Survey
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
Download CLAREMONT McKENNA COLLEGE STOCK MARKET SENTIMENT
Document related concepts
Beta (finance) wikipedia , lookup
Trading room wikipedia , lookup
Syndicated loan wikipedia , lookup
Private equity secondary market wikipedia , lookup
Investment fund wikipedia , lookup
Financial economics wikipedia , lookup
Market (economics) wikipedia , lookup
Stock valuation wikipedia , lookup
Interbank lending market wikipedia , lookup
Stock trader wikipedia , lookup
Transcript
CLAREMONT McKENNA COLLEGE STOCK MARKET SENTIMENT IN CHINA: THE IMPACT ON TIME DEPOSITS AND SHARE PRICES SUBMITTED TO PROFESSOR RICHARD C.K. BURDEKIN AND DEAN GREGORY HESS BY LUKE REDFERN FOR SENIOR THESIS SPRING 2008 APRIL 28, 2008 2 Table of Contents Acknowledgements Abstract ii iii I. INTRODUCTION 4 II. LITERATURE REVIEW 9 III. DATA SECTION 16 Time Deposits 16 Share Discounts 17 IV. RESULTS AND ANALYSIS 23 Time Deposits 23 B-share Discounts 24 H-share Discounts 27 ADR Discounts 30 V. CONCLUSION 34 REFERENCES 36 TABLES 39 FIGURES 47 3 Acknowledgements This work would not have been possible without the constant advice and direction from Professor Richard C.K. Burdekin. The Lowe Institute of Political Economy for their generous support of a Faculty-Student research project in the Fall of 2007 between myself and Professor Burdekin. Nancy Tao, Li Qingyang and Wen Yu for assistance in data collection during the Faculty-Student research project and Jeffrey Redfern for his valuable comments. 4 Abstract This paper uses survey data by the People’s Bank of China from 2003 to 2007 to test the role that investment preferences has on asset allocation decisions within China. This paper finds that rising Chinese sentiment for the stock market has a significant, negative influence on the growth of savings deposits in China after controlling for liquidity measures and changing real interest rates. The analysis continues by also examining the role these investment preferences have on price discrepancies between A-shares trading in China to equivalent shares traded by foreigners in China, Hong Kong, and New York. These foreign securities trade at a persistent discount to their corresponding A-shares, and results indicate these discounts are significantly affected by both Chinese and U.S. investor sentiment. The sample period is limited to 52 monthly observations due to the availability of the Chinese investor sentiment surveys. 5 I. Introduction Decades of growth in China’s broader economy have recently been followed by a rapid expansion in its equity markets. China created the A-share market for local Chinese investors in Shanghai and Shenzhen in 1991, and a year later opened the subsequent Bshare markets where only to foreign investors were allowed to trade. Both markets, Ashare and B-share, had low trading volumes and returns for many years. Government regulations prohibited trading a majority of each company’s shares, while seeking to prevent high returns to assure cheap capital for State Owned Enterprises (SOEs).1 China’s total market capitalization stood at 3.2 trillion renminbi in December of 2005 after five years of negative growth, having been hurt by a government plan announced in June 2001 to sell-off a large portion of its ownership in SOEs.2 But, during 2005-2007, China’s stock market has outpaced other markets worldwide. During 2006 China’s markets more than doubled in size, ending with total market capitalization of 8.7 trillion renminbi. This growth continued in 2007—the Shanghai A Index grew 96% year-overyear and its market capitalization stood at 26 trillion renminbi.3 Since ownership of companies listed on China’s A-share markets has been largely restricted to domestic Chinese, many firms listed themselves on additional markets to allow interested foreign investors to purchase their stock. Common vehicles for Chinese firms to gain access to foreign capital are H-shares in Hong Kong, American Depositary Receipts (ADRs) in the United States, along with B-shares in Mainland China. B-shares, traded in U.S. dollars in Shanghai and Hong Kong dollars in Shenzhen, and H-shares, 1 Wong, Sonia M.L. (2006), pg. 392. Burdekin, Richard C.K. (2008), Chapter 8. 3 Market growth numbers calculated from Shanghai and Shenzhen A Index data retrieved from Bloomberg. 2 6 traded in Hong Kong dollars, are technically equivalent to A-shares and have the same ownership rights and dividends as A-shares. ADRs are certificates of ownership that are actually held by a third party and often represent multiple shares of the underlying asset. Holding an ADR gives indirect ownership of Chinese A-shares and all dividends are received in U.S dollars. While all of these securities are theoretically equivalent to Ashares, as they derive their value from the same cash flows, discount rates, and have the same ownership rights, they frequently trade at a large discount to their firm’s corresponding A-shares. Several academic papers have addressed these discounts, often citing varying transaction costs, exchange rate expectations, and investor sentiment in China and abroad that drives a wedge between A-shares and their equivalent securities in other markets.4 This paper will also examine determinates of these discounts with some similar findings. As the stock market’s growth invariably sparked interest both domestically and abroad; savings deposits growth in China has slowed and actually registered declines in a number of months. During October 2006 they fell by 7.6 billion renminbi. As this trend in savings deposits continued into 2007 the central bank cited enthusiasm for the stock market as the major reason for their decline.5 Only a few months later the central bank was able to quantify this flow of funds, announcing that in the first four months of 2007, over 70 billion renminbi was transferred from savings accounts into the stock market.6 This indicates Chinese citizens were changing their investment preferences. Traditionally, with strong controls and poor returns, Chinese avoided the stock market 4 Refer to Section II for more detail People’s Daily Online (2007a) 6 China Securities Journal (2007) 5 7 and simply deposited their money in savings accounts. The People’s Bank of China (PBoC) increased the one year time deposit interest rate six times in 2007 with the last increase bringing the benchmark rate to 4.14% in December of 2007, stating that these increases were driven by a desire to slow the flow of funds from households and corporations into the stock market. While Chinese were pulling money out of savings accounts seeking higher returns, discounts between A-shares and equivalent foreign securities were, on average, expanding. These trends are represented in Figures 4-6, which show the median discount for B-shares, H-shares, and ADRs from September 2003 to December 2007. These expansions suggest that investors outside of China did not match the enthusiasm of local market participants. Investor sentiment in the context of this study will be defined as the investment preferences that prevail throughout the market. Investors can become overly-enthusiastic or pessimistic about a particular investment vehicle beyond what market fundamentals should dictate, which can eventually affect investment decisions and asset prices.7 With record numbers of Chinese citizens deciding to open brokerage accounts, it becomes important to understand to what extent investor sentiment in China is changing and the effect that this investor sentiment is having on their economy and equity markets. When studying the current determinants of Chinese savings deposits, the real interest rate should be a significant factor as that represents the return on the deposited. With inflation in China reaching 6.5% in August 2007, a ten year high, and subsequently increasing to 8.5% by February 2008, negative real interest rates could be a driving force behind shrinking savings accounts, rather than overly optimistic investors. However, the 7 There has been a lot of literature specifically looking at the affect of sentiment on discounts/premiums on closed-end funds. Refer to Section II for more detail. 8 real interest rate remained positive until December 2006 and by that time savings accounts were already dwindling; suggesting that while recent negative interest rates provide increased incentive to remove funds from savings accounts, it cannot be the only factor behind their decline. Ever since 2003 the PBoC has been surveying households’ investment preferences, and found that, in 2007, record high responses were listing stocks and funds as their first investment choice. In October 2007, 44.3% of households that responded “believed that it pays to invest in the stock market.”8 Figure 1 shows the relationship between the growth of savings deposits and Chinese investor sentiment for stocks as surveyed by the PBoC. While the growth of time deposits is very volatile, there is a clear negative relationship with the investor sentiment series, suggesting that the recent rise in investor sentiment for stocks in China may have been a significant influence in the growth of savings accounts. Similarly, this rise in investor sentiment could be affecting not only the flow of funds within China’s economy, but also the discounts between A-shares and their equivalent B-shares, H-shares and ADRs that trade in foreign markets. While arbitrage should nullify the effect of investor preferences on these discounts, Lee, Shleifer, and Thaler (1991) give a number of reasons why variations in investor sentiment prevent arbitrageurs from taking advantage of price discrepancies, not to mention controls within China that limit foreign ownership of A-shares.9 Also discounts, or premiums, should be expected given changing exchange rate expectations, since these securities are denominated in different currencies. Investors have, on average, expected the renminbi to 8 9 People’s Daily Online (2007b) See Section II for more detail 9 appreciate against the U.S. dollar since 2002.10 These expected appreciations have increased since 2005 as China has allowed more flexibility in its currency and the dollar has weakened worldwide. Therefore, these fluctuations could be driving the expanding discounts. A negative graphical relationship can be observed between investor sentiment and the H-share and ADR discount in Figures 5 and 6.11 In both graphs, when local investor sentiment fell in 2004, the discounts shrank, and when sentiment started to rise rapidly in 2006, the discounts began to expand, suggesting a testable, negative relationship between sentiment and the discount. Figure 4 shows the relationship between investor preferences and the B-share discount, which appears to have a noticeable positive relationship. This could be explained by other factors influencing the discount, such as the rapid growth of the Shanghai B Index. Figures 2 and 3 show the prices and trading volumes for the Shanghai A and B share Index, respectively. The rapid growth in the B-share index should have a positive influence on the discount, i.e. making the discount smaller, because securities are often highly correlated with the market in which they trade. It is also important to note the drastic difference in trading volumes between the A-share market and the B-share market. While trading volumes for the B-shares spiked in June 2001 and June 2007 around 18 billion shares, trading volumes for the A-shares spiked around 870 billion shares in June 2007 and A-share volumes were generally over 50 times that of B-Shares throughout the sample (refer to Figures 2 and 3). These drastic differences in trading volumes could also affect the B-share discount so they need to be 10 The RMB/$ forward market is non-deliverable since capital controls prevent conversion and, therefore, all settlements must be made in cash. However, the RMB/$ forward market is very liquid. Discussion of the change in RMB/$ exchange rate expectations continues in Section III. 11 Reader should note that discounts are negative, so as a discount grows it becomes more negative. 