* Corresponding author: Tel.: #1-530-752-2277; fax: #1-530-752-2924. E-mail address: gnbittlingmayer@ucdavis.edu (G. Bittlingmayer) Journal of Financial Economics 55 (2000) 329}359 DOS Kapital: Has antitrust action against Microsoft created value in the computer industry? George Bittlingmayer!,*, Thomas W. Hazlett",# !Graduate School of Management, University of California, Davis, CA 95616, USA "Agricultural and Resource Economics, University of California, Davis, CA, USA #American Enterprise Institute, USA Received 4 June 1998; received in revised form 26 April 1999; accepted 23 November 1999 Abstract Antitrust enforcement that e$ciently constrains Microsoft's behavior bene"ts "rms supplying complements to and/or substitutes for Microsoft's operating system and applications software. However, from 1991 through 1997, 29 reports of federal antitrust enforcement action against Microsoft were accompanied by declines in the value of an index of 159 computer industry "rms (excluding Microsoft). The mean loss to those "rms exceeded $1 billion per event. Eight retreats or setbacks in enforcement were associated with increased computer sector value. Thus, "nancial markets reveal compelling evidence against the joint hypothesis that (a) Microsoft conduct is anticompetitive and (b) antitrust policy enforcement produces net e$ciency gains. ( 2000 Elsevier Science S.A. All rights reserved. JEL classixcation: G1; K2; L1; L4; L5 Keywords: Financial e!ects of regulation; Antitrust; Monopoly; Computer industry; Microsoft 0304-405X/00/$ - see front matter ( 2000 Elsevier Science S.A. All rights reserved. PII: S 0 3 0 4 - 4 0 5 X ( 9 9 ) 0 0 0 5 3 - 7
31
Embed
DOS Kapital: Has antitrust action against Microsoft created value …masonlec.org/site/rte_uploads/files/Bittlingmayer-Hazlett... · 2020. 7. 22. · increasing returns is crucial
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
DOS Kapital: Has antitrust action againstMicrosoft created value in the computer
industry?
George Bittlingmayer!,*, Thomas W. Hazlett",#
!Graduate School of Management, University of California, Davis, CA 95616, USA"Agricultural and Resource Economics, University of California, Davis, CA, USA
#American Enterprise Institute, USA
Received 4 June 1998; received in revised form 26 April 1999; accepted 23 November 1999
Abstract
Antitrust enforcement that e$ciently constrains Microsoft's behavior bene"ts "rmssupplying complements to and/or substitutes for Microsoft's operating system andapplications software. However, from 1991 through 1997, 29 reports of federal antitrustenforcement action against Microsoft were accompanied by declines in the value of anindex of 159 computer industry "rms (excluding Microsoft). The mean loss to those "rmsexceeded $1 billion per event. Eight retreats or setbacks in enforcement were associatedwith increased computer sector value. Thus, "nancial markets reveal compelling evidenceagainst the joint hypothesis that (a) Microsoft conduct is anticompetitive and (b) antitrustpolicy enforcement produces net e$ciency gains. ( 2000 Elsevier Science S.A. All rightsreserved.
JEL classixcation: G1; K2; L1; L4; L5
Keywords: Financial e!ects of regulation; Antitrust; Monopoly; Computer industry;Microsoft
0304-405X/00/$ - see front matter ( 2000 Elsevier Science S.A. All rights reserved.PII: S 0 3 0 4 - 4 0 5 X ( 9 9 ) 0 0 0 5 3 - 7
1. The value of antitrust
Microsoft's sway over operating systems and applications puts everyone elsein the industry at a disadvantage, said Alan C. Ashton, president of WordPer-fect Corp., Orem, Utah. They are a threat to everybody in the industry.
Wall Street Journal, December 11, 1992, p. A4
Since the advent of the personal computer in 1981, Microsoft's operatingsystems and application programs have claimed impressive shares of a rapidlyexpanding market for PC software. This success has attracted attention not onlywithin the industry but also from state and federal antitrust enforcers, theprivate antitrust bar, and the public at large. Since 1990, federal authorities} "rst at the Federal Trade Commission (FTC) and then at the Department ofJustice's Antitrust Division (DOJ) } initiated a series of antitrust investigationsof Microsoft. These inquiries resulted in a 1995 consent decree between Micro-soft and the DOJ, and a 1997 DOJ suit alleging that Microsoft had violated thedecree. In 1998, the DOJ and 20 states "led a new suit on broader charges.
Is Microsoft &a threat to everybody in the industry'? If so, does that harmconsumers? Will antitrust policy help alleviate that threat? For some analysts,Microsoft's high market share, coupled with certain of the "rm's aggressivemarketing practices, implies predatory conduct (DOJ, 1998; more generally, seeGilbert and Williamson, 1998). For others, Microsoft's success in the highlycompetitive and rapidly growing computer sector o!ers a classic example ofa &good monopoly', one that garners market share by o!ering popular productsat low prices (&A case built on speculation, dubious theories', Wall Street Journal,May 19, 1998, op-ed page; Liebowitz and Margolis, 1999).
Whatever Microsoft's intent or success in monopolizing PC software, it isunclear what kind of policy would promote lower prices for consumers, the goalchampioned by Bork (1978) and other antitrust analysts. A policy that correctsan identi"ed market failure improves welfare only if the value created exceedsthe policy's cost. Even Microsoft's foes fear that intervention by the governmentand the courts will result in net losses: &Supporters and critics of the case bothdread the same doomsday scenarios of the technology industry: that the suitswill ultimately lead to much broader government intervention in the softwarebusiness' (&Microsoft friends and foes alike fear government's intervention', WallStreet Journal Interactive Edition, May 14, 1998).
