Technology and Investment, 2013, 4, 42-53
Published Online Febr uary 2013 (http://www.SciRP.org/journal/ti)
Copyright © 2013 SciRes. TI
The Market Reaction To Stock Splits Used as Dividends
Yang Xiao-Xuan
Guanghua School of Management, Peking University, Beijing 100871, China
Email: yangxiaoxuan@gsm.pku.edu.cn
Received 2012
ABSTRACT
This paper investigates the market reaction to stock splits based on China’s A share companies between 2007 to 2010. I
find significant positive abnormal returns around the announcement date (especially before the announcement date) as
well as four to six days before the execution date of China stock splits. I also observe significant negative abnormal
returns just around the execution date. The above phenomenon is relatively stable even if the selection of samples and
empirical models may vary, but the degree of this phenomenon might change overtime. The cross sectional regression
of the abnormal returns for the announcement date shows that the phenomenon is sensitive to the split ratio and the
market, and it is not sensitive to industry, company size and cash dividends. Therefore, combining with the empirical
data i have constructed a high Sharpe ratio short selling investment strategy around the execution date. Then, the article
further discusses the operability of the investment strategy and its stability over time.
Keywords: China’s A share companies, Stock splits, Stock dividends, Announcement date, Execution date
1. Introduction
There is a common phenomenon about dividend policy
of listed companies, that is, other than cash dividends, a
proportion of the “stock sending ” and “reserve transfer-
ring ” do exist in China’s securities market. Since cash
dividends and stock dividends are coming from the ac-
cumulated undistributed profits of enterprises over the
years. Further, “stock sending” in China is equivalent to
foreign stock dividends. And, in the strict sense, “reserve
tra nsferring ” is not part of the profit allocation, but is
similar to foreign stock splits, due to it comes from the
additional paid-in capital and surplus reserve of a firm.
Yet, as the tradition among Chinese investors and scho-
lars, I regard stock splits as stock dividends in my fol-
lowing study.(XUE Zu-yun/LIU Wan-li, 2009.)
Western scholars have put forward several hypotheses on
the motivation of stock dividends and stock splits beha-
vior. Of those, the signaling hypothesis (Asquith/ Healy/
Palepu [1989]) and the liquidity hypothesis (Baker/ Pow-
ell [1993], Muscarella/ Vetsuypens [1996]) have gained
the most attention. Additionally, some studies find that
the reputation hypothesis, the attention hypothesis and
the optimal trading range hypothesis also provide some
explanation power.
With the deepening of American scholars’ researches,
scholars all over the world are beginning to study the
events of their own stock. Including, Christian Wulff
(2002) found significant positive abnormal returns
around both the announcement and the execution day of
German stock splits, and he also observed an increase in
return variance and in liquidity after the ex-day.
In China’s theoretical circle, this kind of study mainly
focused on two points: the market reaction and factors
that affecting the behavior of stock splits. In the direction
of market reaction research, although different scholars
have varied choices of the announcement date, all of the
studies have found a positive response around the an-
nounc ement day(ZHANG Shui-quan, [1997]; CHEN
Xiao, [1998]; WEI Gang, [1998]; CHEN Lang-nan,
[2000]; YU Qiao, [2001]; KONG Xiao-wen, YU
Xiao-kun, [2003]). On the aspect of influencing factors,
YUAN Hong-qi (2001) found negative correlation be-
tween stock dividends and stock dimensions by analyz-
ing the dividend scheme of China’s listed companies
between 1994 to 1997. YANG Shu-e, CHEN Guo-hui
have obtained the similar results in 2000. ZHAO
Chun-guang(2001) found a substitution relation between
stock dividends and cash dividends by using the annual
report data in 1999. In 2000, by studying the law of stock
dividends from the perspective of ownership structure,
WEI Gang found a positive correlation between stock
dividends and the proportion of tradable shares. In 2003,
after analyzing motives for stock dividends, price illusion
hypothesis was first mentioned by HE Tao, which indi-
cates the behavioral motive of stock splits from the pers-
pective of investors. This hypothesis suggests that the
declined stock price caused by stock splits disturbed the
normal judgment of investors. Specifically, the rising
stock price that caused by those misguided investors who
thought they just found the cheap stock, just meets the
X. X. YANG
Copyright © 2013 S ciRes. TI
companies’ needs of improving their market value. As a
result, this hypothesis suggests that the foreign price
theory(also called as the optimal trading range theory)
and the signaling theory cannot explain the behavior of
stock splits in China. In 2004, by examining the informa-
tion content of dividends of listed firms in China, ZHU
Yun and WU Wen-feng who believe that there is a lack
of consideration of the relationship between dividend
changes and future profits in the current test of China’s
dividend signaling model, concluded that dividends do
not contain the information of future earnings, since
company managers do not formulate dividend policy
according to the expectations of future earnings, and in-
vestors cannot obtain valuable information from the div-
idend policy. Consequently, the signaling hypothesis
does not hold. LIU Wan-li and XUE Zu-yun(2010) firstly
made an empirically research on the influence of stock
price change after stock splits of China’s listed compa-
nies on shareholders’ wealth between 2008 to 2009 based
on the mean comparison and testing method. They found
that the day before ex-day, stock prices decline monoto-
nously faster in 2008, compared to the adjusted stock
prices. Yet, stock prices rise in 2009, which are signifi-
cantly higher than the price on the day before ex-day
after 14th day. Stock prices within 20 days are higher than
the year-end stock price. Their results suggest that com-
pared to the decision-making on the year-end stock price,
stock dividend policy does not reduce stock prices, in
fact, it increases that companies’ total market value and
shareholders’ wealth.
Generally speaking, most of the literature above tested
China’s stock splits phenomenon by using foreign ear-
ly-formed theories and hypotheses. And we are still on
the primary stage of correlation analysis of stock splits
relevant factors at present, namely, we not only have yet
tested the applicability of foreign assumption of behavior
motives to China’s market, but also we have not offered
the assumption for the situation in China. Therefore, in
order to explore, there is the birth of this article.
2. Data and Methodology
2.1. Data Selection and Processing
The event study time of this article is selected from 2007
and 2010. Since during this time interval the financial
market of China and the world were experiencing fluctu-
ation, this study and the follow-up establishment of in-
vestment strategy could be more meanin gf ul.
I construct three data sets(spl_event/index/etdaily) to
comprise a core data set(returns), and the sample collec-
tion interval is selected from January 1, 2006 to Decem-
ber 31, 2010. Considering the buffering effect of
non-trading days on stock prices, I exclude the stock
splits on these days. All of the data used in this paper are
from the CCER financial database.
2.2. Data Set Processing
The spl_event data set is a bonus and dividend data set.
Since the study sample is China’s A share listed compa-
nies, I only choose the stock code starting with the be-
ginning of 0 or 6 as example, because they represent A
share listed companies in Shenzhen and Shanghai stock
market.
The index data is a set of returns of market portfolio,
therefore i select Shanghai and Shenzhen 300 in-
dex(980300) as returns of market portfolio in this article.