10 controlled for before any definitive relationship between the discount and investor sentiment can be established. The goal of this paper is to econometrically determine the connection, if any, between the growth of savings deposits, the discounts of A-shares around the world, and changing investor preferences in China. Recent observations and theory suggest that investor sentiment should have a negative relationship with the growth of savings deposits and A-share discounts, and initial results strongly support these negative relationships even after controlling for other factors commonly related to savings deposits and share discounts. Other research has similarly found a negative relationship between sentiment and A-share discounts; however these studies use proxies for investor sentiment. This study uniquely uses the PBoC investor sentiment survey that directly measures Chinese investor preferences and should offer the most definitive measurement of changing sentiment in China. This paper will proceed as follows. Section II is a review of academic literature on the role of sentiment measures on assets prices and A-share discounts. Section III describes the data and models used to test these hypotheses. Section IV presents the results and analysis of these tests. Section V will conclude the paper and provide possible further areas of study. II. Literature Review In traditional finance theory, investor sentiment does not affect asset prices: arbitrage eliminates the influence of overly-optimistic or pessimistic traders who might allow such sentiment to influence their trading practices. Prior research has explored 11 whether investor sentiment might affect prices in practice. Zweig (1973) and De Long, Shleifer, Summers and Waldman (1990) model the effect of individual investors on asset prices by looking at the discounts on closed-end funds. Zweig develops a theory of investor expectations where professional and non-professional investors interpret information differently and, subsequently, push prices away from their intrinsic values. De Long et al. include a “noise trader sentiment” that changes unpredictably without any tie to traditional fundamentals. Both studies focus on closed-end funds because they provide a unique opportunity to investigate investor sentiment since they frequently trade at a discount or premium to their net asset value (NAV), and individuals, not institutions, make up a majority of their investors. Lee, Shleifer, and Thaler (1991) empirically test previous theories that “noise traders” created the discounts on closed-end funds. Individual investors, posited to be “noise traders,” are the main investors in closed-end funds, while the underlying assets of these funds are held mainly by institutions. Since noise traders buy and sell irrationally, any rational investor would demand a higher rate of return on closed-end funds due to the increased uncertainty of the resale price. As some rational investors hold closed-end funds the price on these funds will fall, creating a discount to their NAV. Theoretically, arbitrage should remove these discounts over time, but Lee at al. maintained that since arbitrageurs do not have infinite time-horizons, they also are exposed to risk from irrational investors and allow the discounts to persist. Empirically, they find that discounts across funds are correlated, indicating there is a common pricing factor and returns on small stocks, which are mainly held by individual investors as well, are inversely related to the discount level. These findings indicate that the overall average 12 discount on closed-end funds could be a proxy for investor sentiment, a systematic risk factor that is priced into securities held mainly by individual investors. Lee et al. acknowledge that their paper only measures the effect of differential sentiment between individual and institutional investors because they use domestic closed-end funds where the underlying assets are also traded in the United States. Bodurtha, Kim, and Lee (1995) correct for this by studying country closed-end funds that invest purely in securities traded in one specific foreign country. They find the share price for country-specific closed-end funds are subject to a U.S. specific risk factor that does not affect the NAV of the assets in the portfolio. This factor cannot be explained with traditional index models, and Bodurtha et al. suggest that changing U.S. sentiment may by the underlying risk factor. Baker and Wurgler (2006) look at the role investor sentiment played in a crosssection of stock returns. They create a SENTIMENT index from six commonly used proxies for sentiment: the closed-end fund discount, NYSE share turnover, the amount and returns on IPOs, percentage of equity in new issues, and the dividend premium. Baker et al. finds if beginning period sentiment was high (low) then stocks prone to speculation—young stocks, growth stocks, non-dividend stocks, distressed stocks, etc., had below (above) average returns. Consistent with previous analysis they find this relationship did not hold for larger, more stable stocks. Neal and Wheatley (1998) explore the ability of different proxies of investor sentiment to predict stock market returns and conclude that the discounts on closed-end funds as well as net redemptions could predict the size premium, but had no ability to predict returns for large firms. With a different proxy for sentiment, Brown and Cliff 13 (2001) investigate the ability of investor sentiment to predict returns and affect asset valuation. They use a survey of market newsletters that had bullish, bearish, or neutral outlooks and define the “bull-bear spread” as a measure of market sentiment. This sentiment could predict stock market returns over a one to three year period and have some explanatory power when assets fluctuate from their intrinsic value as measured by traditional asset pricing models. The ability of investor sentiment to predict returns is still controversial throughout finance. Elton, Martin, and Busse (1998) find that discounts on closed-end funds are not a factor in the return generating process. When they expand the two-index model used by Lee et al. to a five-index model, the sensitivity to sentiment across different size deciles reverses, i.e. larger firms are more sensitive than smaller firms. Chen, Kan, and Miller (1993) look at whether or not closed-end fund discounts actually proxy for investor sentiment. They find, contrary to Lee et al., that discounts and small stock returns are not correlated enough to determine a common pricing factor believed to be investor sentiment. Also, the ability of discounts to predict small firm returns does not remain constant throughout the sample period, calling into question their conclusions even more. These studies show that identifying sentiment in the complicated U.S. market is difficult due to the web of derivatives products and institutional, individual and foreign investors. Now looking at sentiment in China, Bailey (1994) was one of the first to examine the young emerging Chinese equity markets. The segmentation between the domestic Ashare market and the foreign B-share market provides a unique opportunity to test the law of one price. Since corresponding A-shares and B-shares theoretically derive their value from the same expected future cash flows and discount rate, they should fetch the same 14 price; however, B-shares actually traded at a large discount. Bailey first looks at risk factors that influence B-shares versus A-shares and find that, while B-shares trade in mainland China and derive their value from mainland Chinese firms, they have some strong tries to international market indicators such as U.S. Treasury rates and NYSE and Hang Seng market indices. Bailey also finds low correlations between the A-share market and other markets where Chinese stocks are traded, i.e. New York and Hong Kong, revealing segmentation between the domestic Chinese market and the rest of the world. Bailey did not empirically determine factors that affect the B-share discount, although he did find a correlation between discounts on individual securities, indicating a common pricing factor. This could be the result of a systematic political risk premium demanded by foreign investors or possibly “unseasoned or unduly optimistic Chinese investors” pushing up A-share prices.12 Information asymmetries between domestic and foreign investors have also been suggested as a cause for the persistent discounts in the A and B share markets. Gao and Tse (2001) look at the effect of capital controls and the flow of information between A and B shares. They discover the A-share market reacts quicker to earnings announcements than the B-share market and with more trading volume to each announcement. Yet, surprisingly these reactions have little affect on prices. They conclude some local investors have more ready access to information than foreign Bshare investors, possibly the result of higher levels of insider trading. However others such as Chui and Kwok (1998) find evidence suggesting foreigners actually have the ability to access information faster due to information barriers in China, thereby causing 12 Bailey (1994), pg. 257. 