Chinese Dual-Class Shares Listed in Hong Kong and Mainland China

Table of Contents

Abbreviations

Symbols

Figures

Tables

1 Purpose of Research
1.1 Motivation
1.2 Methodology

2 Equity Markets in China and Hong Kong
2.1 Stock Exchanges in Shanghai, Shenzhen and Hong Kong
2.2 Market Regulations and State Ownership
2.3 Dual-Class Shares
2.4 Institutional Investors

3 Literature Review
3.1 Asset Pricing under Foreign Ownership Restrictions
3.2 Competing Concepts in Market Segmentation Theory
3.2.1 Differential Risk Hypothesis
3.2.2 Differential Demand Hypothesis
3.2.3 Liquidity Hypothesis
3.2.4 Asymmetric Information Hypothesis
3.3 Assessing Fundamental Concepts in the Context of China
3.3.1 Market Segmentation Theory and the H-Share Discount
3.3.2 Market Segmentation Theory and the B-Share Discount
3.4 Literature Synopsis and Analytical Implications

4 Revisiting the H-Share Discount Puzzle
4.1 Data Sample
4.2 The H-Share Discount

5 Statistical Analysis
5.1 Preliminary Analysis
5.2 Stock Returns and Market Integration
5.3 Panel Data Analysis of the H-Share Discount
5.3.1 Data and Model Specifications
5.3.2 Results

6 Summary and Conclusion

Appendix

Bibliography

Declaration of Authorship

Abbreviations

illustration not visible in this excerpt

Symbols

illustration not visible in this excerpt

Figures

Figure 1: Development of Major Price Indices at the SHSE, SZSE, and HKSE benchmarked by the MSCI World Index, all values indexed on December 30, 2005 (01/2006-09/2007)

Figure 2: Average Daily Stock Turnover in 100 Million RMB (HKD for HKSE) at the SHSE, SZSE and HKSE, all values indexed on June 30, 2003 (07/2003-07/2007)

Figure 3: Sample Mean H-Share Discount and Mean Prices for H- and A-Shares based on RMB values across Time (01/1995-08/2007)

Figure 4: Sample Market Capitalization based on RMB values by Individual Firms as of September 30, 2007

Figure 5: Mean and Capitalization-Weighed H-Share Discount across Time (01/1995-09/2007)

Figure 6: Distribution of H-Share Discounts and Number of Firms across Time, starting with 10 firms in the sample (01/1996-09/2007)

Figure 7: Average H-Share Discountsillustration not visible in this excerptfor Individual Firms from the Start of each Dual Series (07/1993-09/2007), SZSE Listings are indicated with Cross-stripes and the Sample Mean with Diagonal Stripes

Figure 8: Average H-Share Discount across Time (01/1995-09/2007)

Tables

Table 1: Domestic Market Capitalization and Gross Domestic Product for Selected Countries in Billion USD (12/2005-12/2006)

Table 2: Number of Mainland Chinese Firms Issuing Particular Share Classes (06/2002-06/2007)

Table 3: Market Capitalization, Average Daily Stock Turnover, Trading Volume, Transactions and Major Indices (06/2003-06/2007)

Table 4: Basic Statistics for Share Classes and Capital Structures (06/2002-06/2007)

Table 5: Summary of Key Findings from the Literature Review on the Chinese Discount Puzzle

Table 6: Basic Statistics for the Discounts of Cross-Listed H-and A-Shares (07/1993- 08/2007)

Table 7: Basic Statistics for Attributes of Cross-Listed H- and A-Shares, Average Values Based on Daily Observations Available (07/1993-09/2007)

Table 8: Comovement of H- and A-Share Returns with A- and H-Share Market Indices (07/2003-07/2007)

Table 9: Regression Results Model 1 (Basic)

Table 10: Regression Results Model 2 (Liquidity Proxy 1)

Table 11: Regression Results Model 3 (Liquidity Proxy 2)

Table 12: Regression Results Model 4 (Demand Proxy)

Table 13: Basic H- and A-Share Data, Charts and Scatter Plots for Sample Firms (09/2004-09/2007)

1 Purpose of Research

1.1 Motivation

This paper aims at explaining the phenomenon of price anomalies between dual-class shares of companies located in mainland China (hereafter China). A-shares listed on either the Shanghai Stock Exchange (SHSE) or Shenzhen Stock Exchange (SZSE) command a premium over the price of the corresponding firm’s H-shares traded at the Stock Exchange of Hong Kong (HKSE). This pricing puzzle arises from the segmentation of Chinese equity markets – H-shares may be exclusively acquired by Hong Kong residents and international investors whereas A-shares are restricted to mainland Chinese investors. Although both classes of stock are entitled to the same future cash flows, investors are only willing to buy H-shares at a price significantly lower than that of A-shares. This unique setup offers the opportunity to test competing theories about the effects of market segmentation on asset pricing and to examine the factors that induce the price gap between cross-listed shares on different stock exchanges. Knowledge of the variables determining the price spread between H- and A-shares can make valuable contributions in a number of ways. Firstly, companies in mainland China pursuing initial public offerings (IPO) or seasoned equity offerings (SEO) may base their financing decision on a more thorough understanding of the parameters affecting stock prices of cross-listings in the respective markets. Secondly, policymakers in emerging country stock markets may draw conclusions concerning the design of foreign ownership regulation and investment restraints imposed on domestic and foreign investors. Lastly, international and local investors may build on a more profound understanding of the H- versus A-share discount (hereafter H-share discount) to narrow down attractive investment opportunity sets, especially in the light of the latest regulatory changes on the Chinese equity market.

As of August 2007 the government body monitoring and regulating the national currency, China’s State Administration of Foreign Exchange (SAFE), loosened its rigorous foreign exchange policy. Prior to the recent SAFE ruling, the annual amount to be freely converted from Chinese Yuan Renminbi (RMB) into foreign currencies was capped at a 50,000 United States Dollar (USD) limit. Under the new regime, mainland retail investors are granted unlimited convertibility of RMB into Hong Kong Dollar (HKD) – given that investments flow into the Hong Kong securities market.^[1] In the first place, policymakers might have intended to liberalize the Chinese capital account by gradually increasing the convertibility of China’s currency. Following this announcement, the HKSE experienced an almost 40 percent rally on its H-share listings.^[2]. It is unclear whether this surge was caused by cross-border flows of capital into the Hong Kong market or by speculative activity of Hong Kong and international investors. From the perspective of theory, providing access to the HKSE for Chinese investors has partially integrated the previously separated markets of China and Hong Kong. In such a situation, it should be anticipated that the stock prices of H- and A-shares converge, balance in a new equilibrium and, ultimately, reduce the H-share discount.