The Etdaily data set is a daily yield data set. The CSRS1
is the classification standard to distinguish different in-
dustry. The Tradstat(trading status) is to remove bad
companies, such as ST2, PT3. The Return(daily stock
yield) has already been adjusted, so do not require further
adjustment. The Mktcap4
2.3. Interval Selection and Statistic Interpretation
(total A share stock market
value in circulation) is used to approximate the firm size
in the following cross section regression.
Considering there is usually only one or two months be-
tween the announcement and the execution date, in order
to prevent the overlapping of data while calculating the
CAR, I use the same estimation window to evaluate the
value of
,
αβ
of each event. At last, the estimation
window [-110,-11] is selected before the announcement
day, the event window of announcement day is selected
at the announcement date [-10,10], and the event window
of execution day is selected at the execution date
[-10,10].
Three test statistics are computed in this article in order
to determine statistical significance. The first one is the
1 China Securities Regulatory Commission(CSRC) developed
the standard of industry classification in 1998. See “china listed
corporation classification guidelines (Trial)”, April 7, 1999. No.
5.
2 ST refers to a special treatment for a listed company that has
two consecutive years of losses. Namely, before the name of
special treatment stock there will be a abbreviation ST given by
the Shanghai and Shenzhen Exchange from the beginning of
April 22, 1998.
3 PT refers to the suspension of the listing of a firm’s stock and
the implementation of special transfer services to a firm, which
has three consecutive years of losses, given by the Shanghai
and Shenzhen Stock Exchange according to the company law
and the security law since July 9, 1999. Before the stock’s
name there will be a PT.
4 In this article, total A share stock market value in circula-
tion=yesterday closing price×yesterday total number of shares
in circulation.
43
X. X. YANG
Copyright © 2013 S ciRes. TI
simple t-test, which is under the assumption of same va-
riance. The second one is the
t-test(Brown/Warner[1985]), which is under the assump-
tion of different variance denoted as T(BW) in this paper.
The last one is the nonparametric Wilcoxon signed rank
test, which is to reduce the interference caused by the
extreme data. The p-value(Wt) is the p-value of the non-
parametric Wilcoxon signed rank test.
3. Empirical Results
3.1. Using Market Return mod-
el(
, ,,itii mtit
RR
αβ ε
=++
)5
CASE.1 . Market reaction to the announcement in the
event window [-10,10].(average daily abnormal returns,
average cumulative abnormal returns and their signific-
ance)
Table 1 and 2 and Figure 1 present the abnormal returns
and the cumulative abnormal returns around the an-
nouncement in case 1.
Eventdate is relative span to the announcement or the
execution date. AR refers to the mean of the abnormal
returns. negative AR % are the percentage data of nega-
tive abnormal returns. CAR refers to the mean of the
cumulative abnormal returns. negative CAR % are the
percentage data of negative cumulative abnormal returns.
Significance levels: *** 1% level, ** 5% level, * 10%
level. (the annotation above applies to all of the Tables)
Table 1: Abnormal Returns Around the Announce-
ment in case 1
and sample with no
cash dividend to study the short -term market reac-
tion.
Event
date:
AR :
negative
AR %:
t(BW):
simple-t:
-10
0.31%
51.01%
1.51 1.57 0.54
-9
0.26%
53.77%
1.25 1.25 0.73
-8
0.20%
52.00%
1 0.94 0.91
-7
0.02%
52.74%
0.09 0.08 0.60
-6
0.34%
49.50%
1.65 1.66* 0.52
-5
0.44%
46.80%
2.14
1.93*
-4
0.12%
50.98%
0.06
0.55
-3
0.19%
49.27%
0.92
0.87
-2
0.46%
43.20%
2.23**
2.21**
-1
0.83%
43.00%
4.06***
3.18***
0
0.69%
45.19%
3.37***
1.83*
1 0.09% 54.07%
0.45
0.34
2 0.16% 49.05%
0.79
0.77
3
-0.04%
58.29%
-0.12 -0.24 0.17
5 William F. Sharpe, “A Simplified Model of Portfolio Analysis”,
Management Science, January 1963.
4
-0.13%
53.30%
-0.65
-0.67
0.10
5
-0.29%
55.40%
-1.43
-1.61
0.03**
6
-0.35%
56.07%
-1.69*
-1.63
0.01***
7
-0.08%
48.84%
-0.4
-0.39
0.54
8
0.09%
45.83%
0.44
0.45
0.86
9 0.08% 48.39%
0.39
0.41
0.66
10 -0.18% 54.13%
-0.86
-0.8
0.10
Table 2: Cumulative Abnormal Returns Around the
Announcement in case 1
Event
date:
CAR :
negative
CAR %:
t(BW):
simple-t
p-value
(Wt )
-0 to
0
0.69%
47.47%
3.37*** 1.83* 0.17
-1 to
1
1.62%
46.97%
4.55***
2.64***
0.08*
-2 to
2
2.24% 41.92%
4.87*** 3.22*** 0.00***
-3 to
3
2.38%
38.38%
4.38***
3.15***
0.00***
-4 to
4
2.37%
40.91%
3.85*** 2.86*** 0.00***
-5 to
5
2.52%
41.41%
3.70***
2.93***
0.00***
-6 to
6
2.51%
42.93%
3.39***
2.70***
0.03**
-7 to
7
2.45% 45.96%
3.07*** 2.49* 0.06*
-8 to
8
2.74%
43.43%
3.23***
2.59**
0.03**
-9 to
9
3.08%
41.92%
3.44*** 2.79** 0.01***
-10 to
10
3.21%
43.94%
3.41*** 2.73*** 0.02**
Figure 1: Cumulative Abnormal Returns Around the
Announcement in case 1
Table 1 shows significant positive abnormal returns be-
fore and on the announcement date. The results of both
t(BW) and simple-t test indicate significance 2 days be-
fore and on the announcement date, but insignificance
after the announcement date. Table 2 shows significant
cumulative abnormal returns around the announcement.
According to Table 1, I infer that the cumulative abnor-
44
X. X. YANG
Copyright © 2013 S ciRes. TI
mal returns are composed mostly by the abnormal returns
before and on the announcement date. Figure 1 shows the
cumulative abnormal returns starting 10 days before the
announcement date, from which i can confirm the above
conclusion further: significant positive abnormal returns
before and on the announcement date, but not very sig-
nificant after the announcement date since information
has been fully absorbed after the announcement.
Several hypotheses have been put forward to explain the
motivation behind stock splits in China’s listed compa-
nies. Other than the signaling hypothesis and the liquidity
hypothesis, the price illusion hypothesis(HE Tao/CHEN
Xiao-yue [2003]) suggests that the final goal of stock
splits is to enhance enterprise’s market value with no cost.
Because a company with high market value can not only
manipulate the stock price, but can also offer additional
equity , etc. It is hard to completely enumerate and diffi-
cult to verify. The important condition for listed compa-
nies to achieve this goal is that investors have price illu-
sion to their stocks. Namely, first, investors have only
limited abilities to assess the value of the stock, although
they not only analyze the fundamentals of a firm, but
they will also consider the relative price to the overall
market and the company’s history. Second, at least part
of the new investors, who only judge the absolute stock
value, like to buy low-priced stocks cause by stock splits.
