Market segmentation and price differentials between A shares and H shares in the Chinese stock markets

J. of Multi. Fin. Manag. 16 (2006) 232–248

Market segmentation and price differentials between Ashares and H shares in the Chinese stock markets

Yuming Li a,∗, Daying Yan b, Joe Greco a

a Department of Finance, College of Business and Economics, California State University,800 N State College Building, Fullerton, CA 92834, USA

b School of Economics, Institute of International Economics, Nan Kai University, Tian Jin, PR China

Received 26 July 2004; accepted 29 July 2005Available online 22 September 2005

Abstract

We find that the risk premiums associated with the Hong Kong and mainland Chinese markets in a two-factor model successfully explain the cross section of returns on the A and H shares. Discounts of H-shareprices relative to A-share prices are related to the contemporaneous movements of the H-share local marketindex relative to the A-share local market index, especially during the period of the Asian financial crisis,as well as the spread of savings rates between Hong Kong and mainland China. The evidence suggests thatthe risk premiums associated with the segmented A-share and H-share markets exert crucial impacts on theprice differentials between the two classes of shares.© 2005 Elsevier B.V. All rights reserved.

JEL classification: G14; G15

Keywords: Cross-listed shares; Price discount puzzle; Market risk premiums

1. Introduction

The rapid openness of Chinese capital markets has attracted cross-border portfolio invest-ment, while the explicit institutional barriers may determine the actual efficiency of internationaldiversification. This prompted some recent literature to explore the distinct price behaviors ofstocks that are simultaneously traded in China’s segmented markets. Among these studies, onecontentious issue is the price differentials among different classes of shares. One pioneer work

∗ Corresponding author.E-mail address: [email protected] (Y. Li).

1042-444X/$ – see front matter © 2005 Elsevier B.V. All rights reserved.doi:10.1016/j.mulfin.2005.07.003

Y. Li et al. / J. of Multi. Fin. Manag. 16 (2006) 232–248 233

by Bailey (1994) documents that class B shares traded by foreign investors are sold at discountsrelative to class A shares traded by domestic investors, a phenomenon that is inconsistent with theprice premiums commonly found in other countries (e.g., Bailey and Jagtiani, 1994; Domowitzet al., 1997; Stulz and Wasserfallen, 1995; Bailey et al., 1999).

Several explanations have been provided for this exception. The first is based on the discount-pricing model. Bailey (1994) argues that the lower required returns of Chinese citizens due to thelimited investment opportunities might have an effect. Bailey (1994) further suggests that sincethe political and macroeconomic risks peculiar to the B-share markets are undiversifiable to theprimary foreign investors, they may add potential risk premiums on B shares and discount theprices of B shares heavily as compensation for this systemic risk. Bergstrom and Tang (2001)argue that clientele bias, foreign exchange risks and risk-free return differentials can deter thediversification benefits of foreign investments particularly in the B-share market.

The second hypothesis is invoked by the “investor sentiment” notions introduced by Lee et al.(1991). Bailey (1994) proposes that the unseasoned or unduly optimism of domestic investorsmay drive the overpricing of A-share. A later work by Ma (1996) demonstrates that the A-shareprices are positively related to domestic beta risk, implying the risk-seeking behavior of Chineseinvestors.

The third strand of theories centers on demand and supply of shares in segmented markets.Sun and Tong (2000) argue that the co-existence of H shares and red chips as alternatives to Bshares for foreign investors makes the demand for B shares quite price elastic, leading to lowequilibrium prices of B shares. Indeed, the exceptional discount on foreign-only shares in Chinacomplements the argument of Stulz and Wasserfallen (1995), who attribute the price premium tothe inelastic demand of foreigners.

This article extends the studies on the Chinese stock markets by investigating the pricediscounts of H shares relative to A shares. As documented by Sun and Tong (2000), the pricepremiums or discounts of B shares and H shares tend to move together over time and H sharestraded in the Hong Kong market offer substitutes for B shares traded in mainland China. Hence,an understanding of the price discounts of H shares relative to A shares should help resolve theprice discount puzzle for both B shares and H shares of Chinese firms. Unlike A shares and Bshares which are traded in the same stock exchanges within mainland China, A shares and Hshares are segmented in terms of the stock ownership as well as the listing and trading locations.Specifically, while A shares are traded in mainland Chinese stock exchanges by local investors,H shares are owned and traded by investors in Hong Kong. The unique nature of segmentationbetween A shares and H shares suggests that the price discounts of H shares to A shares may beexplained by the influence of the local market performance.

More formally, we examine whether and how the differences in market risk premiums, inter-est rates and exchange rates could induce the substantive price differentials. Although in ourtime-series model the explanatory factors reflect market-wide information, the tests are mainlyperformed at the company-specific level over a long sample horizon. In this way our results facil-itate closer inspection of stylized price differentials between A shares and H shares without thecost of statistical reliability.1 Our paper indeed reveals that the firm-specific price discounts of Hshares relative to A shares are closely correlated with the difference in the movements between thecontemporaneous Hang Seng Stock Index and the Shanghai Stock Index, and the spread between

1 Bergstrom and Tang (2001) conduct similar regressions, yet only in an aggregate level partly due to the limitation ofthe sample horizon.

234 Y. Li et al. / J. of Multi. Fin. Manag. 16 (2006) 232–248

the Hong Kong and mainland Chinese interest rates, both of which are crucial constituents ofmarket risk premiums.

This paper is also motivated by recent literature on the effects of the Asian financial crisison financial markets. Froot et al. (2001) document that while international investors did notabandon emerging markets during the crisis, daily inflows into all emerging markets over theJuly 1997–July 1998 period averaged only 40% for all emerging markets and 30% for Asiacompared with their pre-crisis (1994–1997) levels. Lin and Swanson (2004), however, find littleevidence of significant crisis effects under information dissemination. We report that the Asianfinancial crisis exerts an economically significant impact on the volatility of the Hong Kongequity market and H-share prices, while the crisis fails to affect the mainland China’s equitymarket and cross-listed A shares traded in mainland China. The evidence is consistent with theasset pricing theory in segmented markets, which suggests that the risk premiums associatedwith the segmented A-share and H-share markets exert crucial impacts on the price differentialsbetween the two classes of shares. Our results confirm the empirical findings that stock returnsfrom different classes of shares are sensitive to distinct markets (Hietala, 1989; Chen et al.,2001). The results are also in accord with empirical findings that asset prices are affected by thelocation of trade (e.g., Froot and Dabora, 1999; Chan et al., 2003).

The rest of paper proceeds as follows. Section 2 briefly outlines the institutional settings ofChinese equity markets and the construction of the sample. Section 3 examines the relationsbetween stock returns and market-wide factors. Section 4 performs analysis of price differentialsbetween A shares and H shares. The final section concludes the article.