During an earlier, likewise transformational, event in 2001 Chinese investors had engaged in bidding up the prices of one share class until prices, which had been remarkably spread before the event, almost reached equality. In that year the China Securities Regulatory Commission (CSRC), the national stock exchange regulation authority, announced that domestic investors would be allowed to purchase B-shares that had formerly been traded by international investors only and exhibited a large discount compared to their respective A-share listings. Instantly, B-share prices rose dramatically, lowering the B-share discount from an average 75 percent to a historic low of 8 percent on average within weeks.^[3] Current market movements in Hong Kong signalled the commencement of an analogous trend for the H- and A-shares, when the average discount dropped by more than 10 percent points in the 5 trading days subsequent to the announcement of the new regulation. Despite this market reaction, opening the Hong Kong market to Chinese investors had less impact on the H-share discount than the historic CSRC move in the B-share segment. This suggests that markets for Hong Kong H-shares and China A-shares in the same firms are strongly segmented. Since the institutional frameworks are similar, it seems a promising approach to build on theories developed in B-share studies, transfer findings with strong verification to the analysis of the H-share discount puzzle in this paper and conduct time-series and cross-sectional regressions to differentiate which concepts have the highest empirical validity.

1.2 Methodology

Compared to the large number of scientific articles on the B- versus A-share discount,^[4] researchers in the field have devoted rather little attention to the subject of the H-share discount.^[5] Nonetheless, numerous insights and hints can be drawn from existing studies treating the B-share pricing anomaly. It is the purpose of this paper to thoroughly review existing literature on the theory of market segmentation and use recent data to test the validity of selected concepts in an empirical analysis of the H-share discount. Hence, this paper is structured as follows: In chapter two, equity markets in China, key players, institutions and regulations are introduced to establish the necessary degree of familiarity with the rules of equity trading in greater China. Some implications of market interferences by government entities and the recent boom in China’s A-shares will be reflected. The literature review in chapter three is, besides a summary of the results from prior B- and H-share studies, dedicated to a discussion of different theories on market segmentation, foreign ownership restriction and asset pricing as well as their applicability in the context of China. In chapter four, empirical evidence on the H-share discount and descriptive statistics on the data sample will be presented. Based on daily share prices, mean and weighted discounts will be calculated to demonstrate the dimension of the price spread. The statistical analysis is covered in chapter five. Regressions of individual stock returns on different market indices will help to clarify which systematic market risks influence H- and A-share prices. Subsequently, data sources, modifications and the model specifications for the statistical analysis derived from the theories in the previous chapters are introduced. The panel data analysis of the time series data for the 37 dual listings of A- and H-shares will be presented and interpreted in order to explain the time-series and cross-sectional variability of H-share discounts. Within chapter six the empirical findings and their interpretation are summarized before conclusions, recommendations and hints on future research areas are outlined.

2 Equity Markets in China and Hong Kong

2.1 Stock Exchanges in Shanghai, Shenzhen and Hong Kong

The HKSE has the longest tradition among Chinese equity markets. Its history goes back as far as to the late nineteenth century, when the exchange started off as the Association of Stockbrokers in Hong Kong, to eventually become what is one of the world’s largest stock exchanges today.^[6] Market capitalization of the 1,023 equity listings at the HKSE was 17,595 billion HKD or, equivalently, 2,225 billion USD in August 2007. All equity trading in China, on the other hand, takes place at the SHSE and SZSE which have both been launched in the early 1990s. While the SHSE started its operations in November 1990, the SZSE soon followed in March 1991, so that the two exchanges could begin to jointly forge an equity market of global scale. As of June 2007 both the SHSE and the SZSE accounted for a combined number of 1,586 listed firms, which account for a total market capitalization of 16,623 billion RMB or 2,211 billion USD (Table 3, p. 51). These numbers seem to support the notion of a stable and strong financial sector in the mainland, considering that China is still perceived to be an emerging country,^[7] although it has reported a gross domestic product (GDP) at current prices of approximately 2.7 trillion USD in 2006.^[8] Nevertheless, such an interpretation may be short-sighted. Before delving into the specifics of equity segmentation and trading in China, it is therefore advisable to briefly recount the development of some performance indicators for Chinese equity markets and pinpoint recent trends that could help to explain the present strength of financial sector variables.

Despite the duration and impressive dimension of GDP growth in China, stock markets have appeared to lag behind industrial progress. To illustrate this lag in financial sector development, domestic market capitalization (DMC) scaled by GDP provides a reasonable estimate.^[9] GDP reflects the total value of final goods and services produced within a country. DMC measures the total market capitalization of listed domestic companies in a country, which is the market capitalization of main and small and medium-sized business markets. Preferred and common stocks are included; investment funds, rights, warrants, convertibles and foreign companies are not.^[10] Dividing DMC by GDP reveals the dimension of a nation’s equity account compared to the sum of all goods and services produced in the national economy.^[11] Markets in countries with a high capital market orientation, like the United States or United Kingdom capitalized more than their respective GDP (Table 1, p. 5). In contrast, the two major Chinese exchanges in the mainland only accounted for roughly 20 percent of the nation’s GDP in market capitalization during 2005.

illustration not visible in this excerpt

In China’s 2005 GDP peer group, Germany and the United Kingdom both had DMC to GDP ratios at least twice as high as that for China. As a consequence, it had to be concluded that China’s equity markets fell short of economic growth. The ratio for Hong Kong, on the contrary, reflected the sophistication of its equity markets with a DMC close to 6 times GDP in 2005. It can be observed for both Chinese positions in the table that DMC has grown significantly within the year 2006. Indeed, at the beginning of 2006 Chinese stock markets experienced tremendous stock price increases. The exceptional uproar in Chinese stock prices can be reviewed by following the Shanghai SE Composite Index and Shenzhen SE Composite Index from January 2006 through August 2007, whose performance dwarfed increases of other major indices like the Hang Seng Index at the HKSE or the MSCI World Index during the same interval (Figure 1, p. 6). The rally on local stocks in China has resulted in total market capitalization at the SHSE and the SZSE rising by more than 400 percent from 3,243 billion RMB in January 2006 to 16,623 until July 2007.^[12]

illustration not visible in this excerpt

Figure 1 : Development of Major Price Indices at the SHSE, SZSE, and HKSE benchmarked by the MSCI World Index, all values indexed on December 30, 2005 (01/2006-09/2007)

