Earnings and Stock Splits

Author(s): Paul Asquith, Paul Healy and Krishna Palepu

Source: The Accounting Review , Jul., 1989, Vol. 64, No. 3 (Jul., 1989), pp. 387-403
Published by: American Accounting Association

Stable URL: https://www.jstor.org/stable/247596

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide
range of content in a trusted digital archive. We use information technology and tools to increase productivity and
facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org.

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
https://about.jstor.org/terms

American Accounting Association is collaborating with JSTOR to digitize, preserve and extend
access to The Accounting Review

This content downloaded from

86.59.13.237 on Fri, 24 Sep 2021 01:52:09 UTC
All use subject to https://about.jstor.org/terms
THIE ACCOUNTING REVIEW
Vol. LXIV, No. 3
July 1989

Earnin s and Stock Slits

Paul Asquith, Paul Healy, and Krishna Palepu

ABSTRACT: This paper examines whether stock splits convey information about earnings.
The results indicate that firms split their shares after a significant increase in earnings. Before
the stock split announcement, the market expects these earnings increases to be temporary.
The split announcement leads investors to increase their expectations that the past earnings
increases are permanent. The evidence also suggests that the market's reaction to split
announcements cannot be attributed to expectations of either future earnings increases or
near-term cash dividend increases.

STOCK splits have long been a puz- rather than due to any intrinsic effects of
zling phenomenon to financial the splits themselves.
economists. Stock splits usually This "dividend hypothesis," however,
occur after an increase in stock prices does not appear to fully explain the ob-
and usually elicit a positive stock price
reaction upon announcement.' The rea- 1 The seminal study by Fama et al. [1969] documents
stock price increases prior to the split. Bar-Yosef and
son for this announcement reaction, Brown [1977], Charest [1978], Foster and Vickrey
however, has not been fully understood. [1978], Woolridge [1983], and Grinblatt et al. [1984] all
Since stock splits themselves do not report positive stock price reactions to split announce-
ments.
directly affect a company's cash flows, 2 Dividend increases, as studies subsequent to Fama et
the increase in a company's stock price al. [1969] showed, cause stock prices to increase (e.g., see
at the time of these announcements must, Aharony and Swary [1980] and Asquith and Mullins
[19831).
assuming market efficiency, reflect the
release of new information. An explana-
tion first proposed by Fama et al. [19691 We wish to thank Robert Holthausen, Robert Kaplan,
Bill Kinney, Richard Leftwich, Robert Merton, Richard
and repeated in subsequent studies is Ruback, Karen Wruck, and two reviewers for their com-
that the market interprets stock split an-ments on an earlier draft of this paper. We also wish to
nouncements as improving the probabil- thank the participants at seminars at Harvard University,
New York University, the University of North Carolina,
ity of near-term dividend increases.2 and the University of Chicago. Finally, we wish to thank
Fama et al. [19691 report data consis- Lydia Magliozzi and Eric Wolff for their assistance in
tent with the dividend explanation: 71.5 data collection.

percent of their sample firms experience

Paul Asquith and Krishna Palepu,
a percentage dividend increase in the
year after the split which is larger than both at Harvard Business School; and
Paul Healy, MIT Sloan School of Man-
the average increase for all securities on
agement.
the New York Stock Exchange. Thus,
they argue that a large price increase at
Manuscript received November 1987.
the time of a stock split is due to altered Revisions received April 1988 and November 1988.
expectations concerning future dividends Accepted January 1989.

387

This content downloaded from

86.59.13.237 on Fri, 24 Sep 2021 01:52:09 UTC
All use subject to https://about.jstor.org/terms
 388 The Accounting Review, July 1989

served market reaction to stock split an- value the pre-split earnings growth. One
nouncements. For example, Grinblatt et managerial motive for splits cited fre-
al. [1984] report a significant stock price quently by practitioners which is consis-
reaction to the announcement of stock tent with this hypothesis is that splits are
splits by firms that do not pay cash divi- used to keep stock prices within an opti-
dends in the three years prior to the split. mal trading range.4
Since only 11 percent of their sample Earlier studies [Lakonishok and Lev,
firms initiate a cash dividend during the 1987; McNichols and Dravid, 19871 find
year following the split, they conclude evidence of dividend and/or earnings
that the valuation changes associated growth surrounding stock split announce-
with stock split announcements cannot ments.5 However, it is not possible to
be attributed totally to revised expecta- conclude from their findings that stock
tions about near-term dividend increases. splits convey information about earnings
An alternative to the "dividend hy- since their sample firms also announce
pothesis" is that the information re- dividend increases prior to, and simul-
vealed by stock splits concerns earnings taneously with, the split announcements.
rather than dividends [Fama, 1976, p. Earlier papers find that dividend in-
164]. Two types of earnings information creases are accompanied by prior earn-
could be conveyed by stock splits. First, ings growth and convey information on
splits could provide favorable informa- subsequent earnings [Offer and Siegel,
tion about improved future earnings per- 1987; Healy and Palepu, 19881. Thus, it
formance. This "signalling hypothesis," is not possible to distinguish the infor-
discussed by Grinblatt et al. [1984] and mation conveyed by stock splits from
Lakonishok and Lev [1987], is that stock that of simultaneous dividend announce-
splits are used by managers to convey ments. Further, Lakonishok and Lev
favorable information to the market. [1987] and McNichols and Dravid [19871
However, splits can provide a credible do not test whether there is a cross-
signal of future performance only if it is sectional relation between the market
costly for firms without favorable infor- reaction to the split announcement and
mation to split their stock.3 the prior earnings growth.
A second type of earnings information This paper reexamines whether stock
that could be conveyed by splits concerns splits convey information about firms'
pre-split earnings. Lakonishok and Lev earnings. In order to mitigate the con-
[1987] find that splitting firms have large founding information effects of simul-
earnings increases prior to the split. Prior taneous dividend changes noted above,
research [Brooks and Buckmaster, 1980; only firms that do not pay cash divi-
Beaver et al., 1980; and Freeman et al.,
1982] indicates that large earnings in- 3 Brennan and Copeland [1987] suggest that one cost
of splits is the increase in investors' transaction costs be-
creases are usually transitory and fol- cause of fixed commissions and the odd lots created by
lowed by earnings decreases. Therefore, the splits.
investors are likely to treat pre-split earn- 4 See, for example, Baker and Gallagher's [1980]
survey of managerial motives for stock splits. Also,
ings increases as largely transitory at the
Lakonishok and Lev [1987] and Defeo and Jain [1988]
time of their announcement. If split deci- provide empirical evidence on absolute share price levels
sions are based on managers' superior which supports the concept of a "normal range."
' McNichols and Dravid [1987] examine analysts'
information that pre-split earnings in-
earnings forecasts around stock splits and finds that
creases are permanent, then the split an- there are statistically significant pre-split upward revi-
nouncement will lead investors to re- sions.