In this paper we ask, Have antitrust enforcement initiatives increased thestock market value of Microsoft's alleged victims? The Microsoft case seemsespecially suited for a stock price study. First, the long history of investigations,"lings, court decisions, and other antitrust announcements o!er a substantialnumber of policy events. Second, a large number of computer industry "rms willprosper if the market for operating systems and major desktop applications
330 G. Bittlingmayer, T.W. Hazlett / Journal of Financial Economics 55 (2000) 329}359
becomes more competitive. This study exploits these circumstances by examin-ing share price reactions for both Microsoft and a portfolio of 159 othercomputer "rms around 54 antitrust enforcement announcements involvingMicrosoft over the seven years 1991}97.
2. The Microsoft question and the stock market
The basic facts are straightforward, but interpretations vary. Microsoft hasa large market share in desktop operating systems and in several important linesof PC software. It has also earned a high rate of return on its past investments.Additionally, the "rm's stock boasts a very high P/E ratio, indicating thatinvestors anticipate unusually high earnings growth in the future. Finally,Microsoft employs aggressive business practices.
To Microsoft's critics, these facts are consistent with the view that Microsoftgained a monopoly through predation and that the stock market expects thisstrategy to produce an even larger #ow of monopoly rents in the future. Thecritics claim that quality-adjusted prices of operating systems, software, andrelated products are higher now } or will be higher in the future } because ofMicrosoft's actions. To Microsoft's supporters, these same facts are consistentwith the view that Microsoft outcompetes rivals by expanding output andlowering prices for consumers and that stock investors expect continued com-petitive superiority in a rapidly growing industry. The supporters claim Micro-soft's practices might have hurt competitors, but not consumers.
Speci"c aspects of the Microsoft debate demonstrate the di$culty of relyingon economic theory alone. For example, after initially giving away its browser,Netscape began to charge $49 per copy. Soon afterwards, Microsoft incorpor-ated its browser, Internet Explorer, with Windows 98, essentially charginga price of zero. Microsoft also required computer manufacturers installingWindows 98 to place an IE icon on the opening screen (Nash, 1996; Quittnerand Slatalla, 1998). Did Microsoft's &tie-in' and/or zero pricing hurt Netscape?Would its actions ultimately hurt consumers by eliminating a rival? It seemsclear that prices for browsers were lower in the short run, but the elimination ofa rival might mean higher prices in the long run. As a matter of economic theory,the ultimate e!ects are unclear.
Another line of controversy concerns &network e!ects' in operating systemsand applications software. Network e!ects arise when a product becomes morevaluable as more people use it. Telephones and fax machines provide classicexamples. When a single "rm controls the underlying standard, the result maybe a &winner-take-all' outcome. Economic analyses (Declaration of Kenneth J.Arrow, U.S. v. Microsoft, Jan. 17, 1995) and popular treatments (&The forceof an idea', The New Yorker, Jan. 12, 1998) of the Microsoft case have high-lighted network e!ects. However, network e!ects cut both ways. &The theory of
G. Bittlingmayer, T.W. Hazlett / Journal of Financial Economics 55 (2000) 329}359 331
increasing returns is crucial to the case against Microsoft2 [but] increasingreturns are equally crucial to the case for Microsoft } as a reason why trying tobreak it up would be a bad thing' (&The legend of Arthur', Slate Magazine, Jan.14, 1998; see also &Soft microeconomics: the squishy case against you-know-who', Slate Magazine, April 23, 1998). Again, theory is inconclusive. Suchambiguity plagues nearly every other aspect of the nearly decade-old discussionabout Microsoft and monopoly.
We will attempt to break this deadlock with the help of some evidence. Weturn to the verdict of the "nancial markets, examining the stock returns of "rmsallegedly hurt by Microsoft's anticompetitive behavior. The use of "nancial datato study the e!ects of regulation is summarized by Schwert (1981). Three types ofstudies are relevant. One type, pioneered by Eckbo (1983) and Stillman (1983),examines the e!ects of mergers or merger policy by examining the stock prices ofcompetitors. Prager (1992) looks at competing "rms in the celebrated 1904Northern Securities decision (which brought mergers under the Sherman Act).Mullin et al. (1995) examine the stock prices of competing and vertically related"rms in the U.S. Steel divestiture suit. Banerjee and Eckard (1998) look at theprices of competitors of merging "rms during the 1897}1903 merger wave.
The second group of event studies examines the stock price e!ects for "rmsthat are actual or potential targets of regulation. Jarrell and Peltzman (1985)investigate the direct and the indirect spillover e!ects of product recalls. Mitchelland Netter (1989) implicate antitakeover legislation as a precipitating factor inthe 1987 stock crash by looking at returns of &in-play' stocks, and Schipper andThompson (1983) analyze securities and tax law changes by looking at the stockprice movements of frequent acquirers. In the case of Microsoft, antitrust actionmay have had legal implications for other "rms. In fact, the FTC has investi-gated both Intel and Cisco since the Department of Justice "led its most recentcharges against Microsoft. America Online (AOL) has been the target of litiga-tion by state attorneys general.
