years. One interpretation of these results is that when firms announce
large earnings increases the market assumes that the increases are at least
partially transitory. When splits are announced, the market revises
upward the probability that the past earnings increases are permanent,
leading to an increase in the firms' stock prices.
5. SUMMARY AND DISCUSSION
Previous studies report significant stock price increases for firms
that announce stock splits. Assuming market efficiency this implies that
stock splits convey new information. This paper examines whether this
information is related to earnings performance surrounding the split.
The analysis leads to several conclusions. First, there are
significant earnings increases in the four years prior to and in the year
of a stock split announcement. These earnings increases appear to be
permanent since the earnings changes after the stock split announcement are
either insignificant or positive for up to five years. Earnings increases
prior to the split announcement are due to industry and firm-specific
factors. Firms that announce stock splits are in industries which perform
well, but outperform their industries in the year prior to the split date.
Second, the stock price reaction to firms' split announcements is
related to their earnings increases in the two years prior to the splits,
and is unrelated to the earnings changes in the years following the splits.
This is consistent with an upward revision in investors' probability
assessments that pre-split earnings increases are permanent rather than
Third, the significant stock price reaction to our sample of split
announcements cannot be solely attributed to expectations of near-term
dividend increases. Our sample firms do not pay dividends prior to the
split announcement, and the fraction of firms that initiate dividend
payments within five years is similar to that for all Compustat firms.
Since our sample of stock splits consists of firms that do not pay
cash dividends, it is possible that the results are not applicable to stock
splits by dividend paying firms. For dividend paying firms the market may
use both dividend changes and stock splits to infer whether earnings
changes are permanent. If so, stock splits by dividend paying firms are
likely to be less informative about earnings than those by non-dividend
paying firms. However, GMT report that the stock split market reaction is
similar for both types of firms. Also LL, whose sample includes dividend
paying firms, report similar earnings patterns to those in this paper.
The above results document that there is earnings information conveyed
by stock splits. However, they do not necessarily explain managers'
motives for splitting their firms' stocks. Managers may not view stock
splits to be a means of communicating earnings information. One frequently
suggested managerial motive for splitting stock is to keep the firm's share
price in a desired trading range. According to this explanation strong
earnings performance of the splitting firm prior to the split date leads to
an increase in the stock price beyond the trading range. If the firm's
managers have information suggesting that the earnings increases are
permanent, the stock is split to bring the price within the desired trading
range. If this explanation is valid, the market will interpret a split as
confirming that past earnings are permanent which is consistent with our
results. However, to fully formulate a trading range hypothesis, further
- 17 -
research is required to document the determinants of a firm's optimal
trading range and the costs associated with trading outside that range.
1. The seminal study by Fama, Fisher, Jensen and Roll (1969) documents
stock price increases prior to the split. Bar-Yosef and Brown (1977),
Charest (1978), Foster and Vickery (1978), Woolridge (1983), and
Grinblatt, Masulis, and Titman (1984) all report positive stock price
reactions to split announcements.
2. Dividend increases, as studies subsequent to FFJR showed, cause stock
prices to increase. See e.g. Aharoney and Swary (1980) and Asquith
and Mullins (1983).
3. Fama (1976) p. 164
4. McNichols and Dravid (1987) examines analysts earnings forecasts
around stock splits and finds that there are statistically significant
pre-split upward revisions.
5. See for example Baker and Gallagher's (1980) survey of managerial
motives for stock splits. Also, Lakonishok and Lev (1987) provide
empirical evidence on absolute share price levels which supports the
concept of a 'normal range."
6. This sample selection does not rule out the possibility that split
announcements implicitly provide information on subsequent dividend
changes. This issue is examined in our empirical analysis (see
7. The differential treatment of retained earnings for stock dividends
creates s potential cost to the firm by making certain debt covenants
more binding. This cost is absent for stock splits. Hence stock
dividends and stock splits may convey different information to the
market. Results of LL (1987) support this proposition.
8. Stock price and return data is collected from the CRSP files,
announcement dates are collected from the WSJ Index and earnings are
collected from the Quarterly Compustat File. Earnings data are not
available for all the sample firms for all the years. For years
before the stock split, data for some sample firms are not available
in the Compustat files primarily because the firm was not listed on
either NYSE or AMEX. For years after the stock split, data are not
available primarily because firms were delisted, merged, or
liquidated. The actual number of firms for which usable earnings data
are available is indicated in the results tables discussed later in
9. The industry earnings data are still defined as fiscal year data
because Compustat does not construct a quarterly Research tape. To
ensure that a complete sample of firms is used to construct industry
medians the annual Compustat research tape, which only has fiscal year
data, is used.
- 19 -
10. We also compute market adjusted returns for the same period. The
results are similar in both size and significance levels to those
11. One possible explanation for negative abnormal returns is that other
information is released simultaneously with the stock split.
