Float, Liquidity, Speculation, and Stock Prices: Evidence from the Share Structure Reform in China

Transcription

1 Float, Liquidity, Speculation, and Stock Prices: Evidence from the Share Structure Reform in China Chuan-Yang Hwang a, Shaojun Zhang b, and Yanjian Zhu c Abstract Prior to April 2005, only one third of the shares issued by exchange-listed companies in China are publicly tradable. The other two thirds, mainly owned by Chinese government agencies or government-linked enterprises, are prohibited from public trading. On April 29, 2005, the Chinese Securities Regulatory Committee announced a reform plan that aims to abolish the split-share structure by converting all non-tradable shares to be publicly tradable. We investigate the consequences of this unique event and shed light on how increase in share float affects liquidity, speculation and stock prices. Firstly, we find that tradable A-shares command a 60% price premium on average over non-tradable A-shares and this price premium contains both liquidity and speculation components. Secondly, the share structure reform increases share turnover and dampens speculative trading. Relative to control firms, share turnover of restructured firms increases substantially after the reform, with the largest increase (107 %) in firms that had low liquidity and low speculative trading before the reform. In contrast, there is no increase in share turnover of firms that had high liquidity and high speculative trading. Thirdly, stock prices drop substantially on the day when the supply of tradable shares increases due to the reform. Moreover, the higher increase in the supply of tradable A-shares, the larger drop in the stock price. This indicates that the short-term demand curve is downward-sloping. Fourthly, despite the fall in stock prices, shareholder wealth increases by 15% on average. We find that the largest price drop and the smallest wealth gain occurs in firms with the highest speculative trading before the reform, which suggests that share structure reform dampens speculative trading in Chinese market. Lastly, split share reform also benefits the B-share market despite that the reform involves only A shares: B-share turnover increases after the reform and the well-known B share price discount narrows substantially. a Nanyang Business School, Nanyang Technological University, S3-01B-46 Nanyang Avenue, Singapore, cyhwang@ntu.edu.sg b Nanyang Business School, Nanyang Technological University, S3-B1A-07 Nanyang Avenue, Singapore, asjzhang@ntu.edu.sg c Nanyang Business School, Nanyang Technological University, S3-B3C-45 Nanyang Avenue, Singapore, zhuy0002@ntu.edu.sg

2 Float, Liquidity, Speculation, and Stock Prices: Evidence from the Share Structure Reform in China Abstract Prior to April 2005, only one third of the shares issued by exchange-listed companies in China are publicly tradable. The other two thirds, mainly owned by Chinese government agencies or government-linked enterprises, are prohibited from public trading. On April 29, 2005, the Chinese Securities Regulatory Committee announced a reform plan that aims to abolish the split-share structure by converting all non-tradable shares to be publicly tradable. We investigate the consequences of this unique event and shed light on how increase in share float affects liquidity, speculation and stock prices. Firstly, we find that tradable A-shares command a 60% price premium on average over non-tradable A-shares and this price premium contains both liquidity and speculation components. Secondly, the share structure reform increases share turnover and dampens speculative trading. Relative to control firms, share turnover of restructured firms increases substantially after the reform, with the largest increase (107 %) in firms that had low liquidity and low speculative trading before the reform. In contrast, there is no increase in share turnover of firms that had high liquidity and high speculative trading. Thirdly, stock prices drop substantially on the day when the supply of tradable shares increases due to the reform. Moreover, the higher increase in the supply of tradable A-shares, the larger drop in the stock price. This indicates that the short-term demand curve is downward-sloping. Fourthly, despite the fall in stock prices, shareholder wealth increases by 15% on average. We find that the largest price drop and the smallest wealth gain occurs in firms with the highest speculative trading before the reform, which suggests that share structure reform dampens speculative trading in Chinese market. Lastly, split share reform also benefits the B-share market despite that the reform involves only A shares: B-share turnover increases after the reform and the well-known B share price discount narrows substantially.

3 1. Introduction An ongoing debate concerns what caused the stock prices in the NASDAQ internet sector to rise dramatically from early 1998 to February 2000 and then fall precipitously in April In the view of Ofek and Richardson (2003), the rapid rise of the internet stock prices during that period occurred because (1) investors had diverse views and there were relatively more individual investors in the internet sector who in general are less sophisticated than institutional investors, (2) short sales constraint prevented the beliefs of pessimistic investors from being incorporated into stocks prices, and (3) shares were in short supply and hard to borrow because of lockup restrictions. They also argue that the collapse of the internet stock price occurred because expiration of lockup agreement greatly increased the number of shares available for directly selling by corporate insiders and for borrowing and short selling by public investors. Cochrane (2003) holds a similar view and concurs with Ofek and Richardson (2003) on the cause of the development and the collapse of the internet bubble. Motivated by empirical evidences on the internet bubble, Hong, Scheinkman and Xiong (2006) develop a theoretical model and formally demonstrate that the combination of short sales constraint, investor heterogeneity and limited public float leads to price bubble and an increase in public float could dampen the bubble. However, Schulz (2006) shows new empirical evidence that the collapse of the internet bubble cannot be explained by the increase in public float due to the expiration of lockup agreements. At the center of the debate lies the question whether an increase in the supply of tradable shares causes speculative prices to fall. In this paper, we contribute to this debate with evidence from a unique event in China. Chinese stock markets provide an ideal setting to study this issue because they are endowed with all the ingredients that Ofek and Richardson (2003) and Hong, Scheinkman 1