The third group of studies focuses on litigation between private parties. Itturns out that the stock gains accruing to a successful litigant are often morethan o!set by the losses to other "rms. Cutler and Summers (1988) look at thePennzoil-Texaco litigation, as do Hertzel and Smith (1993, p. 442), who speci"-cally note &the importance of evaluating industry e!ects when drawing con-clusions about the social costs of litigation'. Similarly, Bizjak and Coles (1995)report that "nancial distress, behavioral constraints, and the costs of follow-onsuits result in a net decline in the total value of the opposing parties in a privatesuit. In the case here, the government suit may represent litigation carried out onbehalf of Microsoft's rivals.
One point deserves emphasis. The stock market will not render a verdict onwhether Microsoft's behavior was anticompetitive. Rather, it will re#ect inves-tors' judgments about the marginal e!ect of antitrust enforcement on theexpected pro"tability of "rms allegedly victimized by monopolization. If stock
332 G. Bittlingmayer, T.W. Hazlett / Journal of Financial Economics 55 (2000) 329}359
values in the computer sector decline with enforcement actions, this could re#ecta belief that antitrust enforcement will impose losses, but not that Microsoft'spractices helped the rest of the industry. Fortuitously, this focuses on the propermargin for policy analysis: of comparison not of an existing market structure toa theoretically improved alternative, but what exists before policy interventionto what is expected after.
3. Testing antitrust policy e4ects
Under anyone's theory of antitrust } the pro-consumer view that antitrustenforcement lowers quality-adjusted prices for consumers (Buchanan and Lee,1992) or the &capture theory' view that antitrust protects competitors (Baumoland Ordover, 1985; Gilligan et al., 1989; Wolf, 1993) } Microsoft share pricesshould fall with unexpected enforcement. Formally, the expected abnormalreturn to Microsoft shareholders will be negative during &pro-enforcement' newswindows: E(R
MS)(0, where R
MSis the return to Microsoft stock during the
event window, adjusted for marketwide returns. Conversely, either view predictsthat setbacks to antitrust actions directed against Microsoft will imply risingshare prices: E(R
MS)'0 during &anti-enforcement' event windows. Antitrust
prosecution brings no bene"ts, while entailing the following: litigation costs,diverted managerial attention, constraints on operations (due to antitrust liabil-ity), lost monopoly pro"ts, and civil penalties levied by a court or agreed to ina consent decree. While Microsoft's stock price reactions do not distinguishbetween the two views of antitrust policy, they do identify the enforcementactions that matter to investors.
Under the pro-consumer view, e!ective antitrust action against Microsoftshould produce three types of bene"ciaries. First, "rms that buy Microsoftproducts will directly bene"t (through lower input costs). Second, "rms thatproduce complementary products will indirectly bene"t (through an outwardshift in demand for their products). Third, according to the predation andvertical foreclosure arguments central to the case against Microsoft, e!ectiveantitrust enforcement will ease barriers to entry for Microsoft's rivals. Thisimplies that, during pro-enforcement event windows, E(R)
NMS'0, where
RNMS
is the abnormal return to non-Microsoft computer companies, and thatduring anti-enforcement event windows, E(R
NMS)(0. Conversely, under the
pro-consumer view, setbacks in enforcement produce the opposite results.Microsoft's customers, the producers of complementary products, and its rivalsshould all lose.
Under the capture view of antitrust, enforcement could hurt consumers byimposing substantial litigation costs, by deterring e$ciency-enhancing behav-ior, by implicitly putting other successful "rms at risk of becoming targets ofantitrust, or by increasing the uncertainty of investment. Stock prices of "rms in
G. Bittlingmayer, T.W. Hazlett / Journal of Financial Economics 55 (2000) 329}359 333
the computer industry could in fact fall, especially the stock prices of customersand producers of complements. Retreats or setbacks in enforcement will haveopposite e!ects, which would be re#ected in rising computer industry stockprices.
4. Antitrust events and the computer portfolio
News of a Federal Trade Commission investigation of Microsoft "rst leakedto the press in March 1991. That investigation led the FTC to vote on a prelimi-nary injunction against the "rm in February 1993, when it deadlocked (hence,"ling no action). It deadlocked a second time in July 1993 on whether it would"le a case. In an unusual policy twist, the Department of Justice immediatelybegan its own investigation utilizing the FTC "les. This inquiry resulted ina July 1994 agreement between Microsoft and the DOJ, "nalized as a consentdecree in August 1995. In the interim, a series of court actions ensued, includingrejection of the proposed consent decree by District Court Judge StanleySporkin (who found the agreement too lenient towards Microsoft), and thenreinstatement of the decree by the D.C. Circuit. The DOJ eventually "led a suitagainst Microsoft on October 21, 1997 accusing the "rm of violating the 1995consent decree. Legal skirmishing, including a much broader antitrust suit "ledby the DOJ in May 1998, continued into 1999. We end our study, however, atDecember 31, 1997.
Table 1 provides details of the 54 events used in this study. The list wascreated from the Wall Street Journal Index, and consists of every breaking newsstory concerning Microsoft and antitrust. Table 1 includes the returns, net of themarket, for both Microsoft and an equally weighted index of computer industry"rms over the one- and three-day windows surrounding publication dates.These are the net returns to Microsoft and to the equally weighted index of othercomputer "rms based on the market model. This index is described in moredetail below. Stock return data used throughout this study are from the Centerfor Research in Security Prices at the University of Chicago. Table 1 also listsother news stories in the Wall Street Journal that could have had a major e!ecton Microsoft share prices on or near event dates.
