Professional Documents
Culture Documents
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
is collaborating with JSTOR to digitize, preserve and extend access to Journal of Economic
Literature
A. CRAIG MACKINLAY
The Wharton School, University of Pennsylvania
Thlanks to John Can ipbell, Briuce G-tr udly, Andrti-ewv Lo, and twvo anonymnotus referees for helpful
comments andcl discussion. Research suLpport from the Rodiney L. WVhite Center for Finiancial
Research is gratefiully acknzowvledged.
13
creased in 57 of the cases and the price with the necessary tools for calculating
declined in only 26 instances. Over the an abnormal return, making statistical in-
decades from the early 1930s until the ferences about these returns, and aggre-
late 1960s the level of sophistication of gating over many event observations.
event studies increased. John H. Myers The null hypothesis that the event has no
and Archie Bakay (1948), C. Austin impact on the distribution of returns is
Barker (1956, 1957, 1958), and John maintained in Sections 4 and 5. Section 6
Ashley (1962) are examples of studies discusses modifying this null hypothesis
during this time period. The improve- to focus only on the mean of the return
ments included removing general stock distribution. Section 7 presents analysis
market price movements and separating of the power of an event study. Section 8
out confounding events. In the late presents nonparametric approaches to
1960s seminal studies by Ray Ball and event studies which eliminate the need
Philip Brown (1968) and Eugene Fama for parametric structure. In some cases
et al. (1969) introduced the methodology theory provides hypotheses concerning
that is essentially the same as that which the relation between the magnitude of
is in use today. Ball and Brown consid- the event abnormal return and firm char-
ered the information content of earn- acteristics. Section 9 presents a cross-
ings, and Fama et al. studied the effects sectional regression approach that is use-
of stock splits after removing the effects ful to investigate such hypotheses.
of simultaneous dividend increases. Section 10 considers some further issues
In the years since these pioneering relating event study design and the pa-
studies, a number of modifications have per closes with the concluding discussion
been developed. These modifications re- in Section 11.
late to complications arising from viola-
tions of the statistical assumptions used 2. Procedure for an Event Study
in the early work and relate to adjust-
ments in the design to accommodate At the outset it is useful to briefly dis-
more specific hypotheses. Useful papers cuss the structure of an event study. This
which deal with the practical importance will provide a basis for the discussion of
of many of the complications and adjust- details later. While there is no unique
ments are the work by Stephen Brown structure, there is a general flow of
and Jerold Warner published in 1980 and analysis. This flow is discussed in this
1985. The 1980 paper considers imple- section.
mentation issues for data sampled at a The initial task of conducting an event
monthly interval and the 1985 paper study is to define the event of interest
deals with issues for daily data. and identify the period over which the
In this paper, event study methods are security prices of the firms involved in
reviewed and summarized. The paper this event will be examined-the event
begins with discussion of one possible window. For example, if one is looking at
procedure for conducting an event study the information content of an earnings
in Section 2. Section 3 sets up a sample with daily data, the event will be the
event study which will be used to illus- earnings announcement and the event
trate the methodology. Central to an window will include the one day of the
event study is the measurement of an ab- announcement. It is customary to define
normal stock return. Section 4 details the event window to be larger than the
the first step-measuring the normal specific period of interest. This permits
performance-and Section 5 follows examination of periods surrounding the
event. In practice, the period of interest choices for modeling the normal re-
is often expanded to multiple days, in- turn-the constant mean return model
cluding at least the day of the an- where Xl is a constant, and the market
nouncement and the day after the an- model where Xl is the market return.
nouncement. This captures the price The constant mean return model, as the
effects of announcements which occur name implies, assumes that the mean
after the stock market closes on the an- return of a given security is constant
nouncement day. The periods prior to through time. The market model as-
and after the event may also be of inter- sumes a stable linear relation between
est. For example, in the earnings an- the market return and the security re-
nouncement case, the market may ac- turn.
quire information about the earnings Given the selection of a normal perfor-
prior to the actual announcement and mance model, the estimation window
one can investigate this possibility by ex- needs to be defined. The most common
amining pre-event returns. choice, when feasible, is using the period
After identifying the event, it is neces- prior to the event window for the estima-
sary to determine the selection criteria tion window. For example, in an event
for the inclusion of a given firm in the study using daily data and the market
study. The criteria may involve restric- model, the market model parameters
tions imposed by data availability such as could be estimated over the 120 days
listing on the New York Stock Exchange prior to the event. Generally the event
or the American Stock Exchange or may period itself is not included in the esti-
involve restrictions such as membership mation period to prevent the event from
in a specific industry. At this stage it is influencing the normal performance
useful to summarize some sample char- model parameter estimates.