4 and Xiong (2006) view as responsible for the forming of a speculative bubble in the NASDAQ internet sector: (1) the market is relatively young and is dominated by inexperienced individual investors who are more likely to hold diverse views on the prospect of stocks; (2) short sale is not allowed; and (3) there is a limited supply of shares available for trade. Prior to April 2005, only one third of the shares issued by exchange-listed companies in China are publicly tradable. The other two thirds, mainly owned by Chinese government agencies or government-linked enterprises, are not publicly tradable. On April 29, 2005, Chinese Securities Regulatory Committee (CSRC) announced a reform plan that aims to abolish the split-share structure by converting all non-tradable shares to be publicly tradable. This event provides us a unique opportunity to study how an abrupt increase in share supply (i.e. public float) affects speculation, liquidity, share prices, and shareholder welfare. More importantly, it provides us an out-of-sample and outside-ofthe-u.s.-market test of the arguments about speculative stock prices in Ofek and Richardson (2003) and Hong, Scheinkman and Xiong (2006). It is interesting to see whether their arguments apply to other settings, different from the internet bubble. According to their arguments, we hypothesize that stock prices in Chinese market are inflated by speculation and the sudden increase in the public float due to the reform dampens speculation. Under normal circumstances, it is almost impossible to directly test whether stock prices are inflated beyond fundamental values by speculation, or whether stock prices are in the midst of forming a speculative bubble, because the fundamental value is simply unobservable. We are now certain there was an internet bubble only because it has collapsed spectacularly. In this paper, we are able to investigate whether stock prices in Chinese markets include a speculative component without seeing the collapse of stock 2

5 prices. Our investigation is based on the belief that the value of non-tradable shares is void of any speculative component because non-tradable shares cannot be traded and hence cannot be speculated upon. Therefore, the price difference between tradable and non-tradable shares can tell the extent to which speculation influences the price of tradable shares. Although this difference cannot be computed directly because the value of nontradable shares is unobservable, the Chinese split-share reform provides us a unique opportunity to measure the price difference between tradable and non-tradable shares. If speculations indeed inflate stock prices in Chinese markets, the price difference we calculated would be positively related to the empirical proxy for the degree of speculation in the cross-section. Consistent with this prediction, we find that the price difference is higher for stocks with higher dollar trading volumes, which proxies the degree of speculation activity, even after controlling for liquidity. We also find that the price difference is about 13% smaller for the stocks with both A and B shares than for the stocks with only A shares. This suggests that speculation drives A share prices even higher for stocks that do not have B shares whose prices serve as an anchor than for stocks that do have B shares. 1 Furthermore, our results show that the increase in share supply due to the split share reform causes speculative prices to fall. Share price drops by 14% on average after the reform, and the drop is larger for firms under more intense speculation before the reform. In addition, although share turnover increases for all firms due to the increase of public float, the increase of share turnover is smaller for firms under more intense speculation before the reform. This again suggests that larger public float dampens speculative trading. 1 A shares and B shares are two classes of shares with identical rights, but all of the B shares are tradable while only a portion of A shares are tradable. Furthermore, A shares can only be held by Chinese investors while B shares can be held by both Chinese and Foreign investors. Until 2001, B shares can only be held by foreign investors. Despite the identical rights, B shares have been traded at a large discount relative to A shares, known as the AB share discount puzzle. 3

6 Other than contributing to the literature on speculative bubbles, our paper also contributes to the following literatures. First, we provide new evidence on whether demand curve is perfectly elastic (i.e., horizontal demand curve). Most of the asset pricing models such as CAPM or APT predicate on the assumption that assets have a perfect or close substitute, thus the demand curve for these assets are nearly horizontal (i.e. perfectly elastic). But there have been evidences showing the demand curve are downward sloping, mainly from examining the return of stocks when they are added to S&P 500 index (for example, Harris and Gruel (1986), Shleifer (1986), Lynch and Mendenhall (1997), Wurgler and Zhuravskaya (2002)) or when the IPO lock-up provisions expire (for example, Field and Hanka (2001), Ofek and Richadson (2000)). However, these studies cannot completely rule out the alternative explanation that their results are due to new information related to these events. Being included into the S&P index may reflect good news despite Standard and Poor s claim to the contrary, while the negative return around the expiration of IPO lock-up provisions may be due to a larger than expected insider selling around expiration dates even though such dates are known well in advance. In contrast, our study is based on a regulatory change event and the public float increases in this event are mandated by regulators on all firms, hence it has the advantage of being free of information contents. Second, our paper shows that liquidity is priced in the market. Both Silber (1991) and Longstaff (1995) find that stocks under trading restrictions are sold at discount relative to unrestricted shares. Amihud and Mendelson (1991), Naik & Radcliffe (1998), Brennen, Chordia and Subrahmanyam (19XX ) find stocks with higher liquidity earn lower expected return. Like Silber (1991) and Longstaff (1995), we find that tradable shares command price premiums over their non-tradable counterparts. We provide direct evidence that these premiums are positively related to the liquidity of tradable shares, 4