Table 1 classi"es three event groups. &Pro-enforcement' events report stricterantitrust enforcement against Microsoft, while &anti-enforcement' events entailclear setbacks for, or withdrawal from, vigorous antitrust enforcement.&Ambiguous' events involve either (1) enforcement actions with unclear implica-tions or (2) another major contemporaneous event likely to substantially a!ectMicrosoft's stock price. To illustrate, the March 12, 1991 revelation that Micro-soft was the target of an investigation had straightforward negative implicationsfor Microsoft shareholders, while no possibly o!setting Journal stories appearsimultaneously. (The closest was a March 20 report on insider sales.) Hence the
334 G. Bittlingmayer, T.W. Hazlett / Journal of Financial Economics 55 (2000) 329}359
Tab
le1
One-
and
thre
e-day
residua
lre
turn
sfo
rm
icro
soft
and
the
com
pute
rin
dust
ry(e
xclu
ding
mic
roso
ft)ar
oun
dda
tes
with=
allStr
eetJo
urna
lne
ws
articl
esin
volv
ing
Mic
roso
ftan
dan
titr
ust
,199
1}19
97.T
he&n
ews
sum
mar
y'co
lum
nin
cludes
ash
ortdes
crip
tion
ofth
eev
entan
dan
yot
her
stor
ypubl
ished
,with
date
ofpu
blic
atio
n,th
atco
uld
hav
ein#ue
nced
retu
rns.&P
ro'ev
ents
are
cate
gorize
das
pro
-enf
orc
emen
t,&a
nti'
are
cate
gorize
das
anti-e
nfor
cem
ent,
and
those
assign
eda
que
stio
nm
ark
are
ambig
uous
eith
erbe
cause
ofth
enat
ure
ofth
een
forc
emen
tac
tion
ordu
eto
aco
ntam
inat
ing
even
t.
Publis
hed
date
Res
idual
retu
rnN
ews
sum
mar
y/C
ateg
ory
1-da
y3-
day
Mic
roso
ft(%
)In
dust
ry(%
)M
icro
soft
(%)
Indus
try
(%)
13/
12/9
1!
2.30
!0.
80!
1.54
!0.
94M
icro
soft
ista
rget
ofF
TC
inve
stig
atio
n.Pro
24/
15/9
15.
69!
1.60
!2.
61!
1.57
FTC
isbr
oade
nin
gits
inve
stig
atio
n.65
%in
crea
sein
earn
ings
(4/1
7).
?
310
/21/
922.
140.
521.
691.
37FT
Chas
subpo
enae
dda
tafrom
Mic
roso
ft.
Pro
412
/11/
92!
2.77
!0.
79!
4.53
!1.
84FTC
sta!
law
yers
sent
report
on
Dec
.4
reques
ting
inju
nct
ion
agai
nst
Mic
roso
ft.
Pro
52/
8/93
!3.
44!
0.58
0.18
!3.
98FTC
split
s2}
2,no
prel
imin
ary
inju
nction
agai
nst
Mic
roso
ft.
?6
7/15
/93
!0.
66!
0.72
!1.
36!
1.85
FTC
ends
its
two-
year
prob
eofIn
tel,
tota
keup
Mic
roso
ftal
lega
tion
s.?
77/
22/9
3!
0.93
0.48
!3.
461.
07FTC
una
ble
tore
ach
aco
ncl
usio
n.
?8
8/2/
93!
2.40
0.61
!7.
820.
98D
OJ
isre
view
ing
docu
men
tsfrom
Mic
roso
ft.
Pro
98/
23/9
31.
110.
990.
74!
0.11
DO
Jre
port
ed8/
20it
will
laun
cha
form
alin
vest
igat
ion
ofM
icro
soft.
Pro
106/
6/94
3.00
0.20
2.06
!2.
23D
OJ
inve
stig
atio
nin
tens
ifyin
g,ta
king
dep
ositio
ns.
No
othe
rre
port
s.Pro
117/
1/94
!3.
41!
1.24
!4.
51!
0.97
Info
rmat
ion-
gath
erin
gpor
tion
ofD
OJ
inve
stig
atio
nis
ove
r.?
127/
18/9
43.
490.
022.
52!
1.85
Mic
roso
ftsign
edco
nse
ntdec
ree.
Ant
i13
10/1
7/94
!1.
440.
02!
1.17
!0.
14M
icro
soft
acqu
isitio
nlik
ely
tosp
ur
antitr
ustpro
be.
?14
10/2
4/94
0.27
0.45
1.77
0.59
DO
Jw
illex
amin
eag
reem
ent
topu
rcha
seIn
tuit.
?15
11/2
2/94
0.37
!0.
30!
0.22
!0.
80M
icro
soft's
pro
pose
dac
quisitio
nofIn
tuit
isfa
cing
continui
ng
scru
tiny
.?
161/
11/9
50.
63!
0.03
1.20
1.30
Com
pet
itor
s"le
brief
,tr
yto
unr
avel
conse
nt
dec
ree.
Pro
171/
16/9
50.
890.
062.
340.
43Sp
orki
nin
vite
sJa
cobo
vitz
and
Reb
ack
topr
esen
tor
alar
gum
ents
.Pro
G. Bittlingmayer, T.W. Hazlett / Journal of Financial Economics 55 (2000) 329}359 335
Tab
le1
(con
tinu
ed)
Publis
hed
date
Res
idual
retu
rnN
ews
sum
mar
y/C
ateg
ory
1-da
y3-
day
Mic
roso
ft(%
)In
dust
ry(%
)M
icro
soft
(%)
Indus
try
(%)
181/
20/9
5!
