Docstoc

The Impact of Institutional Investor on

Document Sample
The Impact of Institutional Investor on Powered By Docstoc
					              Who is a Winner in Volatile Markets, Institutions or
               Individuals? Evidence from Chinese Funds

                                                        Yi Yao**
                                            Department of Accounting
                                             Nankai Business School
                                              Nankai Unversity, 300071
                                                    Tianjin, China
                                            Email: yaoyi88@126.com
                                            Tel:022-85894506



                                                Rong Yang
                            Department of Business Administration & Economics
                                      SUNY – College at Brockport
                                          350 New Campus Drive
                                           Brockport, NY 14420
                                               United States
                                       Email: ryang@brockport.edu




                                                      Zhiyuan Liu
                                          Department of Accounting
                                           Nankai Business School
                                            Nankai Unversity, 300071
                                                   Tianjin, China
                                        Email: liuzy809@yahoo.com.cn

                                           Version of July 3, 2007



                                             **Corresponding author


Preliminary and Incomplete. Please do not cite or circulate without permission.





 We thank the Chinese National Natural Science Foundation Project (CNNSFP, No.70672031), the very important research
sponsor in China, for their financial support. The second author also acknowledges the financial support provided by SUNY at
Brockport. We greatly acknowledge all the comments and suggestions made during the presentation at the 3rd China International
Finance Conference in 2005 sponsored by China Center for Financial Research, Tsinghua University and Sloan School of
Management, Massachusetts Institute of Technology. We also acknowledge the comments made by Sihai Fang, Iftekhar Hasan
and James Cordeiro on the earlier version of this paper. All the data used in this paper are available from public databases. All
errors are ours.




                                                          1
           Who is a Winner in Volatile Markets, Institutions or
            Individuals? Evidence from Chinese Funds

                                            Yi Yao
                                           Rong Yang
                                          Zhiyuan Liu


                                          Abstract

       We examine the impact of mutual and closed-ended fund institutions in volatile

market in China. We find that institutions successfully predict the future tendency by

shifting from negative to positive feedback on markets rising days from a bear market to

a bull market. Meanwhile, due to Chinese special background, institutional investors are

buying against the falling market during both bear and bull markets. Moreover, we find

on bull market rising days the volatile trading volume is not due to institutions, but to

irrational individuals. Taken together, we find the fund institution become a leader in the

Chinese stock market.




Keywords: institutional ownership, returns, turnover, funds, Chinese stock markets.




                                             2
    I.        Introduction
         Who is a winner in the Chinese volatile stock markets, individual or institutional
investors? Whose trading strategy is smarter, individuals or institutions? The answer is
unknown so far, especially with a huge and rapid growth in Chinese funds during the past
decade. Since the Chinese funds consist of the largest group of institutional investors, it is
worthwhile to address this question by investigating institutional trading behavior of
funds (including mutual funds and closed-ended funds) on large market movement days,
defined as the absolute value of the market’s return is three percent1 or more. The reason
that we choose those large market movement days is fund institutions may modify their
expectations and change their trading strategy when facing these extreme market events.

         The issue of whether institutions herd or anti-herd (stabilizes or destabilizes
markets) has not been settled in the literature. For instance, Dennis and Strickland (2002)
provide evidence that institutional investors, particularly from mutual and pension funds,
react more strongly than individuals when the absolute value of the market return is large
on any given days, consistent with positive feedback herding behavior. However, Lipson
and Puckett (2005) find the opposite results that institutional investors do not contribute
to market volatility using a sample of money managers and pension plan sponsors.
Alternatively, because institutional investors dominate trading (Schwartz and Shapiro,
1992), their feedback trading strategy is more likely to have asset-pricing implications.
Badrinath and Wahal (2002) even find a significant difference in trading behavior among
different types of institutions. More importantly, none of prior studies have attempted to
investigate the impact of institutional investors on the Chinese stock market. This
distinguishes this study from others.

         Our main contribution lies in highlighting the relationship between institutional
ownership and abnormal returns and turnover in the Chinese stock market during a period
of 2001-2006. Since the Chinese stock market is highly oriented by regulations issued by
policies and rules, we believe 2004 is a reasonable cutoff point between a bear market
and a bull market due to an essential policy issued by the Chinese State Council in 2004.


1 Same as Dennis and Strickland (2002), three percent cutoff is not arbitrary. We calculate the mean and standard
deviation of daily returns for the GTA value-weighted Shanghai/Shenzhen portfolios from 2001 to 2006 and select days
that are two standard deviations above or below the mean. It corresponded to the days when the return is roughly three
percent above or below the 2001 to 2006 daily mean.




                                                          3
When we split the whole sample into two periods, 2001-2003 and 2004-2006, we find
institutional ownership is significantly related to abnormal returns and turnover. When a
bear market transformed to a bull market after 2004, the role of fund institutions is
changing from net sellers to net buyers on large market-up days. More importantly,
trading decisions made by institutions are exactly the same as the future trend of stock
markets. In particular, institutions adopt a negative trading strategy in a bear market
before 2004 whereas show a positive herding pattern in a bull market after 2004, which
moving the stock index up, and thus become a leading role in the Chinese stock market.
       Second, institutions make large buy decisions by buying against a falling market
no matter the sample period is before and after 2004. That could be explained by a high
proportion of mutual funds listed and individual investors could buyback their funds at
any time in China. Because the redemption rate of Chinese mutual funds is relatively
high, (Yao and Liu,2004), fund managers have to keep the net value of investments in
order to avoid the redemption by institutional investors. Overall, our findings are contrary
to both Dennis and Strickland’s (2002) and Lipson and Puckett’s (2005) findings due to
the complexity of funds and other special situations in the Chinese emerging stock
market, which will be introduced in section II.
     Third, the direction of relationship between fund institutional ownership and
abnormal turnover changes from positive to negative after 2004, and institutional
abnormal turnover significantly declines in a bull market. In particular, on large market-
up days during 2004-2006, the majority of trading transactions are actually taken by
individuals, not by institutions. As a result, market volatility is mainly due to individual
investors on up days in a bull market. Taken together, these results suggest that fund
institutions are a winner in the Chinese capital market, which has an important influence
on the Chinese market movement.
     The remainder of this study is organized as follows. Section II presents an
introduction of institutional background in the Chinese stock market. Section III provides
a brief review of institutional momentum trading and herding behavior. Section IV
describes the data sets used for this study. Section V discusses empirical findings
regarding the institutional ownership and abnormal returns and turnover on large market
movement days. Section VI presents the conclusion of this paper.




                                             4
    II.       Institutional Background
      The Chinese funds consist of mutual funds and closed-end funds, which closed-end
funds launched in 1998 and mutual funds started in 2001. At the end of November 2006,
84.5% of all funds available in the Chinese stock market2 are mutual funds, so mutual
funds become more popular than closed-end funds. Meanwhile, fund institutional
investors have exceeded more than twenty five percent of total liquid shares in China
stock market at the end of 2006. Also all funds has taken over 70% of total market value
of all the institutions available including insurance companies, pension funds, mutual
funds, investment banks, QFII (Qualified Foreign Institutional Investors) at the end of
2006. Since the fund is the largest group of institutions in China, the role of its
institutional investors obviously deserves a closer look.