2. Data and hypotheses

2.1. Institutional settings

Alongside the well-known B shares traded in the Shanghai and Shenzhen stock exchanges(SHSE and SZSE), there is another class of stocks called H shares that allows Hong Kong andforeign investors’ access to mainland Chinese enterprises. Currently H shares refer to stocks ofmainland China-established companies traded directly in Hong Kong Stock Exchange (HKSE).2

A smaller group of Chinese firms are listed on the New York Stock Exchange (NYSE) in the formof ADRs, also known as “N shares”. Although B, H and N shares are all intended to attract foreignfunds, as H and N shares are issued and traded outside mainland China and therefore must complywith the additional legal disclosures set by the trading markets, the co-existence of different classesof shares creates more sophisticated features of market segmentation in Chinese capital markets.In the rest of the paper, we will document some common and distinct price behaviors of the stocksthat are dual-listed in mainland and Hong Kong markets and provide possible explanations forthe evidence on price discounts.

2.2. Data and summary statistics

We first collect daily stock prices of A shares and H shares from Datastream database fromJanuary 1997 to March 2002. Then firms that have issued both A shares in SHSE or SZSE andH shares in HKSE at the beginning of the sample period are selected, yielding an intersection of

2 See Poon and Fung (2000) for an exhaustive listing of all H shares.


13 firms. In spite of the limited size, the sample covers a wide cross section of Chinese industriesand should still be considered as a representative sample of Chinese firms.

Because all H shares are traded in the HKSE and 10 out of 13 A shares in the sample are tradedin the SHSE, we use the Hang Seng Price Index (HSI) and Shanghai Composite Index (SHI) torepresent the local markets for H shares and A shares, respectively. Finally, as the money marketin mainland China is quite thin, we choose to use the monthly mainland Chinese and Hong Kongsavings interest rates from the CEIC database as proxies for risk free rates for A-share and H-shareinvestors, respectively.3 Here, we assume that the interest rates are constant within each month,since the monthly rates in both markets show very low volatility on a short-term basis.

2.3. Summary statistics

Table 1 presents summary statistics of daily excess returns on individual stocks and marketindexes. All returns and savings rates are expressed in Hong Kong dollars. Panel A reports theresults for the full sample while panel B summarizes the results for the period of the Asian financialcrisis: 3 July 1997–31 September 1998. The crisis begins to emerge on 2 July 1997 when Thailandabandoned its currency (Baht) peg to the US dollar, triggering currency devaluations throughoutSoutheast Asia, especially Malaysia, Indonesia, Philippines, South Korea and Taipei. The currencydevaluations had adverse impacts on the stock markets of Asian-Pacific regions, including HongKong, until approximately 30 September 1998.4

One preliminary finding from panel A of Table 1 is that, except for two firms, the average excessreturns on A shares are higher than their H-share counterparts, while the average return on the SHIis higher than that of the HSI. More specifically, the cross-sectional average return on A-sharestocks is nearly zero compared with the corresponding H-share average return of −0.04%/day.Similarly, The SHI generates a positive average daily return in contrast to a negative daily averagereturn on the SHI. Not surprisingly, for all firms without exception, the standard deviations ofreturns exceed their corresponding market index level. Further, the standard deviation of returnson each H-share is higher than that on the corresponding A-share, consistent with evidence that thevolatility of excess returns from the HSI is higher than that of the SHI. Bailey (1994) documentssimilar higher volatility of stock returns in B shares relative to A shares and interprets this asforeign investors typically being less informative of the issuers than local investors.

Table 1 also reports the percentage daily change in the Hong Kong dollar values of mainlandChinese currency (EXCH). The mean of EXCH is close to zero with a daily volatility of 0.05%,reflecting the fact the both currencies are pegged to the US dollars. The correlation between theSHI and the HSI is 9.8%, while the correlations between stock indices and EXCH are muchsmaller in magnitude.

From panel B of Table 1, we observe that during the crisis period, mean returns on all H-sharestocks are negative with a cross-sectional average of −0.45%/day in contrast to an average of−0.01%/day for A shares. The average return on the HSI is also much lower than that on theSHI during the crisis period. Unlike A shares, the volatilities of individual H shares and the HSIare also higher during the crisis period. Most interestingly, we observe an increased volatility of

3 The CEIC database provided by CEIC Data Company Limited offers accurate, timely and comprehensive Asianeconomic data. The data series are based only on official publications or records obtained directly from issuing sources.More than 90,000 raw data series covering 14 countries in the region are available.

4 Froot et al. (2001) and Lin and Swanson (2004) use a similar time period in studying the impacts of Asian financialcrisis on financial markets.


Table 1Summary statistics for excess stock and index returns

Firm name A shares H shares Correlations

Mean (%) S.D. (%) Mean (%) S.D. (%) SHI HIS

Panel A. Full sample: 2 January 1997–28 March 2002Shenzhen stock exchange

1. Jilin Chemical Industrial −0.066 3.338 −0.071 5.8542. Jingwei Textile Machinery −0.033 3.169 0.018 5.5843. Northeast Electrical −0.070 2.974 −0.089 6.253

Shanghai stock exchange4. Beiren printing 0.040 3.000 −0.007 4.6095. Dongfang electrical 0.012 2.974 −0.088 5.6226. Guangzhou shipyard 0.001 2.781 −0.054 5.6577. Kunming machine tool 0.126 3.699 0.046 6.1858. Luoyang glass 0.027 2.933 −0.065 5.8079. Maanshan iron and steel −0.014 2.529 −0.084 5.64110. Nanjing panda 0.021 2.860 0.037 6.35211. Shanghai petrochemical −0.049 2.384 −0.079 5.04212. Tsingtao brewery 0.011 2.558 −0.038 4.95013. Yizheng chemical fibre 0.008 2.909 −0.053 6.036

Average 0.001 2.931 −0.040 5.661

Shanghai Composite Index (SHI) 0.041 1.731Hong Kong Hang Seng Index (HIS) −0.025 2.244 0.098Change in exchange rate (EXCH) 0.001 0.052 −0.023 0.036

Panel B. Asian financial crisis: 3 July 1997–31 September 1998Shenzhen stock exchange

1. Jilin chemical industrial −0.085 3.546 −0.362 7.8592. Jingwei textile machinery −0.167 3.331 −0.465 7.7333. Northeast electrical 0.013 2.965 −0.404 7.767

Shanghai stock exchange4. Beiren printing −0.003 2.986 −0.469 6.0775. Dongfang electrical −0.002 3.072 −0.637 7.1156. Guangzhou shipyard −0.013 2.983 −0.404 7.8327. Kunming machine tool 0.261 3.939 −0.407 8.1818. Luoyang glass 0.038 2.938 −0.548 7.8829. Maanshan iron and steel −0.048 2.541 -0.574 8.03010. Nanjing panda −0.012 3.257 -0.591 7.68311. Shanghai petrochemical −0.106 2.497 -0.299 6.78912. Tsingtao brewery −0.062 2.498 -0.422 7.00713. Yizheng chemical fibre −0.005 3.028 -0.297 9.361

Average −0.009 3.003 −0.452 7.640

Shanghai Composite Index (SHI) −0.021 1.749Hong Kong Hang Seng Index (HIS) −0.232 3.221 0.065Change in exchange rate (EXCH) 0.001 0.077 −0.046 0.106

Excess returns for A shares and H shares, as well as Shanghai Composite and Hong Kong Hang Seng indices, are rawreturns in excess of the mainland Chinese and Hong Kong savings rates, respectively. All returns and savings rates areexpressed in Hong Kong dollars. EXCH is the percentage daily change in the Hong Kong dollar values of mainlandChinese currency.