This content downloaded from

86.59.13.237 on Fri, 24 Sep 2021 01:52:09 UTC
All use subject to https://about.jstor.org/terms
Asquith, Healy, and Palepu 389

dends at the time of the stock split are market's reaction to split announce-
included in the sample. Further, we ments is not due to anticipation of near-
examine the cross-sectional relation be- term dividend increases.
tween the information conveyed by the The paper is organized as follows. The
stock splits and surrounding earnings data and sample are described in the next
changes. section. Section II presents the hypoth-
Our tests, based on a sample of 121 eses and Section III presents the tests and
stock split announcements from the results. The paper's conclusions are dis-
period 1970-1980, lead to several conclu- cussed in Section IV.
sions. Stock splitting firms experience
I. DATA
significant earnings increases for several
years before the stock split. Even when An initial sample of stock splits is
their earnings are adjusted for industry identified from the CRSP Daily Master
performance, splitting firms experience file. To be included in the sample, a firm
significant earnings increases in the year has to: (1) announce a stock distribution
prior to the split. These earnings increases (designated by the firm as either a stock
are permanent, since they are not split or stock dividend) of at least 25 per-
reversed for at least four years subse- cent during the time period 1970-1980;
quent to the split. However, the associa- (2) pay no cash dividends prior to or at
tion between annual changes in stock the time of the stock distribution an-
prices and contemporaneous earnings nouncement; (3) not announce any other
changes for the splitting firms is lower stock distribution of 25 percent or more
in
the two years prior to the split than in in the five years prior to the event; (4) be
"normal" years, indicating that the mar- listed on the New York or American
ket views the earnings increases in these Stock Exchange at the time of the event;
two years as likely to be transitory. Fi- (5) have stock price and return data
nally, cross-sectional tests indicate that available for the announcement date
the larger the earnings changes in the two and two days prior to the event; and (6)
years prior to the split, the larger is the have the stock distribution announce-
market reaction to the split announce- ment date available in The Wall Street
ment. Thus, at the stock split announce- Journal, or on the CRSP Daily Master
ment, investors appear to increase their file.
expectations that past earnings increases Grinblatt et al. [1984] hypothesize that
are permanent rather than transitory. there are potentially different costs asso-
There is evidence of significant earn- ciated with stock distributions greater
ings increases in the year of the split. than 25 percent (stock splits) and those
However, the earnings changes in the less than 25 percent (stock dividends).
four subsequent years are insignificant. While stock dividends generally reduce
Also, we find no relation between stock firms' retained earnings, stock splits do
split announcement returns and post- not. The reduction of retained earnings
split earnings changes. These findings are for stock dividends creates a potential
inconsistent with the hypothesis that cost to the firm by making certain debt
splits signal future earnings increases. covenants more binding. This cost is ab-
Finally, in the post-split years the per- sent for stock splits. Hence, stock divi-
centage of splitting firms initiating cash dends and stock splits may convey differ-
dividends is similar to that for the popu- ent information to the market. Results
lation of all firms, indicating that the of Grinblatt et al. [1984] and Lakoni-

This content downloaded from

86.59.13.237 on Fri, 24 Sep 2021 01:52:09 UTC
All use subject to https://about.jstor.org/terms
390 The Accounting Review, July 1989

shok and Lev [1987] support this propo- four quarterly earnings reports prior to
sition. The present sample comprises only that event; and the earnings for years -6
distributions that do not affect retained to -2 and 1 to 4 are constructed using
earnings. 6 the quarterly earnings data from quar-
Cash dividend paying firms that split ters -24 to -5, and from quarters 4 to
their stock often make contemporaneous 19. Defined this way, the earnings for
announcements concerning dividend years -6 to -1 are announced strictly
changes. The sample is, therefore, re- before the split announcement, and the
stricted to non-cash dividend paying earnings for years 0 to 4 are announced
firms in order to separate the earnings strictly after the split date. This is done
information conveyed by stock splits because stock split announcements occur
from the information conveyed by cash throughout a fiscal year and fiscal year
dividend announcements.7 The restriction earnings include some quarters before
of no other stock distributions in a five- the split and some quarters after the
year period is to insure independence of split. By using quarterly data to create
observations in the sample since five annual event earnings the earnings can
years of prior earnings are analyzed with be aligned with the split announcement
each split. Listing on the AMEX or the and earnings performance before and
NYSE is required to guarantee data after the announcement can be strictly
availability. Announcement dates are separated.
necessary to properly determine the Thus, to be included in the final sam-
market's reaction. ple a firm has to satisfy the following
For each firm in the initial sample the conditions: (1) the date of the quarterly
following additional data are collected: earnings announcement immediately
(1) stock prices at the beginning of each preceding the split announcement is
of the five years before and five years reported in The Wall Street Journal
following the split announcement, and Index, and (2) earnings for two or more
for two days prior to the split announce-
ment; (2) returns on the firm's common 6 See American Institute of Certified Public Accoun-
tants [1953, Chapter 7B1] for a more complete discussion
stock, and on the CRSP equally-weighted of the accounting rules for stock distributions. Compa-
market index, for five years before and nies have some discretion in applying these accounting
five years following the split announce- rules. For example, distributions greater than 25 percent
can be accounted for as large stock dividends, thus
ment; (3) the first quarterly earnings an- reducing retained earnings. However, none of the firms
nouncement date preceding the split an- in the present sample uses this treatment.
nouncement; and (4) earnings per share ' This sample selection does not rule out the possibility
that split announcements implicitly provide information
before extraordinary items and discon-
on subsequent dividend changes. This issue is examined
tinued operations from the 24 quarterly in the empirical test of hypothesis H4.
announcements immediately preceding 8 Stock price and return data are collected from the
CRSP files, announcement dates are collected from The
the split date (to construct earnings for
Wall Street Journal Index, and earnings are collected
years -6 to -1), and the 20 quarterly from Quarterly COMPUSTAT files. Earnings, price,
announcements following the split date and return data are not available for all the sample firms
for all years. For years before the stock split, data for
(to construct earnings for years 0 to 4).'
some sample firms are not available on the CRSP or
Annual earnings are defined so that COMPUSTAT files primarily because the firm was not
year 0 earnings are computed from the listed on either NYSE or AMEX. For years after the
stock split, data are not available primarily because firms
four quarterly announcements subse-
were delisted, merged, or liquidated. The actual number
quent to the split (quarters 0 to 3); year of firms for which usable data are available is indicated in
-I earnings are constructed from the the results tables discussed later in the paper.

