Journal of Financial Economics 11 (1983) 225-240.

North-Holland Publishing Company

EXAMINING ANTITRUST POLICY TOWARDS

HORIZONTAL MERGERS*

Robert STILLMAN
Lexecon Inc., Chicago, IL 60603, USA

Received June 1980, final version received March 1982

A horizontal merger must result in higher product prices to consumers to be anticompetitive or

socially inefficient. Higher product prices, however, imply increased profits for rivals to the
merging firms. Therefore, if a horizontal merger is to reduce consumer welfare, rival firms must
rise in value at the time of events increasing the probability of the merger and fall in value when
the probability of the merger declines. This paper uses daily stock return data from a sample of
rivals to 11 horizontal mergers attempted between 1964 and 1972 that were challenged by the
antitrust enforcement agencies. The paper tests the hypothesis that, but for the government’s
action, these mergers would have resulted in higher product prices. On balance, the data favor
the null hypothesis of no anticompetitive effect.

1. Introduction
Section 7 of the Clayton Act makes illegal any merger where the effect
‘may be substantially to lessen competition or to tend to create a
monopoly’.’ The Celler-Kefauver amendment of 1950 significantly extended
the reach of this statute and, since that time, the Federal Trade Commission
(FTC) and the Department of Justice (DOJ) have brought several hundred
Section 7 cases. In the past, many of these cases have been brought against
vertical and conglomerate mergers. Now, however, with the release of the
Justice Department’s new Merger Guidelines, the government’s enforcement
of section 7 has been refocussed. Summarizing the Antitrust Division’s
present policy, Assistant Attorney General William F. Baxter has stated:
‘Mergers are never troublesome except insofar as they give rise to horizontal
problems’.2
An antimerger policy that focuses exclusively on horizontal mergers is
consistent in principle with economic efficiency. Mergers between current

*This paper is derived from my dissertation research and I would like to thank the members
of my U.C.L.A. committee for their assistance: Harold Demsetz (chairman), Benjamin Klein,
Edward Learner. and Richard Roll. 1 would also like to thank Peter Dodd, Michael Jensen and
Richard Ruback for thier comments. Finally. thanks are due to Laurence Levin for his timely
computer assistance.
‘I5 U.S.C.A. 518.
‘Justice Department’s new merger guidelines may be ready by winter, Baxter indicates’, 1027
Antitrust and Trade Regulations Report A-5, Aug. 13, 1981. The Merger Guidelines were
released on June 14, 1982.

0304-405x/83/$03.00 0 Elsevier Science Publishers

226 R. Stillman, Antitrust horizontal merger policy

producers of close substitutes by definition increase industry concentration,

and economists have long accepted as a theoretical proposition the link
between concentration and collusion (tacit or explicit). The link is traced
most articulately by Stigler (1968) in his theory of oligopoly. Other industrial
organization models suggest a similar basis for concern with horizontal
mergers. For example, in the dominant firm model, the demand facing the
dominant firm becomes less elastic as the market share of the fringe firms
diminishes.3 Thus, the acquisition of a rival by the dominant firm increases
the dominant firm’s market power as measured by the Lerner index and, all
else equal, results in an increased gap between price and marginal cost. In
Cournot-Nash models of firm interaction, the equilibrium price rises as the
number of firms declines. Thus, in these models too horizontal mergers imply
less efficient resource allocation, all else equal.
But while horizontal mergers have the clearest anticompetitive potential,
there are also potential efficiency gains from such mergers that the new
antimerger policy may sacrifice. In addition to the obvious possibility of
complementarities in production and distribution, managers in the same
industry may have a comparative advantage at identifying mismanaged firms.
By foreclosing these managers from the market for corporate control, an
anti-horizontal merger policy may impair efficient allocation of managerial
talent and, perhaps more importantly, weaken significantly the incentive of
incumbent managers to maximize the value of their firms. As Manne (1965)
and others have suggested, the threat of take-over disciplines management
and helps reduce agency costs.
The research reported in this paper contributes to the analysis of whether
there is a social efficiency basis for the antitrust policy against horizontal
mergers. The paper asks: Is there any evidence that the horizontal mergers
challenged in the past by the federal government would have resulted in
higher product prices to consumers? Daily stock market data from a sample
of 11 challenged horizontal mergers attempted between 1964 and 1972 are
studied and the conclusion is that the data do not reject the null hypothesis
of no anticompetitive effect.4
The most significant limitation on the study is that, unavoidably, the
sample includes no horizontal mergers that would have occurred, but were
deterred by antitrust enforcement policy. If antimerger policy succeeds in
deterring firms from even attempting the most anticompetitive mergers,
actual challenges could be almost exclusively against socially efficient mergers
and yet the policy might still be consistent with efficiency. Of course, on the
other hand, the less discerning the government is in its challenges, the more
firms will be deterred from attempting cost-saving, socially efficient mergers
- the deterrent effect cuts both ways.