At last, listed companies use stock splits to enhance their
market value. In short, stock splits finally enhance com-
pany’s market value by disturb investors’ judgment
without changing the fundamentals of the company. That
is why there will be positive cumulative abnormal returns.
At the same time, the significant positive abnormal re-
turns around the announcement shows that stock splits
are welcomed by participants in the stock market.
It is worth noting here that abnormal returns mostly ap-
pear a few days before the announcement date. As usual,
investors could not foresee the future. However, com-
bined with China’s securities market, i deduce that the
investor’s possession of information is asymmetric and
uneven distribution. The message of stock splits may be
revealed to insiders early, so their purchase before the
announcement cause the abnormal returns, and the in-
formation would be digested almost completely on the
announcement date. But several studies said, some in-
vestors’ expectations agree with the plan of stock splits,
or maybe some institutional investors have already
known the plan before the announcement, that is why the
abnormal returns fluctuate lightly around the announce-
ment.
CASE.2 . Market reaction to the execution in the event
window [-10,10].(average daily abnormal returns, aver-
age cumulative abnormal returns and their significance)
Table 3 and 4 and Figure 2 present the abnormal returns
and the cumulative abnormal returns around the execu-
tion in case 2.
Table 3 shows, between 2007 to 2010, significant posi-
tive abnormal returns 4 to 6 days before the ex-day, and
Table 3: Abnormal Returns Around the Execution in
case 2
Event
date:
AR :
negative
AR %:
t(BW):
simple-t:
p-value( Wt):
-10
-0.18%
54.55%
-0.89 -0.77 0.28
-9
-0.30%
60.10%
-1.48
-1.32
0.01***
-8
-0.16%
57.07%
-0.78 -0.68 0.08*
-7
0.26%
45.46%
1.28
1.34
0.18
-6
0.58%
48.49%
2.83*** 2.46** 0.06*
-5
1.33%
36.87%
6.46***
5.05***
0.00***
-4
0.71%
46.97%
3.45*** 3.08*** 0.02**
-3
0.19%
53.54%
0.94
0.83
0.81
-2 -0.51% 62.12%
-2.49* *
-2.36 ** 0.00***
-1
-1.01%
52.02%
-4.90* **
-1.93*
0.24
0 -1.90% 81.31%
-9.27* **
-2.96 ***
0.00***
1
-0.77%
63.64%
-3.76* ** -2.93*** 0.00***
2
-0.64%
60.10%
-3.09* **
-2.44 **
0.00***
3
0.06%
52.02%
0.31 0.26 0.77
4
-0.07%
55.05%
-0.35
-0.28
0.18
5
-0.25%
60.10%
-1.23 -0.97 0.02**
6
-0.19%
59.60%
-0.92
-0.78
0.02**
7
-0.66%
62.63%
-3.22 *** -2.86 *** 0.00***
8
-0.49%
67.17%
-2.40 **
-2.06 **
0.00***
9 -0.05% 55.05% -0. 24 -0.23 0.30
10
-0.49%
61.11%
-2.39 **
-2.15 **
0.00***
Table 4: Cumulative Abnormal Returns Around the
Execution in case 2
Event
date:
CAR :
CAR %:
t(BW):
simple-t:
p-value
(Wt):
-0 to 0 -1.90% 81.31% -9.27*** -2.96 *** 0.00***
-1 to 1
-3.68%
-10.35 ***
-4.16 ***
0.00***
-2 to 2
-4.83%
-10.51 ***
-4.97 ***
0.00***
-3 to 3
-4.57%
-8.41 ***
-4.48 ***
0.00***
-4 to 4
-3.94%
-6.38 ***
-3.76 ***
0.00***
-5 to 5
-2.86%
-4.20 ***
-2.56**
0.00***
-6 to 6
-2.47%
-3.33 ***
-2.11**
0.01***
-7 to 7
-2.86%
-3.60 ***
-2.42 **
0.01***
-8 to 8
-3.52%
-4.15 ***
-2.93 ***
0.00***
-9 to 9
-3.87%
-4.32 ***
-3.14 ***
0.00***
-10 to 10
-4.54%
-4.82 ***
-3.53 ***
0.00***
Figure 2: Cumulative Abnormal Returns Around the
Execution in case 2
significant negative abnormal returns 1to 2 days before
and after the ex-day. Table 4 shows that 81.31% of
45
X. X. YANG
Copyright © 2013 S ciRes. TI
stocks has abnormal returns on the ex-day, and this phe-
nomenon is significant indicated by simple-t and t(BW)
test. Figure 2 shows the most obvious accumulative neg-
ative returns appear 2 days before and after the ex-day.
Since all statistical tests indicate significant AR and CAR
around the ex-day, the execution event does have the
information content. Generally speaking, there would
be a period of time between the announcement date to
the ex-day, so the implied information effect could not
last to the ex-day due to the market efficiency. Yet, ab-
normal price behavior has been found surrounding the
ex-day according to the empirical studies of some foreign
scholars. Such as, Eades, Hess and Kim(1984) find sta-
tistically significant non-zero positive abnormal returns
from day -4 to +3(ex-day is 0 ), based on the study of
1550 ex-right events in New York Stock Exchange be-
tween 1962 to 1980. Woolridge (1983) find the abnormal
returns 9 days before the ex-day differs by almost 4%.
Since the empirical study shows that an abnormal return
of 7.82% appears in the month and previous two months
of the execution, LI Cun-xiu (1990) thinks that the
ex-dividend event does convey a message, namely a
company will tell its investors the change of its future
cash flow by stock dividends. On the other hand, inves-
tors will also infer a company’s information according to
its published dividend rate. Therefore, the signaling ef-
fect of ex-dividend event could appears before or after
the execution. But, the results of this paper show that in
China the signaling effect appears before the ex-day, this
may be due to the ex-dividend news has been disclosed
early, so the market reacts early. I think the expectation
psychology of Chinese investors can perfectly explain
this phenomenon. Specifically, buying before the ex-day
and the behavior of chasing the stock price lead to the
significant positive abnormal returns before the ex-day.
Then at the end of the information effect, stocks have
been sold, and that is why negative abnormal returns
appear after the execution. This process shows that most
Chinese investors tend to short-term speculation, rather
than long-term investment, and they barely consider the
asymmetric information as well. Since the ex-dividend
event is always a bullish factor, it has been in the lime-
light for a long time in China’s equity market. It is note-
worthy that the significant negative abnormal returns
appear before the ex-day, documented in relevant studies
of other countries, can be explained by the tax burden
effect proposed by Elton and Gruber (1970). This effect
states that the higher the tax rate, the higher the abnormal
return rate would be needed for investors to involve in
the ex-dividend. If the tax rate exceeds the market aver-
age, since the abnormal returns will be insufficient to
make up for the tax burden, those investors with high tax
rate are willing to sell stock before the ex-day, which is
the so-called abstention. In general, since investors with
high tax rate who are holding more stocks are mostly the
abstainers, the pressure of selling is greater than that of
buying. Therefore negative abnormal returns appear be-
fore the execution. However, similar to the study of
TIAN Jian-zhong(2007), which indicates that the tax
effect is not significant in China, I find significant posi-
tive abnormal returns 4 to 6 days before the ex-day. As a
result, the expectation psychology is applicable to reac-
tion on the execution in China’s securities market, since
Chinese investors will actively participate in the
ex-dividend, other than absten tion.