Over the same time period, total market capitalization at the HKSE increased by 165 percent from 6.640 to 17.595 trillion HKD. The stunning increase in equity prices but also the tumultuous trading activity of retail investors along with the IPO’s of Chinese financial giants like Industrial and Commercial Bank of China, China Life Insurance, Ping An Insurance, or Bank of China have diverted the attention of the broader public to the case of China.^[13] Additional data from the Chinese stock exchanges might help to undermine why worries about potential bubbles in the stock market have surfaced. Just from June 2006 to June 2007, the daily stock turnover increased by 563 percent and 476 percent at SHSE and SHZE, respectively. During the same period daily trading volume rose from 4.482 to 12.621 billion shares in Shanghai and 2.780 to 6.583 billion shares in Shenzhen. A similar but weaker trend could be observed for the HKSE when daily stock turnover grew by 159 percent from initially 29.535 in July 2006 to 76.744 billion HKD in July 2007 (Figure 2, p. 7 and Table 3, p. 51). Also, during the same time the daily trading volume at the HKSE increased by 159 percent. In June 2007 price earnings ratios for A-shares were 42.72 in Shanghai and 50.90 in Shenzhen when H-shares ratios were 19.96.^[14]

illustration not visible in this excerpt

Figure 2 : Average Daily Stock Turnover in 100 Million RMB (HKD for HKSE) at the SHSE, SZSE and HKSE, all values indexed on June 30, 2003 (07/2003-07/2007)

Were the DMC to GDP ratio calculated as of July 2007, China would score somewhere in the 80 percent region – an enormous improvement with regard to its limping position only a few months ago. This appears even more striking when considering that much structural inefficiency still prevails. Systematic shortages of share supply, low levels of free-float, the scarcity of privately managed institutional investors, and preferred treatment of state-owned companies upon listing inquiry impose massive barriers on capital market efficiency. Rather than stimulating the allocation of capital, primary and secondary equity markets have seemed to fulfill a financing function for state-owned firms that need to be turned around before running into bankruptcy. After listing on the stock exchange, most firms are still government-controlled and might pursue goals other than maximizing shareholder value or firm profits.^[15] The explosive growth in Chinese stocks has therefore not only surprised many market participants but also raised the question whether equity markets have detached from the underlying fundamentals because of massive speculation and hype in the retail-driven Chinese markets or whether they in fact reflect the strong growth perspectives of Chinese firms.^[16]

2.2 Market Regulations and State Ownership

Core regulators on Chinese equity markets are the SAFE and the CSRC. The SAFE is in charge of designing and implementing the balance of payments for China. With the long-term goal of achieving an equilibrium balance of payment position, the SAFE analyzes the capital account and foreign exchange movements in order to assess the feasibility of future moves towards free RMB convertibility. The supervisory activity of the SAFE extends from managing and monitoring inward and outward bound foreign exchange transactions to setting the target exchange rate for the present currency peg. Based on foreign exchange analysis and market research the SAFE drafts recommendations for the foreign exchange policy of the Chinese central bank, the People’s Bank of China.^[17] As a result, China had pegged the RMB to the value of the USD since 1994, permitting minor swings of 0.3 percent around 8.2270 RMB on the USD, because Chinese policymakers intended to induce stable import prices for foreign consumer markets to maintain the growth of the export-driven domestic production. Upon pressure by the United States and neighboring countries, the USD peg was abandoned and transitioned into a currency basket peg. This basket currently contains ten currencies that are dominated by the USD, EUR and JPY. Since inception of the new peg, the exchange rate was allowed to float within a 0.3 percent band on every trading day. Partially unleashing the RMB has resulted in a RMB appreciation of roughly 9 percent versus the USD until September 2007 since the removal of the USD peg in July 2005. Despite this RMB revaluation, it remains to be doubted that the Chinese currency is currently priced at its intrinsic value. The Chinese government has continued to push the RMB/USD relation towards a target rate, frequently buying large quantities of USD, and has accumulated more than 1.3 trillion USD.^[18] World Bank figures document that the RMB/USD exchange rate should be approximately 1.9 based on purchase power parity.^[19] Along with the peg, the SAFE has maintained a policy that does not allow Chinese citizens and firms to freely convert their RMB holdings into foreign currencies except for the annual amount of USD 50,000 whereby every transaction has to be registered with the SAFE. However, over the fruitful years of growth, many individuals and businesses have amassed large quantities of money they must now invest within China because there are only limited opportunities for international assets diversification. First, the managed RMB convertibility isolates Chinese markets from global market mechanisms that balance the demand for and supply of RMB at its efficient price. Second, domestic liquidity faces a restricted set of investment alternatives in China.^[20] Unless they want to invest in low-yield savings accounts whose returns hardly compensate for inflation losses, retail investors and likewise restricted institutional investors will most likely divert their funds into Chinese equity markets. This mechanism could account for some of the excessive increases Chinese equity markets have seen in the last months. It also helps to explain why the government has recently urged large Chinese firms like the Bank of China or China Citic Bank to sell shares to the public, because an increase in shares supplies could take some steam off domestic equity markets.^[21]

In its role as the national capital market regulator, the CSRC closely monitors national securities and futures markets in China. The institution is in charge of overseeing all capital market transactions, e.g. fulfilment of listing requirements at the stock exchanges, trading activity by market participants or compliance with governance principles regarding the disclosure of information. In this function, the CSRC has initiated a reform of the split share structure in Chinese equity markets. The split share structure refers to the existence of tradable and non-tradable shares owned by the state or legal persons. When the share structure reform was launched in May 2005, about two thirds of shares in Chinese companies were non-negotiable. The following low level of free floating shares, limited transfer options for strategic holdings and lacking incentives arising from non-tradable and thus basically unpriced shares imposed enormous barriers on stock transactions and efficient capital allocation. Hence, the CSRC began to liberate markets by initiating a pilot scheme that allowed 4 pioneer firms to convert their non-negotiable shares into freely tradable shares. Shortly after this experimental phase, another 42 pilot firms joined the project before participation was offered to all firms listed at the SHSE and the SZSE.^[22] Until July 2006, more than 80 percent of domestically listed Chinese firms had completed or were in the process of enforcing the state-shareholding reform. To make all their shares tradable, listed companies undergoing the reform had to offer additional shares or funds to private investors to compensate them for potential losses in the value of their portfolios when the publicly-owned shares are introduced to the market.^[23] According to the reform offers made public, the companies or major shareholders, by means of negotiation among stockholders, should offer about three shares per ten shares to tradable shareholders, so as to make all their shares tradable.^[24]

Statistical data from the CSRC, however, shows that the percentage of negotiable shares has only grown by less than 5 percent points between June 2005 and June 2007. Meanwhile, the total number of shares has more than doubled from 788.428 to 1,689.243 billion shares; but this does not explain why the CSRC data is not consistent with the supposed reduction of non-negotiable shares (Table 4, p. 52). In spite of the aforementioned barriers, another reason for the strong stock price growth in China might nonetheless be attributed to rising capital market efficiency, supported by strengthened investor confidence. Fewer state block holdings enable private investors to actively pursue mergers and acquisitions, and inspire new levels of activity in the market for corporate control.

