||jp → cn||6||1||0.6805||0.0140*|
|cn → jp||1||3||0.0043**||0.7686|
|Taiwan (tw)||tw → cn||6||1||0.0762||0.0122*|
|cn → tw||1||1||0.6683||0.5896|
|Hong Kong (hk)||hk → cn||6||1||0.3382||0.0006**|
|cn → hk||4||1||0.1183||0.5979|
|Singapore (sg)||sg → cn||6||1||0.2163||0.0007**|
|cn → sg||5||1||0.5607||0.3510|
the null that the Chinese stock market does not Granger cause any other stock market cannot be rejected. Therefore, unilateral causality exists in each of these six stock markets, except Japan, relative to the Chinese stock market, especially after the crisis.
The results described above suggest the existence of Granger causality between the six stock markets and the Chinese stock market, but cannot explain why the direction is unilateral. The results in Table 9.3 show that H0: b 21 = b22 = … = b2m = 0 and b = 0 is not rejected for the second period except for Japan. This implies that the Chinese stock market does not Granger cause any of the other stock markets. One possible reason is that the Shanghai stock market is a “policy” market. According to a study conducted in China, more than 50 percent of the significant market movements were caused by changes in trading rules, or changes in policies (Jin, 2001). Changes in rules in China could influence the movement of its stock market, but should not have any effect on other markets. If this view is correct, the Chinese stock market does not Granger cause other stock markets. Moreover, there is evidence of a disconnection between stock returns and real economic growth in China. For instance, the annual return of the Shanghai Composite Index was 24 percent in 1998, while the GDP growth rate for that year was 7.8 percent. In 2000, because of a policy shift that was favorable to the stock market, the annual return of the Index was a staggering 52 percent even though China's GDP growth for that year was 8 percent. In 2001, the return swung to 221 percent, when the government sold a huge number of state–owned shares in the secondary market, but GDP growth was 7.3 percent. Another possible reason is the speculative nature of the Shanghai stock market. Stock prices often do not reflect the underlying assets of the firms. Moreover, the Chinese stock market is still in its infancy, not mature enough to Granger cause other markets in the world.
The economies of China, the United States, the United Kingdom, Japan, Taiwan, Hong Kong, and Singapore have become increasingly integrated as a result of growing bilateral trade and direct investment. The purpose of this chapter has been to see whether growing economic integration is reflected in price movements between the Chinese stock market and the other six markets. The results show that after the Asian financial crisis there was significant co–integration between the Chinese stock market and each of the six stock markets studied, between China and the three regional markets together, and between the three world markets together, and between all six markets together. However, before the crisis there was no co–integration between the Chinese stock market and those of the United States, the United Kingdom, Hong Kong, and Singapore, but there was cointegration between China and the three regional markets together, the three world leaders together, and all six markets together. This implies that economic integration has been incorporated into the performance of stock markets in the long–term, especially after the Asian crisis. Furthermore, it has been found that the three regional counterparts— Taiwan, Hong Kong, and Singapore—are more co–integrated with China after
the crisis. This could be explained by geographical proximity, partnerships in trade, and cultural and historical similarities. The results are significant to the exchange rate effect when the Chinese yuan is pegged to the U.S. dollar. It has also been found that all these stock markets, except Japan, Granger cause the Chinese stock market, but not vice versa. This unilateral causal relationship, may be due to the economic relationship, regulatory structures, exchange rate policy, and trade flows between the countries concerned. Thus, the Chinese stock market is still at the developing stage in terms of Granger causing other stock markets.5
5 Note that the co–integration and causality tests employed work well because of the large sample size but may not be applicable when the sample size is small. In this situation, one may use the Modified Maximum Likelihood Estimator approach to modify the tests (Tiku et al., 2000), or the robust Bayesian sampling estimators (Matsumura et al., 1990; and Wong and Bian, 2000). One can also use a “distribution–free” approach (Wong and Miller, 1990), or advanced time series techniques (Penm et al., 1992, 1993, 1995, 1997; and Brailsford et al., 2000, 2001, 2002).