2.24
!0.
29!
1.98
!1.
13Sp
ork
inas
ks
DO
Jan
dM
icro
soft
toex
plai
nw
hy
chan
ges
shou
ldnot
bem
ade
inco
nsen
tde
cree
.Pro
191/
23/9
51.
92!
0.78
!1.
99!
0.87
Spork
inat
logg
erhea
dsw
ith
Bin
gam
anan
dM
Sat
torn
ey;A
ppl
eap
pea
lsca
seto
Supre
me
Court
.Son
yan
dM
Sag
ree
on
par
tner
ship
,free
upg
rades
onW
ord
6.0.
Pro
202/
1/95
!0.
730.
58!
1.80
0.73
DO
Jhas
issu
edsu
bpoe
nas
inM
icro
soft/I
ntuit
inve
stig
atio
n.Pro
212/
15/9
5!
2.63
0.01
!2.
850.
04Sp
ork
inre
ject
sgo
vern
men
t's
cons
ent
asto
ole
nien
t.Pro
222/
23/9
5!
0.59
!0.
340.
66!
1.41
App
leal
lege
sM
icro
soft
thre
aten
edit.
Pro
232/
24/9
5!
0.94
!0.
650.
43!
1.15
Story
aboutan
titr
ust
chie
fBin
gam
anca
lling
Gat
esSa
turd
aynig
ht.
?
243/
8/95
4.44
0.52
6.03
1.19
DO
Jan
dM
icro
soft
ask
feder
alap
pea
lsco
urt
tore
vers
eSpo
rkin's
dec
isio
n.
Ant
i
254/
25/9
52.
95!
0.07
4.67
1.40
Thre
e-ju
dge
appea
lspa
nel
voic
edco
nce
rns
that
Spork
inm
ayha
veex
ceed
edhis
auth
ority
.A
nti
264/
28/9
53.
640.
092.
470.
08D
OJ"le
ssu
itag
ainst
Intu
itdea
l.M
Sis
pre
ssin
gto
com
ple
teth
ede
al.
?
275/
1/95
0.52
!0.
570.
70!
0.50
DO
Junv
eils
key
docu
men
tsin
unusu
alm
ove
.?
285/
22/9
50.
99!
0.18
1.95
0.01
Mic
roso
ftsa
idit
isen
din
gits
pla
nto
acqu
ire
Intu
it.
?29
6/9/
951.
941.
11!
0.11
2.61
DO
Jra
cing
to"nish
anin
vest
igat
ion
aim
edat
MS
plan
tobu
ndl
eac
cess
toits
new
softw
are.
?
306/
12/9
5!
2.17
0.06
!1.
621.
14D
OJ
lookin
gat
Mic
roso
ftst
ipula
tion
agai
nst
pate
ntin
frin
ge-
men
tsu
its.
Pro
316/
19/9
51.
750.
735.
241.
79A
ppea
lsco
urt
rein
stat
esco
nsen
tde
cree
Spor
kin
reje
cted
.A
nti
326/
22/9
5!
0.26
!0.
35!
1.74
0.26
DO
Jissu
essu
bpo
enas
topu
blis
hers
,ot
hers
.Pro
337/
14/9
53.
650.
188.
351.
22D
OJ
unv
eils
antitr
ust
argu
men
ts.N
etin
com
ein
crea
se,(
7/18
).?
347/
24/9
50.
601.
51!
2.27
2.26
DO
Jw
ithd
raw
sbro
adsu
bpo
enas
.A
nti
357/
31/9
5!
2.24
!0.
64!
6.24
!0.
79D
OJ
exte
nds
inve
stig
atio
nto
new
area
-bun
dlin
gofso
ftw
are.
Pro
336 G. Bittlingmayer, T.W. Hazlett / Journal of Financial Economics 55 (2000) 329}359
368/
9/95
3.11
0.96
0.63
0.83
DO
Jsa
ysit
will
take
no
action
prio
rto
Win
dow
s95
ship
men
t.A
nti
378/
22/9
54.
750.
831.
030.
80Ju
dge
form
aliz
esco
nsen
t.A
nti
382/
6/96
!1.
78!
0.17
1.20
!0.
51D
OJ
takes
inte
rest
inM
icro
soft
acquisitio
nofV
erm
eer
Tec
hnol
ogi
es,in
tern
etto
ols.
Pro
399/
20/9
6!
0.58
0.00
!0.
100.
93D
OJ
has
laun
ched
anin
vest
igat
ion
ofM
icro
soft's
Inte
rnet
softw
are
bus
ines
s.Pro
402/
12/9
7!
1.01
!0.
32!
2.40
!2.
59St
ate
ofTex
asha
sla
unc
hed
anan
titr
ust
inve
stig
atio
nof
Mic
roso
ft.
Pro
415/
20/9
72.
000.
182.
320.
44D
OJ
reques
tsad
ditio
nal
info
rmat
ion
about
Mic
roso
ft's
pla
nned
acqui
sition
ofW
ebTV
Net
work
s.M
Shas
sign
edup
mor
eth
ana
doze
nbus
ines
s-in
form
atio
npro
vide
rs(5
/22)
.
Pro
426/
30/9
7!
0.87
0.77
!4.
23!
0.29
Atth
eurg
ing
ofco
mpet
itors
,th
ree
sena
tors
have
aske
dth
eFT
Cto
once
agai
nin
vest
igat
eM
icro
soft's
busines
spr
actice
s,in
clud
ing
com
plia
nce
with
the
1994
cons
ent
decr
ee.