          Insert Figure 1 here

     First, the Chinese stock market was launched in 1991 and has been developed
rapidly during the past fifteen years in terms of the number of companies and market
capitalizations. Before 1998, most participants in the Chinese stock market were
individual investors trading as speculators, causing large market volatility as an emerging
market compared to mature stock markets. The Chinese State Council believes that
increasing funds’ institutions could change the main structure of capital markets, monitor
the stock market by supervising institutions (such as control the fund size, guide
institutional trading strategies on large market movement days), and eventually reduce
the level of market volatility. Due to high turnover rate and high volatility, the Chinese
government has issued several regulations in order to keep this young capital market on a
positive track, and hence the Chinese stock market is highly characterized as “managed-
by-regulations”. For example, on June 24, 2001, the Shanghai stock index went up from
917 to 2,245, the highest index point ever in the Chinese stock market history.
Immediately the Chinese government issued a couple of rules and policies in order to
lower the rapid growth rate of stocks, burst stock bubbles and improve the market
stability. After that day, the Chinese stock market entered into a bear market for a three-
year period of 2001 – 2003. More importantly, a fundamental regulation, “How to
2
 By November 30, 2006, the capital market had issued 81.2 billion shares of closed-end funds and 443.9 billion shares
of mutual funds (File from Shenzhen Stock Exchange, December 2006). Due to the buyback of mutual funds, Chinese
SEC has decided not to issue new closed-end funds any more, and all the issued closed-end funds will be converted to
mutual funds on their maturity date.



                                                          5
improve and stabilize the Chinese stock market”, was issued by the Chinese State
Council on February 1, 2004. This regulation is considered as the most important policy
in the Chinese stock market history. The main goal of this new rule is to emphasize the
importance of capital markets and institutions, and rebuild investors’ faith in the Chinese
stock market. After the release of this regulation, the Chinese stock market fundamentally
transforms from a bear market to a bull market. Although Figure 2 shows the bull market
actually started in 2005, we believe that fund institutions have the ability of predicting the
future trend of stock market in 2004 under the regulation issued, one year ahead the bull
market, before individuals take actions. Therefore, year 2004 is considered as a
turnaround point between two types of security markets, a bear market and a bull market.
        Insert Figure 2 here

        Second, since no stock index futures (or futures options) are available so far in
China, investors can not gain a profit or avoid risks using the difference between spot
market and future market like other stock markets in developed countries. As a result,
institutions can only earn a profit depending on the market-up without the help of stock
index futures. Thus, the limitation of such a market environment leaves institutional
investors only one option, buying stocks at a lower price and selling stocks at a higher
price level.

     Third, individuals can buy or sell funds based on the net value of funds at any time
in the Chinese stock market. That could result in a short-term profit-seeking character of
mutual funds due to the high composition of mutual funds in China. Hence, under such a
special situation in an emerging capital market, fund managers have to pursue an increase
in the net value of investments to avoid the redemption.
     Fourth, there are several studies on herding behavior in the Chinese stock market.
For example, both Wu and He (2005) and Shi (2001) find a strong herd behavior among
Chinese fund institutions using quarterly data during a period of 1999 – 2004. Chang et al.
(2006) investigate the investment behavior of funds from the fourth quarter of 2000 to the
second quarter of 2004. They conclude that investment strategies used by 90 percent of
mutual funds are momentum strategies, and momentum strategies are stronger for buying
than selling decisions. Xu et al. (2003) also examine the investment behavior of funds
from 2001 to 2003, and support a high buying and high selling positive feedback herding



                                              6
strategies. Yang et al. (2006) even provide evidence that institutions show a positive
herding behavior while individuals adopt a negative herding strategy, and especially the
herding levels are higher for individual investors as compared to institutions. However,
none of prior studies has investigated the impact of institutional trading behavior on the
Chinese stock market using daily data. Moreover, the sample period of this study spreads
over six years, 2001-2006, longer than any other relevant prior studies. Considering the
complexity of Chinese funds, we believe a six-year sample period should provide
sufficient and strong empirical results under such a fundamental economic transformation
(from a bear to a bull market). Overall, it is of importance to investigate the effect of
institutional trading behavior on the Chinese capital market on large market movement
days.


   III.      Literature Review
   3.1. Measure base
          A number of prior studies investigating institutional momentum trading have
widely discrepant findings. For example, several studies (Gibson and Safieddine, 2003;
Badrinath and Wahal, 2002; Gompers and Metrick, 2001; and Falkenstein, 1996)
conclude that institutions do not trade momentously whereas others (Bennett, Sias and
Starks, 2003; Chen, Hong and Stein, 2002; Dennis and Strickland, 2002; Cai and Zheng,
2004; Nofsinger and Sias, 1999) find strong evidence of institutional momentum trading.
Also, the short-term momentum trading is anomalous to long-term reversal of stock
market prices (Jegadeesh, 1990; DeBondt and Thaler, 1985; Lo and MacKinlay, 1999;
DeLong et al., 1990; Campbell, 2000; and Hirshleifer, 2001).
          In general there are two types of feedback traders in the market, positive and
negative. As institutional traders purchase more of a firm's stock, the firm’s price rises
even further, enabling the rational speculator to capture additional profits. This positive
feedback trading strategy can create speculative bubbles, drive prices away from their
fundamental values, and thus contribute to market volatility. On the contrary, the negative
feedback trading behavior reduces market volatility and drives stock prices back to their
fundamental values.
          More importantly, why does this study use a “base case” test of institutional
momentum trading? Sias (2007) provides evidence that four factors may account for


                                             7
these discrepancies: “(1) value-weighting versus equal-weighting across stocks, (2)
averaging versus aggregating over managers, (3) disagreement in the signs of measures
of institutional demand, and (4) correlation between current capitalization and both lag
returns and measures of institutional demand.” He concludes that there is a strong
evidence of institutional momentum trading when aggregating across institutions and
treating each stock equally. Based on Sias’s finding, we use the aggregate and equal stock
as the measure base.
      Compared with previous studies use averaging versus aggregating over funds 3 ,
instead we use aggregating over funds to avoid the limitation of institutional ownership
measure. And we study the fund’s trading strategy on large market movement days.which
significantly distinguishes this study from others in the Chinese institutional trading
strategy.


     3.2 Institutional Herding Behavior
     Institutional herding behavior has been actively debated and researched in the
literature. First, Scharfstein and Stein (1990) point out those money managers simply
mimic investment decisions made by other managers (i.e. buying when others are buying
and selling when others are selling) rather than respond to their substantive private
information due to the “sharing-the-blame” effect. This herding behavior could
strengthen the level of market volatility. Falkenstein (1996) also provides empirical
evidence that mutual funds have a significant preference towards stocks with “high
visibility and low transaction costs”, while show an aversion to stocks with “low
idiosyncratic volatility” (small firms with little information). Thus, institutional herding
behavior may be not relevant either to private information held by managers or to the
short-term return.
     Second, Werners (1999) provides additional evidence linking momentum patterns in
stock returns to trading patterns among mutual funds using the trading activity during a


3
  Sias (2007) points out there are some shortcoming using average base to calculate the institutional
demand. Assume that three managers trade a stock this quarter: two managers each buy 100 shares and a
third manager sells 300 shares (100 to each of the two other institutions trading and 100 shares to
individual investors). If one measures institutional demand as changes in the fraction of outstanding shares
held by institutions then, in aggregate, institutions are sellers. Alternatively, if one measures the ratio of the
number of institutional buyers to sellers then, on average, institutions are buyers.