%ee Landes and Posner (1981) for an exposition of the dominant firm model.
4This conclusion is consistent with the findings of Eckbo (1983), who uses a similar
methodology to test the hypothesis.
 R. Stillman, Antitrust horizontal merger policy 221

The body of the paper is divided into three sections. The following section
develops simple testable implications of the hypothesis that the challenged
horizontal mergers would have raised product prices and thereby reduced
consumer welfare. Section 3 follows with a description of the sample of
challenged horizontal mergers studied and how the sample was selected.
Section 4 discusses the empirical technique used to test the inefficiency
hypothesis and then reports the estimation results.

2. Testable implications of the inefficiency hypothesis

As mentioned in the introduction, the link between an antitrust policy

directed against horizontal mergers and economic efficiency is derived from
several models of industrial organization. Stigler’s oligopoly theory, the
dominant firm model (when the merger involves the dominant firm), and
Cournot-Nash interaction models all share the implication that, other things
equal, horizontal mergers tend to result in higher product prices in
equilibrium. If the industry produces a heterogeneous good, the basic
implication is that quality-adjusted prices will tend to be higher post-merger.
Very few challenged horizontal mergers survive government litigation
without the merging firms at least agreeing to a consent decree that in some
way restricts their behavior. Therefore, it is not practicable to test the
inefficiency hypothesis directly by studying the pattern of post-merger
product prices. In the sample of mergers studied in this paper, only two of
eleven mergers resulted in an unambiguous legal victory for the merging
firms. In the remainder of the cases, there either was no post-merger period
to observe (the merger was abandoned or divestiture was ordered) or
behavior in the post-merger period has been constrained by a consent decree.
An alternative and empirically tractable approach to testing the
inefficiency hypothesis builds on an indirect implication of the hypothesis. In
the industrial organization models discussed above, not only do product
prices tend to be higher following a horizontal merger: in addition, a price-
increasing horizontal merger benefits other firms in the industry affected by
the merger. In the dominant firm model, the rivals are sheltered by a price
umbrella. In the tacit collusion models - a label that adequately describes
both Stigler’s oligopoly model and Cournot-Nash models - the rival firms
restrict output and share in the resulting increase in industry-wide profits.
Thus, an implication of the inefficiency hypothesis is that the value of rival
firms should increase as the result of anticompetitive horizontal mergers. This
capital value implication makes stock market data the natural vehicle for
empirical testing. Maintaining a ‘rational expectations’ or ‘efficient markets’
assumption, today’s stock prices represent the capital market investors’ best
estimate of the present discounted value of a share in firms expected future
net income. If the announcement of a horizontal merger changes investors’
expectations about future product prices and hence firms income, the
228 R. Stillman, Antitrust horizontal merger policy

announcement will coincide with an abnormal rise or fall in stock values.