7 indicating liquidity is indeed priced. In addition, we obtain other interesting results on this issue from our event-study investigations: (1) both liquidity and share value increase after the reform and (2) the increase in firm value is related positively to the increase in liquidity. While Amihud and Mendelson (1991), Naik & Radcliffe (1998), and Brennen, Chordia and Subrahmanyam (19XX) study the cross-sectional relations between return and liquidity level, our study examines the relationship between the return and the change of liquidity. Our study of the change in liquidity provides a more powerful test of the relation between liquidity and stock price because change in the liquidity of the same firm controls for other unknown cross-sectional differences in risk that are potentially correlated with the level of liquidity. Finally, our paper contributes to a growing literature of studies on the Chinese financial market. China has one of the fastest growing economies in the world. According to Allen et al. (2005), China is the largest economy on the purchasing power basis second only to the U.S., and if current trends continue, will become the largest economy in the world in ten years. The Shanghai Stock Exchange and the Shenzhen Stock Exchange of China have been growing rapidly since their inception in early 1990s. At the end of 2002, the combined market capitalization of these two exchanges ranked eleventh in the world. We show that the reform aiming at abolishing the split-share structure of Chinese listed firms, which has long been viewed as the major roadblock for further development of Chinese markets, has produced a desirable effect. Evidence in this paper shows that the split share reform has dampened speculative trading and has improved liquidity of shares not only for tradable A shares, but also for B shares which are not involved in the reform. The reform substantially reduced the B shares discount from 44% before the reform to 29% post reform. We expect that there will be a further reduction of the B share discount when all of the previously non-tradable shares are 5

8 available for trade. Currently, only about 15% of the previously non-tradable shares are added to the share supply, while the rest is under a lockup period for one to two years. Our evidence on B shares is consistent with Mei, Scheikman and Xiong (2005) who argue that the B share discount is partly due to the speculative bubble in the A share market. The rest of the paper is organized as follows: Section 2 details the share-structure reform in China; Section 3 reviews related literature and develops our research hypotheses; Section 4 presents our empirical results; Section 5 summarizes and concludes the paper. 2. The share structure reform in China Prior to April 2005, only one third of the shares issued by exchange-listed companies in China are publicly tradable. The other two thirds, primarily owned by Chinese government agencies or government-linked enterprises, are prohibited from public trading. This split-share structure is the result of partial privatization of stateowned enterprises (SOE) by Chinese government. Chinese government started the SOE reform in 1978 by allowing employees and domestic institutions to own shares of SOEs. Because of the desire to conform to the communist public ownership principle, Chinese government kept the majority of shares under direct or indirect state control and disallow these shares from public trading. These shares are known as state shares or legal-person shares. 2 China established the Shanghai Stock Exchange (SHSE) and the Shenzhen Stock Exchange (SZSE) in December 1990 and April 1991, respectively, to facilitate public trading of tradable shares. Only 14 firms were listed on both exchanges at the end of As of April 2005, 1353 firms list their shares on these two exchanges, the total market value tradable shares equal to RMB 987 billion. 2 Walter and Howie (2003) and Sun and Tong (2003) give more detailed account of China s share privatization. 6

9 As Chinese market grew, this split share structure has proven to have many negative effects on the markets' further development. The split share structure has substantially distorted the pricing mechanism in the stock market and added uncertainty to investors' expectations. The majority shareholders who make corporate policies and have the control of the company s profits and assets are indifferent to the fluctuations in stock price. As a result, prices and investor behavior do not reflect fundamental values of listed firm (Allen et al. 2005). The stock market is more like a casino than an intermediary channel of capital. Moreover, since the state owns two-thirds of a listed company, there s no visible owner that holds the controlling interests of that company. The state is simply a notional controlling owner and unable to supervise corporate actions effectively. In addition, there are few market-driven takeovers and mergers, and there is no effective corporate governance, both internal and external. Related party transactions that siphon assets from the company are common. The rights of the minority shareholders are frequently violated. In short, the split share structure hinders China from adopting international norms with respect to ownership structure, corporate governance and professional management, among other significant things. The reform of the split share structure is vital to China s capital market. 3 On April 29, 2005, the government announced a reform plan that aims to abolish the split-share structure by converting all non-tradable shares to be publicly tradable. An important feature of this reform plan is that non-tradable-share owners are required to pay a consideration to tradable-share owners before their shares become tradable. The objective of this requirement is to stabilize stock price and preserve shareholder wealth. Non-tradable shares gain additional value when they become tradable. If non-tradableshare owners do not share this gain with tradable-share owners, tradable-share owners 3 Refer to the speech by Shang Fulin, Chairman of the Chinese Securities Regulatory Committee (CSRC), at the press conference organized by the State Council of China on June 27, 2005, 7