2.3 Dual-Class Shares

Chinese policymakers have indeed taken first steps towards a deregulation of the national financial sector. Nonetheless, some major regulations are yet left untouched. As aforementioned, trading at the SHSE and the SZSE is strictly regulated by the CSRC. One of the CSRC’s fundamental regulations is the implementation of foreign ownership restriction. First of all, shares issued by firms in China belong to different share classes. There are generally three classes of stock: A-, B- and H-shares. A-shares are denominated in RMB and traded at either the SHSE or the SZSE. B-shares are also traded at the two mainland stock exchanges, but prices are quoted in USD in Shanghai and HKD in Shenzhen. H-shares are listed at stock exchanges outside China, and denominated in various currencies, depending on the stock exchange they were introduced to. The majority of H-shares was issued at the HKSE and is therefore denominated in HKD.^[25] H-shares are issued and traded at the HKSE, but the firms have their headquarters and operations in China. With only a few exceptions, firms with H- and A-share listings have usually first issued H-shares before selling A-shares to the public in China. By offering H-shares, firms provide opportunities for investors from abroad to purchase Chinese equity securities without having to deal with ownership regulations or additional trading costs.^[26] Second, certain share categories are restricted with respect to ownership eligibility, depending on the investor’s country of origin.

illustration not visible in this excerpt

In June 2007, the CSRC reported 1,329 firms with only A-shares, 109 firms with only H-shares and 23 firms with only B-shares (Table 2, p. 11). Among the firms that issued shares in Mainland China investors may own A- and B-shares, but they were restricted in their H-share investments until August 2007 due to the foreign exchange limit set by the SAFE.^[27] Foreign and Hong Kong investors are eligible to purchase B- and H-shares. However, current legislation prohibits any A-share trading by non-Chinese investor in more than one category; the distribution shows 86 firms that had both A- and B-shares and 37 firms which had both A- and H-shares.^[28] While the level of B-share issues has more or less stagnated since 2002, the increasing number of IPO’s in the H-share segment shows that H-shares have become more popular among firms which seek equity financing from international sources. With the growing number of H-share issues, dual listings in A- and H-shares have also increased, whereas dual listings in A- and B-shares remained on their 2002 level. Ever since the CSRC granted free access to B-shares, firms have foregone the possibility of pursuing IPO’s in the B-share segment. In August 2007 the total B-share market capitalization at the SZSE and the SHSE was only a very small fraction of A-share market capitalization and accounted for less than 2 percent in total turnover.^[29] For the same month, the HKSE reported a market capitalization of 4.611 trillion HKD in H-shares, which makes up slightly more than one quarter of total market capitalization of the HKSE Main Board.^[30] Clearly, in terms of mere size the HKSE H-share segment casts a vast shadow on China B-share markets. Because of public interest in the subject of H-share discounts, the subsidiary of Hang Seng Bank in charge of maintaining and updating the Hang Seng Index has introduced the Hang Seng China AH Index Series, which comprises the largest and most liquid Chinese companies with both H- and A-share listings. It aims to calculate the weighted average of H- and A-share prices, as well as the premium on A-shares arising from diverging prices. Apart from the B-share indices at the SHSE and SZSE, no similar indices tracking the discounts of B- versus A-shares exist. Briefly put, the above mentioned facts fortify the notion that interest in H-shares has widely risen, while it appears that the matter of the B-share discount has become less relevant to investors.

The important implication here is that markets for H- and A-shares are divided among two groups of market participants: Chinese and international (including Hong Kong) investors. Theory states that if identical claims on a given firm’s cash flows are securitized in stocks, these stocks should consequently trade at equal prices. In China, however, this conclusion no longer holds because there are different shares classes traded by mutually exclusive groups of investors at different stock exchanges. Equity trading in China and Hong Kong takes place in segmented markets. In fact, market segmentation between H-and A-share should be stronger than for B- and A-shares simply because trading locations are also different whereas B- and A-shares have always been traded in the same location. Lastly, there are also different trading hours at the SHSE and SZSE and the HKSE, but they are in the same time zone. The SHSE and SZSE are open from 09:30 a.m. to 11:30 a.m. and 01:00 p.m to 03:00 p.m. while the HKSE opens at 10:00 a.m. to 12:30 p.m. and from 02:30 p.m. to 04:00 p.m.^[31] It could be assumed that asynchronous trading causes the different prices for both share classes, but given the strong overlap of trading times, the price differences are simply too large to explained by non-fundamental attributes.

2.4 Institutional Investors

Just as markets are not perfectly integrated, they are also not completely separated. Some institutional investors are exempt from the SAFE and CSRC trading restrictions. In November 2002, CSRC and the People’s Bank of China launched the Qualified Foreign Institutional Investor (QFII) program. Under the umbrella of this initiative, foreign institutions like mutual funds, commercial banks or insurance companies were allowed to buy equity and debt securities denominated in RMB for the first time. Besides limits on the extent of ownership in a single firm, the CSRC and the SAFE assign investment quotas to the individual QFII. Admission requirements were reviewed and softened in September 2006 to accommodate the growing interest in QFII licensing and to advance in gradually opening up China’s capital account. Some minimum requirements were redefined, such as the one on total assets under management which was lowered to 5 billion USD, or the initial 1 year lock-in for investments, which was reduced to 3 months.^[32] In early 2007, the overall limit set at 10 billion USD was nearly filled because the altogether 49 QFII had been officially approved quotas totalling 9.995 billion USD. Responding to the increasing demand, the investment limit was relaxed by adding another 10 billion USD margin on top of the existing ceiling.^[33]