EXCH and an increased correlation between EXCH and the HSI. Hence, we cannot rule out thepossibility that return and volatility differences for individually dual-listed stocks (A versus H)partially reflect local market conditions (SHSE versus HKSE).

2.4. Hypotheses

Given that the A shares and H shares in our sample are dual-listed in mainland China andHong Kong, we construct two progressive hypotheses to examine the discrepancies in their marketperformances:

Hypothesis 1. We describe the equilibrium excess returns on the two classes of shares by amultifactor asset-pricing model:

E(Rji,t) =

K∑k=1

bj,k,iλjk (1)

where E represents the expectation operator, Rji,t denotes returns on the ith stock of class j (j = A,

H) in excess to the riskless rate at t, bj,k,i the (�) exposure of the stock to the kth risk factor, λjk

represents the risk premium for one unit of kth risk exposure, and K is the number of risk factors.For cross-listed firm i, differences in the risk premiums between λA

k and λHk and cross-sectional

differences in risk exposures between bA,k,i and bH,k,i yield cross-sectional variation in expectedreturns from each pair of A shares and H shares.

We assume that the dominant risk factors are excess returns on the two markets where Ashares and H shares are traded. Allowing for the implicit effect of the risk-free rates on thecost of capital to local investors, the difference in local interest rates is also taken into consid-eration. Solnik (1974) argues that the expected risk premium demanded by a foreign investoron a domestic asset partly depends on the expected devaluation of the currency that denomi-nates the asset, which is considered as exchange risk premium. Earlier empirical work on pricedifferentials (Bailey and Jagtiani, 1994) has documented the impact of this risk factor in othersegmented markets. Because the Chinese yuan (Renminbi) and the Hong Kong dollar are bothpegged with the US dollar, the effects of the exchange rate between the two currencies on theA-share and H-share prices should be minimal in normal market conditions. However, as weobserve the increased volatility of the exchange rates between the two currencies and the increasedcorrelation of exchange rate changes with the Hong Kong market during the Asian financial cri-sis period, we examine the effects of the exchange rate factor on returns from A shares andH shares as well as price discounts of H shares below A shares during and outside the crisisperiod.

Hypothesis 2. Our second hypothesis is that the difference in the required rates of returns on Ashares and H shares causes the time-series variation in the price premiums of H shares relative toA shares. From a present value standpoint, stock prices on A shares and H shares are determinedby expected dividends and required rates of returns on the two classes of shares. As a result, theprice premiums of H shares relative to A shares must reflect the difference in required rates ofreturns since expected dividends from the two classes of shares of the same firm are identical.Particularly to our concern, there should be a noticeable correspondence between price premiumsof H shares relative to A shares and price premiums of the Hong Kong stock market index relativeto Shanghai Stock index, the spread between Hong Kong and mainland Chinese riskfree rates.


This implies the firm-specific price differential can be attributed to the market-level differentialvariables according to the conventional multifactor asset pricing models.

3. Regressions of returns on A shares and H shares

To reveal the sources of price discounts for H shares relative to A shares, we first conductmultiple regression analysis of the systemic risk premiums underlying the price behaviors of eachclass of stocks. Specifically, with a multiple regression, we relate excess stock returns on A sharesand H shares to risk premiums associated with the two market indexes. If the excess returns doexhibit significant difference in risk exposures (�) to each of market risk premiums, then ourexplanation for the price discounts on H shares based on the differentiated risks and requiredreturns in the two markets will be justified.

Our multiple regressions of excess returns on A shares are formulated in the following model:

RAi,t = bi0 + bi1 SHIt + bi2 HSIt + bi3 EXCHt + bi4Dt + bi5(SHIt Dt) + bi6(HSIt Dt)

+ bi7(EXCHt Dt) + bi8RAi,t−1 + εA

i,t (2)

The dependent variable is the excess returns, RAi,t , on the A-share of firm i at time t. The independent

variables include: daily excess returns from the Shanghai Composite Index (SHI) and the HongKong Hang Seng Index (HSI), the percentage daily change in the Hong Kong dollar valuesof mainland Chinese currency (EXCH). To examine the effects of the Asian financial crisis,we also include a dummy variable (D) for the Asian financial crisis period, which takes thevalue of 1 for days: 3 July 1997–31 September 1998 and 0 otherwise, and interactions of thedummy variables with SHI, HIS and EXCH. To control for serial correlations of daily returns, weaugment the model with a lagged dependent variable, RA

i,t−1. The system of equations for all firmsis estimated simultaneously by using the iterated seemingly unrelated regression method (1026daily observations). The t-statistics in parentheses are reported below the associated coefficients.The p-values associated with χ2 (Wald) tests with 13 degrees of freedom are for testing the jointsignificance of coefficients for the same class of shares of all firms.

In Eq. (2), the coefficients bi1, bi2 and bi3 measure the sensitivities of A-share returns to SHI,HSI and EXCH, respectively. The coefficient bi4 measures the change in intercept during thecrisis period, whereas the coefficients bi5, bi6 and bi7 measure incremental sensitivities of A-share returns to the SHI, HSI and EXCH, respectively. Table 2 summarizes the regression results.First, for each of the firms in our sample, the coefficient on the SHI is positive and statisticallysignificant at the 1% or lower level. Consistently, results of χ2-tests also reject the joint hypothesisthat the coefficients for all A shares are zero. However, our regressions do not detect significantrelations between A-share returns and HIS or between A-share returns and EXCH. The dummyvariable and the interactions of the dummy variable with other variables enter the regressionswith insignificant coefficients for most firms. The coefficients on the lagged dependent variableare less than one-tenth in magnitude and statistically significant at the 5% level for only six firms,implying a low level of serial correlations of returns. The adjusted R2 of the regressions are all19% or higher. Overall the results suggest that the A-share market is unaffected by the Hong Kongmarket, the exchange rate changes and the Asian financial crisis, in accord with the hypothesisthat A-share market is segmented from the rest of the world.

