Tugas Perilaku Konsumen 2

Capital controls, market segmentation and stock

prices: Evidence from the Chinese stock

market 1

Introduction

There are two kinds of capital controls in emerging capital markets.

a) Limitations on foreign ownership of domestic equity. Developing countries often impose restrictions on the foreign ownership of domestic equity to ensure domestic control of local firms, especially those firms regarded as 'strategically important' to national interests. In Brazil, for instance, each foreign investor is limited to owning no more than 5% of any company's voting shares. In countries like India and Mexico, the limit is 49%. In China and Philippines, a local firm issues two different types of shares, i.e., A shares and B shares. Foreigners are allowed to hold only B shares. In Thailand, the stock market maintains two separate listings for common stock: one for locals, the 'Main Board'; one for foreigners, the 'Alien Board'. More information about foreign ownership restrictions can be found in Price (1994).

(b) Limitations on domestic investment in foreign capital markets. Developing countries also have different degrees of limitations on capital outflow. For example, domestic citizens may not be permitted to carry foreign exchange abroad freely. This kind of foreign exchange control exits in South Korea, China, Taiwan, and many other developing countries. In China, Chinese citizens cannot buy foreign currency freely. These two forms of capital controls make the emerging capital market segmented from the world capital market.

Some works have been done to analyze the effects of capital controls on stock prices in equity markets. Most of the works fall into international capital asset pricing model literature. Stultz (1981) developed an international asset pricing model in which different countries faced different consumption opportunity sets. He demonstrated that the real expected excess return on a risky asset moved proportionally with the covariance of the return of that asset with changes in the world real consumption rate. The model, however, did not include barriers to international investment. Errunza and Losq (1985) investigated the implications of investment barriers in the international capital markets. The barriers could produce a high risk premium for some securities if there were formal capital controls on foreign portfolio investments. Eun and Janakiramanan (1986) were the first to study the impact of foreign ownership restrictions on security prices. In their two-country world, the domestic investors were constrained to own at most some fraction of the number of shares outstanding of the foreign firms. Two different prices for the same firm's shares might exist when the constraint was binding. Hietala (1989) analyzed asset pricing in the partially segmented Finnish stock market, in which Finnish citizens were allowed to hold only their domestic securities, whereas the foreign investors were essentially allowed to hold all the securities around the world. In equilibrium, Finnish citizens were paying less for their domestic securities than were the foreign investors if the foreign investors required a lower risk premium. Stultz and Wasserfallen (1993) provided another theory of how foreign equity ownership restriction influenced stock prices. In a setting where the domestic investors and foreign investors faced different demand functions for a firm's securities, the firm could discriminate between investors to maximize its profit. The model predicted that the relaxation of foreign equity investment restrictions decreased the value of shares which were available to foreign investors. This prediction was supported by the evidence from Switzerland. Bailey and Jagtiani (1994) studied the effects of ownership restrictions using data from the Stock Exchange of Thailand. They found that cross-sectional differences between local and foreign prices were correlated with proxies for the severity of foreign ownership limits, liquidity, and information availability.

This paper analyzes the effect of both kinds of capital controls in Chinese stock markets. Although China is not the only country with the capital controls we examine, the Chinese stock market is particularly well-suited for our study. In China, a local firm issues two different classes of shares: A shares and B shares. They have the same rights and obligations. The only difference is that A shares are held exclusively by Chinese citizens but B shares are held exclusively by foreigners and traded in foreign currency. Typically, prices of A shares are higher than those of 'B' shares. Because of exchange controls, there is no straightforward arbitrage opportunity to drive prices together. Similar phenomena can be observed in Thailand, Malaysia and Philippines and a few small European markets (Finland and Sweden). A big puzzle, however, is that in those markets the foreign class shares are sold at premia but in China the 'B' shares are sold at a discount. This paper attempts to offer some explanations of this puzzle.

I investigate five possible explanations tbr the puzzle. First, the cost of capital for Chinese citizens may be lower than that for foreigners: the Chinese stock markets may represent the only investment alternative to low-yielding bank deposits. Second, the B shares' discounts may be explained by the investors' attitudes towards risks. The highly speculative behaviors of the Chinese investors may push up the A share prices. Third, the B share market has lower liquidity. The A share market and other foreign markets have higher liquidity and greater depth. Fourth, one reason why foreign investors invest in B shares is that they want to diversify their portfolio. If the prices of B shares are highly correlated to those of foreign shares, foreign investors may not want to invest in B shares because B shares do not have considerable diversification value. If so, little demand for B shares may drive down the B share prices. Fifth, regulatory changes may influence investors' expectations about the future returns. The reactions of domestic and foreign investors to regulatory changes may differ, and thus make the prices of A shares different from those of B shares.

Section 2 introduces the development of Chinese capital markets. In Section 3, an equilibrium asset pricing model is described and its empirical implications are derived. Section 4 tests the five hypotheses empirically. Section 5 concludes.

The Chinese capital markets

The Chinese capital markets are new. Prior to 1979, the capital markets were

almost non-existent because China had a highly centralized financial system. After 1986, money, bond, and equity markets were gradually established throughout the country. China's capital markets currently consists of a wide range of financial instruments, including government bonds; index bonds; municipal government bonds; and bonds issued by specialized banks, non-bank institutions, and industrial enterprises. Stocks are traded on the Shanghai and Shenzhen Exchanges. The Shanghai Securities Exchange was formally established in 1990, and initially eight stocks were listed. The Shenzhen Stock Exchange was also established in 1990. At the beginning, five companies were listed on the Exchange, including the Shenzhen Development Bank, Anda Transport Stock Company, Wanke Enterprise, Gintain Business, and Yuanye Business. In August 1994, the total number of stocks on both markets was 260.