In a similar vein, the government set up a Qualified Domestic Institutional Investors (QDII) program in July 2006. The QDII scheme allows mainland institutions and residents to entrust mainland commercial banks to invest a certain amount of money in financial products overseas, and allowing insurance institutions to invest part of their assets in overseas fixed-income products and money-market products.^[34] Until March 2007, 30 institutions – 11 domestic banks, 7 foreign banks, 11 insurance companies and 1 mutual have fund – had attained the QDII status. These QDII may sell RMB to the SAFE in order to receive foreign currency that they can invest in overseas markets. While investments were initially restricted to money market funds and fixed-return product, the SAFE is about to loosen this regulation by allowing QDIIs to also trade in stocks, fixed-income securities, options, derivatives and mutual funds from abroad. The total investment quota for banks was at 14.8 billion USD in June 2007, of which only 800 to 900 million USD have been used up because their investment scope was too narrow. The rules for insurance firms are also currently subject to renegotiation, and first statements by government officials indicate that insurers will be permitted to invest as much as 15 percent of their total assets in mature stock markets like London and New York. Considering total assets of 1.97 trillion RMB in the insurance sector, around 300 billion RMB could be invested overseas.^[35]

Although QFIIs and QDIIs are exempt from foreign ownership restriction in the A-share and H-share segment, the QFII maximum investment volume of 20 billion USD and the current level of QDII overseas investment rules out any larger impact on the aggregate H- versus A-share discount from QFII and QDII trading activity across both markets. The aggregate market capitalization of the 37 stocks with H- and A-shares included in the data sample for this paper was 1.366 trillion USD at the end of September 2007. It should therefore be hard to argue in favor of the QDII and QFDI legislation significantly affecting H-share discounts, as investment quotas are presently insufficient to cause noticeable increases in market integration. The SAFE decision to open the Hong Kong markets for mainland investors reveals their ambition to move Chinese markets out of their isolated position. The QDII and QFII programs may provide useful vehicles when attempting bolder changes. For instance, investors intending to buy securities in the Hong Kong market without the conversion cap presently need to open brokerage accounts with the Bank of China. The pilot has been implemented in the Tianjin branch of the bank, and further branches can join the initiative by means of contract with this branch. Rolling out the scheme to all institutions qualified as QFDI could severely leverage the impact of the program. Similar initiatives in the future could also be channelled through the QFDI and QFII programs affecting a larger public and providing some momentum for capital market liberalization.

3 Literature Review

3.1 Asset Pricing under Foreign Ownership Restrictions

When analyzing the H- versus A-share discount, scientific literature on asset pricing under foreign ownership restrictions provides the key concepts and theoretical fundament. To begin with, it is worthwhile to ask the question why certain countries enforce restrictions on foreign equity ownership in the first place. As a matter of fact, the inflow of foreign capital increases the capital supply in that country, stimulating equity investments and growth. This is particularly beneficial to countries that suffer from a poor initial capital endowment. Despite the positive effects of such cross-border flows of capital, local governments are also concerned about foreign investors attempting to gain control over companies or certain industries by accumulating the necessary fraction of shares to interfere in management decisions and company strategy, extract private rents, or gain full control for other strategic reasons. Thus, foreign ownership restriction may be motivated by the respective authority’s interest in protecting domestic industries and companies from alien interventions and takeovers.^[36]

In fact, several states make use of such protective legislations. Data from twelve countries provide evidence for statistically significant price differences between dual-class shares that are merely distinct in one feature: One class of stock is officially approved to be traded among international investors, while only domestic traders legally qualify to purchase the other class.^[37] Usually, there are two ways of enforcing the restriction on foreign ownership. One option is that regulation bodies set a limit to the fraction of stock foreign investors can acquire. Once this limit is reached, international investors begin to trade those shares among themselves. To accommodate alien trading, the national stock exchange normally arranges for separate listings or trading boards.^[38] The Thai stock exchange in Bangkok provides an apt example. Foreign ownership is limited to 50 percent for most companies, and to 25 percent in crucial sectors like the banking industry. As soon as foreigners undertook massive investments exceeding the freely available supply of shares, the Main Board of the Bangkok stock exchange was supplemented by an Alien Board where foreign trading was settled.^[39] The alternative to this regulation is to issue nominally different classes of stock in the first place – the most famous example being China with its A-, B- and H-share segmentation at the national stock exchanges.

In either way, what is of importance is that imposing different trading regulations leads to the segmentation of equity markets. In perfectly integrated markets, dual-class share prices should not be affected by different trading arrangements or locations since they depend on the same streams of future cash flows and identical discount rates.^[40] In reality however, huge price gaps exist between dual-class shares. The most comprehensive study in terms of countries covered was conducted by Bailey, Chung and Kang (1999), who examined market data from eleven countries with foreign ownership restriction laws.^[41] Shares available to foreigners generally traded at a higher price compared to the shares traded by local investors. In this context, the shares traded by foreign investors are often referred to as unrestricted, whereas the shares traded by domestic investors are referred to as restricted shares. The calculation of unrestricted share premia is usually based on the following simple formula:

illustration not visible in this excerpt

where S denotes the unrestricted share premium, P U (P R ) the unrestricted (restricted) share price, N the number of firms and T the number of observations available. In their study, that encompasses the timeframe from 1988 until 1995, they observe premia in the range from minus 50 percent and less in China up to plus 130 percent in Malaysia. Moreover, premia are not stable, but fluctuate strongly over time. Domowitz, Glen and Madhavan (1997) equally find statistically significant premia for Mexico; Bailey and Jagtiani (1994) for Thailand; Stulz and Wasserfallen (1995) for Switzerland; Hietala (1989) for Finland as well as Karolyi and Li (2003) and many others for China.^[42] Bailey, Chung and Kang (1999) then construct portfolios for each country that consist of either restricted or unrestricted securities only. They drop stocks from the sample if they exhibit one of the following features: low liquidity, recent IPO, or uncontested foreign ownership limits.

The statistical analysis of the data shows that the rate of return for an unrestricted portfolio – a portfolio composed of shares that are traded by international investors – is not significantly different from the rate of return for the restricted portfolio – a portfolio that contains stocks traded only by domestic investors – of the same country. However, the intranational correlation between rates of return was highly significant, with some of them reaching unity.^[43] Most notable about this study is the observation that foreign investors generally accept a higher price on unrestricted shares. Surprisingly, this does not hold for the Chinese market. With respect to Chinese A- and B-shares – and H-shares as will be shown later in detail – the foreign premium theme reverses. In China, unrestricted shares trade at a discount compared to their restricted equivalents. Moreover, Chinese markets provide the only observations in the sample that exhibit these characteristics.