3.2. Using market-adjusted return mod-
el(
, ,,itmt it
RR
ε
= +
)6
CASE.3 . Market reaction to the announcement in the
event window [-10,10].(average daily abnormal returns,
average cumulative abnormal returns and their signific-
ance)
Table 5 and 6 and Figure 3 present the abnormal returns
and the cumulative abnormal returns around the an-
nouncement in case 3.
Table 5: Abnormal Returns Around the Announce-
ment in case 3
and sample with no cash
dividend to study the sensitivity of the results to the
method.
Event date: AR : negative
AR %: Simple-t: p-value
(Wt):
-10 0.53% 45.45% 2.26*** 0.08*
-9 0.50% 48.99% 2.54** 0.11
-8 0.40% 52.02% 1.86* 0.33
-7 0.27% 50.51% 1.20 0.59
-6 0.49% 49.49% 2.52** 0.13
-5 0.68% 45.45% 3.03*** 0.03**
-4 0.40% 47.47% 1.81* 0.26
-3 0.39% 49.49% 1.79* 0.18
-2 0.62% 41.92% 2.97*** 0.00***
-1 1.08% 41.41% 4.12*** 0.00***
0 0.87% 43.94% 2.39** 0.05**
1 0.33% 53.54% 1.20 0.94
2 0.44% 50.00% 2.16** 0.11
3 0.16% 56.06% 0.88 0.86
4 0.08% 52.02% 0.42 0.61
5 -0.07% 52.02% -0.38 0.59
6 -0.13% 55.56% -0.63 0.10*
7 0.17% 45.96% 0.84 0.52
8 0.34% 47.98% 1.73* 0.21
9 0.29% 49.49% 1.45 0.57
10 0.05% 52. 02% 0.23 0.58
Table 6: Cumulative Abnormal Returns Around the
Announcement in case 3
6 Wugle, J. K. Zhuravskaya, “Does Arbitrage Flatten De-
mand Curves for Stocks”, Journal of Business, 2002.
46
X. X. YANG
Copyright © 2013 S ciRes. TI
Event
date:
CAR :
negative
CAR %:
Simple-t:
-0 to 0
0.87%
43.94%
2.39** 0.05**
-1 to 1
2.28%
42.42%
3.81***
-2 to 2
3.35%
37.88%
4.92*** 0.00***
-3 to 3
3.90%
33.84%
5.33***
-4 to 4
4.38%
32.32%
5.51*** 0.00***
-5 to 5
4.99%
33.33%
6.01***
-6 to 6
5.35%
33.84%
6.09*** 0.00***
-7 to 7
5.79%
31.82%
6.41*** 0.00***
-8 to 8
6.54%
31.82%
6.84*** 0.00***
-9 to 9
7.32%
24.75%
7.34*** 0.00***
-10 to 10
7.90%
29.80%
7.45*** 0.00***
Figure3: Cumulative Abnormal Returns Around the
Announcement in case 3
Table 5 and 6 and Figure 3 show more pronounced posi-
tive abnormal returns both in extent and significance
around the announcement compared to the results in case
1. But at the same time, I find abnormal returns a little
far before the announcement, which are not what i ex-
pected. Therefore, the market return model is more ap-
plicable to the study on the market reaction to stock splits
around the announcement date.
CASE.4 . Market reaction to the execution in the event
window [-10,10].(average daily abnormal returns, aver-
age cumulative abnormal returns and their significance)
Table 7 and 8 and Figure 4 present the abnormal returns
and the cumulative abnormal returns around the execu-
tion in case 4.
Table 7 and 8 and Figure 4 show significant negative
abnormal returns around the ex-day, and significant posi-
tive abnormal returns 4 to 7 days before the execution
date. We can see the overall conclusion does not change,
but it is more pronounced both in extent and significance
than the results in case 2. So, both models work well on
this study.
3.3. Analysis on Sensitivity of the Results to the
Sample Data
CASE.5. From the sample period standpoint, namely
Table 7: Abnormal Returns Around the Execution in
case 4
Event
date:
AR :
negative
AR %:
Simple-t:
p-value( Wt):
-10
-0.04%
48.99%
-0.15
0.73
-9
-0.09%
53.54%
-0.39 0.13
-8
0.12%
50.51%
0.51
0.76
-7
0.46%
43.94%
2.36** 0.02**
-6
0.74%
45.45%
3.13***
0.01**
-5
1.57%
34.34%
6.16*** 0.00***
-4
0.88%
43.94%
3.83***
0.00***
-3
0.39%
50.51%
1.69 0.49
-2
-0.26%
60.61%
-1.22 0.03**
-1
-0.82%
50.51%
1.56*
0.89
0
-1.74%
77.27%
-2.69* ** 0.00***
1
-0.56%
62.12%
-2.19* *
0.00***
2
-0.38%
54.55%
-1.48 0.02**
3
0.26%
51.01%
1.00
0.61
4
0.11%
56.06%
0.42 0.47
5
-0.03%
57.07%
-0.13
0.15
6 -0.02% 56.06% -0.09 0.12
7
-0.46%
61.62%
-2.08 ** 0.00***
8
-0.31%
63.64%
-1.33
0.01***
9
0.14%
53.03%
0.67 0.77
10
-0.27%
61.11%
-1.20 0.03**
Table 8: Cumulative Abnormal Returns Around the
Execution in case 4
Event
date:
CAR :
negative
CAR %:
Simple-t:
p-value( Wt):
-0 to 0
-1.74%
77.27%
-2.69 ***
0.00***
-1 to 1
-3.11%
71.21%
-3.50 *** 0.00***
-2 to 2
-3.75%
66.16%
-3.90 ***
0.00***
-3 to 3 -3.09% 60.61% -3.11*** 0.00***
-4 to 4
-2.10%
61.62%
-2.06 **
0.01***
-5 to 5
-0.56%
53.03%
-0.52 0.46
-6 to 6
0.15%
50.51%
0.13
0.96
-7 to 7
0.15%
53.03%
0.14 0.94
-8 to 8
-0.04%
52.02%
-0.04
0.80
-9 to 9
0.01%
51.52%
0.01 0.53
-10 to
10
-0.30%
53.54%
-0.25 0.33
Figure 4: Cumulative Abnormal Returns Around the
Execution in case 4
47
X. X. YANG
Copyright © 2013 S ciRes. TI
according to annual classification to study the abnormal
returns around the announcement and the execution date
by using the market return model and sample with no
cash dividend.