                                                        8
period of 1975-1994. He finds that herding levels are much higher in trades of small
stocks and in trading of growth-oriented funds due to less information available and
higher information asymmetry. Moreover, He finds that the direction of mutual fund
herding is related to future stock returns after controlling for momentum in stock prices,
although it is no clear how strong the relation is. Further, Lakonishok, Shleifer and
Vishny (1992) find that institutional herding behavior may not contribute to market
volatility based on the 769 equity pension funds from 1985 to 1989. They show that if
institutions need to obtain more information to evaluate the basic value of each stock than
individuals, institutions buy more stocks that have performed poorly and sell more stocks
that have performed well. This herding behavior may be traded off with the irrational
investment decisions made by individual investors, which may push stock prices back
rather than drive them away from their fundamental values.
        More importantly, Dennis and Strickland (2002) find that institutions herd together
and trade with the momentum of the market on large market movement days using
aggregate quarterly institutional holdings data from 1988 to 1996. They conclude that
institutions trade in this fashion implies that, at least in the short term, they contribute to
market volatility, which leading to a rise in stock prices beyond their fundamental values.
However, Lipson and Puckett (2005) revisited this issue and find completely different
results: a negative contemporaneous trading behavior of institutions using daily
institutional trade executions for 716 institutional investors (90 money managers and 620
pension plan sponsors) on large market movement days during the period from 1999 to
2003. Such different findings could be explained by Badrintath and Wahal’s (2002) study:
significant differences in trading practices are due to different types of institutions. So far
the question regarding Chinese institutional trading strategies is still unsolved.


    IV.       Data
          Based on one of Chinese capital market regulations4, all funds are required to
report the content of their holdings to the Chinese government and only the top ten
market value holdings are required to disclose to the public each quarter. Figure 3 shows
the rapid growth of funds from 2001 to 2006 in China.

4
  The Chinese SEC issued a new regulation, Security Funds Management Details, which requires all funds should be
ranked according to their total market value and only the top ten funds should be reported starting from 1997.



                                                         9
                                                 Insert Figure 3 here

         In this study, we use three databases, CCER(Database of China Center for
Economic Research),WIND Database and CSMAR (Database of Financial Data and
Marketing Data of China Capital Market). First we obtain quarterly institutional
ownership data for all Shanghai / Shenzhen listed firms from 2001 to 2006 from CCER
Disclosure. Then from WIND Disclosure, we obtain trading data of each security, such as
Shanghai Composite Index(SHCI)5 , turnover and return rate. The top ten holdings of
each fund’s portfolio are selected from the Chinese stock market during the period from
January 1, 2001 until December 31, 2006. From GTA Disclosure, we obtain a security
identifier and information such as total market liquidity value of each firm. Finally, we
identify volatile markets as days when the absolute value of returns for the SHCI market
index are greater than three percent.
         A potential issue related to the value-weighted days is outliers. Since reported
SHCI of Shanghai/Shenzhen portfolio is a value-weighted average index, very large
positive or negative returns for several big firms may generate large portfolio returns. As
a result, these selected days may contain days when the price change does not reject a
broad market shift. To ensure if this occurs to our sample, we separately calculate the
percentage of firms with positive returns, negative returns, and zero returns. In addition,
we calculate the ratio of firms with positive returns to firms with negative returns on days
when the market return exceeds three percent, and the ratio of firms with negative returns
to firms with positive returns when the market return is less than negative three percent.
Table 1 provides the composition of these large market movement days in details.
                                             Insert Table 1 here
         For value-weighted up days, the smallest percentage of firms with positive returns
is 57.81 percent on January 5, 2004 and the largest percentage of firms with positive
returns is 100 percent on three days,October 23, 2001, June 24, 2002, and June 8, 2005.
On these three extreme days, all stocks moved up in the same direction and each stock
price can not move up more than 10% of the last day’s closing price according to the
10%6 rule. As a result, such extreme-event days can not reflect the effect of institutional

5
  Shanghai Composite Index (SHCI) was launched on Dec.19, 1990 by Shanghai Stock Exchange. The SHCI is
including all the listed companies in Shanghai Stock Exchange, weighted by the outstanding shares of each companies,
which reflects the trading trend of the overall stock market.
6
  One of Chinese stock market regulations requires that a firm’s stock price can not move up or down more than 10% of



                                                         10
ownership on a firm’s return. Therefore, these three days are regarded as outliers and
deleted from the sample. After deleting outliers, the largest percentage of firms with
positive returns is 6.81 percent on January 31,2002 and the mean of the ratio of the
percentage of firms with positive returns to the percentage of firms with negative returns
on these up or down days is separately 95.24 and 34.67, with a mean of 70.75. These
ratios imply that value-weighted positive or negative days in our samples represent
significantly. It also indicates that our results are not driven by outliers.
          Meanwhile, from Table 1, we can see one strong characteristic of the Chinese
stock market: all firms are moving up or down together. This special feature as a young
market is quite different from other stock markets in US or any other developed countries.
In Dennis and Strickland’s (2002) study, the ratio of the percentage of firms with positive
returns to the percentage of firms with negative returns for these days is 1.9 and 3.9 with
a sample mean of 2.8. In contrast, the mean of our sample is 70.75. From Table 1, the
trend of the mean is continually decreasing especially around 2005 and 2006. Although
all firms are sharing up or down together, still there is a difference between markets rising
days and falling days, which obviously important to our research questions. Our final
sample consists of 28 days (4,781 observations) during the market rising days and 19
days (2,829 observations) during the market falling days. We consider these days as
event days.


     V.        Empirical Findings


      Since there are few empirical or theoretical studies on the role of Chinese fund
institutions, we adopt the models from Dennis and Strickland (2002). We discuss
univariate results first and then discuss regression results.


5.1. Univariate Results
     A. Returns
       The market portfolio is defined as the SHCI value-weighted Shanghai / Shenzhen
portfolio. Table 2 presents descriptive statistics for the interest of variables.
                                               Insert Table 2 here

the last day’s closing price. Otherwise, any trading activities should be discontinued.



                                                            11
       Throughout this study t refers to the event day when the absolute value of market's
return is greater than three percent or more. Return is the firm's return on the event day,
abnormal return is an event-day market-adjusted return, and abnormal turnover is an
event-day turnover minus median turnover for days [-250, -20]. The independent
variables as controls in the regression analysis includes Size, defined as the natural
logarithm of the market value of equity for firm i at the end of last quarter prior to day t;
Turnover, defined as the ratio of shares traded to liquid shares outstanding for firm i on
day t; Var, defined as the variance of the market model residual for firm i on day t for the
period from t-250 to t-20 days; Beta, defined as the beta of the firm's daily returns with
the SHCI index for the period from t-250 to t-20 days; and ShareRatio, defined as the
percentage of a firm's liquid shares held by 1997 Rule institutions for firm i on day t. We
calculate the minimum, first quartile, median, mean, third quartile, maximum, and
standard deviation for each of independent variables. We also calculate these statistics for
sub-samples partitioned by the median level of institutional ownership since the overall
level of institutional ownership is increasing during the sample period. As a result, we
partition firms into high and low institutional ownership sub-samples within each
extreme day.
       The first interesting pattern in the return statistics in Table 2 is abnormal returns for
both high and low institutional ownership portfolios. The mean (median) abnormal return
for the whole institutional ownership portfolio is -0.4 (-0.4) percent. This suggests that
actual returns are lower than expected return with high level of institutional ownership as
a whole. The mean (median) abnormal return for the low institutional ownership portfolio
is -0.5 (-0.4) percent while the mean (median) abnormal return for the high institutional
ownership portfolio is -0.3 (-0.3) percent. This suggests the higher the institutional
ownership, the higher abnormal returns and the closer actual returns to expected returns.
We perform a t-test and a simple sign test to determine if the means and medians for the
high and low institutional ownership portfolios are equal. The equality of the means and
medians is rejected at the ten percent level7.