acteristics (e.g., firm market capitaliza- With the parameter estimates for the
tion, industry representation, distri- normal performance model, the abnor-
bution of events through time) and note mal returns can be calculated. Next
any potential biases which may have comes the design of the testing frame-
been introduced through the sample se- work for the abnormal returns. Impor-
lection. tant considerations are defining the null
Appraisal of the event's impact re- hypothesis and determining the tech-
quires a measure of the abnormal return. niques for aggregating the individual
The abnormal return is the actual ex post firm abnormal returns.
return of the security over the event win- The presentation of the empirical re-
dow minus the normal return of the firm sults follows the formulation of the
over the event window. The normal re- econometric design. In addition to pre-
turn is defined as the expected return senting the basic empirical results, the
without conditioning on the event taking presentation of diagnostics can be fruit-
place. For firm i and event date t the ful. Occasionally, especially in studies
abnormal return is with a limited number of event observa-
tions, the empirical results can be heav-
AR1t = R1, - E(RjrjXr) (1)
ily influenced by one or two firms.
where AR,,, Ri,, and E(Ri,IXt) are the ab- Knowledge of this is important for gaug-
normal, actual, and normal returns re- ing the importance of the results.
spectively for time period t. Xl is the Ideally the empirical results will lead
conditioning information for the normal to insights relating to understanding the
return model. There are two common sources and causes of the effects (or lack
of effects) of the event under study. Ad- is Datastream, and the source of the ac-
ditional analysis may be included to dis- tual earnings is Compustat.
tinguish between competing explana- If earnings announcements convey in-
tions. Concluding coiimments complete formation to investors, one would expect
the study. the announcement impact on the mar-
ket's valuation of the firm's equity to de-
pend on the magnitude of the unex-
3. An Example of an Event Study
pected component of the announcement.
The Financial Accounting Standards Thus a measure of the deviation of the
Board (FASB) and the Securities Ex- actual announced earnings from the mar-
change Commission strive to set report- ket's prior expectation is required. For
ing regulations so that financial state- constructing such a measure, the mean
ments and related information releases quarterly earnings forecast reported by
are informative about the value of the the Institutional Brokers Estimate Sys-
firm. In setting standards, the informa- tem (I/B/E/S) is used to proxy for the
tion content of the financial disclosures market's expectation of earnings. I/B/E/S
is of interest. Event studies provide an compiles forecasts from analysts for a
ideal tool for examiining the informnation large number of companies and reports
content of the disclosures. summary statistics each month. The
In this section the description of an mean forecast is taken from the last
example selected to illustrate event month of the quarter. For example, the
study methodology is presented. One mean third quarter forecast from Sep-
particular type of disclosure-quarterly tember 1990 is used as the measure of
earnings announcements-is considered. expected earnings for the third quarter
The objective is to investigate the infor- of 1990.
mation content of these announce- To facilitate the examination of the
ments. In other words, the goal is to see impact of the earnings announcement on
if the release of accounting informnation the value of the firm's equity, it is essen-
provides information to the marketplace. tial to posit the relation between the in-
If so there should be a correlation be- formation release and the change in
tween the observed change of the mar- value of the equity. In this example the
ket value of the company and the infor- task is straightforward. If the earnings
mation. disclosures have information content,
The example will focus on the quar- higher than expected earnings should be
terly earnings announcements for the 30 associated with increases in value of the
firms in the Dow Jones Industrial Index equity and lower than expected earnings
over the five-year period from January with decreases. To capture this associa-
1989 to December 1993. These an- tion, each announcement is assigned to
nouncements correspond to the quar- one of three categories: good news, no
terly earnings for the last quarter of 1988 news, or bad news. Each announcement
through the third quarter of 1993. The is categorized using the deviation of the
five years of data for 30 firms provide a actual earnings from the expected earn-
total sample of 600 announcemnents. For ings. If the actual exceeds expected by
each firm and quarter, three pieces of in- more than 2.5 percent the announce-
formation are compiled: the date of the ment is designated as good news, and if
announcement, the actual earnings, and the actual is more than 2.5 percent less
a measure of the expected earnings. The than expected the announcement is des-
source of the date of the announcement ignated as bad news. Those announce-
ments where the actual earnings is in the For the statistical models, the assump-
5 percent range centered about the ex- tion that asset returns are jointly multi-
pected earnings are designated as no variate normal and independently and
news. Of the 600 announcements, 189 identically distributed through time is
are good news, 173 are no news, and the imposed. This distributional assumption
remaining 238 are bad news. is sufficient for the constant mean return
With the announcements categorized, model and the market model to be cor-
the next step is to specify the parameters rectly specified. While this assumption is
of the empirical design to analyze the eq- strong, in practice it generally does not
uity return, i.e., the percent change in lead to problems because the assumption
value of the equity. It is necessary to is empirically reasonable and inferences
specify a length of observation interval, using the normal return models tend to
an event window, and an estimation win- be robust to deviations from the assump-
dow. For this example the interval is set tion. Also one can easily modify the sta-
to one day, thus daily stock returns are tistical framework so that the analysis of
used. A 41-day event window is em- the abnormal returns is autocorrelation
ployed, comprised of 20 pre-event days, and heteroskedasticity consistent by us-
the event day, and 20 post-event days. ing a generalized method-of-moments
For each announcement the 250 trading approach.
day period prior to the event window is
used as the estimation window. After A. Constant Mean Return Model
presenting the methodology of an event
study, this example will be drawn upon
Let [,u be the mean return for asset i.
