Content uploaded by Graeme Rankine
Author content
All content in this area was uploaded by Graeme Rankine on Aug 28, 2020
Content may be subject to copyright.
Do Stock Splits Really Signal?
Tak Yan Leungb, Oliver Meng Ruic and Steven Shuye Wanga*
This Version: May 4, 2006
a* Corresponding author: Steven S. Wang, School of Accounting and Finance, the Hong Kong
Polytechnic University, Hung Hom, Kowloon, Hong Kong.
Tel.: +852-2766-4952; Fax: +852-2330-9845; E-mail: afwang@inet.polyu.edu.hk
b Department of Accountancy, the City University of Hong Kong.
c Faculty of Business Administration, the Chinese University of Hong Kong.
The authors acknowledge financial support from the Hong Kong Polytechnic University
(Research grant G-T619).
Do Stock Splits Really Signal?
Abstract
Although stock splits seem to be purely cosmetic, there is ample empirical evidence that they are
associated with abnormal returns. This study analyzes the effect of stock splits using intraday
data and insider trading data in Hong Kong from 1980 to 2000. Consistent with the findings of
other countries, we observe positive price reactions in Hong Kong. These positive reactions can
be attributable to improved liquidity and favorable signals. Our microstructural analysis shows
that stock splits improve corporate liquidity. Regression analysis shows the presence of a
possible signaling role for split announcements confounded by increased liquidity. We further
use the abnormal insider trading activity to assess the informativeness of the split signal. We
find abnormally high insider trading activities three to four months before the split
announcement and in the post-announcement period, however, insider trading activities in the
two months immediately before the split announcement is immaterial. Our results suggest that
firms use stock splits to signal in order to increase liquidity.
Keywords: Stock splits, signaling, liquidity, insider trading.
JEL Classifications: G12, G15, G32
1
1. Introduction
Theoretically, stock splits should be cosmetic corporate events as they merely involve the
break up of one share into a certain number of shares and a reduction of a higher to a lower per-
share trading price without changing shareholders’ wealth and relative shareholdings. However,
previous studies provide mixed empirical results. For the widely explored U.S. market, although
early empirical studies find no abnormal performance after stock splits (Fama, Fisher, Jensen and
Roll, 1969), most recent studies (to name a few, Grinblatt, Masulis and Titman, 1984;
McNichols and Dravid, 1990; Maloney and Mulherin, 1992; Ikenberry, Rankine and Stice, 1996;
Ikenberry and Ramnath, 2002; and Byun and Rozeff, 2003) find a positively significant market
reaction to stock split announcements. Stock splits do not appear to be as cosmetic as they
should be. In the finance literature, the role of stock splits remains an enigma. On the other
hand, there are relatively less stock split studies for other countries: the examples include
Kryzanowski and Zhang (1993) for the Canadian market, and Kunz and Majhensek (2004) for
the Swiss market. The finance literature has proposed several hypotheses to explain the stock
split phenomenon, the most popular being the signaling hypothesis, the optimal trading range
hypothesis, and the liquidity hypothesis. However, previous studies provide mixed empirical
evidence for different markets and different sample periods.
As a determination of the robustness of these findings to other markets is of considerable
interest, using a long and recent sample period from 1980 to 2000, this paper studies the effect
of stock splits in Hong Kong with an attempt to differentiate the hypothesis that best explains
the stock split phenomenon. Specifically, our analyses explore stock splits in three respects:
share price performance, corporate liquidity, and insider trading activity. Firstly, we assess the
abnormal market reactions of the splitting firms. We use both the market model and the control
firm approach to measure the abnormal returns. Secondly, we compare the liquidity proxies in
2
terms of bid-ask spread, depth and trading volume in the pre-split and post-split periods. Our
third analysis involves using a prior-period comparison method to examine the insider trading
activity around the stock split announcement. Finally, we conduct a regression analysis to find
the hypothesis that best explains stock splits after controlling for insider trading and other
factors.
This paper contributes to the existing literature in several ways. It extends the
international empirical evidence on stock splits to another important stock market, the Hong
Kong stock market1. It also provides additional insight into the relative explanatory power of
the existing hypotheses. The analysis contributes further to the liquidity hypothesis using
microstructural data and the signaling hypothesis using insider trading data. Stock liquidity has
two inseparable dimensions: the price dimension and the size dimension. Previous studies
focus on the price dimension of liquidity and only give a partial view of liquidity. The
microstructural data in Hong Kong provides us with an opportunity to estimate both dimensions
of liquidity. We use the absolute and relative spreads as our measures of the price dimension,
and volume depth, dollar depth, ask depth, and bid depth as our measures of size dimension.
Using insider trading to study whether a stock split conveys a signal is appealing because we
can bypass a specification of performance benchmark.
We find significant and positive share price performance associated with split
announcements, which indicates that the splitting firms use stock splits to signal favorable
information to the market. The insider trading activity analysis reports that there are both
abnormal buying and selling activities before stock splits. Our microstructural analysis shows
that, in general, stock splits improve corporate liquidity. The post-split depth measures and
1 As of September 30, 2004, in terms of market capitalization, the Hong Kong Stock Exchange was among the ten
largest stock exchanges in the world, and was the second largest in Asia (World Federation of Exchanges:
http://www.fibv.com/publications).
3
trading volume (spread measures) are significantly higher (lower) than those in the pre-split
period. The enhancement of post-split corporate liquidity provides support to the liquidity
hypothesis. Finally, our regression analysis presents evidence for the signaling, optimal trading
range and liquidity hypotheses. As argued by Amihud and Mendelson (1988), the greater the
liquidity of an asset, the greater its value, firms may engage in liquidity-increasing policies to
mitigate the cost and risk of illiquidity. Therefore, firms use stock splits to signal in order to
increase liquidity. Our overall results suggest that stock splits matter. This seemingly non-
economic event serves multiple functions of signaling, realigning trading price and improving
liquidity. There is no single unique motivation for stock splits. Stock split performs a signaling
function of the firms’ liquidity-improvement policy.
This study is structured as follows. Section 2 presents the literature review and
theoretical proposition. The data and methodology are described in Section 3. Section 4 reports
the results, and Section 5 concludes the study.
2. Literature Review and Theoretical Proposition
Since the publication of the classic paper of Fama, Fisher, Jensen and Roll (1969) that
investigate the share price performance of splitting firms, many hypotheses have emerged and
empirical studies have been conducted to explain the puzzling market reaction to stock splits.
The more prominent hypotheses are the signaling hypothesis, the optimal trading range
hypothesis, the liquidity hypothesis, the tax option hypothesis, and the managerial entrenchment
hypothesis.
The signaling hypothesis argues that stock splits convey information about the current
performance and future prospects of the splitting firms (Grinblatt, Masulis and Titman, 1984).
To be a valid and credible signal, the signal has to be costly. Stock splits are costly signals
4
because the fixed component of the brokerage commission increases the post-split per-share
trading cost (such as odd-lot trading costs and administrative cost) of the lower priced shares
(Brennan and Copeland, 1988; Brennan and Hughes, 1991). The presence of positive abnormal
returns around the stock split announcement that is found in many empirical studies (e.g.,
Ikenberry, Rankine and Stice, 1996; Mukherji, Kim and Walker, 1997; Ikenberry and Ramnath,
2002) provides evidence for the signaling hypothesis.
According to the optimal trading range hypothesis, stock splits are used as tools to realign
the share price to a desired price range so that it is more affordable for small investors to buy
round lots of shares. If the pre-split share price is at a high level, then a stock split is justified for
improving the marketability of the shares (Baker and Gallagher, 1980; Lakonishok and Lev,
1987; McNichols and Dravid, 1990). The reduction in trading price through stock splits enables
the post-split shares to become more attractive to previously wealth constrained investors. In
addition, Angel (1997) argues that stock splits can be used to move the share price into the price
range where the institutionally mandated minimum absolute tick size is optimal relative to share
price.
Related to the optimal trading range hypothesis is the liquidity hypothesis. The liquidity
hypothesis is based on the argument that corporate liquidity is affected by the per-share trading
price (Maloney and Mulherin, 1992; Muscarella and Vetsuypens, 1996). If the trading price is
too high, then the liquidity may decline. A low per-share trading price attracts more individual
investors and reduces trading costs. However, the evidence for the liquidity hypothesis is mixed.