Since the SHI is the only significant explanatory variable for A-share excess returns, Eq. (2)amounts to a single-factor or market model for excess returns on A shares. The intercept, bi0,measures the abnormal returns, which should be zero across all firms if the market-based CAPM

Y.Lietal./J.ofM

ulti.Fin.M

anag.16(2006)

232–248239

Table 2Regressions of A-share excess returns

Firm number Independent variables

Constant (%) SHI HIS EXCH D (%) D SHI D HSI D EXCH RAi,t−1 Adj., R2

1 −0.11 (−1.22) 1.13*** (19.78) 0.00 (0.03) −3.08 (−1.47) 0.08 (0.43) −0.07 (−0.63) 0.01 (0.11) 4.24 (1.49) −0.01 (−0.36) 0.262 −0.13 (−1.47) 1.16*** (20.84) 0.05 (0.92) −0.59 (−0.29) 0.01 (0.07) 0.15 (1.44) −0.04 (−0.58) 2.95 (1.07) −0.03 (−1.19) 0.333 −0.13 (−1.41) 0.97*** (17.54) 0.04 (0.87) −0.17 (−0.08) 0.22 (1.30) 0.05 (0.53) 0.02 (0.30) 2.34 (0.85) 0.00 (−0.06) 0.224 0.05 (0.67) 1.20*** (24.43) −0.11** (−2.38) 1.02 (0.56) 0.03 (0.22) 0.06 (0.64) 0.15** (2.43) −0.18 (−0.07) −0.04 (−1.88) 0.405 −0.04 (−0.50) 1.11*** (22.06) 0.06 (1.27) −1.08 (−0.58) 0.09 (0.60) 0.04 (0.39) −0.01 (−0.14) 0.62 (0.25) −0.05** (−2.28) 0.346 −0.08 (−1.03) 1.04*** (22.47) −0.03 (−0.81) 0.29 (0.17) 0.10 (0.72) 0.04 (0.48) 0.04 (0.76) 0.31 (0.13) −0.09 (−4.54) 0.367 −0.02 (−0.18) 1.24*** (17.57) 0.04 (0.61) −0.72 (−0.28) 0.41* (1.88) −0.25* (−1.95) 0.03 (0.30) −1.67 (−0.47) −0.02 (−0.97) 0.198 0.04 (0.39) 1.00*** (18.05) −0.03 (−0.59) −2.39 (−1.16) 0.01 (0.05) −0.03 (−0.30) 0.04 (0.61) 5.08 (1.83) 0.00 (0.16) 0.229 −0.03 (−0.46) 0.84*** (19.30) −0.03 (−0.68) 0.67 (0.42) 0.04 (0.30) 0.10 (1.21) 0.05 (0.98) −1.55 (−0.72) −0.06*** (−2.95) 0.28

10 0.06 (0.64) 0.88*** (15.97) 0.01 (0.26) 0.82 (0.40) −0.05 (−0.31) 0.12 (1.16) 0.05 (0.81) −1.64 (−0.60) 0.02 (0.93) 0.1911 −0.12 (−1.93) 0.90*** (24.27) −0.05 (−1.43) 0.55 (0.40) 0.05 (0.45) 0.09 (1.32) 0.05 (1.01) −0.85 (−0.46) −0.07*** (−3.51) 0.4112 −0.03 (−0.47) 1.06*** (25.36) −0.06 (−1.64) 2.81* (1.82) 0.05 (0.41) −0.16** (−2.07) 0.11** (2.18) −2.20 (−1.06) −0.04** (−2.03) 0.3913 −0.07 (−0.81) 1.03*** (19.81) −0.06 (−1.18) −0.26 (−0.14) 0.10 (0.60) −0.02 (−0.22) 0.06 (0.95) −2.03 (−0.79) −0.06*** (−2.99) 0.27

p-value 0.63 0.00 0.13 0.72 0.95 0.05 0.56 0.64 0.00

The dependent variable is the excess returns, RAi,t , on the A-share of the firm i at time t. The independent variables include: daily excess returns from the Shanghai Composite

Index (SHI) and the Hong Kong Hang Seng Index (HSI), the percentage daily change in the Hong Kong dollar values of mainland Chinese currency (EXCH), a dummy variable(D) for the Asian financial crisis period, which takes the value of 1 for days: 3 July 1997–31 September 1998 and 0 otherwise, interactions of the dummy variables with the SHI,HIS and EXCH, and a lagged dependent variable, RA

i,t−1. The system of equations for all firms is estimated simultaneously by using the iterated seemingly unrelated regression

method (1026 daily observations). The t-statistics in parentheses are reported below the associated coefficients. The p-values associated with χ2-tests with 13 degrees of freedomare for testing the joint significance of coefficients for the same class of shares of all firms.

* Significant at the 10% level.** Significant at the 5% level.

*** Significant at the 1% level.


explains the cross section of excess returns on all A shares. The results in Table 2 indicate thatthe intercepts are all imprecisely estimated and the test of joint significance has a p-value of 0.63,consistent with the hypothesis that the CAPM explains the cross section of returns on A shares.

Next, we consider similar regressions for H-share excess returns:

RHi,t = ci0 + ci1 SHIt + ci2 HSIt + ci3 EXCHt + ci4Dt + ci5(Dt SHIt) + ci6(Dt HSIt)

+ ci7(Dt EXCHt) + ci8RHi,t−1 + εH

i,t . (3)

The dependent variable is the excess returns, RHi,t , on the H-share of firm i at time t. Results of the

regressions are reported in Table 3. We find that returns on six H-share stocks are related to theSHI at the 5% significance level, whereas returns on all H-share stocks are significantly related tothe HIS. The sensitivities on the SHI are noticeably lower than those on the HSI. Accordingly, thejoint hypothesis that the coefficients on the HSI for all firms are zero is rejected at the 1% level,while a similar test of joint significance of the coefficients on the SHI is not reject the 5% level.These results suggest that unlike A shares returns which depend on mainland market exclusively,H-share returns are mostly sensitive to the market information from Hong Kong and to a muchlesser extent responsive to the information from the mainland market.

Similar to A-share returns, H-share returns are not related to exchange rate changes except forone firm, as the coefficients on EXCH and D EXCH indicate. However, the estimated coefficientson D SHI and D HSI reveal that H-share returns of three firms become more sensitive to themainland market whereas H-share returns of all firms become more sensitive to the Hong Kongmarket during the Asian financial crisis. In fact, the coefficients on D HSI are individually andjointly significant at the 1% level and the magnitudes of the coefficients are similar to those on theHSI. For over half of the firms, the sensitivities of H-share returns to the HSI increase by at least100% during the crisis period, suggesting that the Asian financial crisis exerts an economicallysignificant impact on the relation between H-share returns and the Hong Kong equity market,while the crisis fails to affect returns on cross-listed A shares traded in mainland China. In spiteof the fact that H-share returns are related to more explanatory variables, the adjusted R2s of theregressions for the H-share returns are approximately half of those for A-share returns, as a resultof higher volatility of H-share returns.