15 B-share returns to lead A-share returns. Even with these mixed results Chen, Lee, and Rui (2001) find that information asymmetries are not enough to explain the B-share discount, pointing instead to differences in market liquidity. Many others have tried to empirically determine the factors influencing the discounts between Chinese A-shares and equivalent Chinese securities trading on foreign markets. Wang and Jiang (2003) investigate the A-share market in mainland China and Hong Kong’s H-share market. Wang and Jiang claim in cases where markets are segmented “prices are subject to the market-specific risks and investor sentiments in different trading locations.”13 Over the sample period, H-shares trade with an average discount of 69.2% to their corresponding A-share price. They find that H-shares appear to be affected by risk factors in both Hong Kong and mainland China, while A-shares are only affected by the mainland China risk factors. Reduced liquidity in the H-share market and foreign market indices also has a significant influence on the H-share discount. After controlling for varied levels of risk and liquidity, Mei, Scheinkman, and Xiong (2005) find that high turnover rates in the A-share market explain a good portion of the discount between A-shares and B-shares—indicating higher levels of speculation by domestic Chinese investors. Continuing to look at speculation, Ma (1996) argues that varied levels of risk aversion across markets account for higher levels of speculative trading by domestic Chinese, driving up asset prices in the A-share market. Chinese firms have also been listed on the NYSE in the form of American Depository Receipts (ADRs), which are certificates of deposit for A-shares held by a bank in mainland China. Chinese ADRs, like B-shares and H-shares, again trade with a 13 Wang and Jiang (2003), pg 1274. 16 consistent discount to their corresponding A-shares in China. Suh (2003) studies the ADR market in the U.S. and tries to determine if ADR premiums and discounts can be accounted for by variation in U.S. market sentiment. ADRs from a variety of emerging markets are used to test whether aggregate U.S. market returns, a proxy for sentiment, are related to premiums. Each of these countries has a different set of capital controls that prevent arbitrage. Analysis consistently finds a positive relationship with premiums and U.S. market returns when controlling for exchange rate fluctuations and foreign market returns. ADRs also have a very low rate of institutional ownership, which according to Lee et al. could explain their sensitivity to market sentiment. In a recent study, Arquette, Brown, and Burdekin (2008) use expected exchange rate fluctuations and relative price-to-earnings ratios, a proxy for sentiment, to determine discounts on both H-shares and Chinese ADRs. This study allows for both market and firm specific sentiment measures to affect the discount level. The expected future exchange rate fluctuations account for around 40% of the variation while the remaining variation can partially be explained by the market and firm specific sentiment factors. These results indicate that investors in different trading locations are willing to pay different amounts for the same level of expected earnings. In summary, these studies use a variety of different proxies for investor sentiment and repeatedly find sentiment plays a role determining discounts on equivalent assets trading in segmented markets. Sentiment measures also show some signs of influencing returns across certain size classes where there are high levels of individual ownership. 17 III. Data Section Time Deposits The PBoC first reported public sentiment surveys in September 2003. Survey results for individual’s stock propensity have been reported quarterly since September 2003. In these surveys, 20,000 households are questioned from large, medium and small cities throughout the country. Since survey results are compiled throughout each quarter growth between quarters seems to be a reasonable basis for estimating monthly sentiment levels. Sample data for base money growth (M0), time deposits, time deposit rates, percapita monthly income, inflation, and the Shanghai A index were obtained from the Great China Database14. Per-Capita monthly income was only reported on a quarterly basis in 2007 so the average income over each quarter was calculated to acquire monthly data. The real time deposit one year rate was generated by subtracting current period inflation from the current period time deposit one year rate. Money growth measures, time deposits, per-capita income, and the Shanghai A Index were converted into log growth rates for all regressions to assure stationarity. The following is the basic regression model: Growth_of_Time_Depositsi = αi + β1Growth_of_M0i + β2Growth_of_PerCapita_Monthly_Incomei + β3Real_Time_Deposit_Ratei + β4Investor_Sentimenti + β5Lagged_Growth_of_Time_Depositsi + 11 seasonal dummies15 Investor sentiment measures the percent of people that list stocks or funds as their first investment choice, so as investor sentiment increases more people prefer to invest in 14 http://www.finasia.biz/tejonline/tejonline.htm Regression has 45 observations after adjusting for the lagged dependent variable and 28 degrees of freedom. 15 18 the stock market over other investment opportunities, such as savings accounts. The growth of M0 and not M2 was used as an independent variable in our basic regression equation because M2 includes time deposits, our dependent variable. Other controls were added in subsequent regressions, including growth in the Shanghai A index, a trend variable, and price satisfaction. The price satisfaction index indicates people’s level of satisfaction with current prices. The higher the index the more satisfied people are with current price levels. Share Discounts All stock price data have been collected from Bloomberg. While A- and B-shares are traded in both Shanghai and Shenzhen, analysis of the A-share discounts will focus entirely on the larger Shanghai market. The set of Chinese ADRs included in this study is consistent with previous studies (Suh (2003), Arquette, Brown, and Burdekin (2007)), further augmented by various online sources listing Chinese ADRs in the U.S.16 Although the current list of Chinese ADRs in the U.S. is extensive, with over 90 firms now on U.S. exchanges, many of these firms have only been listed in the last year and often are traded over-the-counter, limiting data availability on Bloomberg. These limitations restrict the sample to only 13 Chinese ADRs. The Shanghai B-share Index (SHBSHR) is comprised of 54 firms, all of which have corresponding A-share listings. Since B-shares are traded less frequently than A-shares, the lowest weighted firms in the B-share index were not included to minimize the effect of liquidity on B-share discounts. Also, some firms had limited data availability on Bloomberg or were not listed on both exchanges for the whole 16 Google Search for “Chinese ADRs” returned two sites that had extensive lists of Chinese ADRs, http://stocksabroad.com/modules.php?name=China_ADR and http://www.site-bysite.com/adr/asia/adr_chn.htm, and www.adr.com which Kutan and Zhou (2006) used to identify their list of Chinese ADRs. 19 sample period; therefore, only 34 firms from the index are included. The 13 ADRs used also have corresponding H-share listings and are included in the H-share sample. Additional H-shares listed in previous studies were added to the sample along with some of the top weighted firms in the Hang Seng China Enterprises Index. Altogether 29 Chinese firms are included in the H-share sample. Four of these 29 firms were listed in the last 18 months and, therefore, this set of companies over weights the second half of the sample. In an effort to control this bias, those four firms were excluded from the Hshares empirical work. Refer to Table 1 for a list of all firms included in the sample. A-share, H-share, B-share, and ADR historical monthly prices for each company were pulled from Bloomberg from September 30, 2003 through December 31, 2007. The data range and frequency were both determined by the availability of investor sentiment surveys (described above) from the PBoC. Data were also pulled on trade volumes, shares outstanding, dividends, market capitalization, bid-ask spreads, and P/E ratios for each company. These factors all affect the pricing of each security and have been used in previous studies as determinants of the discount. The discount for each security was calculated as follows: Discount = (Price in Foreign Market / Implied Price of Underlying Security) – 1 The implied price is the A-share price converted into the foreign currency. The implied price for ADRs also needed to be adjusted since ADRs often represent a claim for more than one underlying A-share.17 17 Some of the ADRs are actually claims on the firm’s H-shares and not the A-shares. However since Hshares derive their value from A-shares and U.S. investors can not convert ADRs into A-shares due to capital controls we calculated the ADR discount with the A-share. A similar assumption was employed in Arquette et al. (2008). 20 Persistent discounts or premiums should be expected when considering transaction costs, exchange rate expectations, and convertibility restrictions. The 12 month non-deliverable RMB/$ forward contract and current RMB/$ exchange rates were collected from Bloomberg. Throughout the entire sample, the RMB/$ forward rate was less than the current exchange rate, indicating that investors expected the renminbi to appreciate throughout the sample period. The RMB/$ dollar forward market is “nondeliverable,” because capital controls prevent convertibility and therefore settlement must be made in cash. Yet the market is still liquid, and these settlements represent the difference between the forward rate and the expected future spot rate.18 The expected exchange rate change is calculated as follows (Spot Rate – Forward Rate)/ (Spot Rate). Figure 7 shows the expected exchange rate change over the whole sample period. Expected appreciation first peaked in 2005 around 6%, just before China changed its currency regime to a managed float. In November 2007, expected appreciations peaked to 10.3% as the dollar weakened world-wide. So as when investors expect currency appreciation, they place more value on the security trading in U.S. dollars and, therefore, discounts should shrink. Differences in liquidity can also lead to discounts or premiums. When a security is less liquid it is more likely that when an investor wants to sell few investors will be looking to buy and, therefore, they are not as likely to receive a fair price for their shares. As a result, investors demand a higher return on less liquid securities and prices fall. Trading volumes and equity shares outstanding allowed us to calculate turnover rates for each security. Unfortunately, turnover is not always the best