The inefficiency hypothesis predicts that events that raise the probability of a
merger occurring should coincide with abnormally high returns on the stock
of rivals to the merging firms, while probability-decreasing events (such as
judicial decisions against the merger) should coincide with abnormally low
returns on rivals’ stock.
This set of predictions will be tested in section 4 on a sample of challenged
horizontal mergers described in section 3. It is worth noting here, however,
that the pattern of abnormal returns just described, although necessary for a
horizontal merger to be socially inefficient, is not sufIicient.5 First, as
Williamson (1968) demonstrates, higher product prices can be consistent with
efficiency if accompanied by significant production cost savings. In terms of
the dominant firm model, the merger may both reduce marginal costs and
make the dominant firm’s demand curve less elastic. The equilibrium price
may rise and yet total production costs may decline sufficiently to offset the
deadweight loss from the price increase. Thus, one could observe a pattern of
abnormal returns consistent with the inefftciency hypothesis, and yet the
horizontal merger might be efficient on balance. Second, a merger event may
signal more than a change in the probability of a merger occurring. For
example, a merger announcement may signal the existence of hitherto
unappreciated economies of scale that can be realized by rivals as well as by
the merging firms. Investors might speculate that the rivals themselves are
likely to merge and realize the same scale economies and, thus, investors
could react to news of a socially efficient merger that would actually lower
product prices by bidding up the stock price of rival tirms.‘j

3. Constructing a sample of challenged horizontal mergers

3.1. Sample selection criteria

To test the hypothesis that horizontal mergers challenged by the

government would have reduced consumer welfare had they gone

5More precisely, the pattern of abnormal returns described is a necessary condition of

conventional industrial organization models. There are other models, such as Thompson and
Faith (1979) that emphasize the threat to consumer welfare posed by predatory behavior. If a
horizontal merger facilitates predatory behavior and in this way results in long-run higher
product prices, rival firms obviously need not benefit from efficiency-reducing horizontal
mergers. Firms outside the merger will be victims along with consumers. Whether predatory
behavior is, in fact, ever observed is a long debated subject in industrial organization that is not
addressed here.
6There are other signalling scenarios. For example, a judicial decision against a merger may
establish a precedent and signal that future antitrust actions against firms in the industry are
likely to be decided in favor of the government. This antitrust ‘overhang’ is equivalent in effect
to regulation detrimental to the interests of all firms in the industry. Thus, the partial effect of
blocking a horizontal merger could be to raise product prices (indicating the merger would have
been efficient) and yet rivals’ value might fall with this merger event.
 R. Stillman, Antitrust horizontal merger policy 229

unchallenged, a sample of mergers with the following characteristics is

required. The sample must consist of horizontal mergers (a) challenged by
either DOJ or the FTC; (b) in industries with identifiable rivals that are large
enough to be traded on the New York or American stock exchanges; (c) for
which it is possible to isolate events that unambiguously had an effect on the
perceived likelihood of the merger.
The rationale for the first two criteria is obvious. The study concerns
horizontal mergers, therefore challenged vertical, conglomerate and potential
competition mergers are of no interest. Since the empirical tests will require
stock market data, any merger in industries where firms are not traded on
the two major exchanges, and hence are not on the stock return of the
Center for Research in Security Prices (CRSP) at the University of Chicago,
is also of little value to this study.
The rationale for the third criterion - that there be clearly identifiable
events affecting the perceived likelihood of the merger - is also obvious in
that little would be learned by examining stock price movements of rivals on
dates when investors’ assessments of the probability of a merger were
unchanged. This problem, of course, is encountered in any study using stock
market data and typically the response is to study returns over some longer
period extending before and possibly after the hypothesized ‘event date’ to
ensure that returns on the true event date have been recorded. The cost of
this usually unavoidable empirical compromise is that the power of the
ensuing tests is reduced from what it would be were the true event date
known. By measuring returns over a longer period, the variance about the
predicted normal return expands and it becomes more difficult to reject the
null hypothesis of no abnormal performance. As will be discussed below, one
of the attractive features of this study is that this problem can be
circumvented; it is possible to infer from data on the merging firms returns
whether a hypothesized event actually occasioned investor reaction.