          A second interesting pattern in the return statistics is standard deviations of the
returns for both high and low institutional ownership portfolios. The high institutional

7
 We perform a t-test and a simple sign test to determine if the means and medians of the high and low institutional
ownership portfolios are equal. Generally, the equality of the means and medians is rejected at five percent level.



                                                          12
ownership portfolio that has a lower raw return also has a lower standard deviation during
large market-up days for the sub-samples. The equality of the means and medians is
rejected at the five percent level. This suggests a tighter clustering of returns for the high
institutional ownership portfolio on market-up days. However, high institutional
ownership along with high standard deviation is shown on market-down days. While the
return difference and clustering of returns for portfolios with high institutional ownership
are partially consistent with our hypothesis, they are not conclusive. We will control for
some variables and test our hypothesis in a regression framework after univariate
comparison.
     Third, the mean (median) of institutional ownership is 9 (5.5) percent for the up day
portfolio in Table 2. Also the size statistics are consistent with Lakonishok, Shleifer, and
Vishny’s (1992) findings. Firms with high institutional ownership are significantly larger
than firms with low institutional ownership. The statistics for variance and beta suggest
that firms within the high institutional ownership portfolio have lower idiosyncratic
volatility and systematic risk. The equality of the idiosyncratic volatility means and
medians is rejected at the ten percent level. Besides
      Finally, we calculate descriptive statistics for raw returns besides market-adjusted
returns. While we employ market-adjusted returns in the regressions, the pattern in event-
day raw returns is observed to be more transparent than abnormal returns. On up days the
mean (median) raw return for the low institutional ownership portfolio is 3.9 (3.6)
percent while the mean (median) return for the high institutional ownership portfolio is
3.7 (3.4) percent. This suggests a lower actual return for the higher institutional
ownership portfolio during the market rising days. We perform a t-test and a simple sign
test to determine if the means and medians for the high and low institutional ownership
portfolios are equal. The equality of the means and medians is rejected at the ten percent
level. Moreover, the difference is approximately 20 basis points. There is no substantial
cross-sectional variation in institutional holdings, different from the finding of Dennis
and Strickland (2002), due to the Chinese market up or down less than 10% rule. When
markets are falling, however, it seems that high institutional ownership leads to high raw
returns. The equality of the means (medians) is rejected at the five percent level. The
mean (median) return for the low institutional ownership portfolio is -4.8 (-4.7) percent
while the mean (median) return for the high institutional ownership portfolio is -4.4 (-4.3)


                                              13
percent. This suggests institutions act as liquidity providers and do not enlarge the level
of market volatility when there is a large drop in the Chinese stock market.


    B. Turnover
         Further, trade volume is one possible source of the relationship between the
event-day abnormal returns and institutional ownership (Dennis and Strickland 2002).
We therefore investigate the relationship between abnormal turnover and ownership
structure on these event days.
         As shown in Table 2, the turnover for firms with high institutional ownership is
larger than the turnover for firms with low institutional ownership on up days. Thus, this
reflects stocks with greater institutional ownership are more liquid. However, on down
days, for both turnover and abnormal turnover variables, the equality of the means and
medians is not rejected at the ten percent level. While the difference of turnover and
abnormal turnover for portfolios with high institutional ownership are not consistent with
our expectations, they are not conclusive because univariate tests do not control for the
influence of other extraneous factors. To evaluate this possibility, the multivariate models
are used below.


5.2. Regression Evidence
A. Regression Models
         Although we would like to use all the observations to estimate a pooled time-
series cross-sectional regression, the clustering of the events with certain days may create
a problem. For example, the residual for firm i on day t may be contemporaneously
correlated with the residual for firm j on day t and this correlation of residuals could
understate the true standard errors for the coefficients estimated. To control for both
heteroskedasticity and autocorrelation consistent covariance, we use the technique of
Newey-West method 8 in order to make an adjustment on the standard deviation of
estimated coefficients.
      We adopt Dennis and Strickland’s (2002) model, abnormal return ( ARit ) is defined
as the difference between actual return and expected return based on the value-weighted

8
 Whitney K. Newey, Kenneth D.,A Simple, Positive Semi-Definite, Heteroskedasticity and Autocorrelation Consistent
Covariance Matrix,West Econometrica, Vol. 55, No. 3 (May, 1987), pp. 703-708



                                                       14
market portfolios.
    ARit =  0 +  1 Vari+  2 ShareRatioi +  3 Size +  4 Betai +  5 Turnoverit +ε   (1)
      Where abnormal returns ( ARit ) is the market-adjusted return for firm i on the event
day. The essential independent variable of interest is ShareRatio,the level of institutional
ownership, which used to investigate if the cross-sectional distribution of market-adjusted
returns is related to the level of institutional ownership. If the coefficient on institutional
ownership variable is significant, this would suggest that institutional investors react
strongly to large market price changes. Consistent with Dennis and Strickland (2002), the
rationale for including size,Beta,abnormal turnover and variance in the abnormal return
regression as control variables is due to the institutional preference for large firms with
high idiosyncratic volatility discussed in the previous section.
        Further, we examine the relationship between trade volume and institutional
ownership using model (2) including the same control variables of model (1). Again, we
use the technique of Newey-West method to estimate the following cross-sectional
regression for each event day:
    ATit =  0 +  1 Vari+  2 ShareRatioi +  3 Size +ε      (2)
where abnormal turnover ( ATit ) is defined as the turnover for firm i less the median
turnover for days [-250, -20]. Using abnormal turnover as the dependent variable controls
for the fact that stocks with a high trade volume would normally have a high turnover on
event days. The independent variables in model (2) consist of institutional ownership
(ShareRatio), size and variance, the same definitions as those in model (1).


B. Two-period Regression Evidence
       Since the year 2004 is a cutoff point between two types of security market, a bear
market and a bull market, so not consistence with Dennis and Strickland’s (2002), we
partition the entire sample period (2001-2006) into two periods conducting the
regressions, 2001-2003 and 2004-2006, separately.
                                        Insert Table 3 here
       From Panel A of Table 3, the level of institutional ownership is significantly
negative (-0.02**) related to abnormal returns on up days during 2001-2003. This
provides evidence that institutions make large selling decisions against a rising market in
a bear market, consistent with institutional negative feedback herding behavior. It also


                                                15
reflects that institutions are able to predict the stock market will remain in a bear market
in the near future, and as a result, institutions sell although individual investors make
large buying decisions. From Panel B of Table 3, however, the coefficient on institutional
ownership is significant and positive (0.009*) on market-up days during 2004-2006.
These results suggest institutions experience large buying decisions, consistent with the
positive herding strategy in a bull market. This also indicates that institutions can forecast
the future stock market will continually stay in a bull market for a long period, and hence
make large buying decisions. Therefore, fund institutions make wiser decisions compared
to individuals on up days during the whole sample period, acting as a leader in the
Chinese stock market.
       Meanwhile, the relationship between the level of institutional ownership and
abnormal returns remains significant and positive (0.016* and 0.049***) on down days
from both Panel A and Panel B of Table 3. This suggests institutions make large buying
decisions against a falling market no matter in a bear market or a bull market. That could
be explained by no stock index futures available so far in China as we discussed in
section II. Fund managers have to maintain the net value of investments to avoid the
redemption by individual investors, considering a high proportion of mutual funds and
individuals can buyback their funds at any time in China.
       Taken together, institutions use opposite trading strategies between a bear market
and a bull market during days when markets are rising. In particular, institutions are net
sellers in a bear market and net buyers in a bull market on up days. Furthermore, during
days when markets are falling, institutions act as net buyers in both bear and bull markets.
To summarize, these results suggest that institutions make wiser decisions on large
market movement days than individuals, acting as a leader in the Chinese stock market
by using different trading strategies.
       In addition, our results are inconsistent with either Dennis and Strickland (2002),
or Lipson and Puckett (2005), which could be partially explained by high percentage of
mutual funds listed in the Chinese stock market and fund’s performance is directly
associated with the redemption rate of funds (Yao and Liu 2004). Since the majority of
mutual funds are individual investors focusing more on short-term returns, fund managers
have to pursue an increase in the net value of investments to avoid the redemption and
maintain the stock price.