Conroy, Harris and Benet (1990) show an increase in bid-ask spreads after stock split
announcements. Ferris, Hwang and Sarin (1995) present results of a reduction in depth. Ohlson
and Penman (1985) and Koski (1995) report an increase in return volatility. These results
indicate that corporate liquidity decreases rather than increases after the split. In contrast,
5
Maloney and Mulherin (1992) and Desai, Nimalendran and Venkataraman (1998) observe an
increase in trading volume during the post-split period, and hence provide support for the
liquidity hypothesis of stock splits.
The tax-option hypothesis, proposed by Lamoureux and Poon (1987), suggests that stock
splits increase the return volatility of the splitting firms and hence allow the investors to benefit
from tax-timing options2. The managerial entrenchment hypothesis, put forward by Demsetz and
Lehn (1985), Morck, Shleifer and Vishny (1988), McConnell and Servaes (1990), and Kole
(1995), among others, argues that high shareholdings “entrench” non-wealth maximizing
behavior in management. Lakonishok and Lev (1987) apply the managerial entrenchment
hypothesis to explain stock splits. Managers make use of stock splits to enlarge the ownership
base so that the percentage of shares held by large institutional investors is reduced. In this way,
management makes it more difficult for any one group of shareholders to initiate action against
them. Mukherji, Kim and Walker (1997) find that the number of shareholders increases after a
stock split.
In this study, we focus on the impact of stock splits in several aspects. According to the
signaling hypothesis, there should be positive abnormal returns for splitting firms in our study.
By the optimal trading range hypothesis, if the stock split is a device that brings the stock price
to an acceptable level to attract more investors, particularly wealth constrained investors, then we
predict that the stock split announcements should receive favorable market reactions. We use the
event study methodology to measure the abnormal share price reaction of the splitting firms
around the stock split announcement. In addition, we examine the insider trading activity of the
directors before the stock split. If the stock split conveys an informative and favorable signal to
2 Preferential treatment is given to long-term capital gains according to the U.S. tax code. Short-term capital losses
can be used to offset short-term gains. A security with a price that fluctuates wildly presents its holder with the
6
the market, the directors may make use of their private information advantage to trade before the
news is publicly released. We expect significant insider purchases rather than insider sales
before the stock split. By the liquidity hypothesis, we expect that there should be significant
changes in the liquidity patterns (narrower spreads and wider depth) in the post-split period.3
3. Data and Methodology
3.1. Data
The stock split data are obtained from the Capital Distribution file of the PACAP
database. The PACAP database includes two types of companies, finance companies, and
industrial companies. Our analysis includes only the industrial companies4. The PACAP Capital
Distribution file maintains records of the announcement date, ex-distribution date and adjustment
factor of the stock split. The share price return data and accounting data are retrieved from the
Company Returns file and Financial Statement File of the PACAP database respectively.
We collect the insider trading information from the Inside Trade Asia database
maintained by Primark 5 and the Securities (Disclosure of Interest) Daily Summary and
Directors’/Chief Executives’ Notification Report6. The insider trading records include all types
opportunity to realize losses short term or gains long term to re-establish short-term status. However, investors do
not need to pay any tax on capital gains in Hong Kong.
3Some empirical studies in the finance literature investigate the signaling role of stock splits and stock dividend
jointly (e.g., Grinblatt, Masulis and Titman, 1984; Banker, Das and Datar, 1993). These two events are similar as
they relate to the firm’s stock distribution policy by increasing the number of outstanding shares of the stock
dividend-paying and splitting firms without changing the proportional ownership of shares held by the existing
stockholders, the cash flow, the assets and the liabilities of the firms. However, the market responses to there two
events may be quite different (Lakonishok and Lev, 1987; Rankine and Stice, 1997). In this paper, we concentrate
on the stock splits and insider trading, and we also investigate the possible different market reactions to the stock
dividend and splits announcement in a separate paper.
4 Finance companies are usually the more regulated firms. In addition, as the types of accounting variables in
measuring the performance for finance and industrial companies are substantially different, including the finance
companies in our sample may create problems in our control firm selection for event study.
5 The Inside Trade Asia database maintained by Primark is an electronic version of the trading transactions of
companies directors reported in the Securities (Disclosure of Interest) Daily Summary and Directors’/Chief
Executives’ Notification Report.
6 By the Laws of Hong Kong (Chapter 396 (Disclosure of Interest Ordinance)) and the Listing Rules of the Hong
Kong Exchange, the directors of listed firms are required to disclose their securities transactions in the market within
7
of securities transactions that change the shareholding percentage of the directors. Our analysis
of the insider trading activity around the stock split announcements examines only those inside
transactions that increase or decrease the shareholdings of directors through open market
purchase and sale of shares. Other types of inside transactions such as options and warrants
trading, bonus shares, scrip dividend and gifts are excluded from our sample (Lin and Howe
1990). Our bid-ask records are from the database maintained by the Research and Planning
Division of the Hong Kong Exchange. The database provides intra-day trading information such
as the ask price, bid price, trading price, trading volume, and trading value of all securities that
are traded on the Hong Kong Exchange recorded at 30-second intervals. The types of securities
in the bid-ask data file are ordinary shares, preference shares, warrants, debt securities, and unit
trusts. However, in this study, we measure the changes in the liquidity patterns around the stock
split announcement for ordinary shares only.
Our sample period covers 21 years from 1980 to 2000 and contains 162 stock split
events.7 Of the 162 events, 9 are without valid announcement dates. In addition, 10 stock split
announcements are made by finance companies. We use the control firm approach to measure
the abnormal share price reaction of the firms making the split announcements. The selection
criteria using the control firm approach for event study further removes 23 events from our
sample. Therefore, our final sample consists of a total of 120 cases for event study analysis8.
Table 1 reports the summary statistics of our sample. Of these 120 events, there are 3 from the
five business days from the day they conduct transactions on the Hong Kong Exchange. The Hong Kong Exchange
publishes the trading information of the directors in the Securities (Disclosure of Interest) Daily Summary and
Directors’/Chief Executives’ Notification Report. The information reported includes the name of the director, the
name of the securities traded, the class of the securities, the transaction date, the disclosure date, the number of
shares and the price at which the shares were traded.
7 Following Grinblatt, Masulis and Titman (1984), our sample consists of pure splits. We select the split
announcements that are not contaminated by other announcements over the period around the split announcement
date (one month prior and one month after).
8 To show that our results for stock splits are not affected by reverse splits, we perform a check on stock
consolidation events and find that our sample firms do not conduct stock consolidation during our examination
period.
8
utilities sector, 35 from the properties sector, 43 from the consolidated enterprises sector, 31
from the industrial sector, 5 from the hotels sector, and 3 from the miscellaneous sector. Our
analyses of the insider trading activity and the changes of liquidity patterns around the split
announcements cover the sample period from 1993 to 2000 and from 1996 to 2000 respectively
due to the availability of the insider trading and microstructure data.
The average split factor (the average of the number of new shares exchanged for one old
share of the splitting firms in the sample) and the market capitalization of the splitting firms are
6.9560 and HK$ 4,457,789,000 respectively. Comparing the average number of shares
outstanding between the splitting firms and the corresponding industry mean, the number of
shares outstanding of the industries is two times those of the splitting firms. While the average
number of shares outstanding is higher for the industries, the average share price is higher for the
splitting firms, which is 4.8 times those of the industries. The higher average pre-split share
price of the splitting firms than the industry average suggests that the splitting firms may have
been motivated to use stock splits to realign their share prices to their preferred trading range and
lower their share prices to enhance attractiveness (Lakonishok and Lev, 1987).
*********************
TABLE 1 HERE
*********************
3.2. Methodology
3.2.1. Abnormal Share Price Reaction
We use the event study methodology to evaluate the abnormal share price reaction of the
corporate announcements of stock splits. The event date, t = 0, is the announcement date
recorded in the Capital Distribution file of the PACAP database. We use the market model9 to
9 When using event study methodology to measure the abnormal share price reaction of an event, the clustering of
corporate events is a common problem. The clustering problems of corporate events may create bias in abnormal
return measurement. For robustness purposes, we also compute the abnormal return using the control firm approach
9
estimate abnormal share price reactions to the announcement of stock splits. For the market
model, we use returns on the Hong Kong Hang Seng Index as the proxy of the market returns.
The abnormal return on day t, ARitm, is defined as the difference between the realized returns of
sample firm i and of the market index.