China's equity markets opened to international investors when trading in B shares of Shanghai Vacuum Electron commenced on 21st February 1992. B shares can be owned by foreign investors and traded on the two stock exchanges. The B shares of China Southern Glass listed on 28th February were the first B shares listed on the Shenzhen Stock Exchange. Since then, many additional B shares listings have appeared on the two exchanges. As of August 1994, there was a total of 44 kinds of B shares listed in the two markets. The stock exchange of Hong Kong lists China-related stocks like CITIC Pacific, China Travel International, Denway, Guangdong Investments, Tian An China Land, and other so called 'red chips'. All the 'red chips' companies have extensive business with China. 2

China's market development has been controlled. Companies allowed to list shares have had to fulfill a greater number of requirements when issuing B shares than issuing A shares. In general, companies wishing to list B shares are expected to have an audit performed by international accountants before the initial public offering and produce a stable and adequate supply of foreign currency to pay dividends. In Shenzhen, among a number of additional rules, companies listing B shares must have a minimum return on capital of 10 percent in the year preceding the listing. In Shanghai, among other rules, B share issuers must have been operating profitably for at least two consecutive years prior to listing.

Beijing has imposed a number of regulations aimed at creating orderly markets. The cornerstones are a ban on short-selling, meant to curb speculation, and the requirement that foreign investors must report stock holdings of over 5 percent in any one company (see Price (1994) for details).

For an excellent description of the development of Chinese stock markets, see Bailey (1994).

Since the Renminbi is not internationally convertible, the B share trading takes place in foreign currency. In Shenzhen, B shares are quoted and trades are settled in Hong Kong dollars. In Shanghai, B shares are quoted, and trades are settled in U.S. dollar. Also, the dividends of B shares are paid in foreign currency.

An equilibrium asset pricing model of the Chinese stock markets

The model is an extension of the international asset pricing model developed by Eun and Janakiramanan (1986). In our simplified world, only two countries exist - the domestic country, D, and the foreign country, F. The domestic country is assumed to be small relative to the foreign country. In this respect, the foreign country can be thought to represent the rest of the world's capital markets, of which the domestic country is a relatively minor share.

In order to concentrate on the specific problem of market segmentation and abstract from the concept of exchange risk, a further simplifying assumption of a fixed exchange rate regime is made. In fact, B shares are quoted and trades are settled in foreign currency, so foreign investors do not face the currency exchange problems. In addition, we make the following assumptions:

A.I. There are two groups of investors: domestic investors (D) and foreign investors (F). The stock universe is separated into three mutually exclusive sets: set A contains A shares, set B contains B shares, and set C contains all foreign shares. A shares and B shares have the same rights and obligations. The opportunity set facing domestic investors consists of stocks in set A. The opportunity set for foreign investors consists of stocks in sets B and C. Finally, no investor is allowed to take a short position in any stock in set A or set B. 3

A.2. There exists a perfect competition in each country's capital market, but liquidity is different from one market to another. Amihud and Mendelson (1986) suggest that relatively illiquid stocks have a higher expected return and are thus priced lower to compensate investors for increased trading cost. Investors have to sell their securities at low prices and buy securities at higher prices in the illiquid markets. In our model, we will assume that the trading costs in the A share market different from those in the B share market because B share market is relatively illiquid.

A.3. Investors have homogeneous expectations of securities' risk and return. Security prices are distributed jointly normal.

Table 1

Summary of notations

Notation Explanation

N~

N~

Nc

nkA, nk B, nkc

,% PB, Pc

ra

FB

rc

FA~

FBC

W k

~k

r d, rf

8A

88

8C

Vector of the number of A shares outstanding

Vector of the number of B shares outstanding.

Vector of the number of foreign shares outstanding.

Vectors of the number of A shares, B shares and toreign

shares held by the kth individual, respectively.

The vectors of the current prices of A shares,

B shares and foreign shares, respectively.

The vectors of the random end-of-period prices

of A shares, B shares and foreign shares, respectively.

Vectors of the conditional expected value of the end-of-period

prices of A shares, B shares and foreign shares, respectively.

where f~ is the information at the beginning of the period.

Covariance matrix of the prices of A shares.

Covariance matrix of the prices of B shares.

Covariance matrix of the prices of foreign shares.

Covariance matrix of the prices of A shares and B shares.

Covariance matrix of the prices of B shares and foreign shares.

Investable wealth of investor k at time 0.

Random end-of-period wealth of investor k.

Expected value of 1~,i k.

Risk-free rate in domestic country and foreign country,

respectively.

Vector of per share trading costs of A shares.

Vector of per share trading costs of B shares.

Vector of per share trading costs of foreign shares.

A.4. There are no differential taxes, 4 and all assets are infinitely divisible.

A.5. Investors in both countries can also invest in risk-free assets. The risk-free rates in different countries are different because the capital controls in domestic country block the international capital movements.

The notations summarized in Table 1 will be used in the remainder of the paper.

3.1. The domestic investor's choice problem

Each domestic investor k chooses to invest his (or her) initial wealth W k between the risk-free asset and the risky A shares. Suppose that the investor has a

constant measure of absolute risk aversion, the utility function can be represented

by

Uk(W k) = --exp(--AkW*), (1)

where A* = - U"/U', the Pratt-Arrow measure of risk aversion.

Given the assumptions of jointly normal security returns and exponential utility,

the individual investor maximizes his (or her) certainty equivalent of end-of-period

wealth subject to a budget constraint:

max CEW* = ~k _ ( Ak/2)Var(lg'*) (2)

??kA,Fd

subject to

W k = r/~Ae A + Fd, (3)

~k = gtkA(P. A __ PArd _ gA + Wkrd) , (4)

Var(W*) = n'kAI~AHkA , (5)

where F d is the vector of the number risk-free assets demanded by the investor.