Brailsford, T., J. Penm, and R. Terrell. 2000. A robust algorithm in sequentially selecting subset time series systems using neural networks. Journal of Time Series Analysis 21: 389–412.
Brailsford, T., J. Penm, and R. Terrell. 2001. The adjustment of the Yule–Walker relations in VAR modelling: The impact of the euro on the Hong Kong stock market. Multinational Finance Journal 5 (1): 35–58.
Brailsford, T., J. Penm, and R. Terrell. 2002. Selecting the forgetting factor in subset autoregressive modeling. Journal of Time Series Analysis 23 (6): 629–50.
Cashin, P., S. Kumar, and J. McDermott. 1995. International integration of equity markets and contagion effects. Working Paper 110, International Monetary Fund, Washington, DC.
Chan, K., B. Cup, and M. Pan. 1992. An empirical analysis of stock prices in major Asian markets and the United States. The Financial Review 27: 289–307.
Cheung, Y., and S. Mak. 1992. A study of the international transmission of stock market uctuation between the developed markets and the Asian Paci c market. Applied Financial Economics 2: 43–47.
Chowdhury, A. 1994. Stock market interdependencies: Evidence from the Asian NIEs. Journal of Macroeconomics 16 (4): 629–51.
Dickey, D., and W. Fuller. 1979. Distribution of the estimates for autoregressive time series with unit roots. Journal of the American Statistical Association 74: 427–31.
Dickey, D., and W. Fuller. 1981. Likelihood ratio statistics for autoregressive time series with unit roots. Econometrica 49 (4): 1057–72.
Engle, R., and C. Granger. 1987. Cointegration and error correction: Representation, estimation and testing. Econometrica 55: 251–76.
Eun, C., and S. Shim. 1989. International transmission of stock market movements. Journal of Financial and Quantitative Analysis 24: 41–56.
Garbade, K.D. and Silber, W.L.1979. Dominant and satellite markets: A study of dually–traded securities. Review of Economics and Statistics 61 (2): 455–60.
Granger, C. 1981. Some properties of time series data and their use in economic model speci cation. Journal of Econometrics 29: 121–30.
Granger, C. 1987. Developments in the study of cointegrated economic variables. Bulletin of Economics and Statistics 48: 213–28.
Granger, C. 1988. Some recent developments in the concept of causality. Journal of Econometrics 39: 199–211.
Granger, C., B. Huang, and C. Yang. 2000. A bivariate causality between stock prices and exchange rates: Evidence from the recent Asian u. The Quarterly Review of Economics and Finance 40: 337–54.
Gujarati, D. 2003. Basic Econometrics.4th ed. London: McGraw–Hill.
Hamilton, J. 1994. Time series analysis. Princeton. NJ: Princeton University Press.
Harvey, C. 1991. The world price of covariance risk. Journal of Finance 46 (1): 111–57.
Hsiao, C. 1979. Autoregressive modeling of Canadian money and income data. Journal of the American Statistical Society 74: 553–60.
Hsiao, C. 1981. Autoregressive modeling of money income causality detection. Journal of Monetary Economics 7 (1): 85–106.
Jin, Dehuan. 2001. Signi cant price movements in the Chinese stock markets. Working Paper, Finance and Economics Department, Shanghai University.
Johnson, Robert, and L. Soenen. 2002. Asian economic integration and stock market co–movement. The Journal of Financial Research 25 (1): 141–57.
Kwan, A., A. Sim, and J. Cotsomitis. 1995. The casual relationship between equity indices on world exchanges. Applied Economics 27: 33–37.
Lee, S., and K. Kim. 1994. Does the October 1987 crash strengthen the co–movements among national stock markets? Review of Financial Economics 3 (1–2): 89–102.