                                              16
                                       Insert Table 4 here
        To evaluate the relationship between the level of institutional ownership and
abnormal turnover, Table 4 presents the regression results of abnormal turnover during
two periods, 2001-2003 and 2004-2006, respectively. From Panel A of Table 4, the
coefficient on institutional ownership is significantly positive (0.014***) on down days,
while it is insignificant (0.008) on up days during 2001-2003. Thus, the trade volume
made by institutions is more than individual investors in a bear market, consistent with
the previous explanation that institutions have to maintain the stock market price in order
to reduce the redemption by individuals even in a bear market. From Panel B of Table 4,
however, the coefficients on institutional ownership are significant and negative ( -
0.027*** and -0.066***)during both up and down days in a bull market during 2004-
2006. This indicates the majority of trading transactions are made by irrational individual
investors, not by institutions, in a bull market. Our findings are consistent with Yang et al.
(2006) results, the herding levels are higher for individuals compared to institutions, and
irrational individuals contribute more to market volatility compared to institutions after
2004.
        In addition, we rerun both abnormal returns and abnormal turnover using the
whole period of 2001-2006 all together. We find that not all the coefficients on
institutional ownership are significantly related to abnormal returns or abnormal turnover,
consistent with the notion of year 2004 as a cutoff point. The detailed results are omitted
in the interest of brevity, but the results reported here are not altered in any material way.
        Overall, these results provide evidence that 2004 is a turnaround point in the
Chinese stock market due to the regulation issued at the beginning of 2004. Especially for
institutions, the largest fund in China, their trading strategies have fundamentally
changed after 2004. Our findings provide evidence that fund institutions have made wiser
decisions than individuals, successfully predict the future tendency of stock markets and
become a leader in the Chinese capital market.


C. Summary of Regression Results
        To summarize, we put all the coefficients on institutional ownership in both
regressions of abnormal returns and abnormal turnover together during two sample


                                              17
periods, 2001-2003, 2004-2006 in Table 5.
                                      Insert Table 5 here
       Since we use the aggregated institutional ownership, we examine the impact of
institutions on the Chinese stock market in a short-term period. Our results are partially
consistent with previous findings, a strong herding behavior for fund’s institutions by Wu
and He (2005), Shi (2001) and Xu et al. (2003) in the Chinese stock market. In spite of
that, our results show that institutions show different trading behavior on large market
movement days from Xu et al. (2003) and Change et al. (2006) studies. That may be
attributed to different sample periods used.
       Furthermore, this study is the first research dividing the entire sample into two
periods due to an essential change in the Chinese stock market (from a bear market to a
bull market) using 2004 as a cutoff point. Our empirical results completely support this
turnaround point, which helps to investigate fund institutional trading strategies during a
six-year period in a much clearer fashion. Meanwhile, our findings are inconsistent with
Dennis and Strickland’s (2002), or Lipson and Puckett’s (2005) results, suggesting
different trading strategies among different types of fund institutions under a significantly
different economic and geographical background.


     D. Robustness Tests
       Supplementing the pooled OLS multivariate regression results reported above, we
conducted the regressions by each year separately to justify using 2004 as a cutoff point,
presented in Table 6.
                                      Insert Table 6 here
       From Panel A of Table 6, the coefficients on institutional ownership in
regressions of abnormal returns significantly shift from negative to positive on up days
from 2001-2003 to 2004-2006. From Panel B of Table 6, the relationship between
institutional ownership and abnormal turnover significantly changes from positive to
negative on down days from 2001-2003 to 2004-2006. This indicates using 2004 as a
turnaround point is a valid and effective method to test the role of institutions,
considering the 2004 regulation issued by the Chinese State Council and the complexity
of funds in the Chinese emerging stock market.




                                               18
   VI.      Conclusion
         The primary objective of this study is to examine the impact of fund institutions
on the Chinese stock market using a period of 2001-2006. We find that 2004 is a
turnaround point of institution’s trading strategy from a bear market to a bull market
during this six-year period while the individual investor carried on their inertia falling
thinking. It indicates that institutions are smarter than individuals because institutions
show more rational behavior to grasp the market opportunity. More importantly, we find
that the fund institution trading behavior significantly changes before and after 2004 on
large market movement days. In particular, institutional investors shift from net sellers to
net buyers on markets rising days from a bear market to a bull market. This reflects
institutions change their trading strategies from negative to positive feedback on up days.
Meanwhile, institutional investors are observed to reduce the level of market volatility by
buying against the falling market during both bear and bull markets. This could be
explained by no stock index futures available, and hence managers are forced to pursue
an increase in the net value of investments to avoid the redemption due to the high
percentage of mutual funds available in the Chinese stock market.
         Moreover, the relationship between fund institutional ownership and abnormal
turnover has changed from positive to negative after 2004. Especially the trade volume
by institutions is continually decreasing in a bull market after 2004. In another word, the
volatile trading volume during a period of 2004-2006 is not due to institutions, but to
irrational individuals, consistent with Yang et al. (2006) finding, the herding levels are
higher for individuals than institutions. Overall, fund institutions have been a winner as
compared to individuals in the Chinese stock market.
         This study is important from a public policy perspective because our findings can
help the public to see if those predictive adjustments made by fund institutions are
consistent with the future stock trend, and eventually figure out who dominates the
market movement, institutional or individual investors. Future research might build on
our findings by investigating if different institutions use different trading strategies in the
Chinese stock market. Such research would be especially valuable in the post-2004
regulatory climate.




                                              19
   REFERENCES


Badrinath, S.G. and S. Wahal(2002), ’Momentum trading by institutions’, Journal of Finance,vol.57,
pp.2449–2478.

Bennett, J., R. Sias, and L. Starks(2003), ’Greener pastures and the impact of dynamic institutional
preferences’, Review of Financial Studies, vol.16, pp.1203–1239.

Campbell, J. (2000), ’Asset Pricing at the Millennium’, Journal of Finance , vol.55, pp.1515–1567.

Chen, J., H. Hong, and J. Stein(2002), ’Breadth of ownership and stock returns’, Journal of
Financial Economics, vol. 66, pp.171–205.

De Bondt, W., and R. Thaler(1985),’Does the stock market overreact?’ Journal of Finance , vol.40,
pp.793–805.

DeLong, B.J., A. Shleifer, L. Summers, and R. Waldmann(1990), ’Positive feedback investment
strategies and destabilizing rational speculation’, Journal of Finance , vol.45, pp.379–395.

Dennis, Patrick J. and Deon Strickland(2002), ’Who blinks in volatile markets, individuals or
institutions? ’ Journal of Finance, vol. 57, pp.1923–1949.

Falkenstein, E.G. (1996),’Preferences for stock characteristics as revealed by mutual fund portfolio
holdings’, Journal of Finance , vol.51, pp.111–135.

Faugère, Christophe and Hany A. Shawky(2003), ’Volatility and institutional investors holdings
during a declining market: a study of NASDAQ during the Year 2000’, Center for Institutional
Investment Management University at Albany

Ferson, Wayne, and Rudi Schadt(1996), ’Measuring fund strategy and performance in changing
economic conditions’, Journal of Finance , vol.51, pp.425–462.