There are competing theories that attempt to explain what precisely causes the different asset prices in segmented markets. Basically, four hypotheses have been tested and explored in research on market segmentation: Differential risk, differential demand, stock liquidity and information asymmetry. The grouping of the substantial amount of literature on market segmentation into these four fundamental research streams is owed to Chen, Lee and Rui (2001).^[44] Clustering the competing hypotheses inevitably entails that some studies will be categorized into one group of hypotheses, although its research scope might extend beyond simply examining this particular concept. However, for the sake of clarity, it seems worthwhile to incur this cost in terms of inaccuracy. Below the four major concepts shall be reviewed before recapitulating their empirical validity in the B-share context and judging their potential contribution to an explanation of the H-share discount.

3.2 Competing Concepts in Market Segmentation Theory

3.2.1 Differential Risk Hypothesis

The differential risk hypothesis explains price differences between dual-listed stocks through the different exposure to systematic market risk perceived by investors. In an early theoretical work, Eun and Janakiramanan (1986) suggest that premia for unrestricted shares grow with the aggregate risk aversion of foreign investors, the fraction of shares available to them, the difference in domestic and foreign risk-free interest rates, and the difference in covariance of the security with the foreign market and the covariance of the domestic market portfolio composed of those securities that most highly correlate with the foreign portfolio.^[45] Tests of their framework in a model economy where the authors introduce two countries, relying on the assumptions of capital asset pricing model (CAPM) and imposing investment restrictions on the equity holdings of one country in the other, document that their derivations are valid.

Building on these findings, Hietala (1989) examined the prices of Finnish stocks that are restricted with respect to foreign ownership – only 20 percent of a firm’s stocks could be acquired by foreigners, and those stocks had premia between 12 and 41 percent. Between 1984 and 1985, Finnish investors were not allowed to invest abroad and unrestricted shares traded at a premium over their restricted counterparts. Because of limited diversification opportunities available to domestic investors, they may not diversify away country-specific risk to the same extent as foreign investors. Therefore domestic investors, on the one hand, require a higher return on domestic securities; international investors, on the other hand, discount expected cash flows with a lower rate of return.^[46] An asset pricing model based on the Sharpe-Lintner-Mosse CAPM, which is built on mutually exclusive investment sets available to the investor groups, predicts that unrestricted stock price premia can vary from zero percent upwards. Because both foreign and domestic investors can invest in the unrestricted stocks, there are separate relevant risk premia, as both groups can buy these stocks. Equilibrium prices are attained when the group with the lower discount rate – the international investors as reasoned above – bid up these share prices. In testing the model empirically, the author finds that required rates of return on Finnish stocks are primarily determined by domestic stock betas. These results are in line with the model predictions that higher domestic market betas induce higher premia on the unrestricted stocks. The conclusion is that the differential risk perception of stock returns can partially explain the existence and magnitude of unrestricted stock premia, and that the cross-sectional variation depends on the correlation of stocks with the domestic market. Furthermore, evidence is presented that there is a positive correlation of premia with measures for firm size and liquidity.^[47]

Bodurtha, Kim and Lee (1995) point out that markets in the United States and abroad are affected by different risk factors. They use premia on closed-end foreign country funds, whose domestic share prices exceed their underlying net asset value in the foreign country, to test their hypothesis. Such closed-end funds are priced in the United States, but the value of portfolio assets is determined in the foreign country. A multifactor model that accommodates market segmentation through global and national risk characteristics and proxies for investor sentiment reveals that the foreign and domestic closed-end fund share prices co-move with United States market returns over time. Changes in the net asset value of the foreign fund portfolios, however, are not affected by United States market movement. Hence, it follows that international equity prices are influenced by local risk components or investor sentiments in the country where they are traded.^[48] Inspired by these results, Bailey and Jagtiani (1994) do not find that the betas of firms are related to the premia on unrestricted shares.^[49] Likewise, Bailey, Chung and Kang (1999) conduct tests on the differential risk concept but fail to detect compelling evidence, as there are only little time-varying risk premia or differences in premia between markets in their sample.^[50]

Additional evidence on differential risk is provided by Froot and Dabora (1999), who study the stock price behaviour of multinational firms with dual listings in three countries. Their study enriches the insights from the paper on closed-end fund pricing, because the sample stocks enable the authors to eliminate some potential noise in the data from closed-end country funds like illiquidity, the potential value of the portfolio strategy to investors (which could explain the difference between share prices and the fund’s net asset value), and the prohibitive costs of arbitrage that would arise from replicating the fund portfolios. The analysis shows that stock prices are affected by the location of trade where stocks are most actively traded, and that locally traded stocks are more infected by market-wide shocks.^[51]

3.2.2 Differential Demand Hypothesis

The differential demand hypothesis by Stulz and Wasserfallen (1995) argues that the demand functions for domestic shares are different for local and international investors. They study the empirical implications of their model with data from Switzerland, where firms have the option of restricting a proportion of their stocks to domestic investors. Because there are no legally binding foreign ownership restrictions enforced by the state, investment barriers are set by management. Across the data sample with 19 firms in it, there is an average unrestricted share premium of 71 percent. Their model specifications are as follows: There are deadweight costs associated with holding risky assets, e.g. political risks, taxes on capital gains or dividends, or transactions costs, which differ across countries and investors. Further, the assumption of perfectly elastic demand for capital commonly used in financial research is relaxed to allow for downward-sloping demand curves. If a foreign investor’s demand for domestic stocks is less price elastic than that of local investors, the fraction of shares available to foreigner will trade at a premium. This occurs because profit-maximizing firms engage in price discrimination among foreign and domestic investors. Cross-sectional regression analysis including market betas and a proxy for the scarcity of shares available to foreigner provides evidence that the relative supply of unrestricted shares is significantly and negatively related to premia at the 5 percent level, with a 57 percent adjusted R2 for the model.^[52]

While it might hold true for Swiss firms that the small political risk and preferred tax treatment give them some monopoly power when selling their shares, these conditions cannot be expected to prevail in other countries where unrestricted share premium are observed.^[53] Bailey and Jagtiani (1994) attempt just that in an examination of unrestricted share premia in the Thai market. They argue that international investors who attempt to mimic the domestic country’s portfolio can bring forth demand for unrestricted shares that is not perfectly elastic. Thai firms, which are usually given a higher weight in target portfolios of foreign investors, should have higher premia, even more so if the fraction of unrestricted shares is more limited. The analysis shows that two proxies – ownership limit and the capitalization-weighted fraction of the unrestricted shares, minus the capitalization-weighted fraction out of all shares for a given firm – individually explain up to 10 percent of variability in cross-sectional premia.^[54] Domowitz, Glen and Madhavan (1997) also test the impact of the relative share supplies on the premia on unrestricted shares in Mexico. They use the number of unrestricted shares outstanding, divided by the total number of shares outstanding as a proxy for the scarcity of shares available to foreign investors. A regression of this and other independent variables on the observed premia in the Mexican market between January 1990 and December 1992 yields a negative coefficient that is highly significant. The higher the relative number of unrestricted shares available is, the lower is the premium on that particular firm.^[55]