10 have incentives to sell their holdings before restructuring takes effect, which will drive down stock prices. Chinese regulators learned this lesson from an earlier attempt to float non-tradable shares in In June 2001, Chinese regulators allowed non-tradable-share owners to gradually sell their shares in the public market without paying a consideration to tradable-share owners. By October 2001, the Shanghai Stock Exchange index dropped by 24% and the Shenzhen Stock Exchange index dropped by 29%. The substantial loss of investors wealth forced regulators to delay and eventually stop the selling of non-tradable shares in June In this new attempt, Chinese regulators require listed-firms to take a set of measures in implementing the reform. Figure 1 illustrates the time frame of a typical restructuring process. The process is characterized by two trading suspension periods. Before the first suspension period, the company calls for the major non-tradable-share owners to come up a restructuring plan that proposes a consideration package. Once nontradable-share owners reach an agreement on this plan, the company announces the official start of its restructuring process on day D0 and requests the stock exchange to suspend trading of its shares from the next trading day. During the first suspension period, the company reveals details of the proposed restructuring plan and consults tradable-share owners for their opinions by press releases, road shows, investor symposiums, etc. Nontradable-share owners revise the restructuring plan to reflect the inputs from tradable-share owners. The company will announce the finalized plan on day D1, which marks the end of the first trading suspension. Trading resumes on the next day after day D1 and continues until D2. This period gives investors an opportunity to buy or sell the shares before an extraordinary shareholder meeting is called for the approval of the restructuring plan. Only shareholders who hold shares at the closing of day D2 are registered for the extraordinary shareholder meeting and face the consequences of the restructuring. The 8

11 second trading suspension begins on the day after D2. An extraordinary shareholders' meeting is called for to vote on the plan. Only when over two thirds of all tradable shareholders who participate in the meeting agree with the restructuring plan, can the plan be approved. Once approved, the plan takes effect and trading resumes on day D3. To limit the sudden impact on stock price, although shares held by non-tradable-share owners are legally tradable after reform, they are restricted from trading during the first year after the reform is completed. The supply of tradable shares increased immediately after the completion of the reform because of the bonus shares that non-tradable-share owners transfer to tradable-share owners. Regular Trading Stage 1: Stage 2: Regular Trading Plan Publication and Negotiation Rights Confirmation and Online Voting Regular Trading D0 Trading Suspended D1 Plan Confirmed and Trading Resumed D2 Trading Suspended D3 Plan Effected and Trading Resumed Figure 1 Table 1 illustrates the restructuring process by chronicling the events that happened to TongRenTang, a well-known traditional Chinese medicine and cosmetic company listed in the Shanghai Stock Exchange. On October 15, 2005 (i.e. D0), the company announced the start of its restructuring process. It has 434 million outstanding shares, 64.2% of them are non-tradable and the rest 35.8% tradable, the closing price on that day is RMB On the next trading day, trading is suspended and the company revealed the plan which states that non-tradable-share owners pay 2.2 shares for every 10 shares held by tradable share owners. Tradable share owners did not like this ratio and voiced their opinions 4 It is equivalent to US$2.52 at the exchange rate of RMB8.09 for US$1. 9

12 strongly. As a result, non-tradable-share owners revise the ratio to be 2.5 shares for every 10 shares. The revised plan is confirmed on October 26, 2005, which marks the end of the first trading suspension. Trading resumed on the next day (i.e. D1) with a closing price of RMB19.9. Trading continues until the shareholder registration day November 10, 2005 (i.e. D2) and the closing price is RMB Trading is suspended again on the next day and registered shareholders vote for the approval of the plan. The plan was approved with an approval rate of 98.8% among all tradable-share holders who participated the voting. The plan became in effect on November 25, According to the final plan, nontradable shares account for 55.24% of shares and tradable shares increased by almost 9% to be 44.76%. Trading resumed on November 30 and closed the day at the price of RMB15.40, which is substantially lower than RMB20.38 on the last day before the reform. [ Table 1 is about here. ] To avoid destabilizing the stock market, Chinese government arranged listed-firm to implement this reform in batches. The first batch of four companies started the reform process on May 9, 2005, followed by the second batch of 42 firms starting on June 20. All firms in these two batches except one completed share restructuring within a short time period, and investors supported the reform. The success of the first two batches strengthened regulators confidence and prompted them to extend this reform to all listed companies. Table 2 shows the progress of firms that have started share restructuring process in the first 21 batches. [ Table 2 is about here. ] 3. Research hypotheses and related literature 3.1. Speculation and the price of tradable shares 10