Gibson, S. and A. Safieddine(2003), ’Does smart money move markets?’, Journal of Portfolio
Management, vol. 29, pp.66–77.

Gompers, Paul A., and Andrew Metrick(2001), ’Institutional investors and equity prices’, Quarterly
Journal of Economics , vol.116, pp. 229–259.

Grinblatt, Mark, Sheridan Titman, and Russ Wermers(1995), ’Momentum investment strategies,
portfolio performance, and herding: A study of mutual fund behavior’, American Economic Review
85, pp.1088–1105.

Hirshleifer, D.(2001), ’Investor psychology and asset pricing’, Journal of Finance , vol.56,
pp.1533–97.

Huan, Jing and F. GAO(2006), ’An empirical study of investment fund behavior and



                                                20
performance’,Application of Statistics and Management (125), pp.15–20. (in Chinese)

Jegadeesh, N. (1990), ’Evidence of predictable behavior of security returns’, Journal of Finance ,
vol.45, pp.881–898.

Jiang, Xiaodong(2006), ’Irrationality and limited rationality – empirical research on investors in the
Chinese stock market’,Published by Shanghai University Finance and Economics. (in Chinese)

Kraus, Alan, and Hans R. Stoll(1972), ’Parallel trading by institutional investors’, Journal of
Financial and Quantitative Analysis, vol. 7, pp.2107–2138.

Lakonishok, Josef, Andrei Shleifer, Richard Thaler, and Robert W. Vishny(1991), ’Window dressing
by pension fund managers’, American Economic Review , vol.81, pp.227–231.

Liu,Z.Y. and Yi Yao(2005), ’The Redemption Puzzle of the Open-end Fund’,Security Market
Newsletter, Vol. 2, pp.37–40 (in Chinese).

Lo, A., and A. C. MacKinlay(1999), ’A Non-Random Walk Down Wall Street’. Princeton University,
Princeton, New Jersey.

Ma, Jiujun, Lili Wei(2006), ’Herd Behavior in China’s Investment Fund’, Commercial Research ,
vol.7, pp.160–162. (in Chinese)

Nofsinger, John R., and Richard W. Sias(1999), ’Herding and feedback trading by institutional
investors’, Journal of Finance, vol. 54, pp.2263–2295.

Pettengill, G. N., Sundaram, S., and Mathur I. (1995), ’The Conditional Relation Between Beta and
Returns’, Journal of Financial and Quantitative Analysis , vol.30, pp.101–116.

Schwartz, R.A. and J.E. Shapiro(1992), ’The challenge of institutionalization of the equity markets’,
Recent Developments in Finance, New York University Salomon Center, New York, pp.31–45.

Shi, Donghui(1996), ’The empirical research on the risk of Shanghai stock market’,Economic
Research, Vol. 10, pp.44–48 (in Chinese).

Shi, Donghui(2001), ’The empirical tests on the impact of investment fund’s behavior on market
reactions’, World Economic, Vol. 10, pp.26–31 (in Chinese).

Sias, Richard W.(1996), ’Volatility and the Institutional Investor’, Financial Analysts Journal,
vol.52,March/April, pp.13–20.

Sias,Richard W.(2007), ’Reconcilable differences: momentum trading by institutions’,The Financial
Review,Vol. 42, pp.1–22

Song, Jun and Chongfen Wu(2001), ’The herding behavior in the Chinese stock market and
management issues’,Financial Theory and Practice, Vol. 6, pp.21–27 (in Chinese).

Wu, X. and Peng He(2005), ’The analysis of herding behavior by Chinese mutual funds’,Finance



                                                21
Research, Vol. 5, pp.61–69 (in Chinese).

Xu, Y., J. Lin and X. Qiu(2003), ’The effect of fund’s behavior on stock market’, Working paper (in
Chinese).

Yang, Haicheng, G. Chen and Y. Ren(2006), ’Investor’s behavior and policy implication – the
micro basis of transaction system and production renovation’, working paper, Hong Kong
Polytechnic University.

Yao, Yi and Z.Y., Liu(2004), ’Empirical research on the mutual funds’, Economic Science, Vol. 5,
pp.48–57 (in Chinese).

Wermers, Russ(1999), ’Mutual fund herding and the impact on stock prices’, Journal of Finance e,
vol.54, pp.581–622.

Whitney K. Newey, Kenneth D. (1987), ’A simple, positive semi-definite, heteroskedasticity and
autocorrelation Consistent CovarianceMatrix’,West Econometrica, Vol. 55, No. 3, pp.703–708

Wu, Shinong, Ch. Wu(2003), ’Empirical research on habitual and surplus behavior in the Chinese
stock market’, Economic Science,vol.4, pp.43-50 (in Chinese)




                                               22
                                  40%
                                  35%
                                  30%




                      Ownership
                                  25%
                                  20%
                                  15%
                                  10%
                                   5%
                                   0%
                                                 A        B            C        D

                   Fig.1 The Average Market Value of Major Institutional
                                                 Ownership at the end of 2006
   Figure 1: This figure contains a graph of the average institutional ownership of firms
                 by the end of 2006. Total ownership is including ownership by

A denotes         QFII
B denotes         Insurance companies
C denotes         pension funds
D denotes         Mutual funds and closed-ended funds
File from:Shanghai Finance Newspaper, January 26, 2007, Author: Cheng Chen.




                                  3000
                                  2500
                                  2000
                                  1500                                                        SHCI
                                  1000
                                   500
                                        0
                                            01       02   03         04    05       06   07
                                                                    Year

                            Fig.2 The Trading Trend of Shanghai Composite Index (SHCI)
                                          during 2001-2006




                                                               23
               300,000

               250,000

               200,000




     Million
               150,000

               100,000

                50,000

                     0
                         2001   2002        2003   2004   2005   2006
                                               Year


Fig. 3 The Market Value of Top Ten Funds from 2001 to 2006




                                       24
                                  Tab.1 Market Returns
This table presents dates, market returns, and the fraction of returns that are positive, zero,
and negative when the absolute value of the return of the market portfolio exceeds three
percent. Percent positive is the percentage of firms with returns above zero. Percent zero
is the percentage of firms with returns equal to zero. Percent negative is the percentage of
firms with returns less than zero. Ratio is the ratio of percent positive to percent negative
when the market return is positive and the ratio of percent negative to percent positive
when the market return is negative. The market portfolio is defined as the SHCI value-
weighted Shanghai/Shenzhen portfolio.
                              Panel A: Value-Weighted Up Market
                          Mean       percent percent     percent
             Date                                                          Ratio
                        Return (%) Positive     Zero    Negative
           08/01/01           3.47     96.80      2.07       1.14            85.18
           10/12/01           3.24     91.26      3.97       4.77            19.15
           10/23/01           9.86 100.00         0.00       0.00
           01/23/02           6.35     96.99      0.89       2.13            45.58
           01/31/02           6.81     99.73      0.00       0.27           372.00
           05/21/02           3.02     98.20      0.90       0.90           109.00
           06/06/02           4.05     99.55      0.00       0.45           223.00
           06/21/02           3.07     98.31      0.89       0.80           123.11
           06/24/02           9.25 100.00         0.00       0.00
           01/08/03           3.00     94.39      3.37       2.24            42.05
           01/14/03           5.81     99.65      0.00       0.35           288.00
           04/28/03           3.40     61.96     14.75      23.29             2.66
           11/24/03           3.12     92.91      3.90       3.20            29.07
           12/22/03           3.23     72.48     13.51      14.01             5.17
           01/05/04           3.37     57.81      3.27      38.92             1.49
           09/14/04           3.18     98.95      0.15       0.90           109.58
           09/15/04           4.22     99.70      0.00       0.30           328.50
           09/17/04           3.17     99.24      0.15       0.61           163.75
           09/20/04           3.43     99.54      0.00       0.46           216.83
           09/23/04           3.14     93.51      1.28       5.20            17.97
           11/10/04           3.59     99.33      0.15       0.52           189.71
           02/02/05           5.35     99.05      0.12       0.83           119.00
           04/01/05           3.58     98.20      0.48       1.32            74.55
           06/08/05           8.21 100.00         0.00       0.00
           07/12/05           3.43     92.20      1.12       6.67            13.82
           05/08/06           3.95     94.32      0.95       4.73            19.92
           05/12/06           4.26     91.18      1.38       7.44            12.25
           05/15/06           3.82     93.94      1.21       4.85            19.38
           12/11/06           4.15     95.93      0.86       3.21            29.90
           12/25/06           3.93     61.44      7.04      31.52             1.95
           12/29/06           4.20     76.78      4.68      18.54             4.14