3.2. Identifying challenged horizontal mergers in industries j& which stock

return data are available on identifiable rivals ~

The principal source of data on mergers challenged by the government

under section 7 is the Merger Case Digest compiled by the American Bar
Association (ABA). At the time research began on this study, data were
available in the Digest on section 7 cases filed in 1970 or earlier. For the
years 1971-1974, short summaries of merger cases reported in the Commerce
Clearing House (CCH), Trade Regulation Reporter, were used to identify
horizontal cases. Constructing the sample of 11 mergers reported in table 1
from the ABA and CCH lists was largely a process of excluding mergers for
failing to meet one of the criteria described above. Since the CRSP daily
return tape begins at July 1, 1962, the first reason for eliminating a section 7
230 R. Stillman, Antitrust horizontal merger policy

Table 1
Challenged horizontal mergers in the sample.

Acquiring Merger Complaint

(Acquired) year Industry year Agency Outcome

Chrysler Heavy Merger

(Mack Trucks) 1964 trucks 1964 DOJ abandoned
Schenley Liquor
(Buckingham) 1964 distilling 1966 DOJ Consent
Russell Stover Merger
(Fanny Farmer) 1965 Candy 1965 DOJ abandoned
General Dynamics Pro-def.
(UEC) 1966 Coal 1967 DOJ decision
Sterling Drug Health and Pro-def.
(Lehn and Fink) 1966 beauty aids 1969 FTC decision
Bendix
(Fram) 1967 Filters 1967 FTC Consent
Cooper Industries Natural gas Merger
(Waukesha Motor) 1967 engines 1967 DOJ abandoned
Atlantic Richlield Oil
(Sinclair) 1968 relining 1969 DOJ Consent
Gould National
(Clevite) 1969 Batteries 1969 DOJ Consent
Warner Lambert Ethical Pro-govt.
(Parke Davis) 1970 drugs 1971 FTC decision
Jim Walter Roofing Pro-govt.
(Panacon) 1972 materials 1974 FTC decison

case from the sample was if the merger was announced prior to that date.
This filter eliminated many cases, nevertheless it left 104 DOJ and 59 FTC
cases from the Merger Case Digest as initial candidates for inclusion in the
study.
Next, bank mergers and mergers in heavily regulated industries, such as
telecommunications, were exluded for the reason that the link, if one exists,
between horizontal mergers and anticompetitive behavior does not seem
likely to be as significant in a regulated environment.7 Then, mergers that
according to the ABA or CCH description were not primarily horizontal or
cases that complained of a series of mergers were excluded. The reason for
excluding the multiple merger complaints was that in these the government
was complaining of the cumulative effect of the mergers and did not claim

‘Actually, this is an empirical question that could be tested. As a practical matter, however,
virtually all the mergers excluded for this reason were bank mergers involving local banks. It is
unlikely that the rivals to such firms would be traded on major stock exchanges and therefore
such mergers would drop from the sample for this reason.
 R. Stillman, Antitrust horizontal merger policy 231

that any one merger had a significant effect on product prices. This theory
cannot be tested using the methodology employed here.
These filters reduced the universe of Digest cases to 45 DOJ and 22 FTC
challenged horizontal mergers. Next, mergers in which neither the acquired
nor acquiring firm were traded on the New York or American stock
exchanges were excluded on the grounds that it was unlikely that rivals to
these firms would be on the CRSP tape either. This left 28 DOJ and 17 FTC
cases from the Digest as possible candidates. At this point, an attempt was
made to identify the other firms in the industries that the government alleged
would have been adversely affected by the challenged mergers. The first
source for this information was the published opinions in cases that were
litigated. These decisions often contain a description of the industry and
identify industry members. The other source of data was fact memoranda
prepared by the enforcement agencies in preparation for filing complaints.
These memoranda were obtained by filing a request under the Freedom of
Information Act. In some cases, the memoranda could not be located, while
in others there was no description of other firms in the supposedly affected
industries. After this penultimate filter, the universe of candidate mergers was
reduced to 18.