Eq. (3) is the budget constraint. The domestic investor can only choose A shares

and domestic risk-free assets. By solving this problem, we have the demand for A

shares:

~LA -- PA q -- ~A

Y/kA = Aki- A (6)

3.2. The foreign im, estor's choice problem

The foreign investor q faces a similar choice problem, but he (or she) has a

different opportunity set. The foreign investor chooses his (or her) holdings of B

shares, foreign shares and foreign risk-free assets to maximize his (or her)

certainty equivalent of end-of-period wealth. The foreign investor's choice problem

can be represented as

max CEW q = wq - ( aq/2)Var(l~ q) (7)

ll q B "nq C ' E~

subject to

W q =n qB , p B-i-n, qcPc+Ff,

-- ! !

Var(W ~ )=n qt . r . n q .+z n q n t F , cnqc+nq¢c r cnqc ,

where

(8)

(9)

(10)

F t, is the number of foreign risk-free assets demanded by the foreign

226 X. Ma / Pac~c-Basin Finance Journal 4 (1996) 219-239

investor. Eq. (8) is the budget constraint. By solving this problem, we have the

demands for B shares and foreign shares:

~LB -- PB rf -- ~rl -- AqFBcn q C

nqB = AqFBnB , (11)

P"C -- Pc re- ~c - AqFBcnq B

nqc = ZqFc (12)

3.3. Equilibrium asset pricing

To arrive at equilibrium asset prices, we can aggregate demands for all the

three securities and apply the market-clearing conditions. The market-clearing

conditions require that

Y'~nkA = X A, Y'~nqB = U B, Enqc = U c, (13)

k q q

nkA > 0, nqB > 0. (14)

Condition (13) states that total demand for securities should be equal to supply.

Condition (14) is the short-sale restriction. By summing Eqs. (6), (11) and (12),

and applying the market-cleating conditions, we obtain the equilibrium asset

prices.

Proposition 1. Under capital controls, the equilibrium asset prices are:

PA = l/rd(P"A -- AdFANA -- ~A), (15)

PB = 1/rf(tXB -AFFBNB - AFFBc Nc - gB), (16)

Pc = 1/rf(IXc - AFFc Nc - AFFBc NB -- gc), (17)

where 1/A D = ~Zkl/Ak, l/A v = ~ql/Aq.

3.4. Some empirical implications

Proposition 1 states that under capital controls, prices of A shares are different

from those of B shares. By dividing Eq. (16) by Eq. (15), Eq. 0, we obtain the

price ratio of B share and A share:

PB rd ( IXB--AFFBNB--AFFBCNC--~B )

PA rf ~-£ ~ L ~ ~--Aa " (18)

Eq. (18) provides some possible explanations for the price difference:

(1) The price differences depend on the difference between domestic risk-free rate r d and foreign risk-free rate rf. When the domestic risk-free rate is lower than the foreign risk-free rate, the cost of capital in the domestic market is lower than that in foreign market. The prices of A shares may be higher than those of B shares. In China, the real interest rate has been very low in recent years due to a high inflation rate. A lower interest rate may result in a higher A share price.

(2) The price differences also depend on the liquidities of different shares. Amihud and Mendelson (1986) have suggested that relatively illiquid stocks have higher expected returns and are thus priced lower to compensate investors for increased trading cost. The liquid shares have lower trading costs than the illiquid shares. In fact, B shares in the Chinese stock markets are relatively illiquid because their trading volumes are relatively low. Differential liquidity and trading costs may help explaining B shares' discounts.

(3) The price differences depend on the correlations between B shares and foreign shares which are represented in FBC. When the correlations between prices of B shares and prices of C shares, i.e. FBc, are large, the prices of B shares are low because B shares have low diversification values. From the standard CAPM model, the stock with a higher beta with respect to the market portfolio require a higher return and thus a lower price. If there are some foreign shares highly correlated to B shares, foreign investors may invest in these stocks instead of B shares, and thus the B shares are lower.

(4) The price differences can be affected by the investors' attitudes towards risk. The speculative Chinese investors may be risk-lovers. When A D is negative, the prices of A shares will be high if F A is big. A higher beta of A share means a higher price.

(5) The price differences can be influenced by regulatory changes. Regulatory changes may change investors' expectation about future prices P"A and ~z B. The price difference changes over time as government regulations change. In the next section, we will test these hypotheses empirically.

                                            Empirical tests

4.1. Data and methodology

In China, the two independent stock exchanges are located in Shenzhen (neighbor of Hong Kong) and Shanghai. Both markets are modem exchanges based on computerized order entry and book entry of ownership records. In August 1994, the total number of stocks listed on both markets was 260.

The markets have been open to foreigners since February 1992 when the first B shares were issued by the Shanghai Vacuum and Electronic Devices Company. All data was obtained from the Bridge Information System except for the Chinese interest rate and the consumer price index. Closing A and B share prices were gathered weekly (Monday close) from August 1992 to August 1994. To perform the time series and cross-sectional analysis, we choose 38 companies that have both A and B share listed. The listing dates of the companies are different, so that the numbers of observations of different stocks vary from 20 to 104.

To convert the prices and returns into Chinese yuan, we collected Renminbi/Hong Kong dollar exchange rate and Renminbi/US dollar exchange rates quoted in the swap market for security-related currency transactions.

Additional variables from international capital markets were collected as follows. First, we obtained stock indexes representing Hong Kong (Hang Seng) and the U.S. (SP500), and also, we construct a value-weighted index of the share prices for Hong Kong companies which had extensive business links to China (which are called 'red chips'). 5 Next, we collected time-series macroeconomic variables. They were the three month Chinese deposit rates, the Chinese consumer price index, the U.S. three-month treasury bill rates and the U.S. consumer price index.

Both cross-sectional and time series analysis are conducted. The cross-sectional analysis follows Hietala (1989) and Bailey and Jagtiani (1994) attempting to explain the cross-sectional differences of B shares' discounts. Different betas, liquidity and investors' attitudes toward risk may explain the differences of the relative prices of A shares and B shares across firms.