This content downloaded from

86.59.13.237 on Fri, 24 Sep 2021 01:52:09 UTC
All use subject to https://about.jstor.org/terms
 Asquith, Healy, and Palepu 391

TABLE 1

DISTRIBUTION OF STOCK SPLITS IN THE PERIOD 1970 TO 1980

BY YEAR FOR 121 FIRMS THAT DO NOT PAY CASH DIVIDENDSa

Number of Percent of
Year Firms Total Sample

1970 10 8.3
1971 28 23.1
1972 29 24.0
1973 2 1.7
1974 1 0.8
1975 5 4.1
1976 6 5.0
1977 1 0.8
1978 9 7.4
1979 6 5.0
1980 24 19.8

Total 121 100.0

a To be included in the sam

stock dividend of at least 2
announcement; (3) no other
event; (4) the firm is listed
available in The Wall Street
for the announcement date
the event is reported in Th
available on COMPUSTAT for the six years before and the five years after the event date.

years can be constructed from the Quar- sample firm is defined as the set of firms
terly COMPUSTAT Industrial tapes. listed on Standard and Poor's COMPU-
The final sample consists of 121 firms STAT Industrial tapes with the same
that do not pay cash dividends and an- four-digit SIC Code.9
nounce a stock distribution of at least 25
II. HYPOTHESES
percent during the period 1970-1980.
Table 1 reports the number of stock Since stock split announcements fol-
splits by calendar year. About 55 percentlow and are contemporaneous with stock
of the sample stock splits are from the price increases, assuming market effi-
three-year period 1970-1972; another 20 ciency, they must reveal new informa-
percent are from the year 1980; the re- tion about the firm. This paper examines
maining 25 percent are from the years three types of information that could be
1974-1979. Thus, there is evidence that
the splits are clustered in time. 9 Industry earnings data are constructed from annual,
To test whether any observed earnings rather than quarterly, COMPUSTAT files. Since this file
does not construct a quarterly Research tape, use of
patterns of the sample firms are due to
COMPUSTAT quarterly data will lead to exclusion of
industry and time-related trends, indus- delisted firms from the inustry earnings. This problem is
try data on earnings per share before avoided by using COMPUSTAT Annual Industrial and
Research tapes. One limitation of this approach is that
extraordinary items and discontinued
the time period definitions for the sample firms' earn-
operations are also collected for each ings may differ marginally from those for the industry
final sample firm. The industry for each figures.

This content downloaded from

86.59.13.237 on Fri, 24 Sep 2021 01:52:09 UTC
All use subject to https://about.jstor.org/terms
392 The Accounting Review, July 1989

conveyed by stock splits. First, splits H2: The market reaction to the an-
could provide favorable information nouncement of a stock split is
about improved future earnings perfor- positively related to the earnings
mance. Second, splits could convey fa- changes subsequent to the split.
vorable information on pre-split earn-
As noted above, splits can convey in-
ings. In particular, if splitting firms
formation about pre-split, rather than
experience prior earnings growth, a split
post-split, earnings. Brooks and Buck-
could indicate that managers view these
master [1980], Beaver et al. [1980], and
increases as permanent, rather than tran-
Freeman et al. [1982] report that firms
sitory. Finally, splits could provide
with large earnings increases in one year
information on increases in near-term
are likely to experience earnings declines
dividends. 10
in the following year. In other words,
The first of the above hypotheses is
large earnings changes are usually tran-
discussed by Grinblatt et al. [1984] and
sitory and followed by earnings decreases.
Lakonishok and Lev [1987]. They argue
Beaver et al. [1980] also report that,
that if managers have superior informa-
since large earnings increases are par-
tion to investors on firms' performance,
tially transitory, investors do not fully
stock splits may be used to convey favor-
incorporate them into a firm's future ex-
able information to the market. How-
pected earnings. Hence, if there are earn-
ever, splits can only provide a credible
ings increases prior to the stock split
signal of future performance if it is
announcement, investors may attach a
costly for firms without favorable infor-
low probability to these earnings in-
mation to split their stock.
creases being permanent. The stock split
To test whether stock splits convey in-
may indicate to investors that managers
formation on future performance, earn-
expect pre-split earnings increases to be
ings changes in years subsequent to splits
permanent.
are examined. Prior studies of the usual
To test whether stock splits convey in-
time-series behavior of annual earnings
formation on pre-split earnings, the fol-
(e.g., Ball and Watts [1972] and Watts
lowing three hypotheses are tested:
and Leftwich [1977]) suggest that annual
earnings follow a random walk. Thus, H3: Firms that split stock experience
the average earnings change for a ran- significant earnings increases in
dom sample of firms is expected to be the years prior to the split an-
zero. However, if stock splits communi- nouncement and no earnings de-
cate favorable information about future clines in the post-split years.
earnings, average earnings changes for
H4: The market reaction to the an-
splitting firms subsequent to the split are
nouncement of a stock split is
expected to be positive. Two hypotheses
positively related to the earnings
are tested in the paper to evaluate
increases prior to the split.
whether splits provide information on
subsequent earnings performance: H5: The relation between annual ab-
normal stock returns and con-
HI: Firms that split stock experience
significant earnings increases in
Alternatively, the split could provide information
the years subsequent to the split about the firm's risk. This paper does not investigate this
announcement. hypothesis.

This content downloaded from

86.59.13.237 on Fri, 24 Sep 2021 01:52:09 UTC
All use subject to https://about.jstor.org/terms
Asquith, Healy, and Palepu 393