3.3. Identifying unambiguous merger event dates

The next step in constructing the sample involved collecting dates of

events likely to have affected the perceived probability that the 18 mergers
would ultimately be consummated. The following events were hypothesized
to be of this type: merger rumors and announcements; decisions by the
courts on government motions for temporary restraining orders and
preliminary injunctions; decisions by district courts (DOJ complaints) and
administrative law judges (FTC complaints) on the merits of the merger
complaints; and decisions by appellate courts such as the commissioners of
the FTC, the Court of Appeals, and the Supreme Court. The primary sources
for these dates were the Wall Street Journal and docket sheets kept by the
DOJ and FTC (showing the dates of the complaint and any subsequent
litigation events).
The question then became whether these events were truly events - did
they transmit any new information to the capital markets? The fortunate
feature of this study is that there is an empirical means of answering this
question. Several studies of mergers, most recently Dodd (1980), have found
large, significantly positive abnormal returns accruing to acquired firms on
the day of and day prior to merger announcements. From this, one can infer
that if merger events of the type identified above truly conveyed new
information, then on or about these dates there should have been discernible
abnormal movements in the stock of the would-be acquired firm and, though
 Table 2
Description of merger events in the sample of 11 mergers in the period S/644/72 and demonstration that investors reacted to these
events.a

Upper or lower
Stock Predicted Actual bound of 95%
Merger Event observed effect Date return confidence interval

Chrysler merger Mack + WI64 0.0634 0.0465

Mack announced after 514164 0.0767 0.0424
close of trading,
WSJ 514164
complaint tiled, Mack _ 7131164 -0.1070 -0.0354
7130164
preliminary Mack - 8117164 -0.1243 -0.0364
injunction,
8117164
Schenley merger agree- Schenley + 8126164 0.0581 0.0256
Buckingham ment announced,
WSJ 8127164
Russell Stover merger agree- Fanny Farmer + 214165 0.0291b 0.0365
Fanny Farmer ment announced, 2/S/65 0.0282b 0.0352
WSJ 219165 218165 0.0274b 0.0346
General major stock pur- UEC + 9130166 0.1667 0.0304
Dynamics chase by General
United Dynamics,
Electric Coal WSJ 9129166
Sterling Drug Lehn and Fink an- Lehn and Fink + 2/I/66 0.1475 0.0400
Lehn and Fink nounced receiving
several acquisition
bids, including one,
from Sterling,
WSJ l/31/66

Lehn and Fink Lehn and Fink + 3128166 0.1104 0.0402

approved bid
from Sterling,
WSJ 3128166
 FTC judge Sterling + 5/12/71 0.0361 0.0253
dismissed 5113171 0.0222 0.0218
complaint,
S/12/71
Bendix Fram agreed Fram + 12/28/M 0.0751 0.0373
Fram to merger in t/3/61 0.1132 0.0406
principle,
WSJ l/3/61
Cooper Cooper plans Cooper + 7125161 0.1051 0.0470
Waukesha to acquire
remainder of
shares
WSJ 7/25/61
Atlantic complaint filed, Sinclair _ l/16/69 -0.0955 - 0.0329
Richfield l/15/69
Sinclair Oil
temporary
restraining order,
l/17/69
Gould National merger agree- Clevite + 3/10/69 0.1093 0.0445
Clevite ment announced,
3110169
Warner Lambert Parke Davis Parke Davis + 7/31/70 0.1233 0.0517
Parke Davis agreed to merger
in principle,
WSJ 7/31/70
Jim Walter Jim Walter Jim Walter + 414172 0.0608 0.0409
agreed to buy
89% stock interest,
WSJ 414172

“Sources: Wall Street Journal (WSJ); docket sheets on tile at the FTC and DOJ; and fact memoranda prepared by enforcement
agencies prior to tiling a complaint.
‘The ratio of the three-day cumulative residual to the three-day standard error is 2.73.
234 R. Stihan, Antitrust horizontal merger policy

less certainly, the would-be acquiring firm. (Surprisingly, Dodd also finds
small, but significantly negative abnormal returns on the stock of acquiring
firms over the same announcement period.)
To implement this test, the following market model was estimated by
ordinary least squares for each merging firm having daily stock return data
available:

R[, = cli + BiRmt+ &it,

where

Ri, =return on security i over period t,

R,, =return on value weighted market portfolio over period t,

&it =disturbance term over period t.