Finally, t-tests from the regressions show that the intercepts are not individually significant, andthe χ2-tests indicate the intercepts for all firms are not jointly significant. The results are supportiveof the hypothesis that the multifactor model explains all of the cross-sectional variation in averagereturns on H shares.

4. Regressions of price premiums of H shares relative to A shares

Given the evidence in favor of the market segmentation hypothesis concerning A shares andH shares, the mean–variance portfolio theory implies that the risk-averse investors in each mar-ket should hold diversified portfolios of assets available from their own market. If that is thecase, according to capital asset pricing model (CAPM), the expected returns on each class ofshares should reflect the risk exposure and the risk premium associated with the local mar-ket. Alternatively, under the assumption that asset prices in each market are given by a singleindex model, the arbitrage pricing theory also provides a similar implication for expected returnsfrom each class of shares in each market. Since the prices of each class of shares, like thelocal price indexes of the markets where the shares are traded, mimic the cumulative returns,a further extension from CAPM is that the firm-specific price premium can be related to three

Y.Lietal./J.ofM

ulti.Fin.M

anag.16(2006)

232–248241

Table 3Regressions of H-share excess returns

Firm number Independent variables

Constant (%) SHI HIS EXCH D (%) D SHI D HSI D EXCH RHi,t−1 Adj., R2

1 −0.01 (−0.07) 0.29*** (2.76) 0.76*** (7.63) 0.70 (0.14) 0.01 (0.03) 0.04 (0.16) 0.76*** (5.43) 2.97 (0.45) −0.13*** (−7.67) 0.122 0.17 (0.99) 0.22** (2.14) 0.57*** (5.86) 1.48 (0.30) −0.40 (−1.15) −0.06 (−0.26) 0.91*** (6.72) 5.11 (0.79) −0.09*** (−4.85) 0.103 0.00 (−0.02) 0.19* (1.67) 0.80*** (7.34) 0.65 (0.12) −0.14 (−0.34) 0.26 (1.09) 0.72*** (4.70) 7.39 (1.02) −0.12*** (−6.02) 0.094 0.16 (1.12) 0.14* (1.69) 0.48*** (5.95) 3.40 (0.84) −0.45 (−1.53) 0.18 (0.99) 0.55*** (4.86) 1.55 (0.29) −0.07*** (−3.14) 0.065 0.08 (0.44) 0.30*** (2.93) 0.66*** (6.68) −0.28 (−0.06) −0.40 (−1.11) 0.14 (0.67) 0.55*** (3.97) 2.57 (0.39) −0.05** (−2.32) 0.066 0.06 (0.33) 0.29*** (2.86) 0.57*** (5.77) 1.91 (0.38) −0.24 (−0.66) −0.04 (−0.18) 0.73*** (5.28) −2.24 (−0.34) −0.11*** (−5.49) 0.067 0.28 (1.39) 0.11 (0.94) 0.58*** (5.18) −0.91 (−0.16) −0.48 (−1.18) 0.28 (1.11) 0.59*** (3.75) 5.68* (0.76) −0.08*** (−3.31) 0.018 0.06 (0.31) 0.33*** (3.14) 0.74*** (7.37) 3.50 (0.69) −0.36 (−0.99) 0.36 (1.63) 0.63*** (4.47) 3.98 (0.59) −0.09*** (−5.03) 0.109 0.04 (0.26) 0.23** (2.38) 0.84*** (9.05) 0.52 (0.11) −0.32 (−0.95) 0.41** (2.00) 0.80*** (6.14) 0.85 (0.14) −0.10*** (−6.40) 0.19

10 0.22 (1.11) 0.11 (0.92) 0.86 (7.83) 2.28 (0.41) −0.49 (−1.23) 0.46* (1.9) 0.49*** (3.17) 5.98 (0.81) 0.01 (0.38) 0.0811 −0.02 (−0.11) 0.07 (0.81) 0.74*** (8.78) 2.41 (0.56) 0.09 (0.29) 0.44** (2.34) 0.59*** (4.95) 1.75 (0.31) −0.07*** (−3.81) 0.1412 0.09 (0.57) 0.00 (0.01) 0.54*** (6.35) 9.02** (2.10) −0.21 (−0.68) 0.11 (0.56) 0.76*** (6.37) −6.25 (−1.10) −0.01 (−0.57) 0.1113 0.00 (0.00) 0.11 (1.03) 0.68*** (6.73) 1.09 (0.21) 0.06 (0.17) 0.89*** (3.96) 1.02*** (7.16) 3.78 (0.56) −0.09*** (−5.00) 0.15

p-value 0.97 0.05 0.00 0.87 0.84 0.00 0.00 0.94 0.00

The dependent variable is the excess returns, RHi,t , on the A-share of the firm i at time t. The independent variables include: daily excess returns from the Shanghai Composite

Index (SHI) and the Hong Kong Hang Seng Index (HSI), the percentage daily change in the Hong Kong dollar values of mainland Chinese currency (EXCH), a dummy variable(D) for the Asian financial crisis period, which takes the value of 1 for days: 3 July 1997–31 September 1998 and 0 otherwise, interactions of the dummy variables with the SHI,HIS and EXCH, and a lagged dependent variable, RH

i,t−1. The system of equations for all firms is estimated simultaneously by using the iterated seemingly unrelated regression

method (1026 daily observations). The t-statistics in parentheses are reported below the associated coefficients. The p-values associated with χ2-tests with 13 degrees of freedomare for testing the joint significance of coefficients for the same class of shares of all firms.




market-wide factor differentials. In Appendix A we discuss the details of the mathematicalderivation.

4.1. Definitions of variables and summary statistics

In order to test the validity of this conjecture, we first adopt a firm-specific price premiumsimilar to Bailey (1994):

PRMi,t = PHi,t − PA

i,t

PAi,t

, (4)

where Pji,t is the price of j-share (j = H, A) for company i on day t.

Next, to explore whether and how the time-varying firm-specific price premiums are relatedto different market risk factors associated with the excess returns on A shares and H shares, weconstruct three risk differential variables specifically associated with the markets, interest ratesand exchange rate risk factors.

(i) Index premium (IPRM): A proxy for the market-level price differential, is defined as

IPRMt = PHM,t/P

HM,0

PAM,t/P

AM,0

− 1, (5)

where PHM,t and PA

M,t denote weekly ending values of the Hang Seng Stock Index and theShanghai Composite Index converted to Hong Kong dollars. Both index values are scaledby their beginning-of-sample values in the calculation of the index premium. Because stockreturns on each class of shares are positively related to returns from their own local markets,the price premium should be positively associated with the index premium.