Then the constant mean return model is
to illustrate the execution of a study.
Rit= - i + Git (2)
(2
4. Models for Measuring Normal E(4st) = 0 var (4i) = G7; .
Performance
where Rit is the period-t return on secu-
A number of approaches are available rity i and Cit is the time period t distur-
to calculate the normal return of a given bance term for security i with an expec-
security. The approaches can be loosely tation of zero and variance cy .
grouped into two categories-statistical Although the constant mean return
and economic. Models in the first cate- model is perhaps the simplest inodel,
gory follow from statistical assumptions Brown and Warner (1980, 1985) find it
concerning the behavior of asset returns often yields results similar to those of
and do not depend on any economic ar- more sophisticated mnodels. This lack of
guments. In contrast, models in the sec- sensitivity to the model can be attributed
ond category rely on assumptions con- to the fact that the variance of the abnor-
cerning investors' behavior and are not mal return is frequently not reduced
based solely on statistical assumptions. It much by choosing a more sophisticated
should, however, be noted that to use model. When using daily data the model
economic models in practice it is neces- is typically applied to nominal returns.
sary to add statistical assumptions. Thus With monthly data the model can be ap-
the potential advantage of economic plied to real returns or excess returns
models is not the absence of statistical (the return in excess of the nominal risk
assumptions, but the opportunity to cal- free return generally measured using the
culate more precise measures of the nor- U.S. Treasury Bill with one month to
mal return using economic restrictions. inaturity) as well as nominal returns.
Tr
(estimiation] ( ev7ent I (post-event]
w"indow window] vindow
, (Ri _ -)(R1, _ 9111)
To 0 T9 T A I T(+ 1
13i= lT
Figure 1. Time line for an event study.
, (R ^ 1j)2
t = T(+ 1 (4)
is built around the assumption that RiX and R,, are the return in event pe-
the event impact is captured by the riod t for security i and the market re-
abnormal returns. On occasion, the spectively. The use of the OLS estima-
post event window data is included tors to measure abnormal returns and to
with the estimation window data to develop their statistical properties is ad-
estimate the normal return model. dressed next. First, the properties of a
The goal of this approach is to increase given security are presented followed by
the robustness of the normal market consideration of the properties of abnor-
return measure to gradual changes mal returns aggregated across securities.
in its parameters. In Section 6 ex-
panding the null hypothesis to accom- B. Statistical Properties of Abnormal
modate changes in the risk of a firm Returns
dow market returns, the abnormal re- overall inferences for the event of inter-
turns will be jointly normally distributed est. The aggregation is along two dimen-
with a zero conditional mean and condi- sions-through time and across securi-
tional variance 62(ARic) where ties. We will first consider aggregation
through time for an individual security
G2(AR )=62 + L1 L+ 1 . (8) and then will consider aggregation both
across securities and through time. The
concept of a cumulative abnormal return
From (8), the conditional variance has
is necessary to accommodate a multiple
two components. One component is the
period event window. Define CARi(t1,t2)
disturbance variance G2 from (3) and a
as the sample cumulative abnormal re-
second component is additional variance
turn (CAR) from t1 to t2 where
due to the sampling error in ci and Pi.
This sampling error, which is common T1 <1 <?2 < T2. The CAR from t1 to t
the sum of the included abnormal re-
for all the event window observations,
turns,
also leads to serial correlation of the
abnormal returns despite the fact that tC2
TABLE 1
Market Model
TABLE 1 (Cont.)
tering. That is, there is not any overlap vai(CARCC1t2)) = X2 E 6(t1,t9) (18)
in the event windows of the included se-
curities. The absence of any overlap and
the maintained distributional assumptions For the variance estimators the assump-
imply that the abnormal returns and the tion that the event windows of the N se-
cumulative abnormal returns will be in- curities do not overlap is used to set the
dependent across securities. Later infer- covariance terms to zero. Inferences
ences with clustering will be discussed. about the cumulative abnormal returns
The individual securities' abnormal re- can be drawn using
turns can be aggregated using ARi, fiom (7)
for each event period, t = T, + 1, . . ., T2. CAR(t1l'2) N[O, var(CAR (t1, 2))] (19)
Given N events, the sample aggregated to test the null hypothesis that the ab-
abnormal returns for period t is normal returns are zero. In practice, be-
N cause y2 is unknown, an estimator must
ARIC = NARI, (13) be used to calculate the variance of the
i=l abnormal returns as in (14). The usual
sample variance measure of 2 from the
and for large Li, its variance is
market model regression in the estima-
N tion window is an appropriate choice.
var(APR)=? N2 e, . (14) Using this to calculate va(ARr) in (14),
i=1 Ho can be tested using
Using these estimates, the abnormal re-
turns for any event period can be ana- CAR(t1,9) N(0,1).
vaiCARQ( 1,t)/(
lyzed.