Figure 5: Cumulative Abnormal Returns Around the
Announcement in case 5 (classified in year)
-10 -9 -8-7 -6-5-4 -3-2 -10 1 23 4 5 67 8 910
-4
-3
-2
-1
0
1
2
3
4
5
6
7
8
9
10
11
12
13
14
CAR in %
Day--(Announcement Day)
% (2007)
% (2008)
% (2009)
% (2010)
Figure 6: Cumulative Abnormal Returns Around the
Execution in case 5 (classified in year)
-10 -9 -8 -7 -6 -5-4 -3 -2 -101 2 34 56 7 8910
-10
-8
-6
-4
-2
0
2
4
6
8
10
CAR in %
Day--(Execution Day)
% (2007)
% (2008)
% (2009)
% (2010)
From the comparison of Figure 1 and Figure 5, Figure 2
and Figure 6, we can find that the latter is actually the
weighted average number of the former according to the
annual event number. Specifically, Figure 5 shows posi-
tive abnormal returns 2 days before and after the an-
nouncement date in 2007 and 2008, but this phenomenon
is almost unobvious in 2009 and 2010. Figure 6 shows
obvious negative abnormal returns 2 days before and
after the execution date no matter in which year. At the
same time, it is obvious to see that the abnormal return 5
days before the ex-day mentioned in case 2 is mainly
made up of data in 2007.
Through the discussion of case 5, I find different results
in different years. This also reminds me the stability over
time should be considered cautiously when building a
trading strategy later.
CASE.6 . Analysis on sensitivity of the abnormal returns
around the announcement and the execution date to cash
dividends by using the market return model.
CASE.6.1. Abnormal returns around the announcement
date
Table 9 and 10 and Figure 7 present the abnormal returns
and the cumulative abnormal returns around the an-
nouncement in case 6.1.
Table 9: Abnormal Returns Around the Announce-
ment in case.6.1.
Event
date: AR :
negative
AR %: t(BW): simple-t:
p-value( W
t):
-10 0.13% 54.44% 1.25 1.17 0.35
-9 0.10% 54.58% 0.97 0.95 0.36
-8 0.03% 54.16% 0.33 0.32 0.18
-7 0.10% 50.64% 1.01 1.02 0.81
-6 0.34% 48.94% 3.35*** 3.15*** 0.06*
-5 0.36% 49.22% 3.55*** 3.11*** 0.03**
-4 0.61% 45.70% 6.01*** 5.19*** 0.00***
-3 0.40% 47.11% 3.96*** 3.30*** 0.01***
-2 0.61% 45.13% 6.04*** 4.90*** 0.00***
-1 0.88% 43.87% 8.64*** 6.46*** 0.00***
0 0.15% 52.75% 1.52 0.98 0.72
1 -0.25% 58.96 % -2.48 ** -2.08 ** 0.00***
2 -0.08% 56.14 % -0.83 -0.73 0.02**
3 -0.12% 56.14 % -1.14 -1.09 0.02**
4 -0.16% 56.84 % -1.54 -1.58 0.00***
5 -0.17% 55.43 % -1.65 -1.59 0.01***
6 -0.06% 55.99 % -0.60 -0.56 0.03**
7 0.04% 51.20% 0.42 0.40 0.51
8 -0.23% 55.99 % -2.23 ** -2.28 ** 0.00***
9 -0.18% 55.15 % -1.81* -1.76* 0.02**
10 -0.24% 57.26% -2.33** -2.23** 0.00***
Table 10: Cumulative Abnormal Returns Around the
Announcement in case 6.1.
Event
date:
CAR :
negative
CAR %:
t(BW):
simple-t:
p-value
(Wt):
-0 to 0 0.16% 52. 83% 1.52 0.98 0.71
-1 to 1 0.78% 47. 03% 4.44*** 3.16*** 0.02**
-2 to 2 1.31% 42. 51% 5.78*** 4.21*** 0.00***
-3 to 3 1.60% 46. 89% 5.59*** 4.38*** 0.00***
-4 to 4 2.06% 47. 18% 6.73*** 5.02*** 0.00***
-5 to 5 2.25% 44. 92% 6.66*** 5.00*** 0.00***
-6 to 6 2.53% 45. 34% 6.89*** 5.17*** 0.00***
-7 to 7 2.67% 44.92% 6.79*** 5.09*** 0.00***
-8 to 8 2.48% 46. 47% 5.91*** 4.54*** 0.00***
-9 to 9 2.39% 45. 20% 5.40*** 4.14*** 0.00***
-10 to
10 2.28% 46.05 % 4.90*** 3.83*** 0.00***
48
X. X. YANG
Copyright © 2013 S ciRes. TI
Figure 7: Cumulative Abnormal Returns Around the
Announcement in case 6.1
Compared to CASE 1, the similar shape of Figure 1 and
Figure 7 shows significant positive abnormal returns a
few days before the announcement date with or without
cash dividend. Further, from the comparison of Figure 9,
10 and Figure 1, 2, I find more significant positive ab-
normal returns a few days before the announcement date
with cash dividend. Therefore, cash dividend streng-
thened the market reaction of CASE 1. However, I ob-
serve completely insignificant abnormal returns on the
announcement date with cash dividend. It is possible that
news of cash dividend leaked out very fast. Knowing in
advance by internal people accelerate the uptake of in-
formation.
CASE.6.2. Abnormal returns around the execution date
Table 11 and 12 and Figure 8 present the abnormal re-
turns and the cumulative abnormal returns around the
execution in case 6.2.
Table 11: Abnormal Returns Around the Execution
in case.6.2.
Event
date:
AR :
negative
AR %:
t(BW):
simple-t:
p-value( Wt):
-10
-0.04%
53.03%
-0.38 -0. 35 0.35
-9
-0.08%
52.89%
-0.83 -0.77 0.12
-8
-0.07%
52.19%
-0.69 -0. 65 0.25
-7
-0.13%
55.71%
-1.25 -1. 10 0.02**
-6
0.57%
43.72%
5.62***
4.88*** 0.00***
-5
1.30%
37.80%
12.73***
9.10***
0.00***
-4
0.75%
44.85%
7.35***
5.59*** 0.00***
-3
0.17%
49.65%
1.67*
1.38
0.62
-2
0.22%
53.17%
2.12**
1.08
0.51
-1
-0.95%
56.70%
-9.34* **
-3.73 ***
0.00***
0
-2.73%
75.88%
-26.81 ***
-8.49 ***
0.00***
1
-1.29%
62.76%
-12.67 ***
-5.90 ***
0.00***
2
-0.54%
58.53%
-5.28* **
-3.18 ***
0.00***
3 -0.38% 60.79% -3.70*** -3.06*** 0.00***
4
-0.23%
57.69%
-2.25 ** -1.92* 0.00***
5
0.09%
53.03%
0.86 0.74 0.32
6
-0.30%
58.82%
-2.95 *** -2.54 ** 0.00***
7
-0.12%
54.02%
-1.23 -1. 07 0.06*
8
-0.20%
56.70%
-2.00 ** -1.85* 0.00***
9
-0.48%
63.19%
-4.68 *** -4.01 *** 0.00***
10
-0.48%
62.48%
-4.75 *** -4.15 *** 0.00***
Table 12: Cumulative Abnormal Returns Around the
Execution in case.6.2.