temporaneous earnings changes returns for days AD-240 to AD-121,

is lower than normal in years AD-120 to AD-61, AD-60 to
prior to the split. AD-21, AD-20 to AD-11, and
AD -10 to AD -2 are all positive and
Hypotheses H3 and H4 are similar to hy- significant. The cumulative abnormal re-
potheses HI and H2, but examine pre- turns from AD-240 to AD are 56.8 per-
split, rather than post-split, earnings. cent. The cumulative raw returns for this
Hypothesis H5 examines the relation be- period are 66.4 percent. Thus, stock split
tween annual unexpected returns and announcements are preceded by large in-
contemporaneous earnings changes. Prior creases in the announcing firms' stock
prices.
studies (e.g., Ball and Brown [1968] and
Beaver et al. [1980]) find that there is a During the two days surrounding the
positive relation between unexpected re- split announcement (AD -1 and AD),
turns and earnings. However, if investors the mean market-adjusted return is 3.7
discount pre-split earnings increases as
percent, which is significantly different
largely transitory, there will be a lower from zero at the 1 percent level. The
than normal price to earnings elasticity
median abnormal return is similar (3.5
for these years.
percent). Further, 76 percent of the sam-
In addition to the above earnings hy- ple firms have positive abnormal an-
potheses, HI to H5, this paper also pro- nouncement period returns. Controlling
vides some evidence on the "dividend for other information announced simul-
hypothesis" originally proposed by taneously with the stock split (i.e., earn-
Fama et al. [1969] and subsequently dis- ings, announcements, and financing
announcements) increases the mean mar-
cussed by Grinblatt et al. [1984] and
ket-adjusted return to 4.4 percent for 84
Lakonishok and Lev [1987]. That is:
firms and the percentage of firms with
H6: The stock market responds posi- positive abnormal returns to 78 percent. 12
tively to stock splits because they Finally, the abnormal returns for the 20
usually mean increased near-term days subsequent to the stock split an-
cash dividends. nouncement are insignificant.
The abnormal returns in Table 2 are
III. RESULTS similar to those reported by earlier stud-
ies. For example, Fama et al. [1969] re-
Market Reaction to Stock Split port a cumulative abnormal return of
Announcements 35.8 percent during the 29 months pre-
Table 2 reports market-adjusted and ceding the stock split announcements in
raw returns around the announcement of their sample. Grinblatt et al. [1984]
stock splits. Market-adjusted returns are report a 3.41 percent mean two-day an-
computed using the CRSP equal-weighted
market index.1 The split announcement " We also compute risk-adjusted returns for the same
period. The results are similar in both size and signifi-
day (AD) is the date The Wall Street cance levels to those reported here.
Journal reports the stock split. The mean 12 While the clean announcement sample excludes ob-
market-adjusted returns are insignificant servations with other contemporaneous news releases, it
may also introduce a selection bias, the effect of which is
for the period AD -480 to AD -251 unknown. The discussion of subsequent tests, therefore,
days and significant for days AD -240 emphasizes the full sample results even when clean an-
to AD -2. The holding period abnormal nouncement sample results are also reported.

This content downloaded from

86.59.13.237 on Fri, 24 Sep 2021 01:52:09 UTC
All use subject to https://about.jstor.org/terms
 394 The Accounting Review, July 1989

TABLE 2

RAW AND MARKET-ADJUSTED RETURNS FOR VARIOUS HOLDING

PERIODS SURROUNDING STOCK SPLIT ANNOUNCEMENTS BY
FIRMS THAT DO NOT PAY CASH DIVIDENDSa
(1970-1980)

Raw Returns Market-Adjusted Returns

Holding Periodb Mean t-statistic Mean t-statistic

Panel A. Full Sample:

AD-480 to AD-361 13.4% 3.0, 12.4% 1.5

AD-360 to AD-241 7.6 1.7d 6.6 1.5
AD-240 to AD-121 20.6 4.7c 19.4 4.5c
AD-120 toAD-61 17.1 5.1c 13.4 4.4c
AD-60 to AD-21 18.0 7.1c 14.6 5.8c
AD-20 to AD-11 3.3 2.6d 2.5 2.0d
AD-10 to AD-2 3.7 3.1c 3.2 2.7c
AD-1 to AD 3.7 6.7c 3.7 6.8c
AD+1 to AD+10 0.2 0.2 0.3 0.2
AD+ll to AD+20 -0.1 -0.1 0.1 0.1

Panel B. Sample with No Other Announcements:

AD-480 to AD-361 15.5 % 2.7c 12.8% 2.6c

AD-360 to AD-241 8.3 1.4 6.8 1.4
AD-240 to AD-121 16.7 2.9c 16.3 3.3c
AD-120toAD-61 15.5 3.8c 11.4 3.3c
AD-60 to AD-21 16.6 4.9c 13.1 4.6c
AD-20 to AD-11 4.5 2.7d 3.4 2.4d
AD-10 to AD-2 3.8 2.4d 3.5 2.6d
AD-1 to AD 4.5 6.3c 4.4 7.1c
AD+1 to AD+10 1.6 1.0 1.6 1.1
AD+11 toAD+20 0.2 0.1 0.4 0.3

a The full sample comprises 121

period 1970-1980. The sample wi
computed using equal-weighted market returns.
b The holding periods are for days relative to The Wall Street Journal announcement of the split (AD).
c Significantly different from zero at the 0.01 level (two-tailed test).
d Significantly different from zero at the 0.05 level (two-tailed test).

nouncement return for a large sample of above, all earnings are calculated an-
stock split and stock dividends. nually for the four quarters relative to
the split announcement." To aggregate
Earnings Performance Before and
results across firms, earnings changes are
After Stock Splits
standardized for each firm j in years -5
To determine whether there is a sys- to 4 by its stock price two days before the
tematic earnings pattern for firms that announcement of the stock split, Pj. The
split stocks, earnings changes are esti-
mated for five years before (years -5 to 13 Since stock splits occur throughout a fiscal year and
- 1), the year of the stock split an- since quarterly earnings are used, this method ensures
that all earnings in year 0 are announced after the stock
nouncement (year 0), and four years sub- split. This analysis is repeated using fiscal year earnings
sequent (years 1 to 4). As mentioned data and the results are similar.

This content downloaded from

86.59.13.237 on Fri, 24 Sep 2021 01:52:09 UTC
All use subject to https://about.jstor.org/terms
 Asquith, Healy, and Palepu 395

TABLE 3

SUMMARY STATISTICS ON CHANGE

OF INITIAL EQUITY PRICE FOR YEARS SURROUNDING ANNOUNCEMENTS OF STOCK
SPLITS BY FIRMS THAT DO NOT PAY CASH DIVIDENDSab
(1970-1980)

Period Relative to Number of Student t First Third

Stock Split Firms Mean Probabilityc Quartile Median Quartile

Panel A. Raw Earnings Changes:

Year -5 35 0.61W0 0.57 -0.17% 0.58% 1.4407o

-4 44 1.05 0.02 -0.26 0.41 1.29
-3 61 0.66 0.12 -0.56 0.37 1.63
-2 84 1.24 0.01 0.15 0.92 2.32
- 1 100 2.55 0.01 0.76 1.37 2.52
0 118 2.03 0.01 0.65 1.28 2.88
1 117 0.91 0.17 -0.65 0.93 2.16
2 110 -0.65 0.32 -2.43 0.56 2.28
3 101 2.10 0.11 -1.40 0.55 4.13
4 98 2.82 0.15 -0.87 1.26 4.65