(ii) Interest rate spread (RATE): A proxy for the risk-free rate differential, is measured as

RATEt = SRHKt − SRCH

t , (6)

where SRHKt and SRCH

t refer to monthly savings deposit rates in local currency at the endof week t in Hong Kong and mainland China, respectively. As the interest rate spread is abarometer of the differential cost of capital for investors in Hong Kong relative to those inmainland, the H-share to A-share price premium should be related to the interest rate spread.

(iii) Exchange rate change (EXCH) is calculated as

EXCHt = et

et−1− 1, (7)

where et denotes the number of Hong Kong dollars per unit of Chinese yuan or RMB.

Table 4 reports a descriptive summary of the price premiums and three explanatory variables.Panel A reports the results for the full sample and panel B is for the period of the Asian financialcrisis. We start with a discussion of the results in panel A. Consistent with the results in excessreturns, the average firm-specific price premiums are all negative, indicating the H shares sell atappreciable discounts relative to A shares. The firm-specific discounts are consistent with indexdiscounts, though at different degrees. For example, discounts on individual H shares range from70 to 89%, in contrast to a moderate index discount of 39%. Another interesting fact is that the


Table 4Summary statistics for price premiums

Firm number N Mean (%) S.D. (%) Autocorrelations at lags Correlations

1 10 20 IPRM RATE

Panel A. Full sample: 2 January 1997–28 March 2002

1 1207 −88.6 4.7 0.980 0.832 0.7182 1206 −87.1 7.9 0.988 0.916 0.8613 1201 −88.3 6.4 0.983 0.899 0.8114 1204 −83.5 10.8 0.991 0.950 0.9135 1208 −84.2 12.6 0.990 0.945 0.8996 1211 −81.5 13.1 0.989 0.931 0.8797 1202 −85.6 12.0 0.988 0.938 0.8878 1209 −86.5 10.6 0.989 0.924 0.8609 1208 −76.8 15.0 0.986 0.937 0.884

10 1209 −85.7 7.5 0.983 0.908 0.83011 1207 −69.7 10.7 0.982 0.883 0.78912 1209 −72.3 12.7 0.988 0.925 0.85713 1206 −70.2 12.9 0.983 0.873 0.736

Average −81.5 10.5 0.986 0.913 0.840

IPRM 1215 −38.8 14.8 0.993 0.927 0.837RATE 1215 2.4 1.3 0.999 0.987 0.973 0.000EXCH 1214 0.001 0.052 −0.341 −0.027 −0.014 0.000 0.000

Panel B. Asian financial crisis: 3 July 1997–31 September 1998

1 300 −85.7 7.2 0.980 0.832 0.7182 300 −87.4 8.6 0.986 0.830 0.7283 300 −84.7 9.5 0.985 0.888 0.7814 299 −77.4 13.9 0.989 0.930 0.9005 298 −75.3 15.9 0.983 0.909 0.8496 300 −70.7 18.4 0.984 0.883 0.8117 299 −78.5 16.8 0.988 0.909 0.8338 298 −80.3 14.9 0.982 0.873 0.7779 298 −66.9 19.9 0.982 0.878 0.772

10 299 −85.7 9.3 0.986 0.899 0.81411 300 −65.2 13.7 0.981 0.875 0.81612 300 −66.0 17.5 0.984 0.899 0.80513 299 −67.1 19.6 0.983 0.855 0.673

Average −76.2 14.2 0.984 0.882 0.791

IPRM 300 −38.3 17.7 0.993 0.910 0.820RATE 300 3.2 0.7 0.993 0.930 0.860 −0.001EXCH 300 0.001 0.077 −0.250 −0.086 −0.029 0.000 0.000

The price premium (PRM) of each firm is computed as the percentage difference between H-share prices and A-shareprices. IPRM is the percentage difference between the Hang Seng Index and Shanghai Composite Index, both of whichare scaled to have the same base value at the beginning of the sample. RATE is the difference between the Hong Kongsavings deposit rate and that of mainland China, expressed in annual percentage rates, EXCH is the daily percentagechange of the exchange rate for converting mainland Chinese RMB into Hong Kong dollar, and N is the number of dailyobservations


Fig. 1. Price premiums of H shares relative to A shares. The index premium is the percentage difference between theHong Kong Hang Seng Index and Shanghai Composite Index, both of which are scaled to have the same base value at thebeginning of the sample. JCI: Jilin Chemical Industrial, LG: Luoyang Glass, SP: Shanghai Petrochemical.

standard deviations of firm-specific price discounts are generally lower than that of the indexdiscount, indicating that the price discounts are less volatile at the individual level than the marketlevel. In addition, Table 4 demonstrates that the cost of capital for Hong Kong residence indeedovertly exceeds that of mainland residence in terms of the positive interest rate spread. We alsonote that the interest rate spread is weakly correlated with the index premium and the exchangerate change, with correlations of close to zero. Further, the price premiums for individual firmsand the indexes as well as the interest rate spread are highly serially correlated, suggesting thatthese variables are slow-moving and persistent.

The results in panel B are largely similar to those in panel A, except that the standard deviationsof the price premiums are considerably higher in the crisis period for each firm. On average,the volatility of price premiums increases by approximately 40%. The index premium and theexchange rate change also exhibit increased volatility during this period.

Fig. 1 illustrates the fluctuations in firm-specific price premiums and index premium. Forbrevity, we only pick out three firms whose average price premiums are evenly distributed overthe sample range, which are Jilin Chemical Industrial (JCI), Luoyang Glass (LG) and ShanghaiPetrochemical (SP). All firm-specific price premium series displayed in the figure track the move-ment of the index premium series throughout the sample period, though the absolute levels areall markedly lower than the index premium.

Given the noteworthy co-movement, we turn to an analysis of the time-varying relationsbetween risk factor differentials and firm-specific price differentials.