The average abnormal returns can This distributional result is asymptotic
then be aggregated over the event win- with respect to the number of securities
dow using the same approach as that N and the length of estimation window Li.
used to calculate the cumulative abnor- Modifications to the basic approach
mal return for each security i. For any presented above are possible. One com-
interval in the event window mon modification is to standardize each
abnormal return using an estimator of its
standard deviation. For certain alterna-
CAR('1'l9=2) I ART, (15) tives, such standardization can lead to
X =
more powerful tests. James Patell (1976)
presents tests based on standardization
and Brown and Warner (1980, 1985)
provide comparisons with the basic ap-
var(CAR(Cjt2)) = var (AR). (16) proach.
t = ~1
D. CAR's for the Earnings
Observe that equivalently one can form Announcement Example
the CAR's security by security and then
The information content of earnings
aggregate through time,
example previously described illustrates
N the use of sample abnormal residuals and
CAR(,rl, r) -N,CARi (C I,T2) (17) sample cumulative abnormal returns. Ta-
ble 1 presents the abnormal returns av-
0.025
0.020
0.01
0.005-
-0.005S _
-0.005 _.
-0.015
-0.02
-0.025- I l l l I l l l l l l l l I
-21 -18 -15 -12 -9 -6 -3 0 3 6 9 12 15 18 21
Event Time
* Good News Firms - No News Firms A Bad News Firms
eraged across the 600 event observations deed convey information useful for the
(30 firms, 20 announcements per firm) valuation of firms. Focusing on the an-
as well as the aggregated cumulative ab- nouncement day (day 0) the sample aver-
normal return for each of the three earn- age abnormal return for the good news
ings news categories. Two normal return firm using the market model is 0.965
models are considered; the market percent. Given the standard error of the
model and for comparison, the constant one day good news average abnormal re-
mean return model. Plots of the cumula- turn is 0.104 percent, the value of O1 is
tive abnormal returns are also included, 9.28 and the null hypothesis that the
with the CAR's from the market model event has no impact is strongly rejected.
in Figure 2a and the CAR's from the The story is the same for the bad news
constant mean return model in Figure firms. The event day sample abnormal
2b. return is -0.679 percent, with a standard
The results of this example are largely error of 0.098 percent, leading to 01
consistent with the existing literature on equal to -6.93 and again strong evidence
the information content of earnings. The against the null hypothesis. As would be
evidence strongly supports the hypothe- expected, the abnormal return of the no
sis that earnings announcements do in- news firms is small at -0.091 percent and
0.02--
0.015- /
0.01 - /
-0.015- -
-0.02--
-0?.0?25 f I I I I III I I4
-21 -18 -15 -12 -9 -6 -3 0 3 6 9 12 15
Event Time
-*---- Good News Firms u No News Firms A Bad News Firms
TABLE 2
sample size, that is the number of securi- ues calculated using c(cx/2) and c(1 -
ties for which the event occurs, is ox/2) are -196 and 1.96 respectively. Of
varied from one to 200. The power for course, in applications, the power of the
a test with a size of 5 percent is docu- test should be considered when selecting
mented. With ox = 0.05, the critical val- the size.
1 / ___
0.9 -!
0.8 /
/, / / /
0.7 /
0.6
0.5 /
0.4 1
0.3
o I I I I I I I I I I I I
0 10 20 30 40 50 60 70 80 90 100 110 120 130 140 150 160 170 180 190
Niimber of Secuirities
Figure 3a. Power of event stuidy test statistic 0, to reject the nuill lhypotlhesis tha
squiare root of the average varianice of the abnormal retuirn lacross firms is 2 perc
The power results are presented in Ta- are substantial when the abnormal
ble 2, and are plotted in Figures 3a and return is larger. For example, when the
3b. The results in the left panel of Table abnormal return is 2.0 percent the
2 and Figure 3a are for the case where power of a 5 percent test with 20 firms
the average variance is 0.0004. This cor- is almost 1.00 with a value of 0.99.
responds to a cumulative abnormal re- The general results for a variance of
turn standard deviation of 2 percent and 0.0004 is that when the abnormal return
is an appropriate value for an event is larger than 1 percent the power is
which does not lead to increased vari- quite high even for small sample sizes.
ance and can be examined using a one- When the abnormal return is small a
day event window. In terms of having larger sample size is necessary to achieve
high power this is the best case scenario. high power.