The estimation period was the twelve month period ending one month
after the hypothesized event dates. Examining the residuals from these
regressions, there was a striking correspondence between outliers and
hypothesized event dates in 11 of the 18 mergers. The results from these 11
mergers are reported in table 2, which shows the merger; the nature and date
of the merger event; the stock observed (which is always the stock of the
acquired firm unless no returns were available for this firm); the predicted
effect on the firm’s value given the nature of the hypothesized event; the date
of the outlier; the actual return on that date; and the upper or lower bound
of the 95 percent confidence interval about the predicted normal return for
that date. In each instance reported, actual returns on days on or about the
hypothesized event date were outside the 95 percent confidence interval and,
in most cases, considerably so. In the empirical work reported in section 4,
the dates of these outliers (and only these dates) are considered event dates
- the dates on which new information concerning the mergers reached the
capital markets.

4. Empirical results

4.1. Estimation technique

The inefficiency hypothesis - the hypothesis that challenged horizontal

mergers would have, but for the government’s action, led to increased
product prices - implies that rival firms in industries affected by challenged
mergers should have risen in value when the mergers in the sample became
more likely and depreciated when the probability of these mergers declined.
To test this hypothesis, equally weighted portfolios of the stock of rivals were
formed and, for each merger and significant merger event in the sample, the
following modified market model was estimated by ordinary least squares:
 R. Stillman, Antitrust horizontal merger policy 235

R,,= up+ B,Rmt

+ Y,D,+ ~ptt
where

R,, =return on equally weighted portfolio of rivals over period t,

R,, = return on value weighted market portfolio over period t,
D, =dummy variable that has a value of one if period t is an event date
and zero otherwise,
&Pi
=disturbance term on portfolio of rivals over period t.

The dummy variables in these regressions measure abnormal returns. In

each regression, there is one dummy variable for each merger event date. For
example, in the Chrysler-Mack merger, where the capital markets apparently
reacted to news of the merger on May 1 and May 4, 1964, the regression
would contain two dummy variables, the tirst taking a value of one on May
1 and the second taking a value of one only on May 4. As before, estimation
was over a twelve-month period ending one month after the event date (the
latest event date in cases such as Chrysler-Mack). Because the effects of the
merger information have been purged by the dummy variables, the intercept
and market return coefficients are not influenced by the merger events.

4.2. Estimation results

The underlying objective of this study is to determine whether the antitrust

enforcement agencies have tended to challenge horizontal mergers that would
have in fact resulted in higher product prices. To this end, abnormal returns
on portfolios of the stock of rivals to firms engaged in challenged horizontal
mergers have been estimated for days on which the perceived probability of
the mergers evidently changed. The results are summarized in table 3. The
table lists, for each merger in the sample, the firms in the portfolio of rival$
the merger event; the date of the event; the sign on the rivals’ abnormal
return predicted by the inefficiency hypothesis; the abnormal return estimate;
and the t-statistic on this estimate. Where there are two days on which the
capital markets appear to have reacted to a particular event (e.g., the May 1
and May 4, 1964 reactions to the news of the Chrysler-Mack merger), or
where two merger events occurred in roughly the same time period (e.g., the
complaint filed against the Chrysler-Mack merger on July 30 and the
preliminary injunction against the merger on August 17, 1964) the magnitude
and significance of the sum of the dummy variable coefficients is reported.
These multiple day abnormal return estimates are enclosed in parentheses in
the table. This manner of reporting multiple day results is similar to the

‘The number of rivals varies from merger to merger because the number of identifiable rivals
with traded stock varies.
 Table 3
Abnormal returns on portfolios of rival tirms on days when challenged horizontal mergers became more or less likely for 11
sample mergers in the period _5/644/72.”