The test statistic for the significance of the abnormal return is computed by the standard
deviation measured in the estimation period over 200 days from t = -300 to t = -101 (Brown and
Warner, 1985). Our test period is from t = -60 to t = +360. According to the signaling
hypothesis, the trading range hypothesis, and the liquidity hypothesis, stock splits are expected to
signal favorable information about the value of the splitting firms to the market, bring the share
prices of the splitting firms down to a desired price range, increase the trading volume, enhance
liquidity, and improve marketability. Therefore, we would expect to have a positive market
reaction to the splitting firms around the stock split announcements.
3.2.2. Abnormal Insider Trading Activity
Many studies document that insiders are in possession of private information about the
current and future performance of firms (e.g., Seyhun, 1986; Lin and Howe, 1990). Insider
trading activities have been found around different types of corporate events such as earnings
releases (e.g., Udpa, 1996), seasoned equity offerings (Gombola, Lee and Liu, 1997), mergers
and acquisitions (e.g., Meulbroek, 1992), corporate bankruptcy petitions (e.g., Seyhun and
Bradley, 1997), listing and delisting (e.g., Lamba and Khan, 1999), and analysts’ earnings
(Barber and Lyon, 1997). The control firm approach involves a matching process to choose a control firm that
possesses similar characteristics in terms of market value and book-to-market ratio with a sample firm (Fama and
French, 1992). All firms are categorized into five groups (from 1 to 5) according to the magnitude of the monthly
market values and book-to-market ratios and a sample firm is matched to a control firm if the control firm is in the
same quintile as the sample firm in terms of market value and book-to-market ratio. In addition, as we measure the
insider trading activity around the split announcements, the control firm selected should have neither made the
examined announcement nor conducted inside transactions around the announcement period of the sample firm
10
forecast revisions (e.g., Sivakumar and Vijayakumar, 2001). These studies report that there is a
“regular” trading pattern for insiders, who buy before good news and sell before bad news. In
this study, we examine the abnormal insider trading activity around the impending split
announcement to obtain insights into the authenticity of the split signal from the perspective of
the insiders of the splitting firms.
Our abnormal insider trading activity analysis examines whether the directors use their
inside information about the upcoming announcements of share splits to trade in the market for
their own accounts. Therefore, we assess whether the insider trading activity before the split
announcements (six-month period) is abnormally different from that of the other period. Based
on the methodology of Gombola, Lee and Liu (1997), we use the prior-period comparison
method to measure the abnormal insider trading activity before the announcement. The time-
length of the estimation period for comparison is a 6-month period between m = -12 and m = -7
before the announcement month of share split (m = 0).
The average of the trading measure (the proportion of number of shares traded to number
of outstanding shares, the market value, or the number of transactions) over the estimation period
(-12 ≤ m ≤ -7) is the expected trading level. The abnormal insider trading activity is estimated as
the difference between the actual trading level in the examination period (-6 ≤ m ≤ +6) and the
expected trading level computed over the estimation period. We also measure the standard
deviation in the estimation period (-12 ≤ m ≤ -7) and use it to test the significance of the
abnormal trading activity in the examination period (-6 ≤ m ≤ +6) (Brown and Warner, 1985).
If the stock split is simply cosmetic, then there should be no significant insider trading
activity around the split announcement. If the stock split is not as cosmetic as it appears to be,
then there should be significant insider trading activity, particularly insider buying activity.
(twelve months before and twelve months after). Using the control firm approach, we estimate the abnormal return
11
3.2.3. Liquidity Pattern (Spread and Depth)
We examine the changes of the two dimensions, spread and depth, of the liquidity pattern
around the stock split announcements. The spread measures the price aspect while the depth
measures the size aspect of liquidity. The spread quantifies the cost of trading. A wider spread
level means a higher cost of trading and hence lower liquidity. Depth reveals the effects of the
volume and dollar amounts of trading. Greater depth reflects larger trading volume and dollar
value, and hence higher liquidity. These two dimensions exhibit a negative relation (Lee,
Mucklow and Ready, 1993; Brockman and Chung, 1999): that is, a large (small) spread with a
narrow (wide) depth. We use two spread measures and four depth measures to assess the
changes in liquidity.
Absolute Spread and Relative Spread are our two spread measures. Absolute Spread is
the daily average of the absolute dollar difference of bid and ask recorded at 30-second intervals
on day t. Relative Spread is the daily average of the dollar difference of bid and ask divided by
the bid-ask midpoint recorded at 30-second intervals on day t. The depth is estimated by
Volume Depth, Dollar Depth, Ask Depth, and Bid Depth. Volume Depth is the sum of the
number of shares at the highest bid and the number of shares at the lowest ask recorded (adjusted
by the number of outstanding shares) relative to the number of shares outstanding. Dollar Depth
is the sum of the product of the number of shares at the highest bid and the highest bid price and
the product of the number of shares at the lowest ask and the lowest ask price recorded (adjusted
by the product of price and number of outstanding shares) at 30-second intervals on day t relative
to the market value. Ask (Bid) Depth is the product of the lowest ask (highest bid) price
(adjusted by the number of outstanding shares) and the number of shares at the lowest ask
on day t, ARitc, as the difference between the realized returns of sample firm i and of matched control firm j.
12
(highest bid) price recorded at 30-second intervals on day t relative to the market value
respectively.
One of the explanations for positive market reactions to stock splits is the expected
increased liquidity of the shares of the splitting firms after the splits (Maloney and Mulherin,
1992; Muscarella and Vetsuypens, 1996). This argument is based on the notion that corporate
liquidity may decline if the trading price of the shares is too high. To examine the changes in
corporate liquidity due to stock splits, we compare the liquidity patterns in terms of spread and
depth in the pre-split period and the post-split period. We expect to have a smaller post-split
spread and a larger post-split depth.
3.2.4. Regression Analysis
Table 2 reports that there are significant market reactions for splitting firms around the
split announcements. Numerous hypotheses (the signaling hypothesis, the optimal trading range
hypothesis, the liquidity hypothesis, the tax option hypothesis, and the managerial entrenchment
hypothesis) and empirical studies (Grinblatt, Masulis and Titman 1984; Lakonishok and Lev
1987; Lamoureux and Poon 1987; McNichols and Dravid 1990; Ikenberry and Ramnath 2002;
and many others) have attempted to explain positive abnormal announcement returns. Similar to
those studies, we construct a cross-sectional model to explain the level of abnormal returns of the
splitting firms. The model is defined as:
CAR =
α
0 +
β
1 FACTOR +
β
2 MktValue +
β
3 VolRatio +
β
4 EPSChg +
β
5 Multiple
+
β
6 RetVar +
β
7 PriceDev +
β
8 ShareDev +
ε
(1)
CAR is the cumulative abnormal return over the different periods examined10. FACTOR is the
natural logarithm of the size of the split factor. MktValue is the natural logarithm of the market
10 The CARs of the following periods are estimated (-60 ≤ t ≤ -1, -30 ≤ t ≤ -1, -10 ≤ t ≤ -1, and -10 ≤ t ≤ +10).
13
value (the product of price and number of shares outstanding) of the firm for the month before
the split announcement. VolRatio is the ratio of pre-split trading volume to post-split trading
volume normalized by the number of shares outstanding11. EPSChg is the percentage change in
earnings per share of the current year to those of the previous three years. Multiple is a dummy
variable that takes the value of 1 if there is more than one split announcement over the sample
period from 1980 to 2000. RetVar is the standard deviation of return. PriceDev is the deviation
of share price to the industry median price. ShareDev is the deviation of shares outstanding to
the industry median of shares outstanding.
4. Empirical Results
4.1. Abnormal Share Price Reaction
The signaling hypothesis argues that one of the motivations for firms to split their shares
is that they are optimistic about the future potential increase of their share price. Therefore,
although stock splits appear to be cosmetic, their announcement should lead to positive share
price performance. Table 2 reports the market reaction of the splitting firms over different time
periods from t = -60 to t = +360 and Figure 1 portrays the cumulative abnormal return path. The
abnormal returns are mostly positive from a pre-announcement period of 60 days to a post-
announcement period of 120 days. We find positive abnormal returns, significant at the 0.01
level, for the three days around the announcement day. The 3-day cumulative abnormal return (-
1 ≤ t ≤ +1) is 5.19% using the market model12. The result of positive abnormal returns in this
study is consistent with previous studies for the U.S. market, and suggests that stock splits signal
favorable information to the market. Comparing the abnormal returns around the announcement
11 The number of outstanding shares before the split announcement is different from that after the announcement. To
better compare the changes in the trading activity before and after the announcement, we need to standardize the
trading volume by the number of outstanding shares.