Event
date:
CAR:
negative
CAR %:
t(BW):
simple-t:
p-value
(Wt):
-0 to
0
-2.73%
75.99%
-26.85 ***
-8.49 ***
0.00***
-1 to
1
-4.97%
71.33%
-28.23 *** -10.50*** 0.00***
-2 to
2
-5.30%
69.21%
-23.28 ***
-9.49 ***
0.00***
-3 to
3
-5.50%
67.37%
-20.44 ***
-9.52 ***
0.00***
-4 to
4
-4.98% 65.40%
-16.33 *** -8. 17*** 0.00***
-5 to
5
-3.60%
59.75%
-10.67 ***
-5.66 ***
0.00***
-6 to
6
-3.33%
57.77%
-9.07 *** -5.14*** 0.00***
-7 to
7
-3.58%
59.75%
-9.08 *** -5.41*** 0.00***
-8 to
8
-3.85%
61.16%
-9.18 ***
-5.71 ***
0.00***
-9 to
9
-4.41%
61.16%
-9.95 *** -6.38*** 0.00***
-10 to
10
-4.94%
61.72%
-10.59 ***
-6.97 ***
0.00***
Figure 8: Cumulative Abnormal Returns Around the
Execution in case.6.2
Compared to CASE 2, the similar shape of Figure 2 and
Figure 8 shows significant negative abnormal returns two
days before and after the execution date with or without
cash dividend. Further, from the comparison of Figure 11,
12 and Figure 3, 4 I find, with cash dividend, significant
positive abnormal returns 5 to 7 days before the execu-
tion date and significant negative abnormal returns
around the ex-day. At the same time, I observe greatly
significant negative abnormal returns on the ex-day.
Therefore, cash dividend strengthened the market reac-
tion of CASE 2.
3.4. Conclusion for the Six Cases Above
The results of six cases above show significant positive
abnormal returns around the announcement date with or
without cash dividend using different return models. This
kind of significant positive abnormal return mostly ap-
pears 2 days before and on the announcement date,
which indicates that the effective market reaction to the
information, but the information is also likely to be
49
X. X. YANG
Copyright © 2013 S ciRes. TI
leaked ahead, so investors made action in advance. In
that case, the China’s stock market information disclo-
sure system is still not standardized. Specifically, the
news of dividend distribution had been let out a few days
before the announcement date, some informed investors
made action in advance, which not only did great harm to
the interests of other investors, but also violated the prin-
ciple of the security market(open, fair and just). There-
fore, the relevant departments should further standardize
the information disclosure system of listed companies.
The tests of different years show that although notable
positive abnormal return appears in every year, the am-
plitude of reaction varies considerably, especially the
most obvious abnormal return appears in 2007.
The market reaction around the ex-day shows significant
positive abnormal returns 5 days before the ex-day and
significant negative abnormal returns around the ex-day
with or without cash dividends using different return
models. The tests of different years show that both posi-
tive and negative abnormal returns are very notable in
2007, but after that the significance reduced. This phe-
nomenon can be explained by the rising China stock
market in 2007.
3.5. Factor Analysis on the Announcement Effec t
In this part, I mainly analyze the influence of the split
ratio, cash dividends, firm characteristics and the overall
market condition along with other factors over the an-
nouncement effect. Here the regression model I used is
12
34
**
**
**
...*
AABB
MM
CARRatio Cash
MarCap Market
CSRC CSRC
CSRC
δδ
δδ
δδ
δε
= +
++
++
++ +
In which, the CAR refers to the accumulated abnormal
returns around the announcement date, which is the
summation of a total 11 days of abnormal returns in the
event window[-5, 5] of the announcement date using the
market return model. Ratio corresponds to the split ratio.
Cash refers to cash dividends, which I am using here as a
dummy variable, i.e. 1 stands for with cash dividend, 0
means without cash dividend. MarCap refers to the
company size (unit: one billion), which is an estimation
of the average daily circulated A share market value in
the estimation window[-110, -11]. Market refers to the
overall market situation, which is the summation of daily
stock market returns in the estimation window[-110, -11]
before the announcement. CSRC refers to the industry
classi fication. In order to study the influence of industry
over the abnormal returns, here I run regressions for in-
dustry classification7
7 According to the ”China listed company classification guid-
ance”, the 13 categories of listed companies are: A, farming,
separated as dummy variables
from CSRC A to CSRC M.Table 13 and 14 respectively
represent the regression analysis results of the equation
and tests of the regression parameters and its signific-
ance.
Table 13: Regression Analysis Results of the Equation
Source DF
Sum of
Squares
Mean
Square
F Value Pr>F
Model 17 0.8484 0.0499 3.3600
<.000 1
Error 330 4.9070 0.0149
Uncorrelated
Total 347 5.7555
Root MSE 0.12194
R-Square 0.1474
Dependent
Mean
0.01844 Adj R-Sq 0.1035
Coeff Var 661 .42046
Table 14: Parameter Estimation
Va r i ab l e Regressor Estimate
t-Va l u e Pr> |t |
δ1 Ratio 0.0914 0.0214 4.2 700
<.000 1
δ2 Cash 0.0129 0.0157 0.8200 0.4107
δ3
MarCap
(Billion) -0.0003 0.0005 -0.5900 0 .5558
δ4 Mark e t 0.0591 0.0210 2.8100 0.0052
δA CSRC_A 0.0205 0.0717 0.2900 0.7754
δB CSRC_B -0.016 3 0.0416 -0.3900 0.6958
δc CSRC_C -0 .0541 0.0177 -3.0600 0.0024
δD CSRC_D -0.1849 0.0627 -2.9500 0.0 034
δE CSRC_E 0.0427 0.0395 1.0800 0.2805
δF CSRC_F -0.0328 0.0298 -1.1000 0 .2725
δG CSRC_G -0.0459 0.0290 -1.5800 0.1152
δH CSRC_H -0. 0353 0.0347 -1.0200 0.3087
δI CSRC_I -0.015 4 0.0751 -0.2000 0.8379
δJ CSRC_J -0.0224 0.0289 -0.780 0 0.4379
δK CSRC_K -0.0619 0.0509 -1.2200 0.2252
δL CSRC_L -0.030 3 0.0884 -0.3400 0.7318
δM CSRC_M 0.0208 0.0304 0.6800 0.4 940
Table 14 shows that the higher the split ratio, the greater
the abnormal returns. This can be explained that high
split ratio shows the confidence of a company, and posi-
tive signal has been transmitted to the market.
2
δ
shows that the sensitivity of abnormal returns to cash
dividend is insignificant, namely under the premise of
forest, herd, fishery; B, mining; C, manufacturing; D, electricity,
gas and water production and supply; E, construction; F, trans-
portation, warehousing; G, information technology; H, whole-
sale and retail trade; I, finance, insurance; J, real estate; K, so-
cial services; M, comprehensive category.
50
X. X. YANG
Copyright © 2013 S ciRes. TI
stock splits, cash dividends do not affect abnormal re-
turns notably.
3
δ
shows that the sensitivity of abnormal
returns to firm size is not significant, which means that
the company scale do not affect abnormal returns directly.
Significant positive
4
δ
shows that the better the overall
market condition, like in the bull market, the more posi-
tive abnormal returns. From A
δ
to
M
δ
I find only the
intercepts of the C industry(manufacturing) and the D
industry(electricity, gas and water production and supply)
are significant negative at the 0.05 confidence level,
which means that few of industry itself has a stable ab-
normal returns. Namely, the difference of the response of
abnormal returns around the announcement between dif-
ferent industry is not obvious.