measure of liquidity. Mei, 18 Arquette et al. (2008), pg. 7 21 Sheinkman, and Xiong (2005) find that turnover, while statistically significant in predicting B-share discounts, could also be considered a measure of speculation, so market capitalizations and bid-ask spread were also retrieved as additional liquidity measures.19 Wang and Jiang (2003) find that securities maintain significant exposure to the systematic risks in the location they are traded, along with the systematic risks of the underlying asset. So while ADRs and H-share derive their fundamental values from the risks and cash flows of the firm, their prices are also affected by the performance of other assets in their markets. The Hang Seng Index, Shanghai A and B indexes, and the S&P 500 index were also pulled from Bloomberg over the same time period to control for each security’s exposure to market specific risk. The data that have been retrieved for each type of security constitutes a panel of firms with 52 observations occurring over the specified time period. In many cases these firms are missing some data points or entire series. This creates an unbalanced panel where the number of observations will vary depending on which explanatory variables are included in each regression. Firm specific P/E ratios were often unavailable or periodically missing in the sample on Bloomberg. While this variable is still included, it significantly reduces the number of observations. Most ADRs included in the sample had limited data. Bloomberg typically only reported the price, turnover, and market capitalization for each ADR. While some ADRs are heavily traded with greater data 19 The Bid-Ask spreads were not included in the regressions because as seen in Table 1 some of the spreads from Bloomberg were negative. While a negative spread for a brief period of time is possible, the frequency of negative spreads calls the accuracy of this data into question and was, therefore, removed from the model. When the Bid-Ask spreads were included for test purposes (not shown), results did not change and their coefficients were never significant. 22 availability, they tend to be listed only recently. As a result this study is limited to less liquid ADRs that span the entire time period. See Table 2 for ADR summary stats and the amount of observations available for each variable. The median B-share discount over the period was -47%, although the sample had great variation ranging from a discount of -78% to a premium of 52%. The median discount also fluctuated throughout the sample period as it started with a median discount of -57% on September 30, 2003 that eventually shrank to -38% by December 2007. The median H-share discount throughout the whole sample was -37%, with minimum and maximum values of -90% and 38%, respectively. The median ADR discount throughout the sample was -28%. While the median discount was at -38% at the beginning of the sample it expanded to -56% by the end of the sample period. The ADR discount has a .35 correlation with investor sentiment, while the B and H shares had a .24 and -.086 correlation respectively, Table 3 shows the correlations of all variables in the sample. Figures 4, 5, and 6 display the median discount of each security along with investor sentiment over the whole sample period. Initially these results indicate that the impact of investor sentiment, if any, varies across these securities. Refer to Table 2 for complete summary statistics. This is the basic regression equation for the B-share discount (the basic equation for H-share and ADR discounts are analogous): B_Discountit = αi + βiGrowth_of_Shanghai_Ai + βiGrowth_of_Shanghai_Bi + βiGrowth_of_Market_Cap_A + βiGrowth_of_Market_Cap_B + βiTurnover_A + βiTurnover_B + βiLog(Exchange_Rate_Expectations) + βiLog(Investor_Sentiment) + 23 βiLog(US_Market_Sentiment_index) + βiLagged_Dependent_Variables + βiCompany_Fixted_effects + βiTime_Trend The Shanghai A, Shanghai B, S&P 500 and Hang Seng Indexes are all converted into log growth rates to assure stationarity. The turnover ratio for each security is calculated as in Mei, Sheinkman, and Xiong (2005), where Turnover = Log (1 + Volume/shares outstanding). The relative market P/E ratio is the P/E ratio for the Shanghai A Index divided by the P/E ratio for the Shanghai B index. While the relative firm P/E ratio is the P/E ratio for the firm’s A-shares divided by the P/E ratio for the Shanghai A index. The relative P/E ratios are included to test the robustness of the sentiment index against another commonly used proxy for investor sentiment. Autoregressive terms are also included in each regression such that the Bayesian Schwartz information criterion (BIC) is minimized.20 A U.S. market sentiment index is also included to allow for the possible effect that variations in U.S. sentiment could have on these discounts.21 All regressions are run with and without fixed effects. Fixed effects allow us to control for unobserved variables that vary across companies in the panel but remain constant throughout the sample period. These variables could affect the discount and, therefore, our results; so fixed effects need to be included to test the robustness of each explanatory variable. We will also add an exchange rate regime binary variable that will be 0 before July 2005 and 1 after. This will allow us to control for any possible effect the change to a managed float had on the A-share discounts. 20 Bayesian Schwartz information criterion is used to test which model most accurately fits a dataset, the model that minimizes the BIC is preferred. 21 The U.S. sentiment index is calculated by UBS and available on Bloomberg under the ticker invoovrl index. 24 IV. Results and Analysis Time Deposits: Table 4 shows the regression output for the determinants of the growth of time deposits. Confirming our initial hypothesis, after controlling for the negative real interest rate and seasonal trends there is a strong negative relationship between the growth in time deposits and Chinese investor sentiment, significant at the 95% level. This indicates that, as the Chinese have become more inclined to invest in the stock market, they are substituting away from their traditional concentration in savings accounts. The growth in M0 has a consistent positive relationship with time deposit growth, and was significant at the 99% level. This relationship is expected since as the money supply in the economy expands more money is available for savings deposits. Growth in per capita income has no significant relationship with time deposits over the sample period. The real interest rate also is not significant. In China this latter result is not surprising, since interest rates have rarely influenced investment decisions as the government set deposit rates administratively (while individuals had no investment alternatives). The relationship between investor sentiment and growth in time deposits remains negative and statistically significant, even when controlling for additional factors. In equation (2), which added a time trend control as an independent variable, the significance of investor sentiment fell slightly but remained above 90%. The growth in the Shanghai A index arguably was one of the main drivers behind the recent reduction in savings deposits; however, equation (3) reveals that while the growth in Shanghai A does have a negative sign, it is insignificant and does not change the significance of investor sentiment. This indicates that the recent change in individuals’ investment preferences, 25 and not just the strong performance of the stock market, has influenced the growth of time deposits. When running the same regression but excluding investor sentiment (not shown), the growth in Shanghai A is more significant than in equation (4). This confirms that adding the sentiment measure to the regression, as expected, reduces the impact of the growth in Shanghai A on time deposits. Equation (4) includes another sentiment measure, Price Satisfaction, which actually increases the significance of stock propensity by a small amount. These results all indicate that individuals’ evolving investment preferences have significantly affected the growth of time deposits. B-share Discounts: Table 5 shows regression output for the determinants of the B-share discount. These results support the theory of a negative relationship between Chinese investor sentiment and the B-share discount. Therefore, as investor sentiment in China increases, the discount between A-shares and B-shares also grows.22 Equation (1) represents our basic regression equation for the B-share discount. Investor sentiment has a negative coefficient significant at the 99% level. The B-share discounts are also highly autoregressive. Table 8 shows the AR(1)-AR(5) models for the B-share discounts. The AR(1) model explains 90.25% of the variation in B-share discounts. The subsequent AR models are included to determine the number of lags that most accurately determines the discount. The AR(4) minimizes the BIC at -2.75 and explains 91.15% of the variation in B-share discounts. All regression equations in Table 5 include four lagged dependent variables and a time trend variable, to assure no serial correlation amongst the residuals. 22 It is important to note that a larger discount is more negative. 