Firms in Prediction f-statistic

portfolio of inefficiency Abnormal on abnormal
Merger of rivals Event Date hypothesis return return

Chrysler International merger 5/l/64 + 0.0101 1.53

Mack Harvester announcement
General Motors 514164 0.0062 0.94
(5/t/64 (1) (0.0163) (1.74)
+ S/4/64)
complaint 7130164 _ - 0.0040 -0.60
filed
preliminary g/17/64 _ 0.0005 0.07
injunction
(7/30/64 (-) (-0.0035) (-0.37)
+ 8/l 7/64)

Schenley National merger g/26/64 + 0.0009 0.14

Buckingham Distillers announcement
Heublein
Brown Foreman
American
Distillers
Russell Stover Bartons merger 214165 + 0.0216 0.86
Fanny Farmer Candy announcement
215165 + o.ooo2 0.01
2/g/65 + - 0.0209 -0.83

WW5
t 215165 (+) (0.0010) (0.02)
t 218165)
General Dynamics Conoco stock 9130166 + 0.0041 0.50
United Standard Oil purchase
Electric Coal of Ohio by General
Dynamics

Sterling Drug Warner Lambert Lehn and Fink 2/l/66 + 0.0057 0.98
Lehn and Fink Pfizer received
merger bids
American Lehn and Fink 3128166 + 0.0103 1.77
Cyanamid approved bid
from Sterling
WI66 (+) (0.0160) (1.94)
t 3128166)
FTC judge 5112171 + - 0.0009 -0.12
dismissed 5/13/71 0.0177 2.30
complaint
(5112171 (+) (0.0169) (1.54)
+5/13/71)

Bendix General Motors merger 12128166 + -0.0013 -0.14

Fram agreement
l/3/67 0.0454 4.99
( 12128166 (=) (0.0441) (3.42)
+ l/3/67)
 Table 3 (continued)

Firms in Prediction t-statistic

portfolio of inefliciency Abnormal on abnormal
Merger of rivals Event Date hypothesis return return

Cooper Worthington Cooper 7/25/61 + 0.0020 0.22

Waukesha Caterpillar plans stock
Ingersoll-Rand purchase
Dresser
Industries

Atlantic Richlield Conoco complaint l/16/69 _ 0.0061 1.21

Sinclair Oil Shell filed;
Standard Oil temporary
of Indiana restraining
Exxon order
Texaco

Gould National Union Carbide merger 3110169 + - 0.0057 -0.58

Clevite ESB agreement
PR Mallory

Warner Lambert American Home merger l/3 l/IO + 0.0008 0.09

Parke Davis Products agreement
Upjohn
Smithkline
American Cyanamid

Jim Walter Certain-Teed merger 414172 + 0.0016 0.15

Panacon Johns Manville agreement
Flintkote

“The data in table 2 indicate that, for some mergers, investors appear to have reacted to a single event on more than one
day. In other cases, two merger events occurred very close in time. In such instances, table 3 reports abnormal returns and
t-statistics for the individual days indicated in tables 2 and for the relevant multiple-day period. These multiple-day statistics
are enclosed in parentheses.
 R. Stillman, Antitrust horizontal merger policy 239

cumulative residual approach adopted in other stock market studies,’ except

that it provides a more powerful test of the inefficiency hypothesis; abnormal
returns on days where there is no evidence that the capital markets were
reacting to merger information are excluded from the test statistic.
Of the 11 mergers in the sample and 18 merger event dates, in only two
instances does the stock of rival firms exhibit an abnormal return that is
consistent with the inefficiency hypothesis and significant at the live percent
level. On January 3, 1967, the day that Fram. agreed in principle to merge
with Bendix, the stock of General Motors had an abnormal positive return
of 4.54 percentage points. This abnormal return, which is not obviously
explained by some other event specific to General Motors,” is almost live
times the standard error of the disturbance term. However, General Motors
stock exhibits no similar abnormally high return on December 28, 1966 when
Fram stock rose by 7.51 percent apparently in anticipation of the merger. In
fact, General Motors stock performed abnormally poorly on that day,
although insignificantly so.
The other abnormal return significant in the direction predicted by the
inefficiency hypothesis occurred on May 13, 1971, the day after an FTC
judge dismissed the complaint against the merger between Sterling Drug and
Lehn and Fink. The portfolio of Warner Lambert, Pfizer, and American
Cyanamid had an abnormal positive return of 1.77 percentage points (t-
statistic of 2.30). The significance of this result is somewhat surprising since,
as reported in table 2, the return on Sterling Drug stock for that date was
only marginally abnormal. However, it is interesting to note that when Lehn
and Fink announced approval of Sterling Drug’s bid on March 28, 1966, a
day on which Lehn and Fink stock rose by 11.04 percent against a predicted
normal return of 0.11 percent, the portfolio of rivals also experienced a
marginally significant positive abnormal return. Thus, there seems at least
weak evidence that the returns on the stock of rivals to this merger conform
to the pattern predicted by the inefficiency hypothesis.