14
day and the pre-split period (-60 ≤ t ≤ -1), the magnitude of the abnormal returns is as high as
42%. The high pre-split abnormal returns may be due to the leakage of insider information about
the impending split announcement.
*********************
TABLE 2 HERE
*********************
Asquith, Healy and Palepu (1989) suggest that the information conveyed by stock split is
not short-lived, and may persist for years following a split. Ikenberry, Rankine and Stice (1996)
and Desai and Jain (1997) provide evidence that stock splits result in long-term excess returns.
We also examine the long-term share price performance of the splitting firms and find that, on
one hand, the long-term abnormal returns over the post-split period of +10 ≤ t ≤ +240 and +10 ≤
t ≤ +360, although not significant, are negative; on the other hand, the cumulative abnormal
return from 60 days before to 360 days after the split announcement as portrayed in Figure 1
maintains at a very high level of around 40% persistently for up to 360 days. Our results provide
evidence that stock split affects share price performance of the splitting firms over a long run
from pre-split to post-split period.
*********************
FIGURE 1 HERE
*********************
4.2. Abnormal Insider Trading Activity
The fundamental argument of the signaling hypothesis is that the splitting firms use stock
splits to signal favorable information to the market. We use the abnormal insider trading activity
to assess the informativeness of the split signal. Intuitively, as the insiders are aware of the
12 Although not reported here, we also find significantly positive abnormal returns using the control firm approach.
For example, the 3-day cumulative abnormal returns (-1 ≤ t ≤ +1) is 3.85% (t-statistic = 2.89) using the control firm
approach, which is also significant at 0.01 level.
15
impending corporate news of stock splits and expect positive market reactions to the news, they
should buy the shares of the splitting firms before the stock split announcements. Therefore, we
expect to have significant buying activity from the insiders of the splitting firms. Table 3 reports
our results for abnormal insider trading analysis13. Three variables, “Buy”, “Sell”, and “Net
(difference between purchases and sales14)” are used to assess the intensity of the trading activity
in different trading directions.
In the pre-split period of -6 ≤ m ≤ -1, we find significant buying and selling activities
between m = -6 and m = -3. However, comparing the trading activities between -6 ≤ m ≤ -3 and -
2 ≤ m ≤ -1, the trading activity in -2 ≤ m ≤ -1 is trivial. The immaterial trading activities in the
two months immediately before the split announcement month may be due to the potential threat
of investigation for trading with private and price-sensitive information. Therefore the insiders
may choose to cash in their private information in advance. Indeed, we observe that there are
significant net purchase of shares in m = -4 and consequently significant net sale of shares by
insiders in m = -3.)
For the cumulative abnormal trading activity (-6 ≤ m ≤ -1), the abnormal market values of
purchase and sale are 54.622 and 36.222 respectively, which are both significant at the 0.01 level.
The 54.622 (36.222) abnormal market value of purchase (sale) suggests that the directors have
bought (sold) firms’ shares with a value $54.622 million ($36.222 million) in the examination
period for the six months (-6 ≤ m ≤ -1) higher than in the estimation period for the six months (-
12 ≤ m ≤ -7) before the split announcement. The net (difference between purchase and sale)
cumulative abnormal market value is 2.455, which is insignificant. There are several possible
explanations for the inconsistent trading patterns and lack of significant buying activity. The
13 Table 3 shows the result using the market value as the measure of insider trading activity. To demonstrate
robustness, we also conduct the analysis using the proportion of number of shares traded to number of outstanding
16
insiders may have had no private information about the stock splits or may have had genuine
information about cosmetic stock splits with non-signaling functions. In addition, they may have
intentionally sold their shares (i.e., a contrarian strategy) to justify their securities trading before
corporate announcement to avoid potential accusations of illegal trading with private information.
Therefore, we cannot find significant purchases of shares before split announcement. There are
also abnormal trading activities in the post-announcement period, particularly in +3 ≤ m ≤ +6.
Both the buying and selling activities are significant. The increase in trading activity following
the split announcements can be explained by the fact that the post-announcement period is a
more appropriate time for insiders to conduct securities trading.
*********************
TABLE 3 HERE
*********************
4.3. Liquidity Pattern (Spread and Depth)
The liquidity hypothesis argues that stock splits can lead to improved liquidity.
Following previous studies, we compare the changes in the corporate liquidity in both the price
and size dimensions around the split announcement. To avoid the potential effect of the
temporary increase in trading just before and following the split due to the split announcement
itself, we exclude 20 days before and 20 days after the split announcement from our examination
of the changes in liquidity pattern15. The pre-split period and post-split period are defined as 20
trading days before and following the exclusion period from the split announcement. Both the
parametric two-sample t-test and the non-parametric Mann-Whitney test for sample differences
shares and the number of transactions. The results are qualitatively similar to those that are reported in Table 3.
14 A positive value for “Net” means that there is a higher value for the purchase than for the sale measure.
15 Such an exclusion of a certain period following the split announcement from the analysis is similar to excluding
the period between the split announcement day and the ex-split date from the analysis. Conroy, Harris and Benet
(1990), Ferris, Hwang and Sarin (1995) and Desai, Nimalendran and Venkataraman (1998), in the microstructural
examination of stock splits, also exclude the period around the announcement day and the ex-split date from their
17
are conducted. Table 4 reports the test results16. Improved liquidity is evidenced by a lower
post-split spread and a greater post-split depth.
We find that the two measures of spread, Absolute Spread (0.089 vs. 0.024) and Relative
Spread (0.023 vs. 0.021), decrease from the pre-split period to the post-split period. The
decrease in absolute spread is significant at the 0.01 level in both the parametric and non-
parametric tests for sample differences. Although the fall in Relative Spread is not as
pronounced as that in Absolute Spread, the mean difference of the pre-split and post-split spread
levels is also significant at the 0.01 level using the parametric test. A narrower post-split spread
suggests that there is an increase in liquidity following stock splits.
For the four measures of depth, Volume Depth (0.194 vs. 0.303), Dollar Depth (0.240 vs.
0.475), Ask Depth (0.110 vs. 0.258), and Bid Depth (0.130 vs. 0.217), the increase in depth level
from the pre-split period to the post-split period indicates an improvement in corporate liquidity.
In particular, the post-split increase in depth is statistically significant for the Volume Depth,
Dollar Depth, and Ask Depth.
Besides using the spread and depth to examine the changes in the liquidity pattern, to
measure whether stock splits enhance or reduce trading activity, we compare the trading volume
and market value in the pre-split and post-split periods. Maloney and Mulherin (1992) and Desai,
Nimalendran and Venkataraman (1998) report a post-split increase in trading volume and
conclude that it is evidence of increased liquidity. In Table 4, we observe that both the average
number of shares traded (0.011 vs. 0.034) and the market value of shares traded (0.011 vs. 0.034)
increase substantially. The increase is statistically significant at the 0.01 level using the
analyses to avoid information contamination around the announcement day, transient microstructure effects around
the ex-split date, and distortions due to dual trading in both pre-split and when-issued shares.
16 We show the results using the examination period of 20 days before and 20 days after the exclusion period from
the split announcement. To demonstrate robustness, we repeat the comparison analysis using various examination
windows of ±10 days, ±15 days, ±20 days, ±25 days, ±35 days and ±40 days. The results of a narrower spread and a
greater depth in the post-split period are not affected by the length of the examination window.
18
parametric test. These findings suggest that by reducing the share price to a lower trading range,
the shares become more marketable and hence the trading activity is enhanced.
As the splitting firms use stock splits to reduce their share price to a preferred level, there
is a significant difference in the average pre-split and average post-split share prices. The pre-
split price is 3 times the post-split price (6.583 vs. 2.181). While the share price decreases
substantially in the post-split period, the return volatility increases significantly as a result of the
split (0.007 vs. 0.012). This finding is similar to those of Ohlson and Penman (1985), Koski
(1998) and Gray, Smith and Whaley (2003). As a lower price may improve the attractiveness of
the shares, increased return volatility adversely affects marketability. There are two possible
explanations for the higher return volatility. Ohlson and Penman (1985) and Dravid (1987)
argue that the enlarged return volatility is the result of a wider spread following the split.