Panel B. Industry-Adjusted Earnings Changes:

Year -5 33 0.90qo 0.38 - 1.70qo 0.06%o 1.55qo

-4 42 0.06 0.90 -1.67 -0.60 0.64
-3 56 -0.48 0.37 -2.46 -0.57 1.37
-2 77 0.09 0.83 -1.33 0.26 1.43
- 1 93 2.82 0.01 -0.18 0.74 2.17
0 109 -0.03 0.97 -1.34 0.27 1.80
1 108 -0.45 0.52 -2.39 -0.36 1.15
2 100 -0.63 0.39 -2.99 0.24 2.07
3 91 2.94 0.05 -2.36 1.33 4.26
4 88 1.22 0.58 -4.65 0.75 4.20

a The stock split sample comprises 1

1980. Earnings changes are not avail
available in year - 1. Industry-adjust
ings changes for a given firm in yea
split firm in year t and the median
bEarnings changes are estimated us
the 20 quarters subsequent. Changes
standardized by the firm's stock pric
c Student t-test statistics test the hy
levels are for two-tailed tests of significance.

standardized earnings change in year t, the stock split sample firms are signifi-
AEj,t, is therefore defined as: cantly different from zero in each year.
Panel A in Table 3 reports the mean
AEjt=(Aj,t-Aj,t-1)/Pj
and quartile earnings changes and the
t= -5, . .4 (1) number of firms for which data are
available for the stock split sample in
where Ajt is earnings before
years -5 to 4.14 The extra
mean standardized
nary items and discontinued operations
for firm j in year t. Student t-statistics
14 The number of firms for which there is usable earn-
are calculated to indicate whether the ings data varies across years -5 to 4: the lowest is 35
mean standardized earnings changes of firms in year -5 and the highest is 118 firms in year 0.

This content downloaded from

86.59.13.237 on Fri, 24 Sep 2021 01:52:09 UTC
All use subject to https://about.jstor.org/terms
 396 The Accounting Review, July 1989

earnings changes for years -4 to 0 are median earnings change from the firm's
1.05 percent, 0.66 percent, 1.24 percent, earnings change.
2.55 percent, and 2.03 percent, respec- Panel B of Table 3 reports the mean
tively. These values are significant at the and the quartile industry-adjusted earn-
2 percent level or better in four of the ings changes. The mean industry-ad-
five years; the mean for year - 3 is justed earnings changes are insignificant
significant at the 12 percent level. Mean in all years except -1 and 3. The mean-
earnings changes are insignificant in adjusted earnings change is 2.82 percent
years 1 to 4.15 in year -1 and 2.94 percent in year 3,
The above results indicate that firms both significant at the 0.05 level or
that announce stock splits experience a better. These results indicate that the
significant earnings increase for the four strong earnings performances of the
years before the stock split announce- splitting firms are matched in general by
ment. The largest of these increases their industries' performances. The split
occurs in the year immediately before the firms are in industries that have been
stock split announcement. The earnings experiencing good times. In the pre-split
continue to increase during the year of period, split firms perform significantly
the split. While earnings increases cease better than their industries only in year
in the one year subsequent to the stock -1. This superior earnings performance
split, past increases are not reversed in by splitting firms relative to their indus-
those years. Therefore, the split firms' tries is permanently incorporated into
earnings increases prior to and contem- their earnings levels: in the years subse-
poraneous with stock split announce- quent to the split announcements, the
ments appear to be permanent. split firms do not experience signifi-
Since the splits in the sample are clus- cantly lower earnings growth than their
tered in time, it is possible that the earn- industries.
ings performance of the split firms is due The above results are similar to those
to industry- and time-related factors of Lakonishok and Lev [1987], who ex-
rather than firm-specific factors. To test amine earnings growth rates around stock
this, the split firms' performance is com- splits. However, their sample includes
pared to the performance of their indus- firms that have contemporaneous cash
tries. For each sample firm, industry dividend increases, making it difficult to
earnings changes for years -5 to 4 are relate the earnings growth to the splits
estimated using all firms that (1) have per these. This limitation is mitigated in the
same four-digit SIC code on Standard
and Poor's Annual COMPUSTAT and
The number of observations is low in years -5 to -2
Research Industrial tapes, and (2) have mainly because many of the firms in the sample are newly
earnings data available for two consecu- listed at the time of the stock split and, therefore, do not
have earnings data in these years; the number of observa-
tive years of the 11 years surrounding
tions after year 3 declines principally because of delist-
a split announcement. The earnings ments. Of the 86 firms with unavailable data in year -5,
changes for these comparison firms are 73 are not listed on the NYSE or AMEX and 13 are un-
available on COMPUSTAT. Of the 23 firms with un-
then standardized by their stock prices
available data in year 4, six are acquired, two are
two days prior to the test firm's split liquidated, five are delisted by the exchange, two are
announcement and an industry median is suspended by the exchange, and eight are unavailable on
COMPUSTAT.
constructed. Industry-adjusted earnings
15 The mean earnings changes in years 3 and 4 a
in each year are calculated for each split- large as years - 1 and 0 but not as significant. This is
ting firm by subtracting its industry to increased cross-sectional volatility in earnings.

This content downloaded from

86.59.13.237 on Fri, 24 Sep 2021 0101 Jan 1976 12:34:56 UTC
All use subject to https://about.jstor.org/terms
 Asquith, Healy, and Palepu 397

TABLE 4

SUMMARY OF CORRELATIONS BETWEEN SPLIT ANNOUNCEMENT RETURNS

AND STANDARDIZED EARNINGS CHANGESa SURROUNDING STOCK SPLITS
FOR FIRMS THAT DO NOT PAY CASH DIVIDENDSb
(1970-1980)

Pearson Spearman
Period Relative to Number of Correlation Correlation
Stock Split Firms Coefficient Probability Coefficient Probability

Year -5 35 -0.01 0.98 0.21 0.22

-4 44 0.03 0.83 0.01 0.93
-3 61 -0.05 0.71 -0.16 0.22
-2 84 0.29 0.01 0.32 0.01
- 1 100 0.20 0.04 0.24 0.02
0 118 0.02 0.85 -0.09 0.34
1 117 0.00 0.99 -0.02 0.86
2 110 -0.12 0.25 -0.11 0.29
3 101 0.12 0.30 0.18 0.13
4 98 0.15 0.24 0.18 0.17

a The stock split sample co

1980. Earnings changes are
available in year - 1.
bEarnings changes are estimated using quarterly earnings for the 24 quarters prior to the split announcement and
the 20 quarters subsequent. Changes in earnings per share before extraordinary items and discontinued operations are
standardized by the firm's stock price two days prior to the announcement of the stock split. The announcement
return for each firm is the market-adjusted return, using equal-weighted market returns, for one day prior to and the
day of the stock split announcement.