Pro
437/
14/9
74.
330.
684.
082.
08N
etsc
ape
tryi
ng
toco
nvi
nce
gove
rnm
ent
that
Mic
roso
ftpla
nsto
use
new
Inte
rnet
softw
are
tolo
ckout
riva
ls.M
Sto
announ
ceth
atit
willlink
upw
ith
Disney
and
War
ner
Bro
s.(7
/15)
;N
etin
com
ean
noun
cem
ent
(7/1
8).
?
448/
4/97
0.10
0.21
0.87
2.82
Mic
roso
ftcl
ose
dits
purc
hase
ofW
ebTV
afte
rD
OJ
ends
its
lengt
hy
revi
eww
ithout
takin
gac
tion.
Ant
i
458/
19/9
71.
44!
0.35
!0.
05!
1.63
DO
Jis
revi
ewin
gM
icro
soft's
min
ority
stak
ein
App
leas
wel
las
thre
eunr
elat
edin
vest
men
ts.
Pro
4610
/7/9
7!
0.45
!0.
901.
17!
0.06
Four
mor
est
ates
join
inve
stig
atio
nof
Mic
roso
ft.
Pro
4710
/17/
970.
70!
0.68
!0.
74!
2.21
Mic
roso
ftun
der
inve
stig
atio
nby
Eur
opea
no$
cial
s.Pro
4810
/21/
971.
96!
0.52
!1.
35!
0.40
U.S
.sues
Mic
roso
ftove
rPC
bro
wse
r-m
ove
tore
strict
bun
dlin
gw
ith
Win
dow
s95
hits
key
mar
ket
stra
tegy
.M
Spro"t
tops
anal
ysts'fo
reca
sts.
?
4911
/3/9
7!
0.49
!0.
58!
1.66
0.13
Sena
tein
tern
etpa
nelto
prob
eM
icro
soft's
pow
er.
Pro
5011
/10/
97!
0.33
!0.
181.
40!
1.01
Tex
assu
esM
icro
soft,al
legi
nglic
ense
sim
pede
stat
e's
pro
be.
Pro
5112
/8/9
71.
901.
090.
48!
0.74
Mic
roso
ft,J
ustice
Dep
artm
ent
face
o!,but
hear
ing
ends
withou
tdec
isio
n.?
5212
/12/
97!
1.29
!2.
75!
2.65
!5.
46M
icro
soft
isdea
lta
blo
won
inte
rnet
pla
ns-
judge
ord
ers"rm
tost
opbu
ndlin
gso
ftw
are
with
Win
dow
sope
rating
syst
ems.
Pro
5312
/18/
97!
1.94
!0.
93!
4.90
0.93
U.S
.,M
icro
soft
clas
hon
court
-ord
erco
mplia
nce
.Pro
5412
/30/
970.
30!
0.55
1.06
!0.
24M
icro
soft
isag
ain
assa
iled
byD
OJ,
whic
hsa
ys"rm
thw
arts
court
ord
er.
Pro
G. Bittlingmayer, T.W. Hazlett / Journal of Financial Economics 55 (2000) 329}359 337
Table 2Means, medians and standard deviations of residual returns for Microsoft and the rest of thecomputer industry at announcements of enforcement actions in the=all Street Journal.
1-day 3-day
Microsoft (%) Industry (%) Microsoft (%) Industry (%)
Pro-enforcement Mean !0.45 !0.26 !1.16 !0.53n"29 Median !0.58 !0.29 !1.54 !0.24
SD 1.60 0.73 2.73 1.48
Ambiguous Mean 0.85 !0.13 0.33 !0.26n"17 Median 0.52 !0.18 0.18 !0.40
SD 2.57 0.76 2.99 1.55
Anti-enforcement Mean 2.65 0.59 2.34 1.15n"8 Median 3.03 0.63 1.77 1.29
SD 1.70 0.53 2.82 1.40
All event Mean 0.42 !0.09 !0.16 !0.18n"54 Median 0.29 !0.05 0.06 !0.09
SD 2.23 0.75 2.99 1.57
All days Mean 0.00 0.00 0.00 0.00n"1771 SD 1.75 0.87 2.92 1.72
event is classi"ed as pro-enforcement. In contrast, the April 15, 1991 articleannouncing that the FTC was broadening its investigation (clearly pro-enforce-ment) was followed by an April 17 report that Microsoft earnings increased65%. This event is placed in the ambiguous category.
Table 1 lists 29 pro-enforcement events. These stories report antitrust invest-igations, inquiries proceeding or expanding, pressure on antitrust authorities forstricter enforcement, signals implying a collapse of the negotiated consentdecree, and state-level antitrust investigations of Microsoft. Table 2 reports thesummary statistics for the event dates and all days in the sample. Averageabnormal returns over the 29 pro-enforcement dates for Microsoft equal!0.45% on the day of publication and !1.16% over three days(t!1, t, t#1). The computer industry portfolio also declines: average abnor-mal returns are !0.26% over the one-day window and !0.53% over thethree-day window. On average, both Microsoft and other computer "rm stocksdecline in value when antitrust measures directed against Microsoft are an-nounced in the press.
Table 1 reports eight anti-enforcement events (i.e., positive events for Micro-soft shareholders). Four involve the joint struggle of Microsoft and the DOJwith Judge Sporkin over the proposed consent decree. Table 2 reports Micro-soft's mean abnormal return on those eight news dates as 2.65%. The three-day
338 G. Bittlingmayer, T.W. Hazlett / Journal of Financial Economics 55 (2000) 329}359
return is 2.34%. Average abnormal returns for the rest of the industry are0.59% and 1.15% over one and three days, respectively. Microsoft and othercomputer "rm stocks generally increase in value when antitrust enforcementdeclines.