Karpoff (1987) explains that the increased return volatility may be due to the positive relation
between volatility and trading volume following the split. Jones, Kaul and Lipson (1994) further
argue that enhanced trading activity brings in information to the market and hence affects share
prices and return volatility. As we observe only greater trading volume and not wider spreads
following splits, our result of enlarged return volatility should be mainly due to increased trading
activity.
Our sample comparison results show that narrower spread, wider depth and higher
trading volume follow the split announcement. Consistent with the liquidity hypothesis, our
findings provide evidence that stock splits improve the liquidity of shares.
*********************
TABLE 4 HERE
*********************
4.4. Regression Analysis
19
Table 5 reports the results of the regression model (1).17 The descriptive statistics and
regression results of the variables are shown in Panel A and Panel B, respectively. The p-values
for the coefficients are adjusted for heteroskedasticity using White’s procedure (1980).
The size of the split factor is a signal to the market about the desired trading range in
equilibrium and the extent of firms’ private information about future earnings (Brennan and
Copeland, 1988; McNichols and Dravid, 1990; Brennan and Hughes, 1991). According to the
optimal trading range hypothesis, stock splits are used to bring the share price down to the
desired trading range. The share price of a split event with a large split factor indicates that the
current share price is far outside the favorable trading price range. McNichols and Dravid (1990)
provide strong evidence for the optimal trading range hypothesis and report a positive relation
between the announcement abnormal return and the split factor. As the size of the split factor
signals information to the market, we include FACTOR, which is a measure of the size of the
split factor, in our regression model to control for the effect of the split size. A larger split factor
results in a greater reduction in the ex-split share price and hence a smaller amount for round lot
investment requirements. From both the signaling hypothesis and the optimal trading range
hypothesis, we expect the abnormal market reaction to the splitting firm with a larger split factor
to be more positively significant. In Table 5, FACTOR is positively and statistically significant
related to the announcement returns at the -10 ≤ t ≤ +10 period. The market responds more
favorably to the move of the splitting firms to a lower range of share prices from their current
prices. This result provides evidence that the split factor choice is a signal to the market about
firms’ value.
17 The data of insider trading activity is available from 1993 to 2000, while the intraday data is available from 1996
to 2000. If we include both variables in the regression analysis, then the sample size would be significantly reduced.
Therefore, we use earnings change (EPSChg) as the measure of signaling and trading volume (VolRatio) as the
measure of liquidity.
20
Since small firms have less effective ways to attract investors’ attention and to signal to
the general public, stock split announcement by small firms serves a stronger signaling function.
If a stock split serves a signaling function, then the split informative signal that is conveyed by a
small firm should be stronger than that of a large firm. Therefore, the split announcement
abnormal return should be higher for small firms than for large firms. Ikenberry, Rankine and
Stice (1996) find that the abnormal returns are higher for small splitting firms than for large
splitting firms. We use MktValue, which denotes firm size, as a proxy for information
asymmetry and expect a negative relation between abnormal return and MktValue. As shown in
Table 5, the result for MktValue is mixed. MktValue is negatively related to abnormal returns in
the short-run and positively related to them in the long-run. The negative relation between
MktValue and abnormal return around the split announcement in the short term suggests that
stock splits are a better device for smaller firms than for larger firms in signaling information18.
In the finance literature, trading volume can be used as a proxy of liquidity. Therefore,
we include trading volume in the regression model to examine the liquidity hypothesis. VolRatio
is our liquidity measure. A small value of VolRatio means that there is a great difference in the
trading volume between the pre-split and post-split periods. If the stock split is motivated by
liquidity reasons to lower the share price to enhance trading activity, then we expect an inverse
relation between abnormal return and VolRatio. Muscarella and Vetsuypens (1996) find an
increase in trading activity after the stock split, which is evidence of improved liquidity. In
Table 5, the coefficients on VolRatio are negative around the pre-split and announcement periods.
In particular, VolRatio is negatively significant in the pre-split periods of -60 ≤ t ≤ -1 and -30 ≤ t
18 To show that our sample is not biased in terms of firm size, we perform a check on the size of the firms in our
sample. The firms in our sample are divided into five groups. The number of firms and average price in the first
quintile are 12 (10%) and 3.94, in the second quintile are 26 (21.67%) and 6.54, in the third quintile are 21 (17.50%)
and 8.85, in the fourth quintile are 29 (24.17%) and 10.91, and in the fifth quintile are 32 (26.67%) and 24.92. These
statistics suggest that our sample does not suffer from serious firm size bias.
21
≤ -1, which suggests that the positive pre-split abnormal returns are due partly to the enhanced
liquidity following the split as compared to the pre-split period.
As it is argued that stock splits have signaling effects that shares are undervalued,
Asquith, Healy and Palepu (1989) provide evidence that splitting firms usually have better
earnings performance before the split, and that split announcement return is related to prior
earnings growth. Therefore, we include a variable, EPSChg, which represents the pre-split
earnings performance of the splitting firms, in our regression model. We expect EPSChg to be
positively related to abnormal returns. We observe that the coefficients on EPSChg are mostly
positive and significant in the periods of -60 ≤ t ≤ -1 and -30 ≤ t ≤ -1 in Table 5. The result of
positive coefficients is consistent with that of Asquith, Healy and Palepu in that split return is
correlated with prior earnings performance.
Through the 20 years of our sample period, we observe that some firms conduct more
than one split (9.87%). Pilotte and Manuel (1996) study the effects of recurring split events and
find that the market reaction is more favorable if the current stock split is preceded by a previous
split. If the split is a signaling device, too frequent announcement may be indicative of non-
signaling motivation. According to the efficient market hypothesis, the share price reaction
should be less pronounced for the repeated split announcement with no signaling motivation.
The optimal trading range hypothesis argues that repeated splits imply the superior performance
of the splitting firms in using splits to periodically lower their share prices. To assess the merits
of these arguments and the difference in the abnormal return due to repeated splits, we include
Multiple to examine the effect of split frequency. None of the coefficients on Multiple are
significant. This finding suggests that repeated split announcements provide no additional
signals to the market.
22
Many studies document increases in return volatility following stock splits (Ohlson and
Penman, 1985; Lamoureux and Poon, 1987; Dubofsky 1991; Koski 1998; Gray, Smith and
Whaley, 2003). As shown in Table 5, we find a positive relation between return volatility and
abnormal return in the short-term and a negative relation between them in the long-term. The
positive relation is particularly significant in the pre-split period and around the announcement
day. As mentioned earlier, higher return volatility may be due to wider post-split spread (Ohlson
and Penman, 1985; Dravid, 1987) and greater post-split trading volume (Karpoff, 1987; Jones,
Kaul and Lipson, 1994). Because we observe narrower (increased) rather than wider (decreased)
post-split spread (trading activity) in Table 4, the higher return volatility is the result of enhanced
trading volume. Therefore, we explain our positively significant relation between return
volatility and split announcement return as the higher liquidity of the splitting firms’ shares.
This finding provides evidence for the liquidity hypothesis.
By the optimal trading range hypothesis, there exists a favorable share price range to
improve share marketability. Lakonishok and Lev (1987) suggest that the split factor is driven
by the deviation between the share price of the splitting firm and the market- or industry-wide
average price. Stock splitting is used as a device to bring the share price down to the preferred
share price range the firms consider to be appropriate, a price range which is normally shaped by
the median or average price level of the industry or the market. We use two variables, PriceDev
and ShareDev, to examine the optimal stock price hypothesis. PriceDev is the natural logarithm
of the deviation of share price to industry median price. ShareDev is the natural logarithm of the
deviation of shares outstanding to industry median of shares outstanding. PriceDev and
ShareDev are expected to be positively related to abnormal returns. The larger the deviation
from the median values, the higher the abnormal returns should be. In Table 5, we find that both
PriceDev and ShareDev are positively significant in the pre-split period. This result suggests
23
that the market reacts more positively to the moves of splitting firms whose share prices
(numbers of outstanding shares) are higher than the industry median price (number of
outstanding shares). This result indicates that one of the motivations for stock splits is to return
the price to a level that is consistent with those of other firms in the industry and with market
averages.