3.2.3 Liquidity Hypothesis

According to the liquidity hypothesis, price anomalies between different share classes can be explained by the liquidity of stocks. Stocks that are less liquid inflict higher trading costs on investors. Amihud and Mendelsson (1986) investigate the quoted bid-ask spreads as a measure for the liquidity of stocks. The higher the spread is, the higher are the transaction costs from the investor’s point of view. Thus, investors are going to require a higher rate of return as a compensation for the additional trading costs they incur. The data from 1961 to 1980 used in their study validates the idea that expected returns are a function of price spreads on individual stocks.^[56]

Further evidence on liquidity affecting asset prices is presented by Datar, Naik and Radcliffe (1998), who find that a one percent decrease in turnover rate – the ratio of shares traded to shares outstanding – generally leads to a 4.5 base point increase in average monthly returns. Since the authors control their results for other major determinants of stock returns, like the size of the firm, book-to-market ratio, and firm beta, their model provides support for the adverse effect of illiquidity on asset prices.^[57] Results by Bailey and Jagtiani (1994) also support the notion that there is a liquidity effect in the cross-sectional data on premia in Thai market.^[58] However, Domowitz, Glen and Madhavan (1997) cannot confirm that the liquidity of stocks influences the time-series movement or cross-sectional differences between premia in Mexico. They also apply a vector autoregression model but the results again lack support for liquidity effects.^[59]

3.2.4 Asymmetric Information Hypothesis

The asymmetric information hypothesis refers to accessibility of information by international and local investors. While local investors acquire information about the domestic economy at little cost, this task will – in spite of modern communication infrastructures offered through the internet – easily become more burdensome for foreign investors because of language barriers, unfamiliar accounting standards, or lacking leakage of information that is instead contained in the local country. A survey by Chuhan (1994) concluded that liquidity problems and limited information are two key hurdles for investing in emerging markets.^[60] Motivated to explain the existence of the home bias – the preference for investors to hold domestic assets in their portfolios – Brennan and Cao (1997) develop a model of international equity portfolio investment flows, where foreign and domestic investors have different levels of information. Their analysis of portfolio flows between the United States and other countries underscores the prediction that information asymmetry influences the allocation of funds.^[61]

With respect to foreign ownership restriction, differences in the availability of information should motivate foreign investors to buy stocks of larger firms rather than small-capitalization stocks, which is consistent with the findings of Kang and Stulz (1997) for the stock exchange in Tokyo. International investors avoid buying shares of smaller firms, even at the cost of having a greater volatility in their Japanese portfolio as compared to the Japanese market portfolio.^[62] Similar findings have been provided earlier by Bailey and Jagtiani (1994) for the Thai market.^[63]

3.3 Assessing Fundamental Concepts in the Context of China

3.3.1 Market Segmentation Theory and the H-Share Discount

As aforementioned, only a scarce number of scientific sources address price anomalies between H- and A-shares. Many researchers include H-shares in their depiction of Chinese equity markets, but do not further explore this matter.^[64] Although Sun and Tong (2000) report discounts for the H-share segment which resemble the B- versus A-share pricing discrepancy, their analytical focus remains on the B-share discount.^[65] Fernald and Rogers (2002) are the first to include H-share data when analyzing the pricing puzzles in China. As there is more data on dual B- and A-shares than H- and B-shares, and these subsets are not always reported separately, it is hard to make clear-cut inferences with respect to the H-share discount.^[66]

The first study explicitly treating the H-share discount was published by Wang and Jiang (2003). Based on 16 firms in their sample, with observations in the period from June 1995 to September 2001, the authors measure H-share discounts ranging from 64.4 percent to 89.9 percent. The average discount amounts to 74.7 percent. First of all, they attempt to test if the dynamic movements of H- and A-shares correlate stronger with the SHSE and the SZSE in China or the HKSE in Hong Kong. Since all firms with dual H- and A-share listings are domiciled in China, correlation of both share classes should be highest with the Chinese stock exchanges. If this was indeed the case, international investors would be able to use the HKSE H-share segment as a vehicle to achieve portfolio diversification by including Chinese stocks, but without incurring the costs and administrative barriers of investing directly in China. In any way, the comovement with the different markets should reveal the extent of market segmentation between the share classes. The more markets are integrated, the stronger should be the return correlations between both markets. Hence, they regress daily individual A- and H-share returns on the Hang Seng Index return at the HKSE, as well as the market composite index returns at the SHSE and, where applicable, at the SZSE, and adjust for exchange rate changes. Their results show that for all H-shares HKSE market betas are significantly positive, as well as for 60 percent of H-shares and their corresponding SHSE or SZSE market betas. Above that, they can reject the null hypothesis that market betas are equal between the trading locations in China on the one side and Hong Kong on the other side. A-share market betas are only significantly different from zero with respect to the domestic market betas, which is not surprising, since operations and trading are located in China, and are thus unaffected by Hong Kong market factors. These findings suggest that H-share returns are influenced by the systematic risk of the HKSE and the SHSE or the SZSE whereas A-shares are only subject to market-specific factors at the SHSE or the SZSE. Despite the segmentation of both markets, investors may still benefit from international diversification by adding H-shares to their portfolios, because H-share market betas with the SHSE and the SZSE indices were positive and statistically significant.^[67] More analysis is dedicated to the time-series behavior of the H-share discount. A regression including market index returns for Hong Kong and China, liquidity proxies – the difference between the A- and H-share bid-ask spreads and the volume of H-shares over the total number of shares outstanding – proxies for risk – the ratio of the variance for A- over H-shares – an exchange rate risk measure, and an autoregressive term yields significant results for market returns and liquidity. Chinese market returns have positive coefficients, while Hong Kong returns have negative coefficients. Liquidity also affects the H-share discount negatively where a 1 percent increase in the ratio of H-share volume over total H-shares outstanding reduces the discount by 2 to 3 percent. As for differential risk, changing positive and negative coefficients indicates little support at best. The asymmetric information hypothesis is tested in a causality test that returns no evidence for feedback relations between the two share classes.^[68]

[...]