With time-series analysis, we will attempt to explain the changes of the B shares' discounts over time. Whether the changes of discounts are due to a change in interest rates or a change in government regulations will be explored.

4.2. Cross-sectional analysis

From our theoretical model, we know that the relative price of B share and A share depends on the investment betas. Following Bailey and Jagtiani (1994) and Fama and French (1992), we use full-period weekly returns to estimate the investment betas of A shares and B shares. Chart and Chen (1988) show that full-period 13 estimates for portfolios can work well in testing the CAPM model, even if the true 13's of the portfolios vary through time. The A share betas are estimated as follows:

RA # = OL A j -~- ~A jRAt q- lEA jr, (19)

where Rait = the return on A share j in week t, aAi = the A share 'alpha' for stock j, [3Aj = the A share beta with respect to A share index for stock j,

RAf = the return on A share index in week t.

Similarly, the B share betas can be estimated as

RB # = o£1j ~- ~liR,, + 6tjt, (20)

where c~ U = international 'alpha' for B share j, 13u = international 'beta' for B share j, R U = the return on the world market portfolio in week t. We use Hang Seng, SP500 and a value-weighted red chips index as proxies for world market index.

Table 2 summarizes the average weekly B shares' discounts (= (PB - PA)/PA), 13 estimates, and liquidity proxy (= average ratio of B shares to A shares for weeks with positive B share trading volume). The first finding is that the B shares' discounts vary widely across companies from 0.038 to 0.8761. Our cross-sectional will explain these differences. Second, the betas of A shares with respect to the A share index are all positive, but they vary widely from one firm to the other. The smallest is 0.61015 and the largest is 2.13467. Third, the betas of B shares with respect to the red chips index and the Hang Seng Index are relatively small and not significantly different from zeros. However, the betas of B shares with respect to SP500 vary widely from - 0.5978 to 1.79907 and most of them are positive and significant. The cross-sectional regression is run for the period of January 1994 to August 1994 as follows:

Discounti = a + b * B(A).i + c* B(B) i + d * liquidio, j + ~i' (21)

where

Discount~ = average weekly discount for stock j over period January 1994 to

August 1994, which is (PB -- PA)/PA;

Beta(A)j = A share beta with respect to A share index for stock j;

Beta(B)i = B share beta with respect to International index such as Hang Seng,

SP500 and value-weighted 'red chips' index for stock j;

liquidio'i = liquidity variable of B share j which is measured as a ratio: (average trading volume of B shares/B shares outstanding)/(average trading volume of A shares/A shares outstanding). 6

Table 3 summarizes the results of cross-sectional regressions which investigate the behaviors of B share and A share prices. Three explanatory variables are used: investment betas of A share with respect to the A share index, investment betas of B shares with respect to the SP500 index, a liquidity proxy.

The results are consistent with our risk-lover story. The Chinese markets are highly speculative markets, and the investors might be risk-lovers who want to make money in the short run. In our theoretical model, when the risk-aversion coefficient A D is negative, the prices of A shares become high relative to the prices of B shares if the investment betas of A shares are high. The cross-sectional results indicate that when the investment betas of A shares with respect to the A share index increase, the discounts of B shares increase. Historically, the Hong Kong and Taiwan stock markets in their early stages were highly speculative markets. The new Chinese markets might just follow the paces of Hong Kong and