Louis and Robinson (2005) argue that the incentive to signal private information is
determined by the information environment of the firm and the ability of the insiders to use other
means of communication. Since insiders have stronger incentive to signal to reduce information
asymmetry by communicating favorable private information and we also find significant insider
trading activities around the split announcement, especially three and four months before the
split announcement, to further examine if the split announcement provides a signaling function to
the market after controlling for insider trading, we repeat the regression analysis by including
InsiderD, which is a communication means of the insiders to the market, and InsiderD*MktValue,
in the following regression model:
CAR =
α
0 +
β
1 FACTOR +
β
2 MktValue +
β
3 VolRatio +
β
4 EPSChg +
β
5 Multiple+
β
6 RetVar +
β
7 PriceDev +
β
8 ShareDev +
β
9 InsiderD +
β
10 InsiderD*MktValue +
ε
(2)
InsiderD is a dummy variable which takes the value of 1 if there is insider buying activity around
the split announcement. We also include an interactive variable of InsiderD*MktValue to proxy
the richness of information environment and the signaling hypothesis. Although Model (2) is a
more general model than Model (1), Table 5 shows that the coefficients on InsiderD and
InsiderD*MktValue are not significant while the variable, FACTOR, representing the signaling
hypothesis remains significant.
To show that our regression results are robust to alternative testing methods and
computation of variables, we conduct a number of sensitivity tests on our results. Similar to our
24
analysis for the liquidity pattern reported in Table 4, to demonstrate that our regression results in
Table 5 are not sensitive to the period over which we measure our liquidity variable, VolRatio,
we use different periods to measure VolRatio. In Table 5, VolRatio is computed using the
average trading volume for 60 days in the pre-split and post-split periods. We repeat the analysis
using period lengths of ±10 days, ±15 days, ±20 days, ±25 days, ±30 days, ±35 days, ±40 days,
±45 days, ±50 days, and ±55 days. Similar results are found regardless of the length of period
we use to measure our VolRatio. Lamoureux and Poon (1987) suggest that the measurement for
the difference in the trading volume between pre-split and post-split periods can be adjusted by
the market trading volume. We repeat our regression analysis using this alternative measurement
of VolRatio. Again, we find results similar to those reported in Table 5, which suggests that our
findings are not sensitive to the method by which we compute the trading volume ratio.
The result of ESPChg that is reported in Table 5 is computed using net income (INC9 of
the Financial Statement File of PACAP database). To demonstrate robustness, we use another
variable, income from operation (INC5 of the Financial Statement File of PACAP database) to
repeat the regression analysis. The results are qualitatively the same.
In Table 5, we use PriceDev as our measurement of the preferred share price range.
Peterson and Peterson (1992) use an alternative method to estimate the target share price that the
splitting firms consider to be appropriate. That price is computed as the share price before the
split announcement (t-2) divided by the split factor. We re-run the regression model using the
target share price. Similar results are found, which suggests that our findings supporting the
optimal trading range hypothesis are robust to the different measurements of the desirable
trading price.
*********************
TABLE 5 HERE
*********************
25
5. Conclusion
As stock splits “are just a finer slicing of a given cake” (Lakonishok and Lev, 1987),
without altering the future cash flows of a company one would expect that stock prices would not
react to the announcement of splits. However, many empirical studies in the U.S. and other
countries indicate that splits tend to influence share prices, and the conflict between theory and
practice warrants further study. This study analyzes the effect of stock splits using intraday data
and insider trading data in Hong Kong from 1980 to 2000. We first investigate the abnormal
market reaction of the splitting firms. In line with the results of many other studies in different
capital markets, we find significant abnormal returns around the announcement. The positive
reaction can be attributable to favorable signals and improved liquidity. When observing the
positive price reaction around the split announcement, it is difficult to differentiate signaling
effects from liquidity effects, we then use the abnormal insider trading activity to assess the
informativeness of the split signal. We find abnormally high insider trading activities three to
four months before the split announcement and in the post-announcement period, although
insider trading activities in the two months immediately before the split announcement is
immaterial. This can be explained by the fact that the post-announcement period is a more
appropriate time for securities trading by insiders in order to avoid accusations of illegal trading.
Our microstructural analysis shows that stock split improves the liquidity. We find the post-split
spread measures are significantly lower than those in the pre-split period, and the post-split depth
measures are significantly higher than those in the pre-split period. The regression analysis
shows that the increase in future earnings per share and the increase in trading volume both
occurred following the split. This suggests the presence of a possible signaling role for split
announcements that are confounded with increased liquidity.
26
References
Amihud, Y. and H. Mendelson, 1988, Liquidity and asset prices: Financial management implications,
Financial Management 17, 5-15
Angel, J. J., 1997, Tick size, share price and stock splits, Journal of Finance 52, 655-681.
Asquith, P., P. Healy and K. Palepu, 1989, Earnings and stock splits, Accounting Review 54, 387-403.
Baker, K. and P. Gallagher, 1980, Management’s view of stock splits, Financial Management 9, 73-
77.
Banker, R., S. Das and S. Datar, 1993, Complementarity of prior accounting information: The
case of stock dividends, The Accounting Review 68, 28-47
Barber, B. M. and J. D. Lyon, 1997, Detecting long-run abnormal stock returns: The empirical power
and specification of test statistics, Journal of Financial Economics 43, 341-372.
Brennan, M. J. and T. E. Copeland, 1988, Stock splits, stock prices and transaction costs, Journal of
Financial Economics 22, 83-101.
Brennan, M. J. and P. J. Hughes, 1991, Stock prices and the supply of information, Journal of
Finance 46, 1665-1691.
Brockman, P. and D. Y. Chung, 1999, An analysis of depth behavior in an electronic, order-driven
environment, Journal of Banking and Finance 23, 1861-1886.
Brown, S. and J. Warner, 1985, Using daily stock returns: The case of event studies, Journal of
Financial Economics 14, 3-31.
Byun, J. and M.S. Rozeff, 2003, Long-run performance after stock splits: 1927 to 1996, Journal of
Finance 58, 1063-1085.
Conroy, R. M., R. S. Harris and B. A. Benet, 1990, The effects of stock splits on bid-ask spreads,
Journal of Finance 45, 1285-1295
Demsetz, H., and K. Lehn, 1985, The structure of corporate ownership: Causes and consequences,
Journal of Political Economy 93, 1155-1177
Desai, H. and P. C. Jain, 1997, Long-run common stock returns following stock split and reverse
split, Journal of Business 70, 409-433.
Desai, A. S., M. Nimalendran and S. Venkataraman, 1998, Changes in trading activity following
stock splits and their effect on volatility and the adverse-information component of the bid-ask
spread, Journal of Financial Research 11, 159-183.
Dravid, A., 1987, A note on the behavior of stock returns around ex-dates of stock distributions,
Journal of Finance 42, 163-168.
Dubofsky, D. A., 1991, Volatility increases subsequent to NYSE and AMEX stock splits, Journal of
Finance 46, 421-431.
Fama, E, L. Fisher, M. Jensen and R. Roll, 1969, The adjustment of stock prices to new information,
International Economic Review 10, 1-21.
27
Fama, E. F. and K. French, 1992, The cross-section of expected stock returns, Journal of Finance 47,
427-466.
Ferris, S. P., C. Y. Hwang and A. Sarin, 1995, A microstructure examination of trading activity
following stock splits, Review of Quantitative Finance and Accounting 5, 27-41.
Gombola, M. J., H. W. Lee and F. Y. Liu, 1997, Evidence of selling by managers after seasoned
equity offering announcements, Financial Management 26, 37-53.
Gray, S. F., T. Smith and R. E. Whaley, 2003, Stock splits: Implications for investor trading costs,
Journal of Empirical Finance 10, 271-303.
Grinblatt, M. S., R. W. Masulis and S. Titman, 1984, The valuation effects of stock splits and stock
dividends, Journal of Financial Economics 13, 461-490.
Ikenberry, D. L., G. Rankine and E. K. Stice, 1996, What do stock splits really signal? Journal of
Financial and Quantitative Analysis 31, 357-375.
Ikenberry, D. L. and S. Ramnath, 2002, Underreaction to self-selected news events: The case of stock
splits, Review of Financial Studies 15, 357-375.
Jones, C. M., G. Kaul and M. C. Lipson, 1994, Transactions, volume and volatility, Review of
Financial Studies 7, 631-651.
Karpoff, J. M., 1987, The relation between price and trading volume: A survey, Journal of Financial
and Quantitative Analysis 22, 109-126.