present analysis, since we examine firms market's reaction at the announcement

that do not pay cash dividends. of a stock split and the earnings changes
To summarize, firms that announce documented above (hypotheses H2 and
stock splits experience significant earn- H4). Table 4 reports the Pearson and
ings increases for several years prior to Spearman correlation coefficients be-
the year of the stock split announce- tween the split announcement returns
ments. In the year immediately prior to and the raw earnings changes (standard-
the split announcement, the split firms ized by price) in each of the years -5 to
outperform their industries. This in- 4. The announcement return for each
crease in earnings appears to be perma- firm is the market-adjusted return for
ment and is not reversed for at least fourone day prior to and the day of the stock
years after the split. These findings are split announcement. Standardized earn-
consistent with hypothesis H3. In the ings changes in each year are computed
post-split years, there are significant as described above.
earnings increases only in the year of the The Pearson correlation coefficients
split. This is weakly consistent with HI. are insignificant in years -5 to -3 and 0
to 4. In years -2 and -1, the correla-
Stock Split Announcement Returns
tions are 0.29 and 0.20, both significantly
and Earnings Changes
different from zero. The Spearman rank
This section examines whether there is correlation results are similar, indicating
a cross-sectional relation between the that the Pearson correlations are not

This content downloaded from

86.59.13.237 on Fri, 24 Sep 2021 01:52:09 UTC
All use subject to https://about.jstor.org/terms
 398 The Accounting Review, July 1989

TABLE 5
TESTS OF THE RELATION BETWEEN STO
RETURNS AND EARNINGS CHANGES SURROUNDING THE SPLIT
FOR FIRMS THAT DO NOT PAY CASH DIVIDENDSa b
(1970-1980)

SRET = i+f,31,ALEj2+,f32 AEJ1 + 033Ej 0+34,AEi,1 +EJ

Sample with No Other

Full Sample Announcements

Coefficient t-statistic Coefficient t-statistic

a 0.03 3.7c 0.03 4.3c

0l 0.29 2 .0d 0.27 1.7'
32 0.25 2.Id 0.31 2.2d
03 -0. 14 - 1.17 -0.24 - 1.5
34 0.04 0.62 0.07 1.1
R2 0.115 0.132
n 79 54

a The stock
1980. Forty
available fo
prises 54 of these firms.
I AE., is the change in earnings in year t (estimated from quarterly earnings) standardized by the stock price two
days prior to the stock split announcement; SRETj is the market-adjusted return for firm j for one day prior to and
the day of the stock split announcement, estimated using an equal-weighted market index.
c Significant at the 0.01 level (two-tailed test).
d Significant at the 0.05 level (two-tailed test).
" Significant at the 0.10 level (two-tailed test).

driven by a few outliers. These correla- turn for firm j for one day prior to and
tions indicate that the split announce- the day of the stock split announcement
ment returns are positively related to and AEj1, is the earnings change in year t
earnings increases for the two years prior standardized by the stock price two days
to the split, consistent with H4. There is prior to the stock split announcement.
no relation between the announcement Table 5 reports regression estimates
returns and the earnings changes after for both the full sample and the sam-
the split date even though year 0 has ple of firms that have no other simulta-
a significant earnings increase. This is neous announcements. For the full sam-
inconsistent with H2. ple, the estimated coefficients for earnings
The split announcement returns are changes in years -2 and -1 are 0.29
also regressed on earnings changes in the and 0.25, both significantly different
two years before the split (years -2 and
- 1) and the two years after the split 16 The announcement returns are also regressed on
earnings changes for longer periods. The results are not
(years 0 and 1).16 The following regres-
qualitatively different from those reported below. Only
sion equation is estimated: results for years - 2 to + 1 are reported because the num-
ber of usable earnings observations varies across years
SRET3= CI + flAEJ,2 + f2AEi,- (see Table 4) and these years have the largest number of
+f33AEj,0+341AEI +E (2) observations. Also, as the correlation results in Table 4
show, the other years' earnings changes are not related to
where SRETj is the market-adjusted the
re- announcement returns.

This content downloaded from

86.59.13.237 on Fri, 24 Sep 2021 01:52:09 UTC
All use subject to https://about.jstor.org/terms
 Asquith, Healy, and Palepu 399

from zero at the 0.05 level in a two-tailed where ARj1, is the annual market-
test. These estimates indicate that, all adjusted return for firm j in year t, the
else being constant, a ten percent in- difference between the firm's return and
crease in standardized earnings in year the return on the equal-weighted market
-2 (year -1) is likely to be accompanied index; AE1,, is the change in earnings per
by a 2.9 percent (2.5 percent) market- share for firm j in year t deflated by stock
adjusted return in the two days sur- price at the beginning of the year;18 and
rounding the stock split announcement. DA is an indicator variable that is one in
The coefficients of the earnings changes years -1 and -2, and zero in all other
in years 0 and 1 are insignificant.'7 Re- years.
sults for the subsample with no other To illustrate the method used to com-
simultaneous announcements are similar pute annual returns, consider a company
to those for the full sample. The esti- that announces a stock split on May 15,
mated coefficients for earnings changes 1976. The quarterly earnings announce-
in years -2 and -1 are significant at the ment immediately prior to the split is on
0.10 and 0.05 levels, and the coefficients April 16. The return for year - 1 is the
for earnings in years 1 and 2 are insig- cumulative return for the year ending
nificant. two weeks after the earnings announce-
The above results are consistent with ment date, that is, April 30, 1976. The
hypothesis H4 and inconsistent with hy- earnings for year - 1 are the sum of the
pothesis H2. The information released four quarterly earnings announced dur-
by the stock split announcements is re- ing this period. Defined this way, the
lated to earnings increases in the two year - 1 return includes these four quar-
prior years, but not to subsequent earn- terly earnings announcements. If the
ings growth. One interpretation of these split announcement occurred within two
results is that when firms experience weeks of the earnings announcement,
large earnings increases the market as- year - 1 is defined as ending two days
sumes that the increases are at least par- prior to the split date, May 13, 1976.
tially transitory. When splits are an- Years -5 to -2 are the four years prior
nounced, the market revises upward the to - 1; years 0 to 4 are the five subse-
probability that the past earnings in- quent years. Each of these years is ex-
creases are permanent, leading to an in- pected to include the four quarterly earn-
crease in the firms' stock prices. ings announcements used to compute