Table 2 shows that the 17 ambiguous events are accompanied by a positiveone-day return for Microsoft (0.85%) and a much higher standard devi-ation. Microsoft's median one-day return (0.53%) is substantially closerto zero. Over three days, Microsoft's mean and median returns are even closerto zero (0.33% and 0.18%). Industry returns are negative, especially overthree days.
Note that Microsoft's returns provide a barometer for our choice of eventdates. Three facts support our a priori classi"cation and suggest that theassociated publication dates mark events of importance for Microsoft share-holders. First, Microsoft's one-day return is higher on the eight anti-enforce-ment dates than on the 29 pro-enforcement dates (t"4.74). Second, Microsoft'sone-day return is statistically indistinguishable from zero on the ambiguousdates (t"1.36) and much more volatile. Third, the standard deviation ofMicrosoft's returns is higher on the 54 event dates than for the sample asa whole.
To construct a portfolio of non-Microsoft computer sector stocks, we includeall "rms in Hoover's Guide to Computer Companies 1995 that are publicly tradedin the U.S. and operate primarily in the computer industry. We allocatethe resulting 159 "rms to one of nine industry segments based on their de-scription in Hoover's. We exclude two product lines } electronic contentand mainframes } because each has only two "rms. Not all "rms are publiclylisted throughout the sample period. The Appendix lists the nine industrysegments, their representative products, and the names of the "rms in each.We calculate the simple mean return in each segment and then calculate thesimple mean return across segments. We also utilize an index with equalweighting for each "rm in the industry (save Microsoft), and the results forthat model are reported as well. They do not vary substantially from theresults based on equally weighted segments, which are the results we discussbelow. Either method avoids having a handful of large-capitalization "rmssuch as Intel, Hewlett-Packard, and Compaq dominate a value-weightedportfolio.
5. Empirical results
This paper estimates the marginal e!ect of Microsoft antitrust enforcementnews on equity values in each of nine segments of the computer industry. Weemploy the multivariate regression model used extensively in stock marketstudies of regulation. Binder (1985) discusses the use of the multivariate
G. Bittlingmayer, T.W. Hazlett / Journal of Financial Economics 55 (2000) 329}359 339
1Also see Salinger (1992). Papers using this method include Schipper and Thompson (1983), Rose(1985), Smith et al. (1986), Binder (1988), Kara"ath and Glascock (1989), Cornett and Tehranian(1990), and Alexander and Spivey (1994).
regression model in event studies and credits the method to Gibbons (1980,Appendix D).1
De"ne Dtas a [0, 1] variable that equals unity if a given event occurs on day t;
let Rit
be the return for an index, segment, or "rm i on day t, and let Mtbe the
return on the market on day t. We estimate the following regression:
Rit"a
i#b
iM
t#
5+
k/~5
ci,t`k
Di,k#e
it. (1)
The term biis the coe$cient of the market model. The c
i,t`kcoe$cients yield
estimates of the daily leading, simultaneous, or lagging e!ects of an event, andthe summed values of the estimated c
i,t`k's over k"[!5, 5] yield the cumulat-
ive e!ect. The discussion of results will focus on the three-day window,k"[!1, 1].
5.1. Antitrust events signed a priori
Table 3 presents estimates from Eq. (1) for eight anti-enforcement and 28pro-enforcement events. (While 29 news stories are assigned pro-enforcementstatus, the last event date, December 30, 1997, occurs too late to allow estima-tion of lagging e!ects.) During the latter, Microsoft shares decline by an averageof 1.20% over three days (t"!2.06), while the rest of the industry declines0.71% per event (t"!2.45). Manufacturers of components and networkproducts, as well as distributors, experience statistically signi"cant declines of1% or more. The other sectors are all negative over three days. The top panel ofFig. 1 shows the comparatively strong negative e!ects for Microsoft over these28 events and substantially weaker, though still negative, e!ects for the com-puter industry portfolio.
Also seen in Table 3 are mirrored results during the eight anti-enforcementevents. Microsoft's stock increases by 2.36% on average over the three-daywindows (t"2.19), while the equally weighted nine-segment portfolio experien-ces three-day returns of 1.18% (t"2.21). Firms making components, networkproducts, PC software, and semi-conductors witnessed three-day returns inexcess of 1%. No industry segments exhibit negative three-day returns, while thefew negative returns over one or 11-day windows are statistically insigni"cant.The bottom panel of Fig. 1 graphs the cumulative returns indicating that theanti-enforcement events have strong positive e!ects on Microsoft shares andpositive, although smaller, e!ects on other computer stocks.
340 G. Bittlingmayer, T.W. Hazlett / Journal of Financial Economics 55 (2000) 329}359
Table 3Computer sector returns during Microsoft antitrust news windows
Coe$cients and t-statistics for regressions of "rm or industry segment returns on lagging andleading dummy variables that are coded as shown in Table 1. Coe$cients for the constant and themarket return are calculated but not reported. One pro-enforcement event is omitted due toinsu$cient data for estimation of the lagging terms.