3.6. Investment Strategy Analysis
3.6 1. The Basic Train of Tho ught
The prior empirical research shows significant positive
abnormal returns around the announcement date and sig-
nificant negative abnormal returns around the ex-day.
Therefore, I will build an investment strategy according
to these two phenomenon.
It is more difficult to construct investment strategy
around the announcement date since in the normal cir-
cumstances investors have no internal information,
namely, it is impossible for them to know information of
stock splits in advance. Especially in China, the empiri-
cal results show that most of the positive abnormal re-
turns appear a few days before the announcement date.
Namely, informed investors have done a lot of trade in
advance. As a result, it is hard for a investor with no pri-
vate information to build an investment strategy in this
market.
Considering market reaction around the ex-day, there are
two cases. The first one is significant positive abnormal
returns 4 to 6 days before the ex-day. This is a very good
opportunity for investment. Yet, Figure 6 shows this op-
portunity is the most obvious in 2007, slightly notable in
2008, does not exist in and after 2009. Form today’s
viewpoint, this is most likely due to the overall market
condition or the whole institutional investors use this
strategy. The second case is negative abnormal returns 2
days before and after the ex-day. Figure 6 shows this
phenomenon has been relatively stable from 2007 to
2010. Therefore, a reasonable investment strategy is that
selling short a few days before the ex-day and buying
back a few days after the ex-day. This kind of strategy is
reasonable because of the following two reasons, al-
though short selling is forbidden in China. On the one
hand, the CSRC8
8 China Securities Regulatory Commission.
issued the “The Controls of Experi-
mental Unit of Securities Margin Trading ” in June 30,
2006(effective in August 1, 2006). Then the CSRC an-
nounced the launch of the experimental unit in October
5, 2008. In March, 2010, the CSRC opened partial short
selling. So we have reason to believe that the range of
short selling will be more wide in the next few years. On
the other hand, for fund managers, they may have these
stocks in their own portfolio. They just need to sell them
out a few days before the ex-day and buy them back after
the ex-day.
3.6.1. Investment analysis of risk and revenue
(1) Model construction
Here I construct the following model: selling short x days
before the ex-day and buying back y days after the
ex-day. Assume
0 ,10xy
. I use the equal weighted
investment strategy for convenience, so I can directly
sum stock returns up arithmetically. At the same time,
cash dividends should be considered. Since abnormal
returns are sensitive to the setting of parameters and as-
sumptions, I use the absolute returns of stocks instead of
the abnormal returns. I use the Sharp Ratio9
( )0riskfree
µ
=
, which
means the excess return for every unit of risk, as the
evaluation standard to study the risk and return of this
investment strategy. The specific formula is SR[E(Rp)
Rf]/σp. Here I apply the investment strategy above to
calculate the return around the ex-day[-x, y] given sam-
ple in 2007 to 2010, then I try to find the optimal invest-
ment strategy by comparing Sharp Ratio at different x
and y. In order to calculate simply, I assume the risk-free
return is 0, i.e. .
(2) Data analysis and discussion
Table 15 and 16 present the Sharp index value at differ-
ent x and y with and without cash dividend respectivel y.
Table 15: Sharp Index Value with No Cash Dividend
X/Y 0 1 2 3 4 5 6 7 8 9 10
0
2.77
1.96
1.39
1.07
0.09
0.91
0.84
0.84
0.85
0.89
0.83
1
2.28
1.48
1.25
1.04
0.94
0.96
0.91
0.92
0.91
0.93
0.85
2 2.18 1.47 1.29
1.09
1.01 1.04 1.00 1.00
0.98
0.99 0.90
3 1.26
1.09 1.08
0.95 0.92
0.97
0.95
0.98 0.96
0.95 0.85
4
0.59
0.66
0.74
0.68
0.68
0.73
0.72
0.79
0.79
0.77
0.68
5
0.01
0.17
0.27
0.20
0.21
0.27
0.27
0.36
0.41
0.40
0.36
6 -0.16
0.00 0.10
0.03
0.03
0.08 0.08
0.17
0.23
0.23 0.21
7
-0.28
-0.12
-0.03
-0.10
-0.11
-0.06
-0.06
0.02
0.10
0.10
0.10
8
-0.30
-0.16
-0.08
-0.15
-0.16
-0.11
-0.12
-0.04
0.04
0.04
0.05
9 -0.24
-0.10
-0.02
-0.09 -0.09
-0.05 -0.05 0.02
0.08 0.09 0.09
9 In 1990, the Nobel Prize winner William Sharpe starting
from CAPM(capital asset pricing model ) developed Sharp
Ratio, used to measure the performance of financial assets.
Sharpe, W. F. (1966). "Mutual Fund Performance". Journal of
Business 39 (S1): 119138.
51
X. X. YANG
Copyright © 2013 S ciRes. TI
10 -0.26
-0.14
-0.07 -0.12 -0.13
-0.09 -0.10
-0.03 0.03
0.03 0.04
Table 16: Sharp Index Value with Cash Dividend
X/Y 0 1 2 3 4 5 6 7 8 9 10
0 4.44 2.16 1.71
1.79
1.84
1.60 1.61
1.42 1.15
1.15
1.20
1 4.08
2.28
1.84
1.96 1.94
1.67
1.66
1.46
1.22
1.21
1.26
2 5.93
2.16 1.72
1.86
1.88
1.60
1.61
1.41
1.14
1.14
1.21
3 2.11 1.34
1.19
1.33
1.34
1.13 1.14
1.01 0.83
0.86
0.94
4 0.54
0.63
0.63
0.73
0.74
0.60
0.62 0.54
0.44
0.49
0.56
5 -0.18
0.10 0.18 0.24 0.25 0.15 0.17
0.13
0.08
0.14
0.20
6 -0.42
-0.12
-0.03 0.01 0.02
-0.05
-0.03
-0.06
-0.09 -0.03
0.02
7 -0.34
-0.09
-0.01
0.02
0.03
-0.03
-0.02
-0.04
-0.07 -0.02
0.03
8 -0.33
-0 .11 -0. 04 -0.01
0.00 -0.06
-0.04
-0.06 -0.09
-0.04 0.00
9 -0.34 -0.14 -0.07 -0.04
-0.03
-0.08
-0.07
-0.09 -0.11
-0.06
-0.02
10 0.51
0.55
0.57
0.57 0.57 0.56
0.57 0.56 0.56 0.57 0.57
Table 15 shows that with no cash dividend, i can get the
highest Sharp Index by selling short on the ex-day and
buying back after the ex-day. Table 16 shows that with
cash dividend, i can get the highest Sharp Index by sell-
ing short 2 days before the ex-day and buying them back
on the ex-day. On the whole, I can get higher Sharp In-
dex by using this investment strategy around the ex-day
especially with cash dividends, and this kind of trading
strategy can bring relatively stable and high yield.
In order to further understand annual earnings of this
specific trading strategy, I select the sample whose Sharp
Ratio ≥2.5 to calculate the annual earnings. Table 17
presents the results.