26 The other determinants in this equation have coefficients with the correct sign and are highly significant. Individual securities tend to perform well when the market in which they trade performs well, so as the Shanghai A index grows so do the specific Ashares included in our sample, and subsequently so do their discounts. As expected, the Shanghai A index has a negative relationship and conversely the Shanghai B index has a positive relationship; both are significant above the 99% level. So when the A-share market grows, so does the discount. When the B-share market grows, the discount shrinks. This indicates that a large portion of the discount is determined by market specific risk factors, consistent with the findings by Wang and Jiang (2003). While the S&P 500 index has a positive relationship with the discount, it is not significant, and its influence remains inconsistent across the subsequent regressions. The growth of market capitalization and share turnover for both A- and B-shares controls for varied levels of speculative trading and liquidity across the markets. The growth of the B-share market capitalizations has a positive relationship, significant above the 99% level. This indicates that firms with higher market capitalizations tend to have lower discounts, which is consistent with a view by which a larger firm with a higher market capitalization is more liquid and, therefore, has higher prices. The A-share market capitalization is also highly significant with a negative coefficient. So as the size of the A-share portion increases, the size of the discount increases. This finding is seemingly inconsistent with traditional theory that larger firms should have small price discrepancies across markets, but not necessarily with the speculative trading hypothesis raised by Mei et al (2004) by which firms with more floating shares have lower prices because actual float may still be small. This indicates that A-shares also follow liquidity 27 trading theory. The A-share and B-share turnovers both enter into the equation with above 99% significance and a negative and positive relationship, respectively. These turnover results are consistent with our conclusions above that liquidity drives prices in both A-share and B-share markets. As turnover increases, shares are more liquid and causes their respective asset prices to rise. The change in RMB/$ exchange rate expectations would be expected to have a negative effect on the discount, meaning that as investors expect the renminbi to appreciate (smaller value for the RMB/$ exchange rate), the discount between shares decreases. This occurs because investors place more value on the B-shares that have their underlying value based on renminbi. The actual relationship is not always negative in Table 5, and there is a significant positive coefficient in equation (5). The U.S. investor sentiment index always has a positive and significant coefficient at the 99% level, however. So as sentiment in the U.S. rises, the discount tends to narrow. This indicates that sentiment in both the U.S. and China has a statistically significant effect on the discount between A-shares and B-shares, consistent with our original hypothesis. Most results remain quite robust to adding additional controls, although a few of the relationships do change. When adding market P/E and firm specific P/E ratios, another possible proxy for varied levels of investor sentiment, U.S. market sentiment remains unchanged, with significance above 99% in equations (4)-(6). Chinese investor sentiment loses significance, however, falling to the 90% level, when adding fixed effects in equation (2). Yet the significance returns to the 99% level when adding the exchange rate regime binary variable in equation (3) and remains significant at 99% even after adding the relative P/E measures in equations (4)-(6). All other factors remain significant 28 at the same level and in the same direction as equation (1). Chinese and U.S. investor sentiment basically maintain the same relationship. The Shanghai A and Shanghai B effects remain unchanged, with significance above the 95% level. Share turnover and market capitalizations also remain in line with previous findings and significant at or above the levels in equation (1). The relative market and firm P/E ratios in equation (4) both have negative coefficients significant at the 99% level. The direction and significance of the P/E ratios are mainly unchanged in equations (5) and (6), which include company fixed effects and the exchange rate regime binary variable.23 H-share Discount: The regression analysis of the H-share discounts appears in Table 6.24 These results consistently support our original hypothesis indicating a persistent and significant negative relationship between Chinese investor sentiment and the H-share discount. In equation (1) the coefficient on investor sentiment is -0.028 with a standard error of 0.005, indicating significance above the 99% level. The H-share discount is also highly autoregressive. Table 9 shows the AR(1), AR(2) and AR(3) models for the H-share discount. The AR(1) process explains approximately 90.3% of the variation in discounts. The AR(2) minimized the BIC at -2.33 with an adjusted R-squared of 90.8%, indicating it is the most accurate AR model for the H-share discount. All regressions in Table 5 include an AR(2) along with a time trend variable. These are included to assure no serial correlation in the residuals. 23 The significance of the firm specific P/E ratio falls to the 95% level when company fixed effects are added in equation (4) but return to the 99% level in equation (6). 24 Analysis only includes the 25 firms with data that spans the entire series because including the original 29 firms pulled from Bloomberg over-weights the last 12 months, changing some results. 29 The strong relationship between Chinese investor sentiment and the discount remains negative and significant above the 99% level, with or without company fixed effects, in equation (2). In equations (4)-(6) the relative market and firm P/E ratios are included and significance of investor sentiment remains robust. The significance level of investor sentiment falls slightly, but remains at the 99% level in equation (4) without company fixed effects.25 When company fixed effects are included in equation (5), the significance of investor sentiment is well above the 99% level. Equations (3) and (6) also include the exchange rate regime binary variable along with company fixed effects, to test if our results were affected by the exchange rate regime change in July 2005. In both cases investor sentiment is still significant at above the 99% level. The robustness of investor sentiment across a variety tests strongly indicates that it has a negative influence on H-share discounts. The growth of the Shanghai A index has a negative relationship with the H-share discount and remains significant above the 99% level in all cases. Conversely, the growth of the Hang Seng index has a positive effect on the H-share discount and also remains above the 99% significance level in all scenarios. These results confirm the logical premise that, when the A-share market grows fast, the discounts tend to get larger, and when the Hang Seng market expands, the discounts tend to shrink. The growth of the S&P 500 index has a positive influence on the discount, yet its coefficients remains below the 90% significance level. The U.S. market sentiment index has a positive influence, i.e. in lowering the discount, that is significant over the 99% level in every regression. This result suggests that H-shares are exposed to the U.S. market fluctuations 25 The actual significance level for investor sentiment in equation (4) is 98.95%. 30 and risk factors even though H-shares are traded in Hong Kong and denominated in Hong Kong dollars.26 Since foreigners do not have access to the A-share market and H-shares are a common asset purchased to gain access to the Chinese market, it should be expected to be exposed to some U.S. market risks.27 The H-share discounts are also positively affected by the turnover rate of Hshares. In Table 6, H-share turnover is significant in every equation at the 99% level.28 Since the H-share market is less liquid than the corresponding A-share market, this reduced liquidity affects the price investors are willing to pay. As liquidity, and subsequently turnover increase, the risks associated with the H-shares subside, and prices rise, shrinking the discount. The turnover of A-shares is never significant at any level, however. Another proxy for liquidity and speculation, market capitalization, indicates similar results. The growth of the H-share market capitalization has a strong positive influence on the discounts with significance above the 99% level in equations (3), (5) and (6), and 95% significance in equation (1), (2) and (4). These results indicate that larger, more frequently traded firms have more liquid stocks, and, therefore, have smaller discounts between their H- and A-shares. Conversely, the growth of A-share market capitalization has a strong negative relationship with the discount at the 99% level in every equation. So A-share discounts tend to be bigger when the market capitalizations of the A-share listings are themselves larger. This also indicates that liquidity could be affecting prices in the A-share market. This result is inconsistent with our previous 26 It should be noted that the Hong Kong currency broad ties the Hong Kong dollar to the U.S. dollar. When the H-share discount regressions (1) through (6) are run without the U.S. sentiment index (not shown) the S&P 500 index does have some varying degrees of positive significance. When the U.S. sentiment measure is added the positive influence of the S&P 500 remains but significance falls slightly below the 90% level. 28 Table 5 indicates 95% significance for equations (1), (2), (4) and (5) but rounding p-values gives 99% significance. 27 31 hypothesis that larger firms should have smaller discounts. However, it does not have to be inconsistent with the speculative trading theory raised by Mei et al (2004)—larger market capitalization does not necessarily imply a greater asset float because some shares included in the market capitalization are non-tradable. The H-share discount, as expected from previous studies, has a strong negative relationship with the exchange rate variable, however the significance varies across equations. In equation (1), the coefficient on exchange rate expectations is -0.111 implying that a higher RMB/$ exchange rate raises the discount. This coefficient has a standard error of 0.091, giving a p-value of 22%. In equation (2), with the addition of company fixed effects, the coefficient increases to -0.255 with significance at the 99% level. When the relative market and firm P/E ratios are included and fixed effects are removed in equation (4), its coefficient falls to -0.142, and its significance drops below 90%, but exchange rate expectations regain their significance with the addition of fixed effects and the exchange rate regime binary variable in equations (5) and (6). This generally supports our hypothesis that when the market expects the renminbi to appreciate, investors place more value on H-share securities, because their underlying value is in renminbi, which results in a smaller H-share discount. The relative market and firm P/E ratios in equations (4)-(6) all have the correct negative sign, yet neither are significant in any regression.29 ADR Discounts: The regression output for the ADR discounts appears in Table 7. Consistent with the findings with B-shares and H-shares, rising local Chinese investor sentiment 29 This is contrary to Arquette et al (2008) who find that market and firm P/E ratios have a significant negative effect on H-share discounts. 