4.2. Regression results

To examine the relation between the firm-specific price premium and the simultaneous move-ments of the index premium, the interest rate spread and the exchange rate change, we estimate

Y.Lietal./J.ofM

ulti.Fin.M

anag.16(2006)

232–248245

Table 5Multiple regressions of price premiums

Firmnumber

Independent variables

Constant IPRM RATE EXCH D D IPRM D RATE D EXCH PRM−1 Adj., R2

1 −4.83*** (−9.01) 0.22 (0.63) −1.05*** (−3.19) 0.23 (0.58) 0.91** (2.25) 3.92*** (3.59) 1.44 (0.62) 0.45 (0.84) 0.94*** (9.27) 0.962 −2.20*** (−5.26) −0.70* (−1.74) −1.41*** (−3.41) 0.06 (0.13) 0.47 (1.08) 3.86*** (3.20) 2.28 (0.90) 0.66 (1.11) 0.97*** (5.61) 0.983 −4.22*** (−7.38) 0.68 (1.62) −1.75*** (−4.13) 0.01 (0.01) 1.22** (2.57) 3.23 (2.63) −0.28 (−0.11) 1.01 (1.66) 0.95 (7.51) 0.974 −2.60 (−7.31) 1.92 (3.64) −2.52 (−4.90) −0.04 (−0.07) 0.78 (1.48) 3.59** (2.53) 2.27 (0.74) 1.02 (1.44) 0.96*** (8.15) 0.995 −2.25*** (−6.46) 1.91*** (3.01) −1.94*** (−3.53) 0.04 (0.06) 0.97 (1.56) 3.13* (1.88) 1.05 (0.29) 0.91 (1.09) 0.96*** (7.36) 0.996 −4.40*** (−9.60) 3.23*** (4.12) −3.55*** (−5.02) −0.11 (−0.13) 2.99*** (3.52) 7.24*** (3.31) 0.13 (0.03) 0.53 (0.48) 0.93*** (11.28) 0.987 −3.60*** (−8.59) 1.08* (1.88) −3.41*** (−5.58) 0.10 (0.16) 2.85*** (4.30) 3.66*** (2.14) −5.09 (−1.38) 1.22 (1.44) 0.95*** (9.51) 0.998 −3.87*** (−9.53) 1.89*** (3.14) −2.36*** (−4.36) 0.06 (0.09) 1.74*** (2.74) 4.47*** (2.67) −0.19 (−0.05) 0.90 (1.07) 0.95*** (10.28) 0.989 −2.02*** (−5.80) 3.11*** (3.67) −2.41*** (−3.32) −0.28 (−0.33) 1.34 (1.56) 4.14* (1.79) 1.15 (0.23) 0.83 (0.71) 0.96*** (8.91) 0.98

10 −1.50*** (−3.62) 0.53 (1.19) −0.66 (−1.62) −0.43 (−0.86) −0.11 (−0.22) 2.68** (1.99) 3.03 (1.05) 1.65*** (2.45) 0.98*** (4.31) 0.9811 −1.81*** (−4.79) 2.76*** (3.36) −0.70 (−1.08) −0.08 (−0.10) 0.93 (1.15) 1.45 (0.66) −1.54 (−0.32) 1.87* (1.69) 0.96*** (6.74) 0.9712 −2.51*** (−5.95) 2.87*** (3.34) −2.72*** (−3.54) 0.19 (0.22) 1.08 (1.26) 5.47** (2.39) 3.36 (0.68) 0.26 (0.23) 0.95*** (7.37) 0.9813 −2.54*** (−5.32) 2.75*** (2.87) 0.75 (0.96) −0.22 (−0.22) 2.46*** (2.38) 3.15 (1.17) −6.43 (−1.08) 2.38* (1.76) 0.96*** (7.10) 0.97

p-value 0.00*** 0.00*** 0.00*** 0.99 0.00*** 0.01*** 0.21 0.47 0.00***

The dependent variable is the price premium (PRM) of each firm, computed as the percentage difference between H-share prices and A-share prices of the firm. The indexpremium (IPRM) is the percentage difference between the Hang Seng Index and Shanghai Composite Index, both of which are scaled to have the same base value at the beginningof the sample. The saving rate spread (RATE) is the difference between the Hong Kong savings deposit rate and that of mainland China, expressed in annual percentage rates.The exchange rate change (EXCH) is the percentage daily change in the Hong Kong dollar values of mainland Chinese currency. Each of the three variables IPRM, RATE andEXCH is normalized to have a standard deviation of unity and the corresponding estimated coefficient is multiplied by 1000. D is a dummy variable for the Asian financialcrisis period: 3 July 1997–31 September 1998. PRM−1 refers to the first lag of the PRM. The t-statistics in parentheses are for testing that a coefficient is zero, with exceptionthat for PRM−1 the t-statistics are for testing that each coefficient is unity. The p-values associated with χ2-tests with 13 degrees of freedom are for testing the joint significanceof coefficients for all firms. The system of multiple regression equations is estimated simultaneously by using the iterated seemingly unrelated regression method (1098 dailyobservations).




the following multiple regressions:

PRMi,t = λi0 + λi1 IPRMt + λi2 RATEt + λi3 EXCHt + Dt + λi4(Dt IPRMt)

+ λi5(Dt RATEt) + λi6(Dt EXCHt) + λi7 PRMi,t−1 + εi,t (8)

We also include the dummy variable D for the financial crisis period, and products of the dummyvariable with each of the differential variables, to examine the effects of the crisis on the pricepremiums. To control for serial correlation of the dependent variable, we add the first lag of thePRM, as an explanatory variable. The t-statistics in parentheses are for testing that a coefficientis zero, with exception that for lagged PRM the t-statistics are for testing that each coefficientis unity. The p-values associated with χ2-tests with 13 degrees of freedom are for testing thejoint significance of coefficients for all firms. The system of multiple regression equations isestimated simultaneously by using the iterated seemingly unrelated regression method (1098daily observations).

The estimated slope coefficients on IPRM in Table 5 indicate that firm-specific discounts arepositively and statistically significantly related to the market-level price discount during the fullsample period at the 1% level for eight firms. For four of these firms and four additional firms,the coefficients on D IPRM are significant at the 5% level, suggesting that H-share discounts forthese firms experience increased exposure to the marketwide discount during the Asian financialcrisis. Hence, the increased volatility of the H-share discounts during the Asian financial crisiscan be attributed in part to the increased volatility of the Hong Kong stock market relative to thatof the mainland China. The regression slope coefficients associated with the interest rate spreadare negative for all but one firm and statistically significant at the 5% level for 10 firms. This isconsistent with the hypothesis that H-share discounts are at least partly related to the comparativelylower rate of returns from alternative risk-free investments in the mainland compared with thatin Hong Kong. The coefficients on IPREM, RATE and the dummy variable D are also jointlysignificant at the 1% level and the coefficients on D IPRM are jointly significant at the 5% level.

Consistent with the evidence reported earlier from the regressions of returns, the regressionsof premiums in Table 5 do not show any impact of exchange risk factor on H-share discounts.All of the coefficients on the EXCH are insignificant at the 5% level, and the coefficients on theD EXCH are insignificant except for one firm. We suppose that due to the strict foreign exchangecontrol by the Chinese monetary authority as well as the pegged exchange rate system in HongKong, neither the mainland RMB nor Hong Kong dollar is a genuinely flexible market indicator.As a result, the impact of exchange risk factor is substantially weakened.