Lo, Andrew, and A. MacKinlay. 1988. Stock prices do not follow random walks: Evidence from a simple speci cation test. Review of Financial Studies 1: 41–66.
Manzur, M., W. Wong, and I. Chau. 1999. Measuring international competitiveness: Experience from East Asia. Applied Economics 31: 1383–91.
Masih, Abul, and R. Masih. 1999. Are Asian market uctuations due mainly to intraregional contagion effects? Evidence based on Asian emerging stock markets. Paci c Basin Finance Journal 7 (3–4): 251–82.
Matsumura, E., K. Tsui, and W. Wong. 1990. An extended multinomial–Dirichlet model for error bounds for dollar–unit sampling. Contemporary Accounting Research 6 (2–I): 485–500.
Mclnish, T., and R. Wood. 1992. An analysis of intraday patterns in bid–ask spreads. Journal of Finance 47 (2): 753–64.
Palac–McMiken, E. 1997. An examination of ASEAN stock markets: A cointegration approach. ASEAN Economic Bulletin 13 (3): 299–311.
Penm, J.H.W., J.H.C. Penm, and R. Terrell. 1992. Using the Bootstrap as an aid in choosing the approximate representation for vector time series. Journal of Business and Economic Statistics 10 (2): 213–19.
Penm, J.H.W., J.H.C. Penm, and R. Terrell. 1993. The recursive tting of subset VARX models. Journal of Time Series Analysis 14 (6): 603–19.
Penm, J.H.W., J.H.C. Penm, and R. Terrell. 1995. A note on the sequential tting of subset autoregressions using the prewindowed case. IEEE Transactions on Signal Processing 43 (1): 322–27.
Penm, J.H.W., J.H.C. Penm, and R. Terrell. 1997. The selection of zero–non–zero patterned cointegrating vectors in error–correcting modeling. Econometric Reviews 16 (3): 281–304.
Penm, J.H.W., R. Terrell, and W. Wong. 2003. Causality and cointegration tests in the framework of a single zero–non–zero vector time series modeling. Pakistan Journal of Applied Sciences 3 (4): 247–55.
Phylaktis, K. 1995. Capital market integration in the Paci c region: An analysis of real interest rate linkage. Working Paper 133, International Monetary Fund, Washington, DC.
Sargan, J. 1984. Wages and prices in the United Kingdom: A study in econometric methodology, quantitative economics and econometric analysis. Oxford: Basil Blackwell.
Shanghai Stock Exchange. 2002. Shanghai stock exchange fact book 2002.
Sims, C., J. Stock, and M. Watson. 1990. Inference in linear time series models with some unit roots. Econometrica 58: 113–44.
Tiku, M., and W. Wong. 1998. Testing for unit root in AR(1) model using three and four moment approximations. Communications in Statistics:–b Simulation and Computation 27 (1): 185–98.
Tiku, M., W. Wong, D. Vaughan, and G. Bian. 2000. Time series models with nonnormal innovations: Symmetric location–scale distributions. Journal of Time Series Analysis 21 (5): 571–96.
Wilcox, J. 1992. Taming frontier markets. Journal of Portfolio Management 19 (1): 51–55. Wong, W., and G. Bian. 2000. Robust Bayesian inference in asset pricing estimation. Journal of Applied Mathematics & Decision Sciences 4 (1): 65–82.
Wong, W., and R. Miller. 1990. Analysis of ARIMA–noise models with repeated time series. Journal of Business and Economic Statistics 8 (2): 243–50.
Wong, W., J.H.W. Penm, R. Terrell, and K. Lim. 2004. The relationship between stock markets of major developed countries and Asian emerging markets. Journal of Applied Mathematics and Decision Sciences 8 (4): 201–18.
Yang, J., J. Kolari, and I. Min. 2003. Stock market integration and nancial crises: The case of Asia. Applied Financial Economics 13, 477–86.