Table 17: Annual Rate of Earnings of this Trading
Strategy
2007
2008
2009
2010
mean
std
Sharpe
Ratio
[-0,0]No
Cash
1.51%
1.22%
1.77%
2.72%
1.81%
0.65%
2.77
[-0,0]With
Cash
3.25%
2.76%
1.83%
2.89%
2.68%
0.60%
4.44
[-1,0]With
Cash
3.45%
4.20%
2.23%
3.54%
3.35%
0.82%
4.08
[-2,0]With
Cash
2.74%
3.25%
2.16%
3.02%
2.79%
0.47%
5.93
Table 17 shows that the annual rates of earnings of these
four trading strategies are relatively stable. Among them,
the third strategy gets the highest yield, but at the same
time it associates with the highest risk. Therefore, the
fourth strategy is a better trading strategy since it has a
highest Sharp Ratio.
One of the advantages of this strategy is that capital can
be used repeatedly, which can increase the leverage ratio.
Because the execution distributes over a period of time
instead of focusing on one day, the short margin required
is greatly reduced in this period.
Thi s strategy also associates with the following two risks:
one is the instability over time. Although this strategy
can bring profits to investors from 2007 to 2010, there is
no guarantee that it will work in the future, and the Sharp
Ration will experience a significant slowdown as more
investors adopt this strategy. Secondly, since the strategy
is short selling or closing, the rising market in the cor-
responding period will bring risks. The effect of rising
market over the negative abnormal returns around the
announcement will lead to loss of this strategy. However,
this problem can be solved in two ways. First of all, the
execution distributes along the time line evenly rather
than focusing on one day, and cash will be allocated
equally in different events. This will relieve the rising
market problem. Secondly, this strategy can hedge part
of the risk if it is used by fund managers. Specifically,
they can make money rely on their main position when
the stock market is rising, and they can get more money
in the declined market based on this strategy. In fact, in
this case, the main position of fund managers has an im-
pact of hedging. Of course, it is possible for individual
investors to sell short and buy the index or index futures
to hedge the market risk. But the imperfections and de-
fects of the China capital market system will bring some
difficulties to individual investors.
4. CONCLUSION AND PROSPECT
This paper investigates the market reaction to stock splits
based on China’s A share companies between 2007 to
2010 by using empirical analysis. I find significant posi-
tive abnormal returns around the announcement
date(especially before the announcement date) as well as
four to six days before the ex-right date of China stock
splits. I also observe significant negative abnormal re-
turns just around the ex-right date. The above phenome-
non is relatively stable even if the selection of samples
and empirical models may vary, but the degree of this
phenomenon might change overtime. The cross sectional
regression of the abnormal returns for the announcement
date shows that the phenomenon is sensitive to the split
ratio and the overall market condition, and it is not sensi-
tive to industry, company size and cash dividends.
Therefore, combining with the empirical data I have con-
structed a high Sharpe ratio short selling investment
strategy around the ex-right date. Then, the article further
discusses the operability of the investment strategy and
its stability over time.
The empirical results of this paper with Chinese cha-
racteristics are different from the United States market
and results of Christian Wulff (2002). This is most likely
associated with the one way market structure of no short
selling and the vulnerable internal message. This paper
not only put forward a feasible investment strategy for
the abnormal return phenomenon, but also explore the
underlying reason behind the abnormal returns around
the announcement and execution day. In conclusion, this
paper finds the direction for the future research.
52
X. X. YANG
Copyright © 2013 S ciRes. TI
REFERENCES
[1] .Paul Asquith, Paul Healy, Krishna Palepu. (1989). Earn-
ings and Stock Splits[J]. The Accounting Review, Vol.64,
No. 3, 387403.
[2] BakerH.Kent, PowellE.Gary. (1993). Further evidence on
managerial motives for stock splits[J]. Quaterly Journal of
Business and Economics, Vol. 32, No. 3, 2031.
[3] Chris J. Muscarella, Michael R. Vetsuypens. (1996).
Stock splits: Signaling or liquidity? The case of
ADR`solosplits'[J]. Journal of Financial Economics,
Volume 42, Issue 1, 326.
[4] Christian Wulff.(2002).The Market Reaction To Stock
Splits- Evidence From Germany[J]. Schmalenbach Busi-
ness Review (sbr), LMU Munich School of Management,
vol. 54(3), 270297.
[5] Eisemann P,M.Moses. Stock Dividends:Management’s
View[J]. Financial Analysis Journal,1978.
[6] Eades D ,P.Hess, and H. Kim. On Interpreting Security
Returns during the Ex- dividend period[J]. Journal of Fi-
nancial Economics,1984.
[7] Woolridge R. Ex- date Stock Price Adjustment to Stock
Dividends[J]. A Note. Journal of Finance,1983.
[8] Eades D,P. Hess, and H. Kim .On Interpreting Security
Returns during the Ex- dividend period[J].Journal of Fi-
nancial Economics,1984.
[9] XUE Zu-yun, LIU Wan-li. 2009. An empirical study of
motivations behind stock dividends in China’s listed
companies[J]. Journal of Xiamen University(Arts & So-
cial Sciences), Vol. 5, pp. 114-121.
[10] ZHANG Shui-quan, HAN Zong-de.1997. An empirical
study on the effect of dividend and bonus in Shanghai
stock market. Forecasting. Vol. 2, pp. 28-33.
[11] LIU Wan-li, XUE Zu-yun. 2010. The effect of stock div-
idend on shareholders’ value: evidences from Chinese
listed companies. The Theory And Practice Of Finance
and Economics, Vol. 3, pp. 53-57.
[12] WEI Gang. 1998. An empirical study on dividend distri-
bution of China listed companies. Economics Research
Journal. Vol. 6.
[13] YUAN Hong-qi. 2001. Analysis on the dividend policy of
listed companies in China. The Study of Finance and
Economics. Vol. 3.
[14] YANG Shu-e, WANG Yong, BAI Ge-ping. 2000. An
empirical analysis on the factors that influence the divi-
dend policy. Accounting Research. Vol. 2.
[15] CHEN Guo-hui, ZHAO Chun-guang. 2000. Positive
study on the dividends plicy option for enterprises in the
stock market. Research On Financial and Economic Is-
sues. Vol. 5.
[16] WEI Gang. 2000. An empirical study on stock dividends
of China listed companies. Securities Market Herald. Vol.
11.
[17] LI Kun, SONG Ting-ting. 2005. The effects of stock div-
idends on shareholder structure and the stock liquidity.
Statistics and Decision. Vol. 18.
[18] HE Tao, CHEN Xiao-yue. 2003. A discussion on the
motivation of stock dividend and transfer of reserve to
common shares of china’s listed companies. Journal of
Finance. Vol. 9.
[19] KONG Xiao-wen, YU Xiao-kun. 2003. An empirical
analysis on the signaling effect of listed companies’ divi-
dend policy. Management World. Vol. 6.
[20] ZHU Yun, WU Wen-feng, WU Chong-feng. 2004. Re-
search on the information content of China listed compa-
nies’ stock dividends. Chinese Journal of Management
Science Symposium. Pp. 218-222.
53