32 significantly adds to ADR discounts. This relationship remains negative and significant throughout all equations tested. Equation (1) in Table 6 has a coefficient of -0.040 with a standard error of 0.017 for Chinese investor sentiment, giving a p-value of 1.58%. The significance remains at this same level in equations (2) and (3) that control for company fixed effects and the exchange rate regime binary variable. The significance of investor sentiment falls to the 90% level in equation (4) when relative market and firm P/E ratios are added as additional proxies for investor sentiment. When adding company fixed effects and the exchange rate regime binary variable in equations (5) and (6), investor sentiment is significant at the 95% level and the coefficients increase to -0.097 and 0.120, respectively. The market P/E ratio never enters the equations significantly or with the correct sign. The firm specific P/E ratios do have a significant negative influence in equations (4)-(6), but the coefficients are always smaller than then the coefficient for the Chinese investor sentiment measure.30 This strong and consistent negative relationship again confirms the original hypothesis that, as local Chinese become more accepting and willing to invest in the local Shanghai stock market, the discount between A-shares and their equivalent securities around the world increases. The ADR discounts are also highly autoregressive and, therefore, lagged dependent variables were included in every regression (along with a time trend). Table 10 shows the AR(1), AR(2) and AR(3) for the ADR discounts. The AR(1) explains 77.4% of the variation in ADR discounts. The AR(2) minimized the BIC at -1.903, indicating it is the most accurate AR model for ADR discounts; and, therefore, two lagged terms are included in every regression listed in Table 7. 30 Significance levels varied from above 99% in equation (4) to 98.4% in equation (6). 33 The influence of U.S. market sentiment again has a consistent and significant positive influence, lowering the discount, similar to the relationships determined for Band H-shares. All regression equations in Table 6 have U.S. market sentiment significant above the 99% level with coefficients that range from 0.055 to 0.067. U.S. market sentiment is actually the most constantly significant explanatory variable across the three markets; in every case it has a positive influence significant above the 99% level. Changes in the expected RMB/$ exchange rate have a very strong negative influence on the ADR discount. These results were significant in every ADR regression equation, which was not the case in B-share and H-share analysis. Equation (1) has a coefficient of -0.411 with a standard error of 0.208, giving a p-value of 4.88%. Exchange rate expectations have the most significant economic influence on ADR discounts besides the S&P 500. The significance of exchange rate expectations remains at the 95% level with or without fixed effects and with the addition of relative market and firm P/E ratios.31 The coefficient also increases in the subsequent regression equations, reaching 1.119 in equation (6). This implies that a two standard deviation increase in the expected appreciation of the renminbi would cause the ADR discount to shrink on average by 10 percentage points.32 It is interesting to note the exchange rate regime binary variable for the first time is not significant in equation (6). This indicates the even though exchange rate expectations are an important determinant of the discount, China’s switch from a pegged exchange rate to a managed float does not seem to have affected the process determining the ADR discounts. 31 These results appear in equations (2)-(6) in Table 6. The actual significance of exchange rate expectations in equation (4) is 94.82%. 32 The standard deviation of the Change in exchange rate expectations is 0.0153, multiplied by -1.115 gives -.0499. So two standard deviations implies a 10% shift. 34 The growth of the Shanghai A index and the growth of the S&P 500 index both have significant influences on the ADR discount, with a negative and positive relationship respectively. The Shanghai A index has significance above the 95% level in every regression except equation (5), where it has p-value of 5.99%. The S&P 500 index has the largest economic influence on the ADR discounts, with a coefficient of 0.637 in equation (1), but the coefficient falls slightly to 0.455 by equation (6). These coefficients are significant at 95% in all equations except (5), where the S&P 500 is significant only at the 90% level. These results are consistent with the conclusions for B-shares and Hshares. When the foreign (S&P 500) market is performing well, the ADR discounts tend to be smaller. Conversely, when the local Shanghai A market is performing well, the ADR discounts expand on average. The ADR and A-share market capitalizations have a consistent negative influence on the ADR discount and are significant at the 99% level in every regression in Table 7.33 The negative coefficients on the A-share market capitalizations are consistent with our previous findings, and still imply that there could be some liquidity trading in the A-share market. Companies that are larger tend to be more liquid and have higher prices, which in turn leads to larger ADR discounts. The negative coefficients on the ADR market capitalizations are not consistent with previous findings for B-shares and H-shares. It was expected that ADR market capitalizations would have a positive effect on the discount, since larger companies again are more liquid and more likely to have similar price discrepancies across markets. The consistent and strongly negative coefficients imply that the ADR market is different than either the B-share or H-share markets. According to 33 The significance of the A-share market capitalization in equation (4) in Table (4) is 98.84%. 35 theory developed by Mei et al (2004), a negative coefficient could be explained if the ADR market is driven by speculative trading, i.e. higher market capitalizations leading to less speculative trading and lower prices.34 The ADR and A-share turnover rates could help provide more information concerning the role of speculative and liquidity trades in these markets. Unfortunately, the coefficients are generally not significant and vary in magnitude and direction. V. Conclusion The results strongly confirm the original hypothesis that rising investor sentiment in China has had a negative impact on the growth of savings accounts and heightened Ashare discounts. This indicates that, not only do changing preferences by investors affect the flow of funds within China, but they also affect the relative price of Chinese firms’ equity around the world. The impact of investor sentiment remained consistent and significant in B-share, H-share, and ADR markets, even after controlling for changing exchange rate expectations, liquidity and market specific risks. Unique from other studies, this paper breaks apart the effects of sentiment in China and sentiment in the U.S. on B-share, H-share, and ADR discounts. The discounts were also positively affected by the U.S. sentiment measure, with over 99% confidence in every regression. Again this shows that sentiment is an important factor when pricing securities across markets. The consistent and significant ability of sentiment to affect macroeconomic variables raises the question of whether these changing preferences have negative consequences on the broader economy and equity markets. While investor preferences will always be changing, large and rapid shifts that are not necessarily justified by market 34 Mei et al (2004), pg. 23. 36 fundamentals could lead to market inefficiencies and imbalances.35 Further study is needed on the impact of macroeconomic and market shifts that are caused by investor preferences. It is recognized that there are limitations in this study from a relatively short sample period that is determined by the availability of investor sentiment survey data, and as more data becomes available these tests should be repeated to verify the results. 35 Indeed the strong run up in the Shanghai market was followed by a near 50% drop between October 2007 and April 2008. 37 References: Arquette, Gregory C., Brown, William O. Jr., Burdekin, Richard C.K., 2008. U.S. ADR and Hong Kong H-share discounts of Shanghai-listed firms. Journal of Banking and Finance (forthcoming). Bailey Warren, 1994. Risk and return on China’s new stock markets: Some preliminary evidence. Pacific-Basin Finance Journal 2, 243-260. Baker, Malcolm, Wurgler, Jeffrey, 2006. Investor sentiment and the cross section of stock returns. Journal of Finance 61 (August), 1645–1680. Bodurtha, James N. Jr., Kim, Dong-Soon, Lee, Charlie M. C., 1995. Closed-end Country Funds and U.S. Market Sentiment 8, 879-918. Brown, Gregory W., Cliff, Michael T., 2005. Investor sentiment and asset valuation. Journal of Business 78 (March), 405–440. Burdekin, Richard C.K., 2008. “China’s Monetary Challenges: Past Experiences and Future Prospects.” (New York: Cambridge University Press), forthcoming Chen, Gong-Meng, Lee, Bong-Soo, Rui, Oliver M., 2001. Foreign Ownership Restrictions and Market Segmentation in China’s Stock Markets. Journal of Financial Research 24, 133-155. Chen, Nai-Fu., Kan, Raymond, Miller, Merton H., 1993. Are the discounts on closed-end funds a sentiment index? Journal of Finance 48 (June), 795–800. China Securities Journal, 2007. “Chinese Pour Savings Deposits into Stock Market.” May 14 (http://www.cs.com.cn/english/ei/200705/t20070514_1101649.htm). Chiu, Chien-Liang, Lee, Mingchih, Chen, Chun-Da, 2005. Removal of an investment restriction: The ‘B’ share experience from China’s stock markets. Applied Financial Economics 15 (February), 273–285. Chui, Andy C.W., Kwok, Chuck C.Y., 1998. Cross-Autocorrelation between A-shares and B-shares in the Chinese Stock Market. Journal of Financial Research 21, 333353. De Long, J. B., Shleifer, Andrei, Summers, Lawrence H., Waldmann, Robert J., 1990. Noise Trader Risk in Financial Markets. The Journal of Political Economy 98, 703-738. Elton, Edwin J., Gruber, Martin J., Busse, Jeffrey A., 1998. Do investors care about sentiment? Journal of Business 71 (October), 477–500. 