The results illustrate that when the ab- In the right panel of Table 2 and in
normal return is only 0.5 percent the Figure 3b the power results are pre-
power can be low. For example with a sented for the case where the average
sample size of 20 the power of a 5 variance of the cumulative abnormal re-
percent test is only 0.20. One needs a turn is 0.0016. This case corresponds
sample of over 60 firms before the roughly to either a multi-day event win-
power reaches 0.50. However, for a dow or to a one-day event window with
given sample size, increases in power the event leading to increased variance
0.9
0.8 -
0.7 /
0.6 , -
0.5 /
0.4
0.3 / ,
0 10 20 30 40 50 60 70 80 90 100 110 120 130 140 150 160 170 180 190 200
Niimber of Secuirities
Figure 3b. Power of event stuidy test statistic 01 to reject the nuill hypotlhesis that the
the squiare root of the average variance of the abnormial retuirn across firms is 4 percen
which is accommodated as part of the considered analytically for the given dis-
null hypothesis. When the average vari- tributional assumptions. If the distri-
ance of the CAR is increased from butional assumptions are inappropriate
0.0004 to 0.0016 there is a dramatic then the results may differ. However,
power decline for a 5 percent test. When Brown and Warner (1985) consider this
the CAR is 0.5 percent the power is only possible difference and find that the ana-
0.09 with 20 firms and is only 0.42 with a lytical computations and the empirical
sample of 200 firms. This magnitude of power are very close.
abnormal return is difficult to detect It is difficult to make general conclu-
with the larger variance. In contrast, sions concerning the adequacy of the
when the CAR is as large as 1.5 percent ability of event study methodology to de-
or 2.0 percent the 5 percent test is still tect non-zero abnormal returns. When
has reasonable power. For example, conducting an event study it is best
when the abnormal return is 1.5 percent to evaluate the power given the parame-
and there is a sample size of 30 the ters and objectives of the study. If the
power is 0.54. Generally if the abnormal power seems sufficient then one can
return is large one will have little diffi- proceed, otherwise one should search
culty rejecting the null hypothesis of no for ways of increasing the power. This
abnormal return. can be done by increasing the sample
In the preceding analysis the power is size, shortening the event window, or by
s(K)(= ) L~2 E- N E L 2+
is that there is a positive abnormal re-
'' 1K I rK,t 12+ (25)
turn associated with a given event, the I=T, + I = 1
null hypothesis is Ho:p < 0.5 and the al-
Tests of the null hypothesis can be im-
ternative is HA:P > 0.5 where p =
plemented using the result that the as-
pr[CARi ? 0.0]. To calculate the test sta-
ymptotic null distribution of 0,3 is stan-
tistic we need the number of cases where
dard normal. Corrado (1989) includes
the abnormal return is positive, N+, and
further discussion of details of this test.
the total number of cases, N. Letting 02
Typically, these nonparametric tests
be the test statistic,
are not used in isolation but in conjunc-
tion with the parametric counterparts.
02=L--0.51j -N(O,1). (23) Inclusion of the nonparamnetric tests pro-
vides a check of the robustness of con-
This distributional result is asymnptotic. clusions based on parametric tests. Such
For a test of size (1 - ax), Ho is rejected ifa check can be worthwhile as illustrated
02 > ?-1'("). by the work of Cynthia Campbell and
A weakness of the sign test is that it Charles Wasley (1993). They find that
may not be well specified if the distri- for NASDAQ stocks daily returns the
bution of abnormal returns is skewed as nonparametric rank test provides more
can be the case with daily data. In re- reliable inferences than do the standard
sponse to this possible shortcoming, parametric tests.
0.9
0.8
0.7
0.6
Q~~~~~~~~~~~~~~
0.3
0.2
0.1
0 10 20 30 40 50 60 70 80 90 100 110 120 130 140 150 160 170 180 190
Nuimber of Secuirities
Figure 4. Power of event sttucly test statistic 01 to reject the nuill hypothes
different sampling inter-vals, when the squiare root of the average variance of
percent for the daily inter-val. Size of test is 5 percent.
prQach can be used for inferences and more frequent sampling arises. To ad-
that the t-statistics can be interpreted as dress this question one needs to consider
lower bounds on the true significance the power gains from shorter intervals. A
level of the estimates. comparison of daily versus monthly data
is provided in Figure 4. The power of
10. Other Issues the test of no event effect is plotted
against the alternative of an abnormal re-
A numnber of further issues often arise turn of one percent for 1 to 200 securi-
when conducting an event study. These ties. As one would expect given the
issues include the role of the sampling analysis of Section 7, the decrease in
interval, event date uncertainty, robust- power going from a daily interval to a
ness, and some additional biases. monthly interval is severe. For example,
with 50 securities the power for a 5 per-
A. Role of Sampling Interval
cent test using daily data is 0.94, whereas
Stock return data is available at differ- the power using weekly and monthly
ent sampling intervals, with daily and data is only 0.35 and 0.12 respectively.
monthly intervals being the most com- The clear message is that there is a sub-
mon. Given the availability of various in- stantial payoff in terms of increased
tervals, the question of the gains of using power from reducing the sampling inter-
val. Dale Morse (1984) presents detailed informal procedure works well and there
analysis of the choice of daily versus is little to gain from the more elaborate
monthly data and draws the same conclu- estimation framework.
sion.