6 This liquidity proxy is related but not identical to that used by Bailey and Jagtiani (1994).

Table 2

Summary of data and beta estimates ~

Company ds A shares B shares beta(a) beta(red) beta(hart) beta(sp) vratio vb

cbh -0.17087 104643 161707 0.837805 0.0766 - 0.040985 0.150704 0.2632

cim -0.18444 43968 13000 0.731192 0.062382 0.095287 0.162142 0.28944

cms -0.26033 152100 99000 0.908078 0.352368 0.243366 - 0.151088 0.32603

~t -0.5784 69000 22500 0.767751 -0.23717 0.016699 1.37329 0.19546

hua -0.51654 212330 77145 1.09027 -0.084809 -0.07649 - 0.010156 0.12575

kke -0.044827 162193 78802 0.646837 0.13499 0.128277 1.23684 0.052827

sba - 0.5526 98244 19800 1.00322 0.229278 0.191419 1.86829 0.152

sgl -0.037993 167443 84183 0.840827 0.012838 0.074233 -0.597825 0.18813

spc -0.366 225495 27300 0.942196 - 0.084826 - 0.169618 0.394857 0.13703

s~ -0.23063 843050 100000 0.610288 -0.161001 0.052451 - 0.020283 0.2263

svk -0.18041 187038 55917 0.677368 -0.080016 - 0.163557 0.368794 0.11129

lzp -0.33645 105224 79198 1.03761 0.030025 0.056728 0.589445 0.13661

szp 0.41517 396975 50792 0.601522 0.132986 0.058214 1.0037 0.10072

vic -0.53621 90660 63110 1.13509 0.134442 0.074635 1.10514 0.15642

zch -0.53319 135373 21840 1.00276 0.174545 0.136887 1.40958 0.090908

sde -0.87607 189593 100000 1.19745 0.111578 0.220812 1.79907 0.34866

ctm -0.64563 215423 109200 1.27845 0.17658 0.050223 -0.115071 0.41058

dht 0.38857 65992 60000 1.19782 0.128279 0.120286 0.37727 0.44322

~c -0.51386 459353 35000 1.16242 0.09035 0.300174 - 0.373428 0.36368

~t -0.37808 224150 92430 1.50702 0.35612 0.575439 0.95883 0.50294

rbe -0.76688 58014 33017 1.27546 0.11902 -0.091837 0.13698 0.32167

rcm - 0.68725 119912 50000 1.28965 0.02169 -0.013091 1.45284 0.48347

sai -0.73693 190861 70000 1.36694 0.048065 -0.14885 0.284413 0.25923

sca -0.66343 626380 336000 1.37938 0.126653 0.042957 0.364287 0.73031

sfb -0.57609 171008 60000 0.818777 0.154407 0.209015 0.54442 0.36024

s~ -0.4504 45000 40000 2.13467 0.216243 0.432829 1.10285 0.33324

shp -0.62935 75011 35000 1.26973 -0.214109 - 0.125182 0.592656 0.19924

shx -0.54539 102675 38500 0.765527 0.372694 0.248698 - 0.170043 0.37702

smt - 0.47923 66588 33000 1.49937 - 0.070215 0.0005578 0.628963 (/.35788

118.94286

142.30769

106.11864

29.82

14.64103

16.33929

15.48936

70.60606

33.93023

143.64516

61.52273

37.84615

29.53968

15.74359

17.82692

171.69231

133.98649

79.42254

128.68571

240.84444

43.52113

90.17808

140.8889

333.2027

119.29032

83.87344

44.2

52.22727

80.96875

¢5

2"

4~

ta~

sse -0.31558 203651 70000 0.942621 - 0.080357 -0.240814 0.88884 0.29526 117.57143

ssl -0.61122 158403 80000 1.67996 -0.029697 0.236868 1.36915 0.44898 248.76

jqd -0.4954 50000 143000 0.971006 0.23098 0.646771 1.10644 0.33017 399.18462

stm -0.49012 250000 175000 1.13316 0.284876 0.274921 0.889717 0.48626 274.98649

str -0.43472 587610 221000 1.28459 - 0.211046 -0.262186 0.717955 0.72858 288.47297

sva -0.59683 281339 121000 1.37152 -0.010688 -0.185874 0.576069 0.16635 110.55405

wmc -0.78513 139872 55917 1.17214 -0.019819 0.092572 0.497747 0.51714 317.3913

wss -0.71692 67249 30000 0.332263 0.106579 0.128085 -0.247602 0.3501 32.59722

ypg -0.14446 290000 143000 0.89063 0.032737 0.033259 1.80791 0.44263 185.09677

ds ~ Average weekly discounts of B shares; B share = B shares outstanding; A share = A shares outstanding; beta(a) = betas of A shares with respect to A

share index; beta(red) = betas of B shares with respect to red chips index; beta(han) = betas of B shares with respect to Hang Seng index; beta(sp) = betas of B

shares with respect to SP500 index; vratio = ratio of B shares volume to A shares volume; vb = average weekly trading volume of B shares.

2

4~

~tD

I

232

Table 3

Summary of

X. Ma / Pac(fic-Basin Finance Journal 4 (1996) 219-239

cross-sectional regression a

Constant Beta(a) Beta(sp) Liquidity proxy Adj R 2

- 0.139192 -0.303546

( 1.05778J) (-2.54849)

-0.409220

(7.88219)

-0.446018

(-9.58691)

- 0.123701 -0.276861

(-0.934277) (-2.25586)

-0.087604 - 0.265272

(-0.692103) ( - 2.24444)

- 0.101197

-1.77978)

- 0.031288

(-1.06403)

-0.071625

-1.37914)

- 0.089108 - 0.044755

- 1.67823) (-1.75514)

0.214647

0.065202

-0.013484

0.237738

0.245593

This table reports cross-sectional regression intended to explain average (January 1994 to August 1994) differences in the prices of A shares and B shares of Chinese corporations. The regression equation is Eq. (21). Estimation is by ordinary least square using a correction for heteroskedasticity, which was developed by White (1980).

Taiwan markets. This story explains up to 21% of the cross-sectional variability in

the B shares' discounts.

Our theory predicts that if the investment betas of B shares with respect to the foreign shares are high, the price of B shares should be low and the discounts of B shares should be high, since the diversification value of B shares are low. We use the betas of B shares with respect to SP500, Hang Seng Index and the 'red chips' index as explanatory variables. The coefficients from the estimation are all negative, but not significantly different from zeros when the betas of B shares with respect to Hang Seng and the red chips index are used in the estimation. We do not report these results in Table 3. The estimation results using the betas of B shares with respect to SP500 are reported in Table 3. The negative sign of the coefficient indicates that higher Beta(sp) induces higher discount, which is consistent with our theory. When a Beta(sp) is added as an explanatory variable, in addition to Beta(a), the adjusted R 2 increases from 21% to 24%. The relative liquidities of B shares also explain part of the cross-sectional differences. Our theory predicts that relatively low B share prices may be due to the relative illiquidities of B shares. As we observe, the B share market is relatively thin and trading activity is relatively low. We use the ratio of B share trading volume to A share trading volume as a liquidity proxy. The liquidity proxy explains about 1.5% of the cross-sectional differences, however, the negative sign of the coefficient is not consistent with our theory: higher liquidity should result in higher B share prices and lower B shares' discounts.

4.3. Time series analysis

In this subsection, we will analyze the variability of discounts over time.

Observation 1. The B shares' discounts of different companies move together

over time.

Table 4 represents the correlation matrix of the discounts of B shares for 17 companies. 7 We observe that all the discounts are positively correlated except for two companies. Table 5 represents the correlation matrix of changes of B shares discounts for the same 17 companies. All the changes of the B shares' discounts are highly positive correlated. The comovements of B share discounts may reflect the changes of macroeconomic variables or government's regulations. Our theoretical model predicts that the B shares' discounts will increase with foreign interest rates and decrease with domestic interest rates. On the other hand, the model also predicts that the investors' conditional expectations about future stock price matter. When domestic investors are optimistic about future return, the prices of the A shares increase. On the contrary, the prices of A shares decrease as the investors become pessimistic. In China's stock markets, investors are relatively naive and may overreact to regulatory changes. The rest of this subsection, we will test whether the comovements of the discounts are due to changes in interest rates or government regulations.