38 Gao, Yu, Tse, Yiu K., 2004. Capital Control, Market segmentation and cross-border flow of information: Some empirical evidence from the Chinese stock market. International Review of Economics & Finance 13. Kim, Minho, Szakmary, Andrew C., Mathur, Ike, 2000. Price transmission dynamics between ADRs and their underlying foreign securities. Journal of Banking & Finance 24 (August), 1359–1382. Kutan, Ali M. and Haigang Zhou, 2006. Determinants of returns and Volatility of Chinese ADRs at NYSE. Journal of Multinational Financial Management 16 February: 1-15. Lee, Charles M.C., Shleifer, Andrei, Thaler, Richard H., 1991. Investor sentiment and the closed-end fund puzzle. Journal of Finance 46 (March), 75–109. Ma, Xianghai, 1996. Capital controls, market segmentation and stock prices: Evidence from the Chinese stock market. Pacific-Basin Finance Journal 4 (July), 219–239. Mei, Jianping, Scheinkman, José A., Xiong, Wei, 2005. Speculative Trading and Stock Prices: An Analysis of Chinese A-B-share Premia. NBER Working Paper No. 11362 (May). Neal, Robert, Wheatley, Simon M., 1998. Do measures of investor sentiment predict returns? Journal of Financial and Quantitative Analysis 33 (December), 523–547. People’s Daily Online, 2007a. “The pace of growth in savings seems to be slowing.” (http://english.people.com.cn/200701/05/eng20070105_338558.html). People’s Daily Online, 2007b. “Growth in China's household savings slows down,” (http://english.people.com.cn/90001/90778/6282823.html). Suh, Jungwon, 2003. ADRs and US market sentiment. Journal of Investing 12 (Winter), 87–95. Swaminathan, Bhaskaran, 1996. Time-varying expected small firm returns and closedend fund discounts. Review of Financial Studies 9 (Fall), 845–887. Wang, Steven Shuye, Jiang, Li, 2003. Location of trade, ownership restrictions, and market illiquidity: Examining Chinese A- and Hshares. Journal of Banking & Finance 28 (June), 1273–1297. Wong, Sonia M.L., 2006. China’s Stock Market: A Marriage of Capitalism and Socialism. Cato Journal Fall: Vol. 26, No. 3, pg. 389-423. 39 Zweig, Martin E., 1973. An investor expectations stock price predictive model using closed-end fund premiums. Journal of Finance 28, 67-87. 40 Table 1: Firms Included in Sample Company Name ADRs Aluminum Corp of China Ltd. China Eastern Airlines Co China Life Insurance Co Ltd. China Petroleum & Chemical Corp China Shipping Development Co Ltd. China Southern Airlines Co Guangshen Railway Company Ltd. Huaneng Power International, Inc. Jiangsu Expressway Co Ltd Jiangxi Copper Co Ltd Sinopec Shanghai Petrochemical Co Ltd Tsingtao Brewery Company Limited Yanzhou Coal Mining Company Limited ADR # of firms ACH CEA LFC SNP CSDXF ZNH GSH HNP JEXYF JIXAF SHI TSGTF TZC # of firms B-Shares China First Pencil Co Ltd. China Textile Machinery Danhua Chemical Technology Co Ltd Dazhong Transportation Group Co Ltd. Double Coin Holdings Ltd. Eastern Communications Co Ltd. Huadian Energy Co Ltd. Huangshan Tourism Development Co Ltd. Huaxin Cement Co Ltd. Inner Mongolia Eerduosi Cashmere Product Jinshan Development & Construction Co Ltd. SGSB Group Co Ltd Shanghai Automation Instrumentation Co L Shanghai Baosight Software Co Ltd Shanghai Chlor-Alkali Chemical Co Ltd. Shanghai Dajiang Group Shanghai Diesel Engine Co Ltd. Shanghai Dingli Technology Development Group Co Ltd. Shanghai Erfangi Co Ltd. Shanghai Friendship Group Inc Ltd Shanghai Haixin Group CO Shanghai Highly Group Co Ltd. Shanghai Jinjiang International Investment Holdings Co Shanghai Jinqiao Export Processing Zone Development Co Lt Shanghai Lianhua Fibre Corp Shanghai Material Trading Co Ltd Shanghai Mechanical and Electrical Indus Shanghai Sanmao Enterprise Group Co Ltd Shanghai Waigaoqiao Free Trade Zone Development Co Ltd. Shanghai Wingsung Data Technology Co Ltd. Shanghai Yaohua Pilkington Glass Co Ltd Shanghai Zhenhua Prot Machinary Co SVA Electron Co Ltd. Zhonglu Co Ltd. Shanghai Potevio Co Ltd H-Shares Angang New Steel Co Anhui Conch Cement Co Ltd. Anhui Expressway Beiren Printing Machinery China Eastern Airlines Co China Petroleum & Chemical Corp China Shipping Development Co Ltd. China Southern Airlines Co Dongfang Electrical Machine Guangzhou Pharmaceuticals Guangzhou Shipyard Intl Co Huaneng Power International, Inc. Jiangsu Expressway Co Ltd Jiangxi Copper Co Ltd Jingwei Textile Machinery Luoyang Glass Company Ltd. Maanshan Iron & Steel Ltd. Nanjing Panda Elec Co Ltd. Shandong Xinhua Pharmaceutical Co Ltd. Shenzhen Expressway Co Ltd. Sinopec Shanghai Petrochemical Co Ltd Sinopec Yizheng Chemical Tianjin Capital Environmental Protection Tsingtao Brewery Company Limited Yanzhou Coal Mining Company Limited # of firms ANGGY B-Share H-share A-share Start End 2600 670 2628 386 1138 1055 525 902 177 358 338 168 1171 601600 600115 601628 600028 600026 600029 601333 600011 600377 600362 600688 600600 600188 4/30/2007 9/30/2003 1/31/2007 9/30/2003 9/30/2003 9/30/2003 12/29/2006 9/30/2003 9/30/2003 9/30/2003 9/30/2003 9/30/2003 9/30/2003 12/30/2007 12/30/2007 12/30/2007 12/30/2007 12/30/2007 12/30/2007 12/30/2007 12/30/2007 12/30/2007 12/30/2007 12/30/2007 12/30/2007 12/30/2007 600612 600610 600844 600611 600623 600776 600726 600054 600801 600295 600679 600843 600848 600845 600618 600695 600841 600614 600604 600827 600851 600619 600650 600639 600617 600822 600835 600689 600648 600613 600819 600320 600602 600818 600680 9/30/2003 9/30/2003 9/30/2003 9/30/2003 9/30/2003 9/30/2003 9/30/2003 9/30/2003 9/30/2003 9/30/2003 9/30/2003 9/30/2003 9/30/2003 9/30/2003 9/30/2003 9/30/2003 9/30/2003 9/30/2003 9/30/2003 9/30/2003 9/30/2003 9/30/2003 9/30/2003 9/30/2003 9/30/2003 9/30/2003 9/30/2003 9/30/2003 9/30/2003 9/30/2003 9/30/2003 9/30/2003 9/30/2003 9/30/2003 9/30/2003 12/31/2007 12/31/2007 12/31/2007 12/31/2007 12/31/2007 12/31/2007 12/31/2007 12/31/2007 12/31/2007 12/31/2007 12/31/2007 12/31/2007 12/31/2007 12/31/2007 12/31/2007 12/31/2007 12/31/2007 12/31/2007 12/31/2007 12/31/2007 12/31/2007 12/31/2007 12/31/2007 12/31/2007 12/31/2007 12/31/2007 12/31/2007 12/31/2007 12/31/2007 12/31/2007 12/31/2007 12/31/2007 12/31/2007 12/31/2007 12/31/2007 000898 600585 600012 600860 600115 600028 600026 600029 600875 600332 600685 600011 600377 600362 000666 600876 600808 600775 000756 600548 600688 600871 600874 600600 600188 9/30/2003 9/30/2003 9/30/2003 9/30/2003 9/30/2003 9/30/2003 9/30/2003 9/30/2003 9/30/2003 9/30/2003 9/30/2003 9/30/2003 9/30/2003 9/30/2003 9/30/2003 9/30/2003 9/30/2003 9/30/2003 9/30/2003 9/30/2003 9/30/2003 9/30/2003 9/30/2003 9/30/2003 9/30/2003 12/31/2007 12/31/2007 12/31/2007 12/31/2007 12/31/2007 12/31/2007 12/31/2007 12/31/2007 12/31/2007 12/31/2007 12/31/2007 12/31/2007 12/31/2007 12/31/2007 12/31/2007 10/31/2006 12/31/2007 12/31/2007 12/31/2007 12/31/2007 12/31/2007 12/31/2007 12/31/2007 12/31/2007 12/31/2007 13 34 900905 900906 900921 900903 900909 900941 900937 900942 900933 900936 900916 900924 900928 900926 900908 900919 900920 900907 900902 900923 900917 900910 900914 900911 900913 900927 900925 900922 900912 900904 900918 900947 900901 900915 900930 25 347 914 995 187 670 386 1138 1055 1072 874 317 902 177 358 350 1108 323 553 719 548 338 1033 1065 168 1171 CEA SNP CSDXF ZNH HNP JEXYF JIXAF SHI TSGTF TZC 41 Table 2: Summary Statistics Observations Median Mean Min Max St Dev Growth in Shanghai A 51 3.29% 2.65% -20.12% 24.38% 8.06% Growth in Shanghai B 51 1.76% 2.56% -19.00% 33.29% 10.83% Growth in S&P 51 1.20% 0.76% -4.50% 5.35% 2.28% Growth in Hang Sang Index 51 2.12% 1.78% -9.22% 14.42% 4.55% Change in renminbi/USD exchange rate expectations 52 4.46% 4.56% 1.38% 10.28% 1.53% U.S. investor sentiment Index 52 74.0 73.4 34.0 108.0 17.0 China Investor Sentiment Survey 52 9.7% 15.2% 5.1% 44.3% 12.3% B Share Sample Observations Median Mean Min Max St Dev Discount 1755 -48.4% -45.3% -83.6% 51.8% 19.9% B Share Turnover 1806 7.3% 13.0% 0.0% 139.1% 17.1% A Share Turnover 1807 6.1% 12.6% 0.0% 911.6% 26.6% B Share Bid-Ask Spread 1793 0.001 0.000 -0.255 0.033 0.016 A Share Bid-Ask Spread 1754 0.001 -0.006 -0.706 0.046 0.049 B Share Market Cap (Millions of dollars) 1799 360 540 24 11,770 818 A Share Market Cap (Millions of Renminbi) 1773 2,951 4,317 218 87,953 6,217 Relative Market P/E 1820 0.71 0.72 0.34 1.17 0.23 Relative Firm P/E 1260 1.99 6.25 0.33 185.88 14.44 U.S. Market Sentiment 1820 74.0 73.4 34.0 108.0 17.0 China Investor Sentiment 1820 9.7% 15.2% 5.1% 44.3% 12.3% Notes: This data is from 35 firms, with 52 monthly observations for each from Sept 30, 2003 to Dec. 31, 2007 H Shares Sample Observations Median Mean Min Max St Dev Discount 1339 -37.1% -37.1% -90.3% 38.5% 24.4% H Share Turnover 1476 15.7% 19.8% 0.0% 168.6% 16.9% A Share Turnover 1351 5.8% 9.7% 0.0% 83.0% 11.0% A Share Bid-Ask Spread 1351 0.010 -0.007 -3.820 1.000 0.212 H Share Bid-Ask Spread 1335 0.025 0.040 0.000 1.000 0.060 A Share Market Cap (Millions of Yuan) 1467 13,568 59,140 1,031 1,993,339 184,754 H Share Market Cap (Millions of HKD) 1349 14,211 61,494 511 2,067,388 183,848 Relative Market P/E 1508 0.55 0.57 0.38 0.82 0.11 Relative Firm P/E 1204 0.96 2.06 0.22 62.36 5.27 Note: This data is for all 29 firms with 52 monthly observations for each from Sept 30, 2003 to Dec. 31, 2007 42 ADRs Observations Median Mean Min Max St Dev Discount 549 -28.5% -28.7% -98.9% 25.4% 20.0% ADR Turnover 671 2.2% 3.6% 0.0% 59.5% 5.2% A Share Turnover 556 4.5% 7.6% 0.0% 42.9% 7.8% ADR Bid-Ask Spread 62 0.500 1.566 0.150 19.750 3.388 A Share Bid-Ask Spread 546 0.010 -0.004 -1.890 0.300 0.162 ADR Market Capitalization (Millions of Renminbi) 650 5,288 206,456 949 115,000,000 4,507,795 A Share Market Capitalizations (Millions of Renminbi) 555 31,478 141,702 7,538 5,404,439 415,965 Relative Market P/E 728 1.32 1.44 0.88 2.49 0.45 Relative Firm P/E 469 0.81 1.24 0.22 15.66 1.73 Note: This data is for all 13 firms with 52 monthly observations for each from Sept 30, 2003 to Dec. 31, 2007 Table 3: Correlation Matrices B Share Sample Discount A Share Mkt Cap B Share Mkt Cap Shanghai A Index ShanghaigB Index Expectations A Share Turnover B Share Turnover Relative Firm P/E Relative Market P/E Sentiment U.S. Investor Sentiment H Share Sample Discount A Share Mkt Cap H Share Mkt Cap Shanghai A Index S&P 500 Index HSI Indexg Expectations A Share Turnover H Share Turnover Relative Firm P/E Relative Mkt P/E Sentiment U.S. Investor Sentiment ADR Sample Discount A Share Mkt Cap ADR Mkt Cap Shanghai A Index S&P 500 Index g Expectations ADR Turnover A Share Turnover Relative Firm P/E Relative Mkt P/E Sentiment U.S. Investor Sentiment Discount 1 0.06 0.07 0.11 0.15 0.16 0.32 0.15 -0.23 0.24 0.25 0.12 A Share Mkt Cap B Share Mkt Cap 1 0.77 0.51 0.57 0.10 0.31 0.26 -0.01 0.19 0.18 0.15 1 0.44 0.53 0.09 0.27 0.23 -0.02 0.17 0.16 0.14 ∆ Exchange Shanghai Shanghai Rate A Share A Indes B Index Expectations Turnover 1 0.66 0.08 0.33 0.28 0.03 0.34 0.28 0.31 1 0.09 0.30 0.44 0.05 0.28 0.28 0.31 1 0.22 0.19 -0.02 0.55 0.55 -0.11 B Share Turnover Relative Firm P/E Relative Market P/E 1 0.03 0.49 0.58 0.30 1 -0.02 -0.01 -0.01 1 0.93 0.16 1 0.57 -0.03 0.49 0.56 0.11 Chinese U.S. Investor Investor Sentiment Sentiment 1 0.06 1 ∆ Exchange Chinese H Share A Share Mkt H Share Mkt Shanghai A S&P 500 Rate A Share H Share Relative Firm Relative Mkt Investor U.S. Investor Discount Cap Cap Index Index HSI Index Expectations Turnover Turnover P/E P/E Sentiment Sentiment 1 0.02 1 0.02 0.96 1 0.05 0.48 0.45 1 0.01 0.20 0.20 0.25 1 0.03 0.18 0.18 0.39 0.57 1 -0.02 -0.25 -0.24 -0.33 0.11 -0.15 1 -0.08 0.26 0.25 0.29 0.01 0.09 -0.42 1 -0.06 0.20 0.22 0.09 0.09 0.11 -0.13 0.22 1 0.03 0.06 0.05 0.07 -0.01 0.02 -0.09 -0.03 0.23 1 0.17 -0.25 -0.24 -0.39 