C. Robustness
A sampling interval of one day is not
the shortest interval possible. With the The statistical analysis of Sections 4, 5,
increased availability of transaction data, and 6 is based on assumption that re-
recent studies have used observation in- turns are jointly normal and temporally
tervals of duration shorter than one day. independently and identically distri-
However, the net benefit of intervals less buted. In this section, discussion of the
than one day is unclear as some compli- robustness of the results to departures
cations are introduced. Discussion of us- from this assumption is presented. The
ing transaction data for event studies is normality assumption is important for
included in the work of Michael Barclay the exact finite sample results to hold.
and Robert Litzenberger (1988). Without assuming normality, all results
would be asymptotic. However, this is
B. Inferences twith Event-Date
generally not a problem for event studies
Uncertainty
because for the test statistics, conver-
Thus far it is assumed that the event gence to the asymptotic distributions is
date can be identified with certainty. rather quick. Brown and Warner (1985)
However, in some studies it may be diffi- provide discussion of this issue.
cult to identify the exact date. A com-
D. Other Possible Biases
mon example is when collecting event
dates from financial publications such as A number of possible biases can arise
the Wall Street Journal. When the event in the context of conducting an event
announcement appears in the paper one study. Nonsynchronous trading can in-
can not be certain if the market was in- troduce a bias. The nontrading or non-
formed prior to the close of the market synchronous trading effect arises when
the prior trading day. If this is the case prices, are taken to be recorded at time
then the prior day is the event day, if not intervals of one length when in fact they
then the current day is the event day. are recorded at time intervals of other
The usual method of handling this prob- possibly irregular lengths. For example,
lem is to expand the event window to the daily prices of securities usually em-
two days day 0 and day +1. While there ployed in event studies are generally
is a cost to expanding the event window, "closing" prices, prices at which the last
the results in Section 6 indicated that transaction in each of those securities oc-
the power properties of two day event curred during the trading day. These
windows are still good suggesting that closing prices generally do not occur at
the costs are worth bearing rather than the same time each day, but by calling
to take the risk of missing the event. them "daily" prices, one is implicitly and
Clifford Ball and Walter Torous (1988) incorrectly assuming that they are
have investigated the issue. They de- equally spaced at 24-hour intervals. This
velop a maximum likelihood estimation nontrading effect induces biases in the
procedure which accommodates event moments and co-moments of returns.
date uncertainty and examine results of The influence of the nontrading effect
their explicit procedure versus the infor- on the variances and covariances of indi-
mal procedure of expanding the event vidual stocks and portfolios naturally
window. The results indicates that the feeds into a bias for the market model
beta. Myron Scholes and Williams (1977) have been in the area of corporate fi-
present a consistent estimator of beta in nance. Event studies dominate the em-
the presence of nontrading based on the pirical research in this area. Important
assumption that the true return process examples include the wealth effects of
is uncorrelated through time. They also mergers and acquisitions and the price
present some empirical evidence which effects of financing decisions by firms.
shows the nontrading-adjusted beta esti- Studies of these events typically focus on
mates of thinly traded securities to be the abnormal return around the date of
approximately 10 to 20 percent larger first announcement.
than the unadjusted estimates. However, In the 1960s there was a paucity of
for actively traded securities, the adjust- empirical evidence on the wealth effects
ments are generally small and unimpor- of mergers and acquisitions. For exam-
tant. ple, Henry Manne (1965) discusses the
Premn Jain (1986) considers the influ- various arguments for and against merg-
ence of thin trading on the distribution ers. At that time the debate centered on
of the abnormal returns from the market the extent to which mergers should be
model with the beta estimated using the regulated in order to foster competition
Scholes-Williams approach. When com- in the product markets. Manne argued
paring the distribution of these abnormal that mergers represent a natural out-
returns to the distribution of the abnor- come in an efficiently operating market
mal returns using the usual OLS betas for corporate control and consequently
finds that the differences are minimal. provide protection for shareholders. He
This suggests that in general the adjust- downplayed the importance of the argu-
ments for thin trading are not important. ment that mergers reduce competition.