Kole, S., 1995, Measuring managerial equity ownership: A comparison of sources of ownership data,
Journal of Corporate Finance 1, 413-435
Koski, J. L., 1998, Measurement effects and the variance of returns after stock splits and stock
dividends, Review of Financial Studies 11, 143-162
Kryzanowski, L. and H. Zhang, 1993, Market behavior around Canadian stock-split ex-dates,
Journal of Empirical Finance 1, 57-81.
Kunz, R.M. and S. Majhensek, 2004, Stock split signaling: Much ado about nothing in Switzerland?
Working paper of University of Zurich.
Lakonishok, J. and B. Lev, 1987, Stock splits and stock dividends: Why, who and when, Journal of
Finance 62, 913-932
Lamba, A. S. and W. A. Khan, 1999, Exchange listings and delistings: The role of insider
information and insider trading, Journal of Financial Research 22, 131-146.
Lamoureux, C. and P. Poon, 1987, The market reaction to stock split, Journal of Finance 62, 1347-
1370.
Lee, C., B. Mucklow and M. Ready, 1993, Spreads, depths and the impact of earnings information:
An intraday analysis, Review of Financial Studies 6, 345-374
Lin, J. and J. S. Howe, 1990, Insider trading in the OTC market, Journal of Finance 45, 1273-1284.
28
Louis, Henock and D. Robinson, 2005, Do managers credibly use accruals to signal private
information? Evidence from the pricing of discretionary accruals around stock splits, Journal of
Accounting and Economics 39, 361-380
Maloney, M. and J. Mulherin, 1992, The effects of splitting on the ex: A microstructure
reconciliation, Financial Management 21, 44-59.
McConnell, J., and H. Servaes, 1990, Additional evidence on equity ownership and corporate value,
Journal of Financial Economics 27, 595-612.
McNichols, M. and A. Dravid, 1990, Stock dividends, stock splits and signaling, Journal of Finance
45, 857-879.
Meulbroek, L., 1992, An empirical analysis of illegal insider trading, Journal of Finance 47, 1661-
1700.
Morck, R., A. Shleifer, and R. Vishny, 1988, Management ownership and market valuation: An
empirical analysis, Journal of Financial Economics 20, 293-316.
Mukherji, S., Y. H. Kim and M. C. Walker, 1997, The effects of stock splits on the ownership
structure of firms, Journal of Corporate Finance 3, 167-188.
Muscarella, C. J. and M. R. Vetsuypens, 1996, Stock splits: Signaling or liquidity? The case of ADR
“Solo splits”, Journal of Financial Economics 25, 3-26.
Ohlson, J. and S. Penman, 1985, Volatility increases subsequent to stock splits: An empirical
aberration, Journal of Financial Economics 14, 251-266.
Peterson, D.R., and P.P. Peterson, 1992, A further understanding of stock splits: The case of reverse
stock splits, Journal of Financial Research 15, 189-205.
Pilotte, E. and T. Manuel, 1996, The market’s response to recurring events: The case of stock splits,
Journal of Financial Economics 41, 111-127.
Rankine, G. and E. R. Stice, The market reaction to the choice of accounting method for stock
splits and large stock dividends, Journal of Financial and Quantitative Analysis 32, 161-182
Seyhun, H. N., 1986, Insider profits, costs of trading, and market efficiency, Journal of Financial
Economics 16, 189-212.
Seyhun, H. N. and M. Bradley, 1997, Corporate bankruptcy and insider trading, Journal of Business
70, 189-216.
Sivakumar, K. and J. Vijayakumar, 2001, Insider trading, analysts’ forecast revisions, and earnings
changes, Journal of Accounting, Auditing and Finance 16, 167-187.
Udpa, S. C., 1996, Insider trading and the information content of earnings, Journal of Business
Finance and Accounting 23, 1069-1095.
White, H., 1980, A heteroskedasticity-consistent covariance matrix estimator and a direct test for
heteroskedasticity, Econometrica 48, 817-838.
29
Table 1
Summary Statistics of the Stock Split Sample
Sample
Size
Mean (Median)
Split
Factora
Market
Capitalization
(HK$’000)
Number of Shares
Outstanding of
Splitting Firms
(‘000)
Share Price
of Splitting
Firms
Average Number of
Shares Outstanding of
Each Industry Sector
(‘000)
Average Share
Price of Each
Industry
Sector
Utilities 3 4.6667 (2.00) 36,289,748 559,382 64.50 2,124,766 7.97
Properties 35 8.2350 (5.00) 2,753,490 308,808 17.97 581,837 3.72
Consolidated
Enterprises 43 6.1977 (4.00) 3,410,268 427,709 10.36 851,299 4.26
Industrial 31 5.5161 (5.00) 5,379,383 332,302 22.52 627,735 2.32
Hotels 5 15.8000 (10.00) 2,515,497 171,295 43.23 294,739 3.69
Miscellaneous 3 5.3333 (4.00) 1,237,815 136,093 10.67 170,076 11.12
Average 6.9560 (5.00) 4,457,789 353,700 18.46 706,568 3.84
Total 120
a Split factor is defined as number of new shares exchanged for one old share.
30
Table 2
Abnormal Returns around Stock Split Announcements
Event Abnormal Return
Window (t-statistics)
-60, -1 0.4235
(12.25)**
-30, -1 0.2528
(10.34)**
-10, -1 0.1028
(7.28)**
-1, +1 0.0519
(6.71)**
0 0.0145
(3.24)**
-3, +3 0.0860
(7.28)**
-5, +5 0.0983
(6.64)**
-10, +10 0.1270
(6.21)**
+1, +10 0.0098
(0.69)
+10, +60 0.0572
(1.80)
+10, +120 0.0099
(0.21)
+10, +240 -0.0360
(-0.53)
+10, +360 -0.0856
(-1.02)
** Significant at the 0.01 level.
* Significant at the 0.05 level.
Figure 1
Cumulative Abnormal Return Path
0.000
0.100
0.200
0.300
0.400
0.500
0.600
-60
-40
-20
0
20
40
60
80
100
120
140
160
180
200
220
240
260
280
300
320
340
360
Event Day
Cumu lative A bnorm al Retur n
31
Table 3
Abnormal Insider Trading Activity in terms of Market Value around
Stock Split Announcements
The buy subsample consists of events where there is a purchase of shares. The sell subsample consists of events
where there is a sale of shares. The net subsample consists of events where there is a net purchase of shares (the
quantity of purchased shares exceeded the quantity of sold shares). A positive (negative) value for “Net” means that
the number of purchased shares is higher (lower) than the number of sold shares. The 54.622 (36.222) abnormal
market value of purchase (sale) suggests that the directors have bought (sold) firms’ shares with a value $54.622
million ($36.222 million) in the examination period for the six months (-6 ≤ m ≤ -1) higher than in the estimation
period for the six months (-12 ≤ m ≤ -7) before the split announcement. The net cumulative abnormal market value,
which is the difference between purchase and sale, is 2.455.
Buy Sell Net
Event Abnormal Insider Trading Activity
Month (t-statistics)
-6 0.0933 -7.8310 6.0820
(0.05) (-2.12)* (2.34)*
-5 -1.3650 1.9202 -2.5185
(-0.73) (0.52) (-0.97)
-4 40.1978 -3.1780 24.1388
(21.40)** (-0.86) (9.28)**
-3 20.7692 47.2326 -23.7649
(11.06)** (12.78)** (-9.14)**
-2 -1.7300 -1.1785 -0.0610
(-0.92) (-0.32) (-0.02)
-1 -3.3431 -0.7431 -1.4220
(-1.78) (-0.20) (-0.55)
0 -0.5055 -0.1048 -0.2009
(-0.27) (-0.03) (-0.08)
+1 -2.9532 -1.1664 -0.7269
(-1.57) (-0.32) (-0.28)
+2 -0.4615 -0.2376 -0.1183
(-0.25) (-0.06) (-0.05)
+3 20.5351 23.7863 -4.7064
(10.93)** (6.44)** (-1.81)
+4 -3.3266 -0.4243 -1.8513
(-1.77) (-0.11) (-0.71)
+5 -0.1210 26.5397 -18.3355
(-0.06) (7.18)** (-7.05)**
+6 -3.8751 -3.2408 -0.0322
(-2.06)* (-0.88) (-0.01)
-6 to -1 54.6222 36.2222 2.4545
(11.87)** (4.00)** (0.39)
* Significant at the 0.05 level.