13 Chinese markets are known to be highly speculative. Share turnover in Chinese market is extremely high; the aggregate turnover in 2002 is 224% of the total market capitalization (Allen et al. 2005). By comparison, in the same year, the aggregate turnover as a percentage of the total market capitalization is only 160% on Nasdaq and 95% on NYSE. But it is unclear whether such a high level of trading indeed inflates prices beyond the fundamental value. Mei, Scheikman and Xiong (2005) study 75 Chinese stocks that have both A- and B- shares and find that, relative to B-shares, the price of A-shares is inflated by speculative trading. However, the price difference between A-shares and B- shares may be due to different level of information asymmetry and liquidity in these two segmented markets. Chan et al. (2006) find that difference in the level of information asymmetry between these two markets explains a substantial portion of the B-share price discount. In the following, we derive a formula that links the price difference between tradable A-shares and non-tradable A-shares to the consideration that non-tradable-share owners pay to tradable-share owners in the share restructuring reform. This provides a means to study whether speculative trading in tradable shares leads to a large price difference between tradable and non-tradable shares. Let and P represent the prices of non-tradable and tradable shares before share P10 20 structure reform; and N represent the number of non-tradable shares and number of N1 2 tradable A-shares before share structure reform; b is the number of bonus shares as the consideration that non-tradable-share owners pay to tradable-share owners for every tradable share; P is the price of all shares after share restructuring completes. Let Z be the total gain to the company value due to restructuring, Z 1 be the gain to non-tradable shareholders, and Z 2 be the gain to tradable shareholders. After restructuring, the number of shares held by original non-tradable shareholders will 11

14 decrease to N1 b N 2 and the number of shares held by original non-tradable shareholders will increase to N1 + b N 2. Taking this into account, we can calculate the three gains as follows Z = N + N ) P ( N P + N ), (1) ( P20 Z 1 N1 b N 2 ) = ( P N P, (2) 1 10 Z 2 N 2 + b N 2 ) = ( P N P. (3) 2 20 Assuming that the total gain Z is shared by non-tradable and tradable shareholders in proportion to the number of shares they own after restructuring, we have N1 b N 2 Z1 = Z (4) N + N 1 2 Z 2 = N + b N N + N Z (5) Substituting Equations (1) and (2) into Equation (4) and rearranging, we obtain a ( 1 a) P P (6) 2 10 = b a (1 a) P10 a (1 a) P20 + b (1 a) 20 where a N N + N 1 = is the proportion of non-tradable shares. 1 2 Let L represent the percentage price difference between tradable and non-tradable shares, that is, L P P = (7) P 10 By combining Equations (7) and (8) and canceling the unobservable value of nontradable shares P, we can write L as follows 10 b L = a b + a b (8) 12

15 The above formula shows that, for every firm, the Price Premium of tradable shares relative to non-tradable shares is related to the bonus share ratio b published in the share structure reform plan and the proportion of non-tradable shareholders a. According to this formula, as long as a a b <, the price premium is positive. The upper bound 1 a 1 a is larger than 1 if and only if the percentage of non-tradable shares a is greater than 50%. Since the percentage a is more than 50% for most Chinese listed firms and at the same time the bonus shares ratio b must be less than 1, the formula must yield a positive price premium for Chinese firms. According to extant literature, tradable shares are more valuable than non-tradable shares because of liquidity. Longstaff (1995) conducts a theoretical analysis within an option pricing framework and finds that trading restriction reduces share values. His theory shows that the loss in value is positively related to volatility of the share price and duration of the trading restriction. Silber (1991) studied restricted stocks that are issued by publicly traded firms, not registered with the SEC, and sold through private placements to investors under SEC Rule 144. Restricted stocks cannot be resold in the open market in the first year of placement. When a restricted stock is issued, the issue price is set much lower than the observable prevailing market price. Silber found that the median discount for restricted stock is 33.75%. In China, although non-tradable shares cannot be publicly traded, they are allowed, under certain circumstances, to be transferred from one owner to another through private placements and auction. Chen and Xiong (2001) collect the auction and private placement prices of non-tradable shares of 258 Chinese firms and compare them with the market prices of tradable shares. They report that the price discount on non-tradable shares is 78% for auctions and 86% for private placement. The value premium of tradable shares over non-tradable shares is likely dependent on a few factors. Silber (1991) finds that the price discount on non-tradable shares varies 13

16 inversely with earnings, total revenues, and market capitalization of the issuing firm. Chen and Xiong (2001) show that the discount is related to tradable shares volatility, firm size, leverage ratio, return on equity, book-to-price ratio, and earnings-to-price ratio. Amihud and Mendelson (1986) suggest that the price of a tradable share would reflect the present value of the expected transaction costs in future. Using bid-ask spread as the measure of transaction costs and show that small spreads can translate into big illiquidity discounts on the price of tradable shares. Amihud and Mendelson (1991) compare the yields on treasury bonds with less than six months left to maturity with treasury bills that have the same maturity. Consistent with their illiquidity discount theory, the yield on the less liquid treasury bond was 43 basis points higher on an annualized basis than the yield on the more liquid treasury bill. Datar, Naik & Radcliffe (1998) use the turnover ratio as a proxy for liquidity. After controlling for size and the market-to-book-equity ratio, they find that illiquid stocks of the lowest 10% turnover ratio have annual returns about 3.25% higher than liquid stocks of the highest 10% turnover ratio. Recent studies suggest that speculative trading affects the price of tradable shares. Owners of tradable shares have the freedom to resell shares at favorable prices. Harrison and Kreps (1978) and Scheinkman and Xiong (2003) suggest that investors pay prices that exceed their own assessment of fundamental value as they anticipate finding a buyer willing to pay even more in the future. This resale option effect can create a speculative component of stock price. Miller (1977) and Chen et al. (2001) show that when investors hold heterogeneous beliefs and there exists short-sale constraints, stock price only reflects the beliefs of the optimistic group as the pessimistic group lack of means to participate. This optimism effect causes price to be biased upward. Hong, Scheinkman and Xiong (2006) demonstrate that in the presence of heterogeneous belief and short-sale constraints, limited float magnifies both the optimism effect and the resale option effect because price 14