The methodology used to compute the At the conclusion of his article Manne
cumulative abnormal returns can induce suggested that the two competing hy-
an upward bias. The bias arises from the potheses for mergers could be separated
observation by observation rebalancing by studying the price effects of the in-
to equal weights implicit in the calcula- volved corporations. He hypothesized
tion of the aggregate cumulative abnor- that, if mergers created market power,
mal return combined with the use of one would observe price increases for
transaction prices which can represent both the target and acquirer. In contrast,
both the bid and the offer side of the if the merger represented the acquiring
market. Marshall Blume and Robert corporation paying for control of the tar-
Stambaugh (1983) analyze this bias and get, one would observe a price increase
show that it can be important for studies for the target only and not for the ac-
using low market capitalization firms quirer. However, Manne concludes, in
which have, in percentage terms, wide reference to the price effects of mergers,
bid offer spreads. In these cases the bias that "no data are presently available on
can be elimninated by considering cumu- this subject."
lative abnormal returns which represent Since that time an enormous body of
buy and hold strategies. empirical evidence on mergers and ac-
quisitions has developed which is domi-
11. Concluding Discussion nated by the use of event studies. The
general result is that, given a successful
In closing, examples of event study takeover, the abnormal returns of the
successes and limitations are presented. targets are large and positive and the ab-
Perhaps the most successful applications normal returns of the acquirer are close
to zero. Gregg Jarrell and Poulsen (1989) -3.56 percent. In contrast, when firms
document that the average abnormal re- decide to use straight debt financing, the
turn for target shareholders exceeds 20 average abnormal return is closer to
percent for a sample of 663 successful zero. Mikkelson and Partch (1986) find
takeovers from 1960 to 1985. In contrast the average abnormal return for debt is-
the abnormal returns for acquirers is sues to be -0.23 percent for a sample of
close to zero. For the same sample, Jar- 171 issues. Findings such as these pro-
rell and Poulsen find an average abnor- vide the fuel for the development of new
mnal return of 1.14 percent for acquirers. theories. For example, in this case, the
In the 1980s they find the average abnor- findings motivate the pecking order the-
mal return is negative at -1.10 percent. ory of capital structure developed by Ste-
Eckbo (1983) explicitly addresses the wart Myers and Nicholas Majluf (1984).
role of increased market power in ex- A major success related to those in the
plaining merger related abnormal re- corporate finance area is the implicit ac-
turns. He separates mergers of compet- ceptance of event study methodology by
ing firms from other mergers and finds the U.S. Supreme Court for determining
no evidence that the wealth effects for materiality in insider trading cases and
competing firms are different. Further, for determining appropriate disgorge-
he finds no evidence that rivals of firms ment amounts in cases of fraud. This im-
merging horizontally experience negative plicit acceptance in the 1988 Basic, In-
abnormal returns. From this he con- corporated v. Levinson case and its
cludes that reduced competition in the importance for securities law is discussed
product market is not an important ex- in Mitchell and Netter (1994).
planation for merger gains. This leaves There have also been less successful
competition for corporate control a more applications. An important characteristic
likely explanation. Much additional em- of a successful event study is the ability
pirical work in the area of mergers and to identify precisely the date of the
acquisitions has been conducted. Mi- event. In cases where the event date is
chael Jensen and Richard Ruback (1983) difficult to identify or the event date is
and Jarrell, James Brickley, and Netter partially anticipated, studies have been
(1988) provide detailed surveys of this less useful. For example, the wealth ef-
work. fects of regulatory changes for affected
A number of robust results have been entities can be difficult to detect using
developed from event studies of financ- event study methodology. The problem
ing decisions by corporations. When a is that regulatory changes are often de-
corporation announces that it will raise bated in the political arena over time and
capital in external markets there is, on any accompanying wealth effects gener-
average, a negative abnormal return. The ally will gradually be incorporated into
magnitude of the abnormal return de- the value of a corporation as the prob-
pends on the source of external financ- ability of the change being adopted in-
ing. Asquith and Mullins (1986) find for creases.
a sample of 266 firms announcing an eq- Larry Dann and Christopher James
uity issue in the period 1963 to 1981 the (1982) discuss this issue in the context of
two day average abnormal return is -2.7 the impact of deposit interest rate ceil-
percent and on a sample of 80 firms for ings for thrift institutions. In their study
the period 1972 to 1982 Wayne Mikkel- of changes in rate ceilings, they decide
son and Megan Partch (1986) find the not to consider a change in 1973 because
two day average abnormal return is it was due to legislative action. Schipper
and Thompson (1983, 1985) also encoun- Based Accounting Research," J. Acc. Res., 1987,
25(1), pp. 1-48.
ter this problem in a study of merger BLUME, MARSHALL E. AND STAMBAUGH,
related regulations. They attempt to ROBERT F. "Biases in Computed Returns: An
circumvent the problem of regulatory Application to the Size Effect," J. Finan. Econ.
Nov. 1983, 12(3), pp. 387-404.
changes being anticipated by identify- BOEHMER, EKKEHART; MUSUMECI, JIM AND
ing dates when the probability of a POULSEN, ANNETTE B. "Event-Study Method-
regulatory change being passed changes. ology under Conditions of Event-Induced Vari-
ance," J. Finan. Econ., Dec. 1991, 30(2), pp.
However, they find largely insignificant 253-72.
results leaving open the possibility the BROW7N, STEPHEN J. AND WARNER, JEROLD B.