0.06 -0.09 0.68 -0.40 -0.22 -0.07 1 -0.10 0.22 0.21 0.27 -0.12 0.20 -0.91 0.46 0.19 0.07 -0.77 1 -0.13 0.19 0.17 0.31 0.14 0.13 0.07 0.13 0.18 0.01 -0.33 0.06 1 ADR Discount 1 -0.15 -0.11 -0.03 0.07 -0.24 -0.33 -0.17 -0.20 -0.47 -0.41 -0.09 A Share Mkt ADR Mkt Shanghai Cap Cap A Index 1 0.22 0.60 0.30 0.08 0.15 0.32 0.01 0.28 0.27 0.29 1 0.15 0.02 0.13 0.00 0.04 -0.01 0.14 0.12 0.00 1 0.28 0.03 0.08 0.33 0.05 0.33 0.29 0.31 ∆ Exchange S&P 500 Rate ADR Index Expectations Turnover 1 -0.14 -0.03 0.03 0.04 -0.07 -0.11 0.12 43 1 0.32 0.15 0.01 0.60 0.59 -0.19 1 0.13 -0.05 0.47 0.43 0.06 A Share Turnover Relative Firm P/E Relative Mkt P/E 1 0.09 0.40 0.44 0.20 1 0.08 0.09 -0.02 1 0.07 -0.21 Chinese U.S. Investor Investor Sentiment Sentiment 1 0.11 1 Table 4: Time Deposit Regression Results Dependent Variable = Growth of Time Deposits (1) (2) (3) (4) 0.1556*** (4.54) 0.1545*** (4.65) 0.1648*** (4.73) 0.1561*** (4.74) Growth Rate of PerCapita Monthly Income 0.0220 (1.50) 0.0209 (1.34) 0.0315 (1.56) 0.0210 (1.35) Real Time Deposit 1-yr Rate 0.0003 (0.26) -0.0004 (-0.20) 0.0008 (0.55) 0.0002 (0.17) Independent Variables Growth rate of M0 Growth rate of M2 Growth rate of Shanghai A Index Stock Propensity -0.0321 (-0.75) -0.0456** (-2.02) -0.0591* (-1.75) -0.0396** (-2.02) Price Satisfaction -0.0002 (-0.62) Time Lagged Growth of Time Deposits Observations Adjusted R-squared -0.0535** (-2.03) 0.0001 (0.48) 0.0932 (0.54) 0.1051 (0.60) 0.0902 (0.50) 0.0957 (0.55) 45 45 0.868 45 0.872 45 0.869 0.872 Notes: Regressions include 11 seasonal dummies and constant (not shown); T-stat is in parentheses; Heteroskedastic Robust Standard Errors Used; * denotes significance at 90% level, ** denotes significance at 95% level, *** denotes significance at 99% level 44 Table 5: B-share Discount Regression Results (1) Dependent Variable = B-Share Discount (2) (3) (4) (5) (6) Independent Variables Growth of Shanghai A Index -0.166* (0.045) -0.159* (0.043) -0.145* (0.043) -0.132* (0.051) -0.126* (0.048) -0.120* (0.048) Growth of Shanghai B Index 0.361* (0.042) 0.347* (0.041) 0.344* (0.041) 0.349* (0.055) 0.334* (0.055) 0.337** (0.054) Growth of S&P 500 Index 0.012 (0.047) 0.003 (0.045) -0.003 (0.045) -0.021 (0.050) -0.036 (0.049) -0.047 (0.051) Growth of Market Cap A Share -0.224* (0.032) -0.219* (0.032) -0.215* (0.031) -0.228* (0.041) -0.219* (0.042) -0.217* (0.042) Growth of Market Cap B Share 0.030* (0.011) 0.030* (0.011) 0.031* (0.011) 0.027* (0.010) 0.028* (0.011) 0.028* (0.011) Log(1+A Share Turnover) -0.060** (0.024) -0.070** (0.033) -0.069** (0.033) -0.064** (0.012) -0.090* (0.015) -0.093* (0.016) Log(1+B Share Turnover) 0.125* (0.020) 0.134* (0.023) 0.135* (0.023) 0.118* (0.022) 0.136* (0.025) 0.137* (0.026) -0.120*** 0.039 -0.161** -0.028 0.131** -0.031 (0.069) (0.061) (0.077) (0.070) (0.062) (0.076) -0.019* (0.003) -0.007*** (0.004) -0.012* (0.004) -0.040* (0.007) -0.022* (0.008) -0.022* (0.008) Log(Market P/E) -0.029* (0.008) -0.024* (0.006) -0.018** (0.007) Log(Firm P/E) -0.004* (0.001) -0.005** (0.002) -0.005* (0.002) Change in Exchange Rate Expectations Log(Investor Sentiment) Log(U.S. Market Sentiment) 0.033* (0.008) 0.031* (0.007) 0.029* (0.007) 0.039* (0.010) 0.036* (0.009) 0.034* (0.009) Constant -0.205* (0.040) -0.247* (0.039) -0.251* (0.039) -0.310* (0.064) -0.332* (0.056) -0.322* (0.057) Break Trend Lagged Dependent Variables Fixed Effects Observations Adjusted R Squared -0.024* (0.004) -0.019* (0.006) 0.001* (0.000) 0.001* (0.000) 0.002* (0.000) 0.001* (0.000) 0.001* (0.000) 0.002* (0.000) Yes Yes Yes Yes Yes Yes No Yes Yes No Yes Yes 1530 0.947 1530 0.949 1530 0.949 1106 0.954 1106 0.956 1106 0.956 Note: W hite Period Robust Standard Errors In Parentheses, *** denotes 90% confidence, ** denotes 95% confidence, * denotes 99% confidence 45 Table 6: H-Share Discount Regression Results (1) Dependent Variable = H Share Discount (2) (3) (4) (5) (6) Independent Variables Growth of Shanghai A Index -0.311* (0.032) -0.268* (0.031) -0.243* (0.031) -0.329* (0.033) -0.282* (0.032) -0.270* (0.032) Growth of Hang Seng Index 0.514* (0.050) 0.473* (0.045) 0.450* (0.048) 0.530* (0.057) 0.480* (0.056) 0.477* (0.057) Growth of S&P 500 Index 0.143 (0.104) 0.123 (0.095) 0.155 (0.096) 0.118 (0.113) 0.101 (0.103) 0.116 (0.103) Log(1+A Share Turnover) -0.019 (0.027) -0.005 (0.040) -0.002 (0.039) -0.017 (0.027) -0.015 (0.041) -0.010 (0.040) Log(1+H Share Turnover) 0.070** (0.028) 0.107** (0.043) 0.118* (0.043) 0.074** (0.030) 0.133** (0.046) 0.142* (0.045) Growth of Market Cap A -0.784* (0.277) -0.798* (0.262) -1.037* (0.255) -0.774* (0.293) -0.804* (0.275) -1.085* (0.276) Growth of Market Cap H 0.643** (0.280) 0.651** (0.264) 0.890* (0.256) 0.645** (0.298) 0.670* (0.280) 0.952* (0.278) Market P/E -0.030 (0.020) -0.026 (0.021) -0.056 (0.021) Firm P/E -0.001 (0.003) -0.001 (0.004) -0.001 (0.004) -0.048* (0.010) -0.066* (0.010) -0.084* (0.011) Log(Investor Sentiment) Change in Exchange Rate Expectations -0.042* (0.006) -0.061* (0.006) -0.067* (0.006) -0.111 -0.255* -0.536* -0.142 -0.315* -0.673* (0.091) (0.092) (0.098) (0.087) (0.100) (0.109) Log(U.S. Market Sentiment) 0.060* (0.010) 0.056* (0.009) 0.052* (0.009) 0.053* (0.010) 0.047* (0.010) 0.043* (0.010) Constant -0.407* (0.061) -0.491* (0.050) -0.492* (0.050) -0.403 (0.063) -0.483* (0.057) -0.524* (0.059) Break Trend -0.034* (0.006) -0.042* (0.006) 0.002* (0.000) 0.003* (0.000) 0.004* (0.000) 0.002 (0.000) 0.003* (0.000) 0.004* (0.000) Lagged Dependent Variables Yes Yes Yes Yes Yes Yes Fixed Effects No Yes Yes No Yes Yes 1206 0.929 1206 0.932 1206 0.933 1067 0.935 1067 0.938 1067 0.939 Observations Adjusted R-squared Note: This is a limited sample of firms, 25, that have data over the whole sample period. White Period Robust Standard Errors in Parentheses, *** denotes 90% confidence, ** denotes 95% confidence, * denotes 99% confidence 46 Table 7: ADR Discount Regression Results Dependent Variable = ADR Discount (1) (2) (3) (4) (5) (6) Growth of Shanghai A Index -0.182** (0.072) -0.145** (0.066) -0.137** (0.069) -0.163** (0.080) -0.136*** (0.072) -0.141** (0.071) Growth of S&P 500 Index 0.637** (0.262) 0.543** (0.254) 0.544** (0.257) 0.511** (0.302) 0.465*** (0.281) 0.455** (0.291) Growth of A Market Cap -0.225* (0.086) -0.213* (0.076) -0.215* (0.077) -0.230** (0.091) -0.214* (0.074) -0.216* (0.073) Growth of ADR Market Cap -0.038* (0.009) -0.035* (0.010) -0.035* (0.010) -0.038* (0.009) -0.036* (0.010) -0.036* (0.010) Log(1 + ADR Share Turnover) -0.038 (0.009) -0.203*** (0.108) 0.077 (0.112) -0.152 (0.096) -0.290** (0.133) -0.277** (0.128) Log(1 + A Share Turnover) 0.009 (0.047) 0.068 (0.078) 0.077 (0.080) -0.035 (0.067) 0.077 (0.069) 0.096 (0.067) Log(Market P/E) 0.030 (0.036) 0.009 (0.041) 0.042 (0.048) Log(Firm P/E) -0.018* (0.004) -0.029** (0.012) -0.028** (0.011) Independent Variables Log(Investor Sentiment) -0.040** (0.017) -0.085** (0.036) -0.088** (0.037) -0.063*** (0.034) -0.097** (0.048) -0.120** (0.060) Log(U.S. Market Sentiment) 0.057* (0.013) 0.067* (0.018) 0.065* (0.018) 0.055* (0.019) 0.066* (0.024) 0.064* (0.023) -0.411** -0.539** -0.691** -0.644** -0.759*** -1.119** (0.208) (0.260) (0.320) (0.330) (0.388) (0.564) Change in Exchange Rate Expectations Break -0.020** 0.010 -0.034 0.021 Constant -0.390* (0.099) -0.614* (0.187) -0.613* (0.187) -0.468* (0.178) -0.656* (0.256) -0.713* (0.288) Trend 0.002* (0.001) 0.004* (0.001) 0.004* (0.001) 0.003* (0.001) 0.004** (0.002) 0.006** (0.003) Yes Yes Yes Yes Yes Yes No Yes Yes No Yes Yes 439 0.837 439 0.848 439 0.848 363 0.814 363 0.826 363 0.826 Lagged Dependent Variables Fixed Effects Observations Adjusted R-squared Note: White Period Robust Standard Errors In Parentheses, *** denotes 90% confidence, ** denotes 95% confidence, * denotes 99% confidence 47 Table 8: B-Share Autoregressive Regressions Dependent Variable = B-Share Discount (1) (2) (3) (4) (5) Independent Variables B-Share Discount(-1) -0.022 0.004 0.892 0.026 0.875 0.027 0.859 0.027 0.850 0.028 B-Share Discount(-2) 0.950 0.008 0.065 0.026 -0.084 0.037 -0.071 0.037 -0.081 0.038 0.175 0.030 0.047 0.037 0.047 0.038 0.135 0.026 0.080 0.038 B-Share Discount(-3) B-Share Discount(-4) B-Share Discount(-5) 0.080 0.028 Constant -0.022 0.004 -0.019 0.004 -0.015 0.004 -0.012 0.004 -0.010 0.005 Adjusted R-Squared BIC 0.9025 -2.70 0.9060 -2.72 0.9086 -2.73 0.9115 -2.75 0.9118 -2.73 1531 1531 1531 1531 1531 Observations Table 9: H-Share Autoregressive Regressions Dependent Variable = H-Share Discount (1) (2) (3) Independent Variables H-share Discount(-1) 0.949 0.008 H-share Discount(-2) 0.819 0.034 0.809 0.035 0.140 0.034 0.032 0.043 H-share Discount(-3) 0.121 0.032 Constant -0.019 0.004 -0.017 0.004 -0.015 0.004 Adjusted R-Squared BIC 0.903 -2.288 0.908 -2.333 0.908 -2.326 Observations 1254 1220 1186 48 Table 10: ADR Autoregressive Regressions Dependent Variable = ADR Discount (1) (2) (3) 0.901 0.025 0.747 0.080 0.736 0.088 0.186 0.081 0.130 0.090 Independent Variables ADR Discount(-1) ADR Discount(-2) ADR Discount(-3) 0.077 0.062 Constant -0.033 0.009 -0.025 0.009 -0.023 0.009 Adjusted R-Squared BIC 0.774 -1.847 0.789 -1.903 0.787 -1.872 531 513 496 Observations Date Growth in Time Deposits 49 Investor Sentiment May-07 Mar-07 Jan-07 Percent who Prefer Stocks and Funds 0% Nov-06 -8% Sep-06 5% Jul-06 -6% May-06 10% Jan-06 -4% Mar-06 15% Nov-05 -2% Sep-05 20% Jul-05 0% Mar-05 25% May-05 2% Jan-05 30% Nov-04 4% Jul-04 35% Sep-04 6% May-04 40% Mar-04 8% Jan-04 45% Nov-03 10% Sep-03 Growth Rates Figure 1: Investor Sentiment Surveys and Growth of Time Deposits Figure 2: Shanghai A Index Prices and Trading Volume 7000 4000 3500 6000 3000 2500 4000 2000 3000 1500 Shanghai A Index Price Trading Volume (Billions) 5000 2000 1000 1000 500 Shanghai A Index Volume M ar -0 8 -0 7 M ar M ar -0 6 -0 5 M ar -0 4 M ar M ar -0 3 -0 2 M ar M ar -0 0 M ar M ar -9 9 -9 8 M ar M ar -0 1 0 -9 7 0 Shanghai A Index Price 350 60 300 50 250 40 200 30 150 20 100 10 50 0 0 50 -0 7 ar M ar M ar M ar M Shanghai B Index Price -0 6 -0 5 -0 4 M ar -0 ar M ar M -0 3 2 -0 1 ar M ar M ar M -0 0 9 -9 8 M ar -9 -9 ar -0 Shanghai B Index Volume Shanghai B Index Price 70 8 400 7 80 M Trading Volume (Billions) Figure 3: Shanghai B Index Prices and Trading Volumes Figure 4: B-Share Median Discount and Investor Sentiment .6 .4 Discount .2 .0 -.2 -.4 -.6 -.8 2004 2005 2006 2007 Median AB Discount Investor Sentiment Figure 5: H-Share Median Discount and Investor Sentiment .6 .4 Discount .2 .0 -.2 -.4 -.6 -.8 2004 2005 2006 Investor Sentiment Median AH Discount 51 2007 Figure 6: ADR Median Discount and Investor Sentiment .6 .4 Discount .2 .0 -.2 -.4 -.6 2004 2005 2006 2007 Median Investor Sentiment Median DISCOUNT Figure 7: Expected RMB/$ Exchange Rate Change .12 Percent .10 .08 .06 .04 .02 .00 2004 2005 2006 Expected Exchange Rate Change 52 2007