17 can easily be moved by the optimistic group when float is small. Since non-tradable shares have no resale value, speculative trading will affect the price difference between tradable and non-tradable shares to the same extent as it affects the price of tradable shares. Chinese market is characterized by limited float, prohibition of short selling, and a great number of inexperienced investors. Therefore, we hypothesize the following. Hypothesis 1 Cross-sectionally, the price premium of tradable shares in Chinese markets is related to proxies for the level of speculative trading after controlling for difference in liquidity, marketability restriction, and transaction costs Impact of the reform on share turnover One of the main objectives of the share-structure reform is to diminish speculative trading in Chinese market. According to Hong et al. (2006), the value of the resale option is negatively related to the public float. Increase in public float reduces the option value and lowers investors incentives to buy shares, which should lead to less speculative trading. Therefore, we hypothesize the following. Hypothesis 2 Turnover increases after the reform. Stocks that were lack of liquidity before the reform have a larger increase than those that had sufficient liquidity. Stocks that were subject to speculative trading have a smaller increase than those without speculative trading Impact of the reform on stock price and shareholder wealth Literature suggests that the short-term demand curve of stock is downward sloping. Harris and Gurel (1986), Shleifer (1986), Lynch and Mendenhall (1997), Wurgler and 15

18 Zhuravskava (2002), and others document significant abnormal returns of around 3% when stocks are added to the S&P 500 index. In contrast, stocks have significant negative abnormal returns around removal from the S&P 500 index. Similar results are reported for stocks that are added to or deleted from foreign indexes. Since the share restructuring reform will cause the supply of shares to increase abruptly, a downward sloping demand curve dictates a drop in price. However, it does not necessarily mean that the share restructuring reform will erode shareholder s wealth. The value of non-tradable shares increases because of the liquidity they acquire. Thus owners of non-tradable shares gain from this reform. The reform requires that non-tradable-share owners transfer part of this gain with tradableshare owners. So long as tradable-share owners receive adequate consideration, their wealth may increase despite the drop in the market price of tradable shares. We expect the negotiation process required by the regulator will lead to a win-win outcome for both groups. Therefore, we expect the following. Hypothesis 3 The share restructuring reform will cause stock price to drop, but shareholder wealth will increase. 4. Empirical Results 4.1. Data Sources Table 2 shows that 509 firms have started the share structure reform before March Thirty-one firms are still in the process of the share structure reform as of March 25, Out of the 478 firms that completed restructuring, 89 firms adopted a consideration package that is more complex than paying bonus shares. We also exclude 16

19 IPO firms in the year before March and firms whose price data are not available from Bloomberg terminal. Our final sample includes 309 firms. We collect the information about share restructuring from the information disclosure website officially designated by the CSRC. We also collect profitability measures ROE and ROA for the fiscal year ending on December 31, We collect stock prices and trading volume data from the Bloomberg system Summary statistics about the price premium of tradable shares Table 3 reports the summary statistics of the price premium of tradable shares and related variables. On average, firms in our sample have 136 million A-shares outstanding before share restructuring. The proportion of tradable A-shares to total A-shares is on average 35%. Non-tradable-share owners in these firms paid on average shares to tradable-share owners for each share in their hand. The price premium of tradable shares is based on Equation (8) and the average price premium is 60.8%. This means that tradable shares are about 60.8% more valuable than non-tradable shares. In addition, Table 3 reports the characteristics of firms in our sample. The market price of tradable shares ranges from 1.72 RMB to 27.6 RMB at the closing of April , the day before CSRC announced the plan to reform split-share structure. The average price is 5.5 RMB. Market capitalization is calculated as the closing price on April times the number of tradable A-shares before reform. It ranges from 87 million RMB, to 8.1 billion RMB, with a mean of 731 million RMB. The average daily dollar volume for the one-year period before April 29, 2005 varies considerably across our sample firms, with the minimum of million RMB, the maximum of million RMB and the mean of 9 million RMB. The majority of our firms are profitable; the first quartiles of the Return on Equity (ROE) and Return on Assets (ROA) in our sample are 17