Evidence from Three Decades," Financial Man- MYERS, STEWART C. AND MAJLUF, NICHOLAS S.
agement, Autumn 1989,18(3), pp. 12-19. "Corporate Financing and Investment Deci-
JENSEN, MICHAEL C. AND RUBACK, RICHARD S. sions When Firms Have Information That In-
"The Market for Corporate Control: The Scien- vestors Do Not Have," J. Finan. Econ., June
tific Evidence," J. Finan. Econ., Apr. 1983, 1984, 13(2), pp. 187-221.
11(1-4), pp. 5-50. PATELL, JAMES M. "Corporate Forecasts of Earn-
KANE, EDWARD J. AND UNAL, HALUK. "Change ings Per Share and Stock Price Behavior: Em-
in Market Assessments of Deposit-Institution pirical Tests," J. Acc. Res., Autumn 1976, 14(2),
Riskiness," J. Finan. Services Res., June 1988, pp. 246-76.
1(3), pp. 207-29. PRABHALA, N. R. "Conditional Methods in Event
LANEN, WILLIAM N. AND THOMPSON, REX. Studies and an Equilibrium Justification for Us-
"Stock Price Reactions as Surrogates for the ing Standard Event Study Procedures." Work-
Net Cash-Flow Effects of Corporate Policy De- ing Paper. Yale U., Sept. 1995.
cisions," J. Acc. Econ., Dec. 1988, 10(4), pp. RITTER, JAY R. "Long-Run Performance of Initial
311-34. Public Offerings," J. Finance, Mar. 1991, 46(1),
LINTNER, JOHN. "The Valuation of Risky Assets pp. 3-27.
and the Selection of Risky Investments in Stock Ross, STEPHEN A. "The Arbitrage Theory of
Portfolios and Capital Budgets," Rev. Econ. Capital Asset Pricing," J. Econ. Theory, Dec.
Stat., Feb. 1965, 47(1), pp. 13-37. 1976, 13(3), pp. 341-60.
MACKINLAY, A. CRAIG. "On Multivariate Tests of SCHIPPER, KATHERINE AND THOMPSON, REX.
the CAPM," J. Finan. Econ., June 1987, 18(2), "The Impact of Merger-Related Regulations
pp. 341-71. on the Shareholders of Acquiring Firms,"
MALATESTA, PAUL H. AND THOMPSON, REX. J. Acc. Res., Spring 1983, 21(1), pp. 184-
"Partially Anticipated Events: A Model of Stock 221.
Price Reactions with an Application to Corpo- . "The Impact of Merger-Related Regula-
rate Acquisitions," J. Finan. Econ., June 1985, tions Using Exact Distributions of Test Statis-
14(2), pp. 237-50. tics," J. Acc. Res., Spring 1985, 23(1), pp. 408-
MANNE, HENRY G. "Mergers and the Market for 15.
Corporate Control," J. Polit. Econ., Apr. 1965, SC,HOLES, MYRON AND WILLIAMS, JOSEPH T.
73(2), pp. 110-20. "Estimating Betas from Nonsynchronous Data,"
MCQUEEN, GRANT AND ROLEY, VANCE. "Stock J. Finan. Econ., Dec. 1977, 5(3), pp. 309-
Prices, News, and Business Conditions," Rev. 27.
Finan. Stud., 1993, 6(3), pp. 683-707. SC,HWERT, G. WILLIAM. "Using Financial Data to
MIKKELSON, WAYNE H. AND PARTCH, MEGAN. Measure Effects of Regulation," J. Law Econ.,
"Valuation Effects of Security Offerings and the Apr. 1981, 24(1), pp. 121-58.
Issuance Process," J. Finan. Econ., Jan./Feb. SHARPE, WILLIAM F. "Capital Asset Prices: A
1986, 15(1/2), pp. 31-60. Theory of Market Equilibrium Under Condi-
MITCHELL, MARK L. AND NETTER, JEFFRY M. tions of Risk," J. Finance, Sept. 1964, 19(3), pp.
"The Role of Financial Economics in Securities 425-42.
Fraud Cases: Applications at the Securities and . Portfolio theory and capital m-arkets. New
Exchange Commission," Business Lawyer, Feb. York: McGraw-Hill, 1970.
1994, 49(2), pp. 545-90. SHARPE, WILLIAM F.; ALEXANDER, GORDON J.
MORSE, DALE. "An Econometric Analysis of the AND BAILEY, JEFFERY V. Investments. Fifth
Choice of Daily Versus Monthly Returns In Ed. Englewood Cliffs, NJ: Prentice-Hall,
Tests of Information Content," J. Acc. Res., 1995.
Autumn 1984, 22(2), pp. 605-23. WHITE, HALBERT. "A Heteroskedasticity-Consis-
MYERS, JOHN H. AND BAKAY, ARCHIE J. "Influ- tent Covariance Matrix Estimator and a Direct
ence of Stock Split-Ups on Market Price," Har- Test for Heteroskedasticity," Economtetrica,
vard Bus. Rev., Mar. 1948, 26, pp. 251-55. May 980, 48(4), pp. 817-38.