Abnormal Returns and Earnings 17Specification tests are conducted to assess whether
Changes Surrounding Stock Splits the residuals are homoscedastic, normally distributed,
and cross-sectionally dependent. One cannot reject the
To investigate whether the market dis- hypotheses that the residuals are homoscedastic and nor-
mally distributed at the 0.05 level. To test whether the
counts earnings changes in the two years
residuals are cross-sectionally dependent, regression ob-
prior to a stock split, the relation be- servations are sorted by split announcement date. The
tween annual abnormal stock returns and Durbin-Watson statistic indicates that there is no signifi-
cant residual cross-correlation.
earnings changes for years surrounding
18 Annual earnings are constructed using quarterly
the stock split is estimated as follows: earnings aligned in relation to the stock split announce-
ment, as discussed above. However, in previous tests
ARjt=Xo+XAEj,t earnings changes were deflated by the stock price two
+ X2AEjtDt + Ej,t days prior to the split announcement, whereas earnings
changes in these tests are standardized by the beginning
t = - 5, . . . ,4 (3) of each year price.

This content downloaded from

86.59.13.237 on Fri, 24 Sep 2021 01:52:09 UTC
All use subject to https://about.jstor.org/terms
 400 The Accounting Review, July 1989

TABLE 6

TESTS OF THE RELATION BETWEEN ANNUAL MARKET-ADJUSTED RETURNS

AND EARNINGS CHANGES IN YEARS SURROUNDING STOCK SPLITS
FOR FIRMS THAT DO NOT PAY CASH DIVIDENDSa b
(1970-1980)

ARjt =XO+ XIAEJ.t+ X2AEj.tDt + ej.,

Coefficient t-statistic

X0 0.46 2.8c
)x, 5.12 8.0c
X2 -5.18 -2.Od
Adjusted R2 0.083
n 705

a The stock split

1980.
b A Ei, is the ch
stock price at th
taking the value one in years -2 and -1, and zero otherwise.
c Significant at the 0.01 level (two-tailed test).
d Significant at the 0.10 level (two-tailed test).

earnings for that year. Actual earnings two years prior to the stock split, the
announcement dates in these years are relation between earnings and price
not collected since there is evidence from changes is significantly lower, as indi-
prior studies that earnings announce- cated by the negative estimate for X2.
ment dates are stable across years During these two years, there is no linear
[Chambers and Penman, 1984]. relation between abnormal returns and
The coefficient Xh is the average price changes in earnings.20 This result is con-
to earnings elasticity in years -5 to -3 sistent with hypothesis H5, that investors
and 0 to 4. Earlier studies find that this view the earnings increases in these two
coefficient is positive [Ball and Brown, years as transitory, rather than perma-
1968]. The coefficient X2 measures the ment.21
deviation of the price to earnings elas-
' Specification tests are conducted to assess whether
ticity in years -2 to - 1 relative to other the residuals are homoscedastic, normally distributed,
years. If, as hypothesized in the previous and cross-sectionally independent. One cannot reject
section, the market discounts large earn- the hypotheses that the residuals are homoscedastic and
cross-sectionally independent at the 0.05 level, but one
ings increases experienced by the split can reject the hypothesis that the residuals are normally
firms in years -2 and - 1, X2 will be distributed. This implies that the least squares estimates
negative. are still best linear unbiased and consistent, but the
standard formulae for t- and F-tests are inappropriate.
Table 6 presents results of the estima- However, the central limit theorem implies that standard
tion of equation (3). 9 The estimate for statistical tests are approximately correct for reasonably
X, is 5.12 and is significant at the 0.01 large samples, as in this case (see Pindyck and Rubinfeld
[1976]).
level, indicating that on average there is
20 To determine whether there is a nonlinear relation
a positive association between price and between earnings and price changes in these years, we re-
earnings changes. This implies that nor- gressed the absolute value of AR1., on the log of the abso-
lute value of AEj.,. The estimated slope coefficient is
mally a one dollar increase in earnings is
positive and significant at the 0.01 level.
accompanied by about a five dollar in- 21 We also test whether there is a difference in the rela-
crease in stock price. However, in the tion between stock returns and earnings changes in years

This content downloaded from

86.59.13.237 on Fri, 24 Sep 2021 01:52:09 UTC
All use subject to https://about.jstor.org/terms
 Asquith, Healy, and Palepu 401

Dividend Initiations Subsequent to sions. There are significant earnings in-

Stock Splits creases in the four years before the stock
split announcement. These earnings in-
Hypothesis H6 suggests that the mar-
creases appear to be permanent since the
ket's reaction to split announcements is
earnings changes after the stock split an-
due to an implicit signal about increased
nouncement are either insignificant or
near-term cash dividend payments. Since
positive for up to five years. Pre-split
our sample consists of firms that do not
earnings increases are due to both indus-
pay cash dividends at the time of the
try and firm-specific factors. Firms that
split, an increase in cash dividends im-
announce stock splits are in industries
plies dividend initiation. In order to in-
which perform well, but outperform their
vestigate this possibility, the sample is
industries in the year prior to the split
followed for five years subsequent to the
date. However, the market discounts the
split and cash dividend initiations are re-
pre-split earnings growth, since the rela-
corded. Of the 121 firms in the sample, 81
tion between stock price and earnings
(67 percent) do not initiate a cash divi-
changes during the two years before the
dend payment during the five years after
split is significantly lower than normal.
the split announcement. Of the remain-
The stock price reaction to firms' split
ing 40, the mean (median) lag between
announcements is related to their earn-
the stock split announcement and the
ings increases in the two years prior to
payment of the first cash dividend was
the splits, consistent with the hypothesis
32 (34) months. Only 11 (9 percent) sam-
that split announcements lead to an up-
ple companies pay cash dividends within
ward revision in investors' probability
one year from the date of the stock
assessments that pre-split earnings in-
split.22
creases are permanent rather than tran-
The percentages of splitting firms that
sitory.
initiate cash dividends are similar to
The significant stock price reaction to
those for the population of all firms. For
the sample of split announcements
the sample period 1970-1980, 10 percent
cannot be attributed to investors' expec-
of the non-cash dividend paying firms
tations of post-split earnings or dividend
listed on COMPUSTAT initiate divi-
dends in any given year. Furthermore, 70
percent of the firms which do not pay
cash dividends do not initiate dividends o and 1 for firms that have positive and negative earnings
during the next five years. These results changes. If splits convey information on permanence of
earnings levels, there should be no difference in this rela-
are inconsistent with hypothesis H6.
tion, since the market expects post-split earnings to have
a mean of zero. If, on the other hand, splits provide in-
IV. SUMMARY AND DISCUSSION formation on permanence of earnings growth, the mar-
ket will expect post-earnings changes to be positive. Price
Previous studies report significant changes are, therefore, likely to be larger for earnings de-
clines than earnings increases. The test results indicate
stock price increases for firms that an- that the association between price changes and negative
nounce stock splits. Assuming market changes in years 0 and 1 is not significantly different
efficiency this implies that stock splits from the normal association between price and earnings
changes.
convey new information. This paper ex- 22 If the market responds to a 9 percent increase in the
amines whether this information is re- probability of dividend initiation in one year with a 3.2
lated to earnings performance either percent abnormal return, this implies an extremely large
response to dividend initiation announcements. How-
before or following the split, or to cash
ever, Asquith and Mullins [1983] report that dividend ini-
dividend payments after the split. tiation announcements elicit an abnormal return of only
The analysis leads to several conclu- 3.7 percent.