Observation 2. The B shares' discounts do not have a constant long-run mean.

Another interesting question we want to investigate is whether the B shares' discounts are stable over time. One would expect the investors' attitudes toward risks or the correlation between B shares and foreign shares to be relatively stable over time. These explanations would suggest the existence of long-run mean. We investigate the hypothesis of a constant B shares' discount in Table 6 using unit root tests for the ratios of the A shares' prices to the B shares' prices. The following equation is estimated using ordinary least squares:

Alog( PB/PA )it = [3o + [31 log(PB/PA) ir-I + [32 t + A log( PB/PA )it 1

+ A log( PB/PA);,_2 + A log( PB/PA)i, 3

+ A log( PB/PA)it_4 + Et, (22)

where

A log( PB/PA)i, = log( P,/PA);,- log( PB/PA)i,_~.

The augmented Dickey-Fuller (DF) test is used to test the null hypothesis that natural logarithms of price ratios have unit roots. The null hypothesis that the price ratios have unit roots can be accepted for 16 out of 17 firms at the 0.05 level.

4~

Table 4

Correlation matrix of discounts of 17 companies

CBHDS CTMDS DHTDS HUADS KKEDS RBEDS SCADS SBADS SGLDS PCDSS STMDS STRDS SVADS SZPDS VICDS WSSDS ZCHDS WVDS

CBHDS 1

CTMDS 0.47399 l

DHTDS 0.813 0.63455 I

HUADS 0.77027 0.23801 0.51021

KKEDS 0.33069 0.20344 0.39561

RBEDS -0.5188 0.11392 -0.5248

SCADS 0.75619 0.53455 0.77984

SBADS 0.84884 0.4135 0.76167

SGLDS 0.87389 0.49521 0.85403

SPCDS 0.89607 0.39297 0.78799

STMDS 0.70427 0.73346 0.82687

STRDS 0.80089 0,74839 0.89865

SVADS 0.26359 0.19324 I).27966

SZPDS 0.80505 0.47613 0.74756

VICDS 0.87184 0.42357 0,8173

WSSDS I).4231 0.1464 0.34643

ZCHDS 0.84875 0.42048 0.79724

WVDS 0.32214 0.3928 0.38788

I

0.28245 I

- 0.28457 - 0.407 I

0.56688 0.19469 -0.16014 I

0.64712 0.48849 0.30218 0.82584 I

0.63094 0.54816 0.73925 0.61476 0.75635 1

0.74926 0.56337 -(I.51325 0.75883 0.9018 0.88058 I

0.38227 0.30671 0.58053 0.50251 0.47976 0.81225 0.56975 1

I).48696 0.15031 -0.38383 0,7804 0.67622 0.76963 0,7064 0.84581 I

I).36483 0.16422 0,39268 0.60895 0.48271 0.14525 0.3731 0.015336 0.257381

0.64267 0.67411 0.58075 0.59581 0.75041 0.86375 0.84477 0.67595 0.62592 0.17401 I

0.7103 0.38447 0.3317 0.86048 0.93964 0.76614 0.9116 0.51177 0.76009 0.5046 0.7239 I

0.49683 0.4706 0,58838 -0.3632-0.469 -0.5567 0.6225 0.1553 0.1639 0.1666 0.5239 0.51711

0.59752 0.3613 0,3444 0.85854 0.89497 0.75935 0.84956 0.52869 0.75672 0.51438 0.73411 0.9255 0.44733

0.32487 0.04826 0.1106 0.57161 0.40377 0.20522 0.37756 0.21351 0.4451 0.461 0.18863 0.43939 -0.10190.34368 I

5"

4~

I

Table 5

Correlation matrix of changes of discounts of 17 companies

CBHDSC CTMDSC DHTDSC IIUADSC KKEDSC RBEDSC SCADSC SBADSC SGLDSC SPCDSC STMDSC~7 STRDSC SVADSC SZPDSC VICDSC WSSDSC ZCHDSC WVDSC

CBHDSC I

CTMDSC 0.55571 I ~

DHTDSC 0.50174 0.74947 1

HUADSC 0.7601 0.59325 0.53185 I

KKEDSC 0.56447 0.39216 0.2749 0.54321 I 2'

RBEDSC 0.43369 0.77769 0.64861 0.51617 0.298 I ~.

SCADSC 0.58794 0.84164 0.81536 0.60297 0.34139 0.75442 1

SBADSC 0.80926 (I.63844 0.59084 0.73419 0.54254 0.55616 0.69382 I

SGLDSC 0.79852 0.55121 0.56521 0.70647 0.54009 0.41029 0.59679 0.74807 I

SPCDSC 0,7902 0.63372 0.61762 0.754 0.61263 0.49305 0.69452 0.79729 0.77033 1 ~"

STMDSC 0.46494 0.76033 0.81274 0.5439 0.30869 0.68251 0.76002 0.54444 0.57496 0.51408 I

STRDSC 0.51152 0.75872 0.84167 0.57008 0.29 0.68831 0.8857 0.63992 0.5555 0.67434 0.76023 I ~,~

SVADSC 0.58714 0.75299 0.71536 0.59583 0.35467 0.77587 0.82548 0.62738 0.60781 0.62402 0.68756 0.7506 I "~

SZPDSC 0.55766 0.40955 0.33481 0.55977 (1.51291 (I.35042 0.32928 0.48654 0.55798 0.55023 0.41371 0.31083 0.29891 I

VICDSC 0.79884 0.57421 0.68508 0.65865 (}.4236 0.46538 0.70496 0.82121 0.68959 0.80501 0.52421 0.70704 0.60801 0.43309 I ~~'x~