20 4.47% and 1.76%, respectively. The percentage of zero return trading days in the oneyear period before April has the minimum of 0.42%, the maximum of 11.5%, and the mean of 3.69%. The return volatility is the standard deviation of daily returns in the one-year period before April , and the mean volatility is 2.4%. [ Table 3 is about here. ] 4.3. Liquidity premium and speculation component in the price of tradable shares In this section, we report evidence on liquidity premium and speculation component in the price of tradable shares. More specifically, we study the relation between the price premium of tradable shares and the following factors: the reciprocal of the closing price (in RMB) on April , the logarithm of the average daily trading volume (in million RMB) for the one-year period before April , the logarithm of the market capitalization (in million RMB) calculated as the closing price on April 28, 2005 multiplied by the number of tradable A-shares before share restructuring, return on equity (ROE) and return on assets (ROA) for the 2004 fiscal year, the percentage of zero return days in the one-year period before April , and the daily return volatility measured by the standard deviation of daily returns in the one-year period before April 29, [ Table 4 is about here. ] Panel A of Table 4 reports correlation coefficients between these variables. It is not surprising to see that trading volume and market capitalization are highly correlated with a correlation coefficient of 0.76, and that ROE and ROA are also highly correlated with a correlation coefficient of The other correlation coefficients seem reasonable and unlikely to cause the multicollinearity problem in regression. 18

21 Panel B of Table 4 reports the estimated coefficients from four regressions of the price premium of tradable shares on these factors. Because ROE is highly correlated with ROA and trading volume is highly correlated with market capitalization, we include them in the separate regressions. In all regressions, the profitability measures, ROE and ROA, are significantly negative. This is consistent with the evidence in Silber (1991) and Chen and Xiong (2001). Non-tradable shares of profitable firms tend to have a better chance of finding a potential buyer through private placement or auction and thus are more valuable, which leads to smaller price premium of tradable shares. The coefficient of return volatility is insignificant in all regressions. Longstaff (1995) predicts that return volatility has an influence on the discount. The option pricing framework of Longstaff (1995) assumes that stock price is exogenous, i.e., not affected by the trading restriction imposed on a small number of restricted shares relative to total shares. However, in our case, non-tradable shares account for such a large proportion that liquidation of these shares will have an impact on the supply of tradable shares and the market price. This suggests that the theoretical framework of Longstaff (1995) does not apply in our study. Therefore, the insignificance of daily return volatility in our regressions does not contradict with Longstaff (1995) s prediction. The coefficient of the percentage of daily zero return days is significantly negative at 5% level in all four regressions. Lesmond et al. (1999) show that the percentage of zero-return days is a reliable proxy for the level of transaction costs and is highly correlated with other proxies such as quoted bid-ask spread and Roll s measure of the effective spread. Stocks that have high percentage of zero-return days tend to have high transaction costs. According to Amihud and Mendelson (1986), stocks of high transaction costs tend to have a lower price on tradable shares than those of low transaction costs. On the other hand, transaction costs have no impact on the value of non-tradable shares. 19

22 Therefore, the price premium of tradable shares over non-tradable shares is negatively related to the percentage of zero-return days. The coefficients of both trading volume and market capitalization are significantly positive at 5% level in all regressions. Since trading volume and market cap are highly correlated, it is not surprising that they affect the price premium in the same direction. Trading volume is often viewed as a proxy for liquidity. Stocks of low trading volume tend to have low liquidity, which reduces the price of tradable shares according to Amihud and Mendelson (1986). 5 This provides an explanation of the positive coefficient of trading volume in our price premium regressions. However, trading volume is also related to the level of speculative trading. Hong et al. (2005) show that speculative trading generates trading volume. Firms of high trading volume are more likely to have speculative trading. If speculative trading pushes up the price of tradable shares, trading volume will have a positive effect on the price premium. Both liquidity and speculative trading are plausible explanations for the positive coefficient of trading volume. The coefficient of the reciprocal of share price is significantly positive at 1% level in all four regressions. A few studies, including Stoll and Whaley (1983) and Bhardwaj and Brooks (1992), show that low-price stocks have higher transaction costs measured by both bid-ask spread and commission fees than high-price ones. Since high transaction costs lower the price of tradable shares, low-price stocks should have lower price premium than high-price stocks. This is in contradiction to the positive coefficient on the reciprocal of share price. On the other hand, low-price stocks are very likely traded by individual investors. In Hong et al. (2006) s model, the resale option value increases in the overconfidence parameter, a measure of the degree of behavioral bias. Many individual investors in China have limited experience of investing in stock market. To the extent that 5 Empirically, Brennan et al. (1998) and Chordia et al. (2001) document a negative relation between stock returns and trading volume. 20