5. Conclusion

The Justice Department’s new Merger Guidelines make horizontal mergers

the principal target of merger antitrust policy. The topic of this paper is the
record of past governmental efforts to block horizontal mergers. In
particular, is there any evidence that horizontal mergers challenged in the
past by the Department of Justice and Federal Trade Commission would

‘See, for example, Fama, Fisher, Jensen and Roll (1969), Mandelker (1974), and Ellert (1976).
“On January 4 3 1967 1the Wall Street Journal printed two stories concerning General Motors:
one reporting that 1966 production statistics were below the 1965 level; the other reporting
plans to expand a Toledo, Ohio transmission plant.
240 R. Stillman, Antitrust horizontal merger policy

have resulted, but for the government’s action, in higher product prices to
consumers?
Daily stock market data from a sample of 11 challenged horizontal
mergers attempted between 1964 and 1972 are used to test this hypothesis. If
a horizontal merger raises product prices, other firms in the industry affected
by the merger benefit. Therefore, if the challenged mergers in the sample
would have been socially inefficient, rivals to the merging firms should have
risen in value on days of events that increased the probability of the mergers
and depreciated on days of events that decreased the probability of the
mergers.
The results reported in section 4 indicate that rivals in only one merger
(Sterling Drug - Lehn and Fink) in the sample of 11 exhibited a pattern of
abnormal returns generally consistent with the predictions of the inefficiency
hypothesis. The rival in one other merger (Bendix-Fram) revealed a mixed
pattern of abnormal returns: significant in the direction of the ineftIciency
hypothesis at the time of one event, insignificant at the time of another event.
The rivals in the other nine mergers exhibited no abnormal returns of any
kind. If the sample studied here is representative of the universe of challenged
horizontal mergers, these findings suggest that on balance the government
has brought Section 7 cases against horizontal mergers that were not
expected by investors to have any appreciable effect on product prices.

American Bar Association, 1970, Merger case digest (American Bar Association, Chicago, IL).
Commerce Clearing House, Trade Regulation Reporter (Commerce Clearing House, New York).
Demsetz, H., 1973, Industry structure, market rivalry, and public policy, Journal of Law and
Economics 16, l-10.
Dodd, P., 1980, Merger proposals, management discretion and stock holder wealth, Journal of
Financial Economics 8, 105-137.
Eckbo, B., 1983, Horizontal mergers, collusion, and stockholder wealth, Journal of Financial
Economics, this issue.
Ellert, J., 1976, Mergers, antitrust law enforcement and stockholder returns, Journal of Finance
31, 715-732.
Fama, E., L. Fisher, M. Jensen and R. Roll, 1969, The adjustment of stock prices to new
information, International Economic Review 10, 1-21.
Landes, W. and R. Posner, 1981, Market power in antitrust cases, Harvard Law Review 94,
937-996.
Mandelker, G., 1974, Risk and return: The case of merging firms, Journal of Financial
Economics 1, 303-336.
Manne, H., 1965, Mergers and the market for corporate control, Journal of Political Economy
78, 11&120.
Stigler, G., 1968, The organization of industry (Richard D. Irwin, Homewood, IL).
Thompson, E. and R. Faith, 1979, A model of non-competitive interdependence and anti-
monopoly laws, Discussion paper no. 143 (University of California, Los Angeles, CA).
Williamson, O., 1968, Economies as an antitrust defense: The welfare tradeoff, American
Economic Review 58, 18-36.