                                              25
              Panel B: Value-Weighted Down Market
           Mean
                      percent    percent   percent
  Date     Return                                     Ratio
                     Positive     Zero    Negative
            (%)
01/15/01      -3.44       3.42       2.77     93.81    27.43
07/30/01      -5.27       1.21       5.03     93.76    77.67
08/06/01      -3.91       2.55       3.05     94.40    37.08
08/27/01      -3.16       2.31       4.02     93.67    40.57
10/10/01      -3.33       2.87       1.02     96.11    33.50
10/22/01      -3.29       5.75       4.96     89.30    15.53
11/07/01      -4.62       0.72       0.00     99.28   138.00
01/14/02      -3.29       3.05       3.49     93.46    30.60
01/17/02      -4.06       1.90       3.46     94.63    49.68
01/21/02      -3.42       5.25       6.77     87.99    16.77
01/28/02      -6.33       1.96       1.78     96.27    49.23
05/16/02      -3.06       2.53       5.24     92.23    36.45
05/13/03      -3.04      11.54       2.29     86.17     7.47
10/14/04      -3.88       2.40       0.45     97.15    40.47
08/18/05      -3.76       8.16       1.13     90.71    11.12
05/16/06      -3.05      24.24       0.64     75.11     3.10
05/23/06      -3.21      13.54       0.73     85.73     6.33
06/07/06      -5.33       4.76       0.00     95.24    20.02
07/13/06      -4.84       5.32       0.25     94.43    17.75




                            26
                    Tab. 2 Descriptive Statistics for High and Low Institutional Ownership

          This table presents event-day sample descriptive statistics. The event-day is defined as a
          trading day when the absolute value of the return of the market portfolio exceeds three
          percent. The market portfolio is defined as the SHCI value-weighted Shanghai/Shenzhen
          portfolio. The variables are size which is the natural logarithm of the market value of
          equity the last quarter-ended prior to the event-day, turnover which is daily volume
          expressed as a percentage of liquid shares outstanding on the event-day, variance which
          is the market model residual variance for days [-250, -20], beta which is computed using
          returns for days [-250, -20] for the SHCI index, Share ratio which is the percentage of a
          firm's liquid shares held by institutions of 1997 Rule, return which is the firm's return on
          the event-day,abnormal return which is event-day market-adjusted return ,and abnormal
          turnover which is event-day turnover minus median turnover for days [-250, -20].
          Levene's Test for Equality of Variances represents a rejection at the one percent level of
          equality of the statistic for the sub-samples.

                                                                                          Standard
 Variable        Partition      Min       25th       Median     Mean      75th    Max                    N
                                                                                          Deviation
                                                  Panel A: Up Market
ARit          Whole Sample     -0.144    -0.017      -0.004     -0.004   0.010   0.078      0.027        4781
              io< Median       -0.139    -0.018      -0.004     -0.005   0.008   0.074      0.026        2390
              io≥ Median       -0.144    -0.016      -0.003     -0.003   0.011   0.078      0.028        2391
ATit          Whole Sample     -0.025     0.001       0.007      0.015   0.020   0.661      0.025        4781
              io< Median       -0.023     0.001       0.007      0.016   0.020   0.661      0.028        2390
              io≥ Median       -0.025     0.001       0.007      0.014   0.020   0.170      0.020        2391
Vari          Whole Sample      0.000     0.000       0.000    0.00058   0.001   0.007      0.000        4781
              io< Median        0.000     0.000       0.001    0.00059   0.001   0.007      0.000        2390
              io≥ Median        0.000     0.000       0.000    0.00058   0.001   0.005      0.000        2391
Share
              Whole Sample
Ratioi                          0.001    0.021        0.055    0.090     0.123    0.705     0.099        4781
              io< Median        0.001    0.011        0.021    0.023     0.034    0.055     0.015        2390
              io≥ Median        0.055    0.081        0.123    0.157     0.201    0.705     0.102        2391
Sizei         Whole Sample      8.395    9.004        9.201    9.230     9.406   10.669     0.328        4781
              io< Median        8.467    8.955        9.150    9.170     9.345   10.327     0.298        2390
              io≥ Median        8.395    9.048        9.255    9.289     9.471   10.669     0.347        2391
Betai         Whole Sample     -0.149    0.892        1.055    1.064     1.216    3.211     0.279        4781
              io< Median       -0.149    0.936        1.088    1.099     1.239    2.535     0.261        2390
              io≥ Median        0.166    0.838        1.011    1.030     1.193    3.211     0.293        2391
Returni       Whole Sample     -0.100    0.022        0.035    0.038     0.051    0.102     0.027        4781
              io< Median       -0.100    0.023        0.036    0.039     0.052    0.102     0.027        2390
              io≥ Median       -0.063    0.021        0.034    0.037     0.050    0.101     0.026        2391
Turnoveri     Whole Sample      0.000    0.008        0.017    0.025     0.032    0.672     0.026        4781
              io< Median        0.001    0.008        0.016    0.026     0.033    0.672     0.030        2390
              io≥ Median        0.000    0.009        0.017    0.023     0.031    0.193     0.022        2391




                                                         27
                                                                                    Standard
 Variable     Partition    Min      25th     Median    Mean       75th     Max                   N
                                                                                    Deviation