23 they have a larger overconfidence parameter, low-price stocks that they actively trade are likely to have a larger speculative component and thus a higher price premium of tradable shares. The last variable in the regression provides further evidence that the price of tradable shares is inflated by speculative trading. This dummy variable is equal to 1 when the firm has both A- and B-shares. Twenty-seven firms in our sample have both A- and B-shares. Since the B-share price serves as an anchor that helps to prevent price from deviating too far away from the fundamental value, firms with B-shares should have a smaller price premium. In fact, Table 4 shows that after controlling for the effect of transaction costs, profitability and liquidity, the dummy variable is significantly negative in all four regressions Impact on share turnover and trading volume We study the impact of share restructuring on share turnover in this section. There are 309 firms that completed share restructuring before March 25, 2006, paid only bonus shares as a consideration, and have the required data. For each restructured firm, we find a control firm that has not started share restructuring, is in the same industry as the restructured firm and has the closest market capitalization. Since only 296 control firms have the required price and trade data in the Bloomberg system, our final sample includes the 296 pairs of restructured and control firms. For both restructured and control firms, we measure the change in daily turnover as (Average daily turnover after reform) / (Average daily turnover before reform) -1. Daily turnover is equal to the daily number of A-shares traded divided by the total number of tradable A-shares. The average daily turnover before reform is based on the 30 trading days immediately before April 29, The average daily turnover after reform is based 21

24 on the first 30 trading days after the restructured firm completed the reform. The excess change in daily turnover of a restructured firm is equal to the change in daily turnover of the restructured firm minus that of the control firm. Compared with the pre-reform period, the share turnover post reform increased by 1.78 times on average for all restructured firms, whereas share turnover increased by only 1.40 times on average for all control firms. The median excess change of daily turnover is statistically significant at 1% level. We classify the restructured firms by the level of liquidity and the level of speculative trading before reform. Firms of high liquidity before reform have a percentage of zero return days below the median, whereas firms of low liquidity have a percentage of zero return days above the median. Firms of high speculative trading before reform have the average daily dollar volume above the median, whereas firms of low speculative trading have the average daily dollar volume above the median. This classification splits the whole sample into four subsamples. Table 5 reports the mean and median excess change in daily turnover of restructured firms in these four subsamples, and the twosample t and Wilcoxon statistics to test whether the difference in mean and median between subsamples are statistically significant. [ Table 5 is about here. ] Comparing firms that had high liquidity before restructuring with those that had low liquidity, it is evident that firms of high liquidity experienced smaller increase in share turnover. The mean (median) excess change of share turnover is (0.239) for high liquidity firms and (0.768) for low liquidity firms. The difference between high and low liquidity firms is statistically significant at 10% level in both mean and median. Next, we compare firms that had high speculative trading before restructuring with those that had low speculative trading. Despite the increase in the supply of tradable shares, restructured firms of high speculative trading on average experienced smaller 22

25 increase in share turnover than their control firms did. The mean excess change in share turnover of restructured firms of high speculative trading is only In contrast, restructured firms of low speculative trading on average experienced larger increase in share turnover than their control firms did; the mean excess change in share turnover is The difference between firms of high and low speculative trading is statistically significant at 1% level. Furthermore, we compare the four subsamples sorted by the level of liquidity and the level of speculative trading. We find that the restructured firms that had high liquidity and high speculative trading before reform, on average, had no increase in share turnover relative to their control firms. Other restructured firms that had low liquidity or were less speculated all experienced significant increase in turnover relative to their control firms. In summary, the above evidence shows that relative to control firms, restructured firms that had low liquidity or low speculative trading experienced significantly higher increase in turnover, whereas restructured firms that had high liquidity and high speculative trading did not. The evidence suggests that the share restructuring reform reduce the share turnover due to speculative trading. Besides share turnover, daily dollar volume is often used to proxy for the level of speculative trading. We examine the impact of the reform on trading volume in the following. For both restructured and control firms, we measure the change in daily dollar volume as (Average daily dollar volume after reform) / (Average daily dollar volume before reform) -1. The average daily dollar volume before reform is based on the 30 trading days immediately before April 29, The average daily dollar volume after reform is based on the first 30 trading days after the restructured firm completed the reform. The excess change in daily dollar volume of a restructured firm is equal to the change in daily dollar volume of the restructured firm minus that of the control firm. 23

26 Table 6 reports the mean and median of excess change in daily dollar volume of restructured firms in every subsample, and the two-sample t and Wilcoxon statistics to test whether the difference in mean and median between subsamples are statistically significant. [ Table 6 is about here. ] Table 6 shows a similar pattern as Table 5. Comparing firms that had high speculative trading before reform with those that had low speculative trading, despite the increase in the supply of tradable shares, restructured firms of high speculative trading on average experienced smaller increase in trading volume than their control firms did. The mean excess change in trading volume of restructured firms of high speculative trading is only In contrast, restructured firms of low speculative trading on average experienced larger increase in trading volume than their control firms did; the mean excess change in trading volume is The difference between firms of high and low speculative trading is statistically significant at 1% level. Among the four groups sorted by the level of liquidity and the level of speculative trading, only the restructured firms that had high liquidity and high speculative trading before reform, on average, had smaller increase in trading volume than their control firms. The other three groups all had significant increase in trading volume than their control firms. In summary, the evidence for both share turnover and trading volume shows that the reform reduced the level of speculative trading, and the effect is more significant for firms that were subject to speculative trading before reform Impact on stock price and downward sloping demand curve 24