^[1] See Wood (2007), http://www.ft.com.

^[2] See Wang, Q. (2007), http://www.morganstanley.com.

^[3] See Karolyi and Li, L. (2003), p. 21.

^[4] See Bailey (1994), Ma, X. (1996), Charkravarty, Sarkar and Wu, L. (1998), Chui and Kwok (1998), Poon, Firth and Fung (1998), Su, D. (1999), Fung, Lee, W. and Leung (2000), Sun and Tong (2000), Chen, G., Lee, B. and Rui (2001), Eun, Jakinaramanan and Lee (2001), Fernald and Rogers (2002), Yang (2002) as well as Karolyi and Li, L. (2003).

^[5] See Wang, J. and Jiang (2003), Li, Y., Greco and Chavis (1999) as well as Zhao, Z., Ma, Y. and Liu (2005).

^[6] See Hsu (1997), p. 672.

^[7] See Kosyrev (2006), http://www.wpherald.com.

^[8] See National Bureau of Statistic of china (2007), http://www.stats.gov.cn.

^[9] See Wu, H., Chen, G. and Shiu (2007), p. 404.

^[10] See World Federation of Exchanges (2007), p. 61.

^[11] It must be noted that the data for Hong Kong are excluded from China GDP calculations. Likewise, Chinese DMC is based on data from the SHSE and SZSE, excluding the HKSE. Data on DMC for the selected countries does not necessarily summarize the total value of all national equity because some stock exchanges are not included, e.g. the data for Germany relies on reportings of Deutsche Börse only, and does not include some smaller German stock exchanges. Yet, this inaccuracy should not impair the quality of the point that is to be made.

^[12] See CSRC (2007a), http://www.csrc.gov.cn.

^[13] See Wagner (2007a), pp. 76-78 as well as Wagner (2007b), pp. 92-94.

^[14] See Ulrich (2007), http://www.ft.com.

^[15] See Zhao, Z., Ma, Y. and Liu (2005), pp. 18-19.

^[16] See Fernald and Rogers (2002), p. 419.

^[17] See SAFE (2007), http://www.safe.gov.cn.

^[18] See Wagner (2007b), pp. 92-94.

^[19] See World Bank (2006), http://devdata.worldbank.org.

^[20] See Bailey (1994), pp. 243-244.

^[21] See Wagner (2007a), pp. 76-78.

^[22] See Chen, F. (2007), http://english.gov.cn.

^[23] See Yan (2006), http://english.gov.cn.

^[24] See Pan (2005), http://english.gov.cn.

^[25] Apart from 112 H-share listings in Hong Kong only, there are further two in Singapore and
18 simultaneous listings in Hong Kong and the USA or Hong Kong and London.

^[26] See Wang, S. and Jiang (2003), p. 8.

^[27] Legalization of domestic ownership in B-shares has taken effect in February 2001. Formerly,
local mainland investors were denied access to accounts that would enable them to trade B-
shares, see Karolyi and Li, L. (2003), p. 5.

^[28] Please refer to Table 4 in the appendix for further data.

^[29] See SHSE (2007), http://www.sse.com.cn as well as SZSE (2007), http://www.szse.cn.

^[30] See HKSE (2007a), http://www.hkex.com.hk.

^[31] See Chong and Su (2005), p. 72.

^[32] See Zhu (2006), http://english.gov.cn.

^[33] See Runping (2007), http://english.gov.cn.

^[34] See Pan (2006), http://english.gov.cn.

^[35] See Li, L. (2007), http://english.gov.cn.

^[36] See Ma, X.(1996), p. 220

^[37] These countries are China, Finland, Indonesia, Japan, Malaysia, Mexico, Netherlands, Norway, Philippines, South Korea, Switzerland, Singapore, Taiwan and Thailand. Further countries that have protective legislations in place are Australia, Burma, Canada, India, France, Spain and Sweden; see Eun und Janakiramanan (1986), p. 899. Domestic investors may often trade the unrestricted share class but restricted shares are equivalent substitutes at lower prices.

^[38] See Bailey, Chung and Kang (1999), p. 490.

^[39] See Bailey and Jagtiani (1994), p. 59.

^[40] See Wang, S. and Jiang (2003), p. 2.

^[41] This study provides an ideal primer in market segmentation theory and helps to clarify the basic research terminology.

^[42] See Domowitz, Glen and Madhavan (1997), p. 1071 f., Bailey and Jagtiani (1994), p.66 f., Stulz and Wasserfallen (1995), p. 1042, Hietala (1989), p. 700 as well as Karolyi and Li, L. (2003), p. 11.

^[43] See Bailey, Chung and Kang (1999), p. 499.

^[44] See Chen, G., Lee, B. and Rui (2001), p.134.

^[45] See Eun and Janakiramanan (1986), pp. 904-908.

^[46] See Hietala (1989), pp. 697-698.

^[47] See IBIDEM, pp. 703-708.

^[48] See Bodurtha, Kim and Lee, C. (1995), pp. 880-884

^[49] See Bailey and Jagtiani (1994), pp. 77-79.

^[50] See Bailey, Chung and Kang (1999), pp. 501-502.

^[51] See Froot and Dabora (1999), pp. 190-214.

^[52] See Stulz and Wasserfallen (1995), pp. 1020-1023.

^[53] See IBIDEM, p. 1054.

^[54] See Bailey and Jagtiani (1994), pp. 70-72.

^[55] See Domowitz, Glen and Madhavan (1997), pp. 179-180.

^[56] See Amihud and Mendelson (1986), pp. 245

^[57] See Datar, Naik and Radcliffe (1998), p.216.

^[58] See Bailey and Jagtiani (1994), p. 84.

^[59] See Domowitz, Glen and Madhavan (1997), p. 161.

^[60] See Chuhan (1994), pp. 33-35.

^[61] See Brennan and Cao (1997), pp. 1875-1876.

^[62] See Kang and Stulz (1997), p. 27.

^[63] See Bailey and Jagtiani (1994), p. 84.

^[64] See Poon, Firth and Fung (1998), p. 308; Su (2000), p. 39 as well as Yang (2002), p. 4.

^[65] See Sun and Tong (2000), p. 1888.

^[66] See Fernald and Rogers (2002), p. 413.

^[67] See Wang, S. and Jiang (2003), pp. 12-16.

^[68] See Wang, S. and Jiang (2003), pp. 17-25.