                                         Panel B: Down Market
ARit        Whole Sample   -0.080   -0.021      -0.004   -0.004    0.011    0.150        0.029   2829
            io< Median     -0.077   -0.021      -0.005   -0.006    0.009    0.139        0.027   1413
            io≥ Median     -0.080   -0.022      -0.003   -0.003    0.013    0.150        0.031   1416
ATit        Whole Sample   -0.030   -0.001       0.004    0.014    0.021    0.228        0.025   2829
            io< Median     -0.019   -0.002       0.002    0.013    0.019    0.211        0.026   1413
            io≥ Median     -0.030    0.000       0.006    0.016    0.024    0.228        0.025   1416
Vari        Whole Sample    0.000    0.000       0.000    0.001    0.001    0.006        0.000   2829
            io< Median      0.000    0.000       0.000 0.00052     0.001    0.006        0.000   1414
            io≥ Median      0.000    0.000       0.000 0.00059     0.001    0.006        0.000   1415
Share
            Whole Sample   0.000    0.018      0.048     0.078     0.104    0.705        0.089
Ratioi                                                                                           2829
            io< Median      0.000    0.011     0.018      0.021    0.031    0.048        0.013   1413
            io≥ Median      0.048    0.068     0.104      0.136    0.163    0.705        0.095   1416
Sizei       Whole Sample    8.395    9.004     9.196      9.220    9.398   10.561        0.316   2829
            io< Median      8.467    8.978     9.151      9.180    9.343   10.327        0.295   1413
            io≥ Median      8.395    9.029     9.233      9.260    9.457   10.561        0.332   1416
Betai       Whole Sample    0.143    0.897     1.057      1.052    1.199    2.260        0.233   2829
            io< Median      0.143    0.921     1.081      1.068    1.212    1.920        0.221   1413
            io≥ Median      0.173    0.859     1.041      1.037    1.190    2.260        0.243   1416
Returni     Whole Sample   -0.109   -0.067    -0.045     -0.046   -0.027    0.100        0.031   2829
            io< Median     -0.101   -0.068    -0.047     -0.048   -0.029    0.100        0.030   1413
            io≥ Median     -0.109   -0.066    -0.043     -0.044   -0.025    0.100        0.033   1416
Turnoveri   Whole Sample    0.000    0.005     0.011      0.023    0.033    0.250        0.028   2829
            io< Median      0.000    0.005     0.009      0.022    0.029    0.230        0.029   1413
            io≥ Median      0.000    0.006     0.014      0.025    0.035    0.250        0.027   1416




                                                 28
Tab. 3 Event-Day Abnormal Return Regressions on Institutional Ownership and
Control Variables,2001-2003 and 2004-2006 Separately

This table contains coefficient estimates from pooled time-series cross-sectional
regressions using the following model:
        ARit =  0 +  1 Vari+  2 Share ratioi +  3 Sizei +  4 Betai +  5 Turnoverit +ε   (1)

The dependent variable is the event-day market-adjusted abnormal return. The event-day
is defined as a trading day when the absolute value of the return of the market portfolio
exceeds three percent. The market portfolio is defined as the SHCI value-weighted
Shanghai/Shenzhen portfolio. The independent variables are size which is the natural
logarithm of the market value of equity the last quarter-ended prior to the event-day,
turnover which is daily volume expressed as a percentage of liquid shares outstanding on
the event-day, variance which is the market model residual variance for days [-250, -20],
beta which is computed using returns for days [-250, -20] for the SHCI index, and
ShareRatio which is the percentage of a firm's liquid shares held by institutions of1997
Rule.

                     Panel A: Regressions of Abnormal Returns, 2001-2003
                         Up Days                          Down Days

                        Coefficient         t-value       Coefficient     t-value
     Intercept                -6.228***         -3.213        -14.433*** -7.155
       Vari                        0.033         0.858              0.020   0.455
    ShareRatioi                -0.020**         -2.379             0.016*   1.729
       Sizei                   0.833***          4.120          1.252***    6.073
       Betai                  -1.948***         -5.790          2.501***    7.264
        Tit                    0.311***          5.540             -0.011 -0.119
    Adjusted R2            0.100                            0.059
                     Panel B: Regressions of Abnormal Returns, 2004-2006
                         Up Days                          Down Days

                          Coefficient           t-value            Coefficient    t-value
     Intercept                   3.798**             2.523            -18.623*** -6.242
       Vari                    -0.075***            -4.060              -0.116*** -2.682
    ShareRatioi                    0.009*            1.677               0.049***   4.912
       Sizei                        0.014            0.085               1.575***   5.026
       Betai                   -3.677***           -15.660               2.518***   4.978
        Tit                        -0.046           -1.427               0.220***   3.347
    Adjusted R2             0.215                                    0.091
* denotes significance at the 10% level
** denotes significance at the 5%
*** denotes significance at the 1%level




                                                 29
 Tab.4 Event-Day Abnormal Turnover Regressions on Institutional Ownership and
             Control Variables,2001-2003 and 2004-2006 Separately

This table contains coefficient estimates from pooled time-series cross-sectional
regressions using the following model:
                 ATit =  0 +  1 Vari+  2 Share ratio +  3 Sizei +ε   (2)
The dependent variable AT is the event-day abnormal turnover, defined as the turnover
for firm i on day t less the median turnover for days [-250, -20]. The event-day is defined
as a trading day when the absolute value of the return of the market portfolio exceeds
three percent. The market portfolio is defined as the SHCI value-weighted
Shanghai/Shenzhen portfolio. The independent variables are size which is the natural
logarithm of the market value of equity the last quarter-ended prior to the event-day,
variance which is the market model residual variance for days [-250, -20], and
ShareRatio which is the percentage of a firm's liquid shares held by institutions of 1997
Rule.

                   Panel A: Regressions of Abnormal Turnover, 2001-2003
                              Up Days                        Down Days

                      Coefficient        t-value        Coefficient              t-value
   Intercept                   -2.956        -1.598                -1.547          -1.252
     Vari                       0.037         1.466                 0.010           0.813
  ShareRatioi                   0.008         1.326              0.014**            2.232
     Sizei                   0.397**          1.978                 0.186           1.392
  Adjusted R2                    0.005                          0.005
                   Panel B: Regressions of Abnormal Turnover, 2004-2006
                               Up Days                       Down Days

                       Coefficient        t-value           Coefficient          t-value
   Intercept                  3.738**          2.090                6.219***        2.578
     Vari                     0.036**          2.076                0.083***        2.860
  ShareRatioi              -0.027***          -6.835               -0.066***       -9.517
     Sizei                      -0.201        -1.055                   -0.334      -1.284
  Adjusted R2                     0.018                              0.086

* denotes significance at the 10% level
** denotes significance at the 5%
*** denotes significance at the 1%level




                                             30
Tab. 5 Summary of Coefficients on Institutional Ownership during
                   2001-2003 and 2004-2006

 Dependent
                   Days        2001--2003         2004--2006
 Variables

                    UP          Net sellers       Net buyers
  Abnormal
   returns
                  DOWN         Net buyers         Net buyers

                                                 Decrease trade
                    UP         Insignificant
                                                    volume
  Abnormal
  Turnover
                              Increase trade     Decrease trade
                  DOWN
                                 volume             volume




                               31
        Tab. 6 Coefficients on Institutional Ownership of Event-Day Abnormal
                Return,Abnormal Turnover Regressions by Each Year

Using model (1) and (2), we rerun both regressions of abnormal return and abnormal
turnover on institutional ownership by each year.



                     Panel A: Regressions of           Panel B: Regressions of Abnormal
                       Abnormal Returns                           Turnover

                 Coefficient        P value            Coefficient     P value
 2001up            0.006              0.239              0.012          1.289
 2002up            -0.026         -2.756***   -          -0.003        -0.404
 2003up            -0.025          -2.509**   -          -0.007         -0.74
 2004up            0.021           4.024***               -0.01        -1.194
                                              -
 2005up            0.004              0.597   -           -0.01      -1.814***
                                              +
 2006up            0.025           2.720***   -           -0.04      -4.784***
2001down           0.003               0.19              0.012        2.742***
                                                                                   +
2002down           0.036            2.561**              -0.006        -1.081      -
2003down           -0.047            -1.509              0.032          1.067
                                                                                   -
2004down            0.03             1.630*              -0.025      -2.155***
                                                                                   -
2005down           0.001              0.098              -0.037      -3.130***     -
                                                                                   -
2006down           0.055           4.181***              -0.064      -6.004***

* denotes significance at the 10% level
** denotes significance at the 5%
*** denotes significance at the 1%level




                                                  32

				
DOCUMENT INFO
Shared By:
Categories:
Tags:
Stats:
views:2
posted:11/3/2012
language:English
pages:32