WSSDSC 0.23183 0.52586 0.36258 0.25177 0.15782 (].57899 (I.36599 0.23099 0.18764 0.23736 0.43519 0.2679 0.412 0.25826 0.11514 I

ZCFIDSC 0.79138 0.51122 0.47211 0.57036 0.59263 0.43376 (}.55156 (I.75607 0.73478 0.77176 0.42546 0.50598 0.51342 0.55507 0.74837 0.17317 I

WVDSC 0.46149 0.605 0.73168 0.52519 0.25826 0.46836 (I.737 0.54305 0.55911 0.56746 0.62294 0.72368 0.73262 0.16931 0.6025 0.1069 0.36269 I /

236

Table 6

Unit root tests a

X. Ma / Pacific-Basin Finance Journal 4 (1996) 219-239

Company Augmented Dickey-Fuller statistics

Critical value

5% - 3.47

1% -4.07

cbh -4.14

ctm - 3.35

dht - 2.50

hua - 2.81

kke - 1.84

rbe - 3.29

sca - 2.88

sba - 2.67

sgl -2.63

spc - 2.47

stm 1.42

str - 2.45

sva - 2.96

szp - 2.48

vic - 2.97

wss -2.32

zch - 3.09

The test is a unit root test in the natural logarithm of the price ratios of B shares to A shares. Unit root is not rejected if the Augmented Dickey-Fuller statistics is smaller in absolute value than the critical value. Observations are weekly from January 4, 1993 to August 22, 1994.

These tests generally reject the hypothesis that the B shares' discounts are stationary and have long-run means. The evidence suggests that the investors' attitudes toward risks and the correlation between B shares and foreign shares cannot explain the time-series variability of the B shares' discounts.

Obserrvation 3. The regulatory changes can explain part of the time-series variability of the B shares' discounts.

From Eq. (18), Eq. (), we know that the price ratio of B shares to A shares over time may depend on the domestic interest rate and the foreign interest rate. Additionally, the investors' expectations about future prices P~A and ~z B can change the price ratios over time. One would expect that government policies' changes and regulatory changes may change investors' expectations about the future return, and thus the B shares' discounts. On June 30, 1993, the Chinese government adopted new measures to control inflation: forbidding bank loans to institutional investors who trade stocks, requiring some institutions to buy treasury bonds, forbidding institutional investors to channel funds from the public directly, and increasing interest rates. Since then the stock markets experienced a year-long

25

20

15

-¢ 10

o

-5

"--~iscount

I-- --A share index

h B share index /\

,/ //,\/,.,/

\I "" ~\ , ",

,,,

Time (Jan., 93 - Aug., 94)

Fig. I. Discounts, A share and B share indexes.

decline until the government announced a market-rescue plan on July 30, 1994. The plan included a temporary freeze on issues and listing of new shares and the development of multi-channels for pumping funds into the stock markets. After the plan was announced, the Shanghai A share index increased more than 100% in a week (see Fig. 1). It is interesting to test whether the changes of price ratios of B shares to A shares over time are due to the changes of interest rates or to the regulatory changes. We use the U.S. three-month treasury rate as a proxy of the foreign interest rate and use the Chinese three-month deposit rate as the domestic interest rate. Monthly data from August 1992 to July 1994 are used in the estimations. The following equation is estimated:

(.o) "~A t = [~0 ~- ~1 * r + ~2 * dummy, + ~,, (23)

where r d = real interest rate in China; rf = real interest rate in U.S.; dummy = dummy variable reflecting regulatory changes, and it is zero from August 1992 to June 1993 and one from July 1993 to July 1994.

This equation is based on the theoretical model [see Eq. (18)] in which the price ratio is positively correlated with the interest rate ratio. We also regress the value-weighted A share and B share indexes on the interest rates and the dummy variable to test the hypothesis.

Table 7

Time series regressions a

Dependent variable Constant rinc rinu rinc/rinu dummy Adj R 2

PA 12.4024 -- 1.60316 -- 3.93711 0.4398

(15.9625) (--0.329677) (--3.97726)

PB 5.75309 -- 9.82576 -- 1.00078 0.2965

(22.2023) (-- 1.04446) (--2.80303)

PB/PA 0.52945 --0.00084 0.224669 0.3209

(11.2906) (--0.3946) (3.47352)

Regression of the value-weighted price ratios, A share index and B share index on real interest rates

of China and the U.S. and the regulatory dummy variable, rinc = real interest rate of China, rinu = real

interest rate of U.S.

Table 7 summarizes the regression results from which we can see that the dummy variable explain most of the changes of prices over time. The coefficients of the interest rate variables are not significantly different from zero. When the Chinese government changed its policy to a relatively 'tight' policy, both the domestic investors and foreign investors would expect that the future prices should decline. As a result, prices of both A shares and B shares decreased. However, the A share prices decreased more than did the B share prices as regulations changed. We can see this from the values of the two coefficients of the dummy variables in the first two regression equations of Table 7 (-3.93711 and - 1.00078 respectively). This implies that the reactions to the regulatory changes were much stronger in the A share market than in the B share market. Consequently, the price ratio of B shares and A shares increased as the government changed its policies (the coefficient of the dummy is 0.224669).

Conclusion

Economic reforms in developing countries - including equity market openings and international equity offerings have encouraged large increases in foreign purchases of emerging market equity in recent years. The evidence from China's new stock markets suggest that the prices of the same stock may differ if the stocks is traded in a segmented market. The price difference may be due to the investors' attitudes toward risks, regulatory changes and the diversification value of the stocks in emerging markets. The lesson from this study to the international portfolio investors is that the prices of the stocks in emerging markets may be overvalued because of the speculative behaviors, and regulatory changes may create an extra risk in emerging markets.