** Significant at the 0.01 level.
32
Table 4
Sample Comparison of Liquidity Pattern around Stock Split Announcements
The Pre-split Period is a 30-day period before the stock split announcement day. The Post-Split Period is a 30-day
period after the ex-split date. PRICE is the daily average trading price taken at the 30-second intervals. RETURN is
estimated by taking the natural log of the contemporaneous average bid-ask to its respective lagged average taken at
the 30-second intervals. VOLATILITY is the standard deviation of the daily continuous return. VOLUME is the
daily total trading volume adjusted by the number of outstanding shares. Market Value is the daily market value of
traded shares adjusted by the market value of the firm. Absolute Spread is the daily average of the absolute dollar
difference of ask and bid recorded at 30-second intervals on day t. Relative Spread is the daily average of the dollar
difference of ask and bid divided by the bid-ask midpoint recorded at 30-second intervals on day t. Volume Depth is
the sum of the number of shares at the highest bid and the number of shares at the lowest ask recorded at 30-second
intervals on day t (adjusted by the number of outstanding shares). Dollar Depth is the sum of the product of the
number of shares at the highest bid and the highest bid price and the product of the number of shares at the lowest
ask and the lowest ask price recorded at 30-second intervals on day t (adjusted by the market value of firm). Ask
Depth is the product of the number of shares at the lowest ask and the lowest ask price recorded at 30-second
intervals on day t (adjusted by the number of outstanding shares). Bid Depth is the product of the number of shares
at the highest bid and the highest bid price recorded at 30-second intervals on day t.
Pre-Split Post-Split t-statistics for
Period Period Mean Difference
PRICE 6.583 2.181 16.257**b
RETURN -0.0001 -0.001 2.317*b
VOLATILITY 0.007 0.012 -2.845**b
VOLUME 0.011 0.034 -5.065**a
Market Value 0.011 0.034 -5.055**a
Absolute Spread 0.089 0.024 23.684**b
Relative Spread 0.023 0.021 2.918**
Volume Depth 0.194 0.303 -2.163*b
Dollar Depth 0.240 0.475 -3.296**a
Ask Depth 0.110 0.258 -4.545**b
Bid Depth 0.130 0.217 -1.495
a The mean difference between the Pre-Split Period and the Post-Split Period is significant at the 0.05 level by the
Mann-Whitney test.
b The mean difference between the Pre-Split Period and the Post-Split Period is significant at the 0.01 level by the
Mann-Whitney test.
* Significant at the 0.05 level.
** Significant at the 0.01 level.
33
Table 5
Regression Analysis
CAR =
α
0 +
β
1 FACTOR +
β
2 MktValue +
β
3 VolRatio +
β
4 EPSChg +
β
5 Multiple
+
β
6 RetVar +
β
7 PriceDev +
β
8 ShareDev +
ε
(1)
CAR =
α
0 +
β
1 FACTOR +
β
2 MktValue +
β
3 VolRatio +
β
4 EPSChg +
β
5 Multiple
+
β
6 RetVar +
β
7 PriceDev +
β
8 ShareDev +
β
9 InsiderD +
β
10 InsiderD*MktValue +
ε
(2)
CAR is the cumulative abnormal return over the different periods examined (-60 ≤ t ≤ -1, -30 ≤ t ≤ -1, -10 ≤ t ≤ -1, -1
≤ t ≤ +1, -3 ≤ t ≤ +3, -5 ≤ t ≤ +5, -10 ≤ t ≤ +10, +1 ≤ t ≤ +10, +10 ≤ t ≤ +60, +10 ≤ t ≤ +120, +10 ≤ t ≤ +240 and +10
≤ t ≤ +360). FACTOR is the natural logarithm of the size of the split factor. MktValue is the natural logarithm of the
market value (the product of price and number of shares outstanding) of the firm for the month before the split
announcement. VolRatio is the ratio of pre-split trading volume to post-split trading volume normalized by the
number of shares outstanding. EPSChg is the percentage change in earnings per share of the current year to the
previous three years. Multiple is a dummy variable that takes the value of 1 if there is more than one split
announcement over the sample period from 1980 to 2000. RetVar is the standard deviation of return. PriceDev is the
natural logarithm of the deviation of share price from industry median price. ShareDev is the natural logarithm of
the deviation of shares outstanding from the industry median of shares outstanding. InsiderD is a dummy variable
which takes the value of 1 if there is insider buying activity around the split announcement. InsiderD*MktValue is
an interactive variable of InsiderD and MktValue. The p-values for the coefficients are adjusted for
heteroskedasticity using White’s procedure (1980).
Panel A: Descriptive Statistics
FACTOR MktValue VolRatio EPSChg RetVar PriceDev ShareDev
Mean 1.5061 13.5455 3.3391 0.0377 0.0445 1.4604 12.1847
Median 1.6094 13.3043 1.3481 0.1534 0.0357 1.3894 12.4372
Standard Deviation 0.9312 1.5763 5.3115 0.6149 0.0321 1.6813 1.1356
Maximum 4.6052 18.2418 35.3581 0.9565 0.1876 5.4951 15.7564
Minimum -3.6889 10.2736 0.0005 -0.9634 0.0055 -2.4572 8.3700
34
Table 5 (continued)
Regression Analysis
Panel B: Regression Result
Model (1) Model (2)
-60 ≤ t ≤ -1 -30 ≤ t ≤ -1 -10 ≤ t ≤ -1 -10 ≤ t ≤ +10 -60 ≤ t ≤ -1 -30 ≤ t ≤ -1 -10 ≤ t ≤ -1 -10 ≤ t ≤ +10
Beta Coefficient
(p-value)
Intercept -0.0202 0.0042 0.0094 -0.0254
-0.0216 0.0057 0.0059 -0.0327
(0.09) (0.78) (0.65) (0.50)
(0.07) (0.71) (0.78) (0.38)
FACTOR -0.0013 -0.0009 -0.0007 0.0098
-0.0011 -0.0009 -0.0006 0.0096
(0.19) (0.38) (0.78) (0.05)*
(0.22) (0.38) (0.83) (0.03)*
MktValue -0.0005 -0.0019 -0.0037 0.0002
-0.0003 -0.0020 -0.0034 0.0002
(0.47) (0.01)** (0.01)** (0.90)
(0.65) (0.01)** (0.02)* (0.90)
VolRatio -0.0015 -0.0015 -0.0002 -0.0009
-0.0015 -0.0015 -0.0002 -0.0010
(0.00)** (0.01)** (0.79) (0.26)
(0.00)** (0.01)** (0.81) (0.20)
EPSChg 0.0046 0.0049 0.0021 0.0048
0.0048 0.0049 0.0023 0.0048
(0.01)** (0.03)* (0.51) (0.23)
(0.01)** (0.03)* (0.48) (0.24)
Multiple -0.0015 -0.0021 0.0004 0.0072
-0.0009 -0.0027 0.0019 0.0102
(0.56) (0.42) (0.92) (0.21)
(0.74) (0.29) (0.68) (0.07)
RetVar 0.3288 0.2694 0.4561 0.4087
0.3305 0.2679 0.4599 0.4310
(0.00)** (0.00)** (0.00)** (0.01)**
(0.00)** (0.00)** (0.00)** (0.02)*
PriceDev 0.0015 0.0020 0.0040 -0.0019
0.0013 0.0022 0.0036 -0.0026
(0.01)** (0.02)* (0.01)** (0.33)
(0.03)* (0.02)* (0.03)* (0.18)
ShareDev 0.0017 0.0014 0.0022 0.0003
0.0016 0.0014 0.0022 0.0011
(0.04)* (0.24) (0.20) (0.86)
(0.07) (0.26) (0.21) (0.56)
InsiderD
0.0279 -0.0222 0.0467 -0.0222
(0.32) (0.48) (0.51) (0.77)
InsiderD*MktValue -0.0020 0.0017 -0.0034 0.0008
(0.33) (0.45) (0.47) (0.87)
Adj R2 0.6566 0.5284 0.4716 0.4289
0.6438 0.5087 0.4490 0.4597
F-statistic 12.9480 8.0035 6.1317 5.5994 10.0375 6.1770 4.6673 5.0832
(0.00)** (0.00)** (0.00)** (0.00)**
(0.00)** (0.00)** (0.00)** (0.00)**
* Significant at the 0.05 level.
** Significant at the 0.01 level.