5. Conclusions

Inspired by the existing studies on price differentials between A shares and B shares in thesegmented Chinese capital market, this paper conducts an exploratory study of price discountson H-share relative to A shares using the conventional asset pricing theory. By studying theprice behaviors of 13 firms dual-listed on mainland and Hong Kong stock markets over January1997–March 2002, we find that A-share excess returns are primarily explained by the market riskpremium from the mainland China while H shares excess returns embody risk premiums fromboth Hong Kong and mainland China’s markets, with a larger portion pertaining to the former.We find that the risk premiums associated with the Hong Kong and mainland China’s marketsin a two-factor model successfully explain the cross section of returns on the A and H shares.Consistently, we show that discounts on H shares relative to A shares are highly related to the


contemporaneous discounts of H-share local market relative to the A-share local market, as wellas the spread of the savings rates between Hong Kong and mainland China. The results of thisstudy indicate that the price differentials in the Chinese dual-listed A shares and H shares aremainly attributable to the deviation in the systemic risk premiums of the local markets.

We also report that the increased volatility of the H-share discounts during the Asian financialcrisis can be attributed in part to the increased volatility of the Hong Kong stock market relative tothat of the mainland China. The exchange rate change between the currencies of Hong Kong andmainland China, though experienced increased volatility and increased correlation with the HongKong market during the crisis period, does not have any significant effects on the price discountsof H shares below A shares.

Appendix A

In this appendix we obtain a model for price premiums of H shares relative to A shares based ona single factor model for A shares and a two-factor model for H shares. Let rA

i,t ≡ ln(PAi,t/P

Ai,t−1)

and rAM,t ≡ ln(PA

M,t/PAM,t−1) denote returns on the A shares of firm i and mainland market returns,

respectively, all expressed in Hong Kong dollars. Similar variables are defined for H-share returnsand Hong Kong market returns. The single-factor model assumes that the returns on the A sharesis given by

rAi,t = aA,i + bAA,ir

AM,t + εA

i,t (A.1)

for simplicity, we assume that prices include all reinvested dividends. Then

ln

(PA

i,t

PAi,0

)= rA

i,1 + rAi,2 + · · · + rA

i,t , (A.2)

ln

(PA

M,t

PAM,0

)= rA

M,1 + rAM,2 + · · · + rA

M,t. (A.3)

Substituting Eq. (A.1) into (A.2) and using (A.3), we have

ln

(PA

i,t

PAi,0

)= aA,it + bAA,i ln

(PA

M,t

PAM,0

)+

t∑s=1

εAi,s (A.4)

Eq. (A.5) implies that the price of an A-share today reflects a cumulative gain in the mainlandmarket index, given the initial A-share price and the level of the market. Similarly under a two-factor model for H shares:

rHi,t = aH,i + bHA,ir

AM,t + bHH,ir

HM,t + εH

i,t , (A.5)

we obtain

ln

(PH

i,t

PHi,0

)= aH,it + bHA,i ln

(PA

M,t

PAM,0

)+ bHH,i ln

(PH

M,t

PHM,0

)+

t∑s=1

εHi,s (A.6)


Note that in Eq. (A.4), bAA,i = bHA,i + (bAA,i − bHA,i). Taking the difference between Eqs. (A.4)and (A.6) implies

ln

(PH

i,t

PAi,t

)= ln

(PH

i,0

PAi,0

)+ (aH,i − aA,i)t + bHA,i ln

(PH

M,t/PHM,0

PAM,t/P

AM,0

)

− (bAA,i − bHA,i) ln

(PA

M,t

PAM,0

)+ bHA,i ln

(PA

M,t

PAM,0

)+

t∑s=1

(εHi,s − εA

i,s).

(A.7)

In Eq. (A.7), if bAA,i ≈ bHH,i and bHA,i is relatively small compared with bHH,i, then the index pre-mium ln[(PH

M,t/PHM,0)(PA

M,0/PAM,t)] along with a time trend should be an important determinant of

the H-share premium. Here the error term∑t

s=1(εHi,s − εA

i,s) is serially correlated but uncorrelatedwith the index premium under the conventional assumption underlying the factor models.

References

Bailey, W., 1994. Risk and return on China’s new stock markets: some preliminary evidence. Pacific-Basin Finance Journal2, 243–260.

Bailey, W., Jagtiani, J., 1994. Foreign ownership restrictions and stock prices in the Thai capital market. Journal ofFinancial Economics 36, 57–87.

Bailey, W., Chung, P., Kang, J.K., 1999. Foreign ownership restrictions and equity price premiums: what drives the demandfor cross-border investments? Journal of Financial and Quantitative Analysis 34, 489–511.

Bergstrom, C., Tang, E., 2001. Price differentials on different classes of stocks: empirical study on the Chinese stockmarket. Journal of Multinational Financial Management 11, 407–426.

Chan, K., Hameed, A., Lau, S.T., 2003. What if trading location is different from business location? Evidence from theJardine group. Journal of Finance 58, 1221–1246.

Chen, G.M., Lee, B.S., Rui, O., 2001. Foreign ownership restrictions and market segmentation in China’s stock markets.Journal of Financial Research 24, 123–133.

Domowitz, I., Glen, J., Madhavan, A., 1997. Market segmentation and stock prices: evidence from an emerging market.Journal of Finance 52, 1059–1085.

Froot, K.A., Dabora, E.M., 1999. How are stock prices affected by the location of trade? Journal of Financial Economics53, 189–216.

Froot, K.A., O’Connell, P., Seasholes, M.S., 2001. The portfolio flows of international investors. Journal of FinancialEconomics 59, 151–193.

Hietala, P.T., 1989. Asset pricing in partially segmented markets: evidence from the Finnish market. The Journal ofFinance 44, 697–718.

Lee, C., Shleifer, A., Thaler, R., 1991. Investor sentiment and the closed-end fund puzzle. Journal of Finance 46, 75–109.Lin, A.Y., Swanson, P.E., 2004. International equity flows and developing markets: the Asian financial market crisis

revisited. Journal of International Financial Markets, Institutions and Money 14, 55–73.Ma, X., 1996. Capital controls, market segmentation and stock prices: evidence from the Chinese market. Pacific-Basin

Finance Journal 4, 219–239.Poon, W., Fung, H.G., 2000. Red chips and H shares: which China-backed securities process information the fastest?

Journal of Multinational Financial Management 10, 315–343.Solnik, B.H., 1974. The international pricing of risk: an empirical investigation of the world capital market structure.

Journal of Finance 29, 365–377.Stulz, R., Wasserfallen, W., 1995. Foreign equity investment restrictions, capital flight and shareholder wealth maximiza-

tion: theory and evidence. Review of Financial Studies 8, 1019–1057.Sun, Q., Tong, W.H.S., 2000. The effect of market segmentation on stock prices: the China syndrome. Journal of Banking

and Finance 24, 1875–1902.

Documents

Market segmentation and price differentials between A shares and H shares in the Chinese stock markets