For further reading

Adler and Dumas, 1983, Chen et al., 1986, Cutler et al., 1989, Fama and MacBeth, 1973, Gale, 1992, Harvey, 1993, Hirshleifer, 1988, Lee et al., 1991, Mullin, 1993, Merton, 1973

References

Adler, Michael and Bernard Dumas, 1983, International portfolio choice and corporate finance: A

synthesis, Journal of Finance 38, 925-984.

Amihud, Yakov and Haim Mendelson, 1986, Asset pricing and the bid-ask spread, Journal of

Financial Economics 17, 223-247.

Bailey, Warren and Julapa Jagtiani, 1994, Foreign ownership restrictions and stock prices in the Thai

capital carket, Journal of Financial Economics.

Bailey, Warren, 1994, Risk and return on China's new stock markets: Some preliminary evidence,

Pacific Basin Finance Journal.

Chan, K.C. and Nai-fu Chen, 1988, An unconditional asset-pricing test and the role of firm size as an

instrumental variable for risk, Journal of Finance 43, 309-325.

Chen, Nai-fu, Richard Roll and Stephen Ross, 1986, Economic forces and the stock market, Journal of

Business 59, 451-471.

Cutler, David M., James M. Poterba and Lawrence H. Summers, 1989, What moves stock prices,

Journal of Portfolio Management, Spring, 1-12.

Errunza, Vihang and Etienne Losq, 1985, International asset pricing under mild segmentation: Theory

and test, Journal of Finance 40, 105-124.

Eun, Cheol S. and S. Janakirmanan, 1986, A model of international asset pricing with constraint on

foreign equity ownership, Journal of Finance 41, 897-914.

Fama, Eugene F. and James MacBeth, 1973, Risk, return, and equilibrium: Empirical test, Journal of

Political Economy 38, 607-636.

Fama, Eugene F. and Kenneth R. French, 1992, The cross-section of expected stock returns, Journal of

, 427-465.

Gale, Douglas, 1992, Standard securities, Review of Financial Studies 59, 731-755.

Harvey, Campell R., 1993, Predictable risk and returns in emerging markets, Unpublished working

paper (Duke University, Durham, NC).

Hietala, Pekka, 1989, Asset pricing in partially segmented markets: Evidence from the Finnish market,

Journal of Finance 44, 697-718.

Hirshleifer, David, 1988, Residual risk, trading costs, and commodity futures risk premia, Review of

Financial Studies 1, 173--193.

Lee, Charles, Andre Shleifer and Richard Thaler, 1991, Investor sentiment and the closed end fund

puzzle, Journal of Finance 40, 75-110.

Mullin, John, 1993, Emerging equity markets in the global economy, FRBNY Quarterly Review,

Summer, 54-83.

Melton, Robert C., 1973, An intertemporal capital asset pricing model, Econometrica 41, 867-887.

Price, Margaret M., 1994, Emerging stock markets (McGraw-Hill, New York).

Stultz, Rene, 1981, A model of international asset pricing, Journal of Financial Economics 9, 383-406.

Stultz, Rene and Walter Wasserfallen, 1993, Foreign equity restrictions and shareholder wealth

maximization: Theory and evidence, Unpublished working paper (Ohio State University, Columbus,

OH).

White, H., 1980, A heteroskedasticity-consistent covariance matrix estimator and a direct test for

heteroskedasticity, Econometrica 48, 149-170.

Tugas Softskill :

Latar Belakang :

Ada dua jenis kontrol modal di pasar modal negara berkembang.

a) Pembatasan kepemilikan asing ekuitas dalam negeri. Negara-negara berkembang sering memberlakukan pembatasan kepemilikan asing ekuitas dalam negeri untuk memastikan kontrol domestik perusahaan lokal, terutama perusahaan-perusahaan dianggap sebagai 'strategis penting' untuk kepentingan nasional. Di Brasil, misalnya, setiap investor asing dibatasi untuk memiliki tidak lebih dari 5% dari hak suara setiap perusahaan. Di negara-negara seperti India dan Meksiko, batas adalah 49%. Di Cina dan Filipina, sebuah perusahaan lokal menerbitkan dua jenis saham, yaitu, saham saham A dan B. Asing diizinkan untuk memegang saham hanya B. Di Thailand, pasar saham memelihara dua daftar terpisah untuk saham biasa: satu untuk penduduk setempat, yang 'Main Board'; satu untuk orang asing, yang 'Alien Dewan'. Informasi lebih lanjut tentang pembatasan kepemilikan asing dapat ditemukan di Harga (1994).

(b) Pembatasan investasi domestik di pasar modal asing. Negara-negara berkembang juga memiliki derajat yang berbeda dari pembatasan arus modal keluar. Misalnya, warga domestik mungkin tidak diizinkan untuk membawa devisa luar negeri bebas. Semacam ini keluar kontrol devisa di Korea Selatan, Cina, Taiwan, dan negara-negara berkembang lainnya. Di Cina, warga Cina tidak dapat membeli mata uang asing secara bebas. Kedua bentuk kontrol modal membuat pasar modal muncul tersegmentasi dari pasar modal dunia.

Kesimpulan :

Reformasi ekonomi di negara-negara berkembang - termasuk pembukaan pasar ekuitas dan penawaran ekuitas internasional telah mendorong peningkatan besar dalam pembelian asing ekuitas pasar berkembang dalam beberapa tahun terakhir . Bukti dari pasar saham baru China menunjukkan bahwa harga saham yang sama mungkin berbeda jika saham diperdagangkan di pasar tersegmentasi . Perbedaan harga mungkin karena sikap investor terhadap risiko , perubahan peraturan dan nilai diversifikasi saham di pasar negara berkembang . Pelajaran dari penelitian ini untuk investor portofolio internasional adalah bahwa harga saham di pasar negara berkembang dapat dinilai terlalu tinggi karena perilaku spekulatif , dan perubahan peraturan dapat membuat risiko tambahan di pasar negara berkembang .