Controlling for other explicit Information releases, i.e. earnings,
announcements, financing announcements etc. , increases the mean
risk-adjusted return to 4.6% for 84 firms and the percentage of firms
with positive abnormal returns to 78%. All subsequent tests are
repeated using this clean announcement sample and the results are
similar to those reported in the paper.
12. If the market responds to a 9% increase in the probability of divided
initiation in one year with a 3.9% abnormal return, this implies an
extremely large response to dividend initiation announcements.
However, Asquith and Mullins (1983) report that dividend initiation
announcements elicit an abnormal return of only 3.7%.
13. Since stock splits occur throughout a fiscal year and since quarterly
earnings are used, this method ensures that all earnings in year are
announced after the stock split. This analysis is repeated using
fiscal year earnings data and the results are similar.
14. The number of firms for which there is usable earnings data varies
across years -5 to 4: the lowest is 35 firms in year -5 and the
highest is 118 firms in year 0. The number of observations is low in
years -5 to -2 mainly because many of the firms in our sample are
newly listed at the time of the stock split and therefore do not have
earnings data in these years; the number of observations after year 3
declines principally because of delistments. Of the 86 firms with
unavailable data in year -5, 73 are not listed on the NYSE or ASE and
13 are unavailable on Compustat. Of the 23 firms with unavailable
data in year 4, 6 are acquired, 2 are liquidated, 5 are delisted by
the exchange, 2 are suspended by the exchange and 8 are unavailable on
15. The mean earnings changes in years 3 and 4 are as large as years -1
and but not as significant. This is due to increased
cross-sectional volatility in earnings.
16. The announcement returns are also regressed over earnings changes for
longer periods. The results are not qualitatively different from
those reported below. Only results for years -2 to +1 are reported
because the number of usable earnings observations varies across years
(see Table 4) and these years have the largest number of observations.
Also, as the correlation results in Table 4 show, the other years'
earnings changes are not related to the announcement returns.
Aharoney, J. and I. Swary, 1980, "Quarterly Dividend and Earnings
Announcements and Stockholders' Returns: An Empirical Analysis,"
Journal of Finance , 31, 1-12.
Asquith, P. and D.W. Mullins, 1983, "The Impact of Initiating Dividend
Payments on Shareholders' Wealth," Journal of Business 56, 77-96.
Baker, H.K. and P.L. Gallagher, 1980, "Management's View of Stock Splits,"
Financial Management , 9, 73-77.
Ball, R. , and P. Brown, 1968, "An Empirical Analysis of Accounting Income
Numbers," Journal of Accounting Research , 6, 159-178.
Ball, R. , and R. Watts, 1972, "Some Time-Series Properties of Accounting
Income," Journal of Finance . 27, 663-682.
Bar-Yosef, S., and L. Brown, 1977, "A Re-examination of Stock Splits Using
Moving Betas," Journal of Finance , 32, 1069-1080.
Beaver, W. , R.Lambert, and D. Morse, 1980, "The Information Content of
Security Prices," Journal of Accounting and Economics , 2, 3-28.
Brooks, L.D., and D.A. Buckmaster, 1980, "First-Difference Signals and
Accounting Income Time-Series Properties," Journal of Business Finance
and Accounting (Autumn 1980), 437-454.
Charest, G. , 1978, "Split Information, Stock Returns, and Market
Efficiency," Journal of Financial Economics , 6, 265-296.
Fama, E. , 1976, Foundations of Finance , Basic Books, New York.
Fama, E. , L. Fisher, M. Jensen, and R. Roll, 1969, "The Adjustment of Stock
Prices to New Information," International Economic Review , 10, 1-21.
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 , 20, 639-653.
Foster, T. and D. Vickery, 1978, "The Information Content of Stock Dividend
Announcements," The Accounting Review , 53, 360-370.
Healy, P. and K. Palepu, 1987, "Earnings Information Conveyed by Dividend
Initiation and Omissions," MIT Working Paper.
Grlnblatt, M.S., R.W. Masulis, and S. Titman, 1984, "The Valuation Effects
of Stock Splits and Stock Dividends," Journal of Financial Economics ,
Lakonishok, J. and B. Lev, 1987, "Stock Splits and Stock Dividends: Why,
Who, and When," Journal of Finance , 42, 913-932.
- 21 -
Miller, M. and K. Rock, 1985, "Dividend Policy under Asymmetric
Information," Journal of Finance . 40, 1031-1051.
McNichols, M. and A. Dravid, 1986, "Stock Dividends, Stock Splits and
Signalling," Stanford University Working Paper.
Watts, R.L., and R.W. Leftwich, "The Time Series of Annual Accounting
Earnings," Journal of Accounting Research . 15, 253-271.