This content downloaded from

86.59.13.237 on Fri, 24 Sep 2021 01:52:09 UTC
All use subject to https://about.jstor.org/terms
 402 The Accounting Review, July 1989

increases. First, in the post-split period stock splits by dividend-paying firms are
the firms show significant earnings in- likely to be less informative about earn-
creases only in the year of the split, and ings than those by non-dividend-paying
these are unrelated to the split announce- firms.
ment return. Second, the sample firms The above results document that there
do not pay dividends prior to the split is earnings information conveyed by
announcement, and the fraction of firms stock splits. However, they do not neces-
that initiate dividend payments within sarily explain managers' motives for
five years is similar to that for all COM- splitting their firms' stocks. Managers
PUSTAT firms. may not view stock splits as a means of
Since the present sample of stock splits communicating earnings information.
consists of firms that do not pay cash Irrespective of their motives, if managers
dividends, it is possible that the results use their superior information on the
are not applicable to stock splits by divi- permanence of pre-split earnings growth
dend-paying firms. For dividend-paying in making split decisions, investors will
firms the market may use both dividend interpret a split as confirming that past
changes and stock splits to infer whether earnings are permanent, consistent with
earnings changes are permanent. If so, our results.

REFERENCES
Aharony, J., and I. Swary, "Quarterly Dividend and Earnings Announcements and S
turns: An Empirical Analysis," Journal of Finance (March 1980), pp. 1-12.
American Institute of Certified Public Accountants, Accounting Research Bulletin No.
Asquith, P., and D. W. Mullins, Jr., "The Impact of Initiating Dividend Payments on
Wealth," Journal of Business (January 1983), pp. 77-96.
Baker, H. K., and P. L. Gallagher, "Management's View of Stock Splits," Financial Management (Sum-
mer 1980), pp. 73-77.
Ball, R., and P. Brown, "An Empirical Evaluation of Accounting Income Numbers, " Journal ofA ccount
ing Research (Autumn 1968), pp. 159-178.
, and R. Watts, "Some Time Series Properties of Accounting Income," Journal of Finance (June
1972), pp. 663-681.
Bar-Yosef, S., and L. D. Brown, "A Reexamination of Stock Splits Using Moving Betas," Journ
Finance (September 1977), pp. 1069-1080.
Beaver, W., R. Lambert, and D. Morse, "The Information Content of Security Prices," Journal of A
counting and Economics (March 1980), pp. 3-28.
Brennan, M. J., and T. E. Copeland, "Stock Splits, Stock Prices, and Transaction Costs," Working pape
(UCLA, 1987).
Brooks, L., and D. Buckmaster, "First-Difference Signals and Accounting Income Time-Series
erties," Journal of Business Finance and Accounting (Autumn 1980), pp. 437-454.
Chambers, A. E., and S. H. Penman, "Timeliness of Reporting and the Stock Price Reaction to
Announcements," Journal of Accounting Research (Spring 1984), pp. 21-47.
Charest, G., "Split Information, Stock Returns, and Market Efficiency: Part I," Journal of F
Economics (June/September 1978), pp. 265-296.
Defeo, V. J., and P. C. Jain, "Stock Splits: Price per Share and Trading Volume," Advances in Quanti-
tative Analysis of Finance and Accounting (1988).
Fama, E., Foundations of Finance: Portfolio Decisions and Securities Prices (Basic Books, 1976).
L. Fisher, M. C. Jensen, and R. Roll, "The Adjustment of Stock Prices to New Information,"
International Economic Review (February 1969), pp. 1-21.
Foster, T., and D. Vickrey, "The Information Content of Stock Dividend Announcements," THE Ac-
COUNTING REVIEW (April 1978), pp. 360-370.

This content downloaded from

86.59.13.237 on Fri, 24 Sep 2021 01:52:09 UTC
All use subject to https://about.jstor.org/terms
 Asquith, Healy, and Palepu 403

Freeman, R. N., J. A. Ohlson, and S. H. Penman, "Book Rate-of-Return and Prediction of Earnings
Changes: An Empirical Investigation," Journal of Accounting Research (Autumn 1982), pp. 639-653.
Grinblatt, M. S., R. W. Masulis, and S. Titman, "The Valuation Effects of Stock Splits and Stock Divi-
dends," Journal of Financial Economics (December 1984), pp. 461-490.
Healy, P. M., and K. G. Palepu, "Earnings Information Conveyed by Dividend Initiations and Omis-
sions, " Journal of Financial Economics (September 1988), pp. 149-175.
Lakonishok, J., and B. Lev, "Stock Splits and Stock Dividends: Why, Who, and When," Journal of Fi-
nance (September 1987), pp. 913-932.
McNichols, M., and A. Dravid, "Stock Dividends, Stock Splits, and Signalling," Working paper (Stan-
ford University, 1987).
Pindyck, R. S., and D. L. Rubinfeld, EconometricModels and EconomicForecasts (McGraw-Hill, 1981).
Watts, R. L., and R. W. Leftwich, "The Time Series of Annual Accounting Earnings," Journal of Ac-
counting Research (Autumn 1977), pp. 253-27 1.
Woolridge, J. R., "Stock Dividends as Signals," Journal of Financial Research (Spring 1983), pp. 1-12.

This content downloaded from

86.59.13.237 on Fri, 24 Sep 2021 01:52:09 UTC
All use subject to https://about.jstor.org/terms