Woolridge, R. , 1983, "Stock Dividends as Signals," Journal of Financial
Research . 6, 1-12.
Distribution of stock splits in the period 1970 to 1980
by year for 121 firms that do not pay cash dividends^
Number Percent of
Year of 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
^ To be included in the sample, firms are required to meet the following requirements: (1) the firm
made a stock split or stock dividend of at least 25%; (2) the firm did not pay cash dividends prior to
or at the time of the dividend announcement; (3) no other stock dividend or split of 25% or more had
been made in the five years prior to the event; (4) the firm is listed on the New York or American
Stock Exchange; (5) the stock split announcement date is available in the Wall Street Journal, or on
the CRSP Daily Master File; (6) Stock price and return data are available for the announcement date
and two days prior to the event; (7) the first quarterly eamings announcement preceding the event is
reported in the Wall Srreet Journal; and (8) at least two years of consecutive quarterly eamings data
are available on Compustat for the six years before and the five years after the event date.
Abnormal returns for various holding periods for 121
firms that announce stock splits but do not pay cash dividendsS'''
Panel A: Abnormal returns surrounding the split announcement
Holding period Mean t statistic
AD-480 to AD-361 7.0% 1.5
AD-360 to AD-241 2.8 0.6
AD-240 to AD-121 17.8 3.9
AD-120 to AD-61 12.8 3.9
AD-60 to AD-21 14.1 5.3
AD-20 to AD-11 2.5 1.9
AD-10 to AD-2 3.3 2.6
AD-1 to AD 3.9 6.9
AD+1 to AD+10 1.6 1.2
AD+11 to AD+20 1.1 0.8
Panel B: Abnormal returns for AD-1 to AD
Percent positive 75.8%
3 The sample comprises 121 firms that announce stock splits or stock dividends of 25% or nvjre in the period 1970-
^ Risk-adjusted returns are cumulative retums for days relative to the Wall Street Journal announcement of the split
(AD) with pre-split market model parameters estimated using returns for days -630 to -AS^ and post-split
paran^eters estimated using returns for days 241 to 390.
Summary statistics on changes in earnings per share as a percentage
of initial equity price for years surrounding announcements of stock
splits by firms that do not pay cash dividends^-''
Period relative Number Student t First Third
to stock split of firms Mean probability'^ quartile Median quartiie
Panel A: Raw earnings changes
Panel B: Industry-adjusted earnings changes
3 The stock split sample comprises 121 firms that announce stock splits of 25% or more in the period 1970 to 1980.
Earnings changes are not available for three of the split Firms in year sir>ce they do not have earnings data available In
year -1. Industry-adjusted earnings changes are available for 109 of these firms. Industry-adjusted earnings
changes for a given firm in year t are defined as the difference between the standardized earnings change for the split
firm in year t and the median standardized earnings change for otfier firms In tlie same Industry in that year.
^ Earnings changes are estimated using quarterly earnings for the 24 quarters prior to the split announcement and tfie
20 quarters subsequent. Changes in earnings per share before extraordinary Items and discontinued operations are
standardized by tfie firm's stock price two days prior to the arviouncement of tt\e stock split.
^ Student t test statistics test the hypothesis that the mean earnings changes are dfferent from zero. The probability
levels are for two-tailed tests of significance.
Summary of correlations between standardized earnings changes
surrounding stock split announcements by firms that do not pay
cash dividends, and split announcement returns^
to stock split
3 The stock split sample comprises 121 firms that announce stock splits of 25% or more In the period 1970 to 1980.
Earnings changes are r>ot available for three of the split firms in year since they do not have earnings data available in
^ Earnings changes are estimated using quarteriy earnings for the 24 quarters prior to the split announcement and ttw
20 quarters subsequent. Changes in earnings per share before extraordinary hems 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 risk-adjusted return for one day prior to and the day of the stock split announcement.
Tests of the relation between stock split announcement
returns and standardized earnings changes in surrounding years^
SRETj = a + Pi5E.2.j + PaSE.ij + PaSEoj + p45Ei j + ej b
Coefficient t statistic
a 0.025 3.65=
Pi 0.306 2.12d
p2 0.241 2.07d
p3 -0.148 -1.22
p4 0.012 0.21
^ The stock split sample comprises 121 firms that announce stock splits of 25% or more in the period 1970 to 1980.
Forty-two firms are excluded from the regression since they do not have earnings data available for each of the lour years
surrounding the split announcement.
^ 5E( j 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; SRET; Is the risk-adjusted return for firm j for one day prior to and the day of the
stock split announcement.
*- Significant at the one percent level using a two-tailed test.
*^ Significant at the five percent level using a two-tailed test.
III l\ -^ U l^JtJC,