Firm or Segment 8 Anti-enforcement events 28 Pro-enforcement events
The dollar values implied by these estimates are of interest. The 159 computer"rms used in this study have an aggregate market capitalization of $141 billionin January 1991 and $754 billion in December 1997. (This increase re#ects thegrowth in value of the "rms as well as an increase in the number of "rms.) The
G. Bittlingmayer, T.W. Hazlett / Journal of Financial Economics 55 (2000) 329}359 341
Fig. 1. Cumulative residual returns for Microsoft, equally weighted average of all computer "rms inthe sample (excluding microsoft), and equally weighted average of all segments (excluding Micro-soft). The upper panel shows the cumulative residual returns from day !5 through day 5 around 28&pro-enforcement' antitrust event dates involving Microsoft. The sold line shows the cumulativereturns for Microsoft, the long dashes for an equally weighted portfolio of all other "rms in oursample, and the short dashes for a portfolio of equally weighted segments (themselves equallyweighted averages). The lower panel shows similar cumulative returns for eight &anti-enforcement'antitrust event dates involving Microsoft. The list of dates appears in Table 1. (a) Pro-enforcementevents (n"28), (b) Anti-enforcement events (n"8).
mean three-day return per event of !0.59% for the average of all computer"rms implies a loss of several hundred million to several billion dollars perevent. In fact, this group experiences mean market-adjusted declines in value of$1.2 billion on the three days surrounding each of 29 pro-enforcement events.The cumulative decline is $35 billion. Conversely, the same group of "rms hasa mean increase in value of $8.8 billion on the three days surrounding each of the
342 G. Bittlingmayer, T.W. Hazlett / Journal of Financial Economics 55 (2000) 329}359
eight anti-enforcement events. The cumulative increase is $70 billion. Clearly, inview of the large standard errors attached to these estimates, they should betaken with a grain of salt. The fact that the cumulative increase is greater thanthe cumulative decline might only re#ect the possibility that good news forMicrosoft came as more of a surprise than bad news. Still, these dollar estimateso!er some idea of the importance for the computer industry as a whole ofpolicies directed against Microsoft.
5.2. Antitrust events dexned by Microsoft stock price reactions
As an alternative to a classi"cation system that relies on the investigator'sjudgment, we allow Microsoft stock price movements themselves to identifyannouncements as good or bad news for Microsoft. A positive event forMicrosoft, which we classify as anti-enforcement, occurs on one of the 54 datesin Table 1 when either a) Microsoft's cumulative residual return exceeds2% over the "ve-day interval [!4, 0] or b) Microsoft's daily residual returnexceeds 2% on any day of the three-day interval [!2, 0]. The longer,"ve-day criterion based on cumulative returns is intended to capture slowreleases of information, and the shorter, three-day criterion is intended tocapture sudden releases of information. We de"ne a negative (pro-enforcement)event symmetrically. For the 1771 days in the sample Microsoft's cumulative"ve-day residual excess returns exceed two percent 26.7% of the time and fallbelow minus two percent 28.9% of the time. Microsoft's daily excess returnsexceed 2 percent with probability 16.6%, and fall below !2% with probability11.7%.
Twenty-four of the 54 events are accompanied by increases in Microsoft stockas de"ned here; of these, six are among the eight de"ned a priori as anti-enforcement events in Table 1. We eliminate the July 14 and July 24, 1995 eventbecause they coincide with a sharp increase and then collapse of Microsoft'sstock price that was linked with shifting earnings expectations in connectionwith the July 18 publication of earnings news. The two event dates are separatedby only "ve trading days. Twenty-three of the 54 events in Table 2 are #agged bythreshold declines in Microsoft stock; of these, 15 are among the 29 de"ned aspro-enforcement events in Table 1.
The results in Table 4 show that negative movements of Microsoft stock at thetime of enforcement actions coincide with negative returns for rest of theindustry. Not surprisingly, Microsoft stock declines by 2.18% over three days(t"!3.49) and by 4.52% over 11 days (t"!3.89). However, the industry asa whole declines by 0.62% over three days (t"!2.00) and by 1.05% over 11days (t"!1.83). Fig. 2 shows this pattern and that both Microsoft and othercomputer stocks continue to decline after the event dates.
On news accompanied by positive Microsoft stock movements, the rest of theindustry prospers. Microsoft's stock price increases an average of 1.49% during
G. Bittlingmayer, T.W. Hazlett / Journal of Financial Economics 55 (2000) 329}359 343
Table 4Computer sector returns during microsoft antitrust news windows accompanied by increases ordecreases in Microsoft stock price
Coe$cients and t-statistics for regressions of "rm or industry segment returns on lagging andleading dummy variables equal to unity if (a) an antitrust event listed in Table 1 occurred and (b)Microsoft&s cumulative residual return increased by more than 2% over the "ve days [!4, 0] orincreased by 2% on any day [!2, 0] (to identify implied anti-enforcement events), or Microsoft'scumulative residual return declined by more than 2% over the "ve days [!4, 0] or declined by 2%on any day [!2, 0] (to identify implied pro-enforcement events).
Firm or Antitrust news & Antitrust news &Industry Microsoft increases (n"24) Microsoft decreases (n"23)
the three-day windows. Note that the cumulative return over the longer 11-dayinterval is only 0.76%. Fig. 2 shows that for this cluster of events, Microsoftstock declines after day zero, largely reversing earlier gains. (The decline over
344 G. Bittlingmayer, T.W. Hazlett / Journal of Financial Economics 55 (2000) 329}359
Fig. 2. Cumulative residual returns for Microsoft, equally weighted average of all computer "rms inthe sample (excluding Microsoft), and equally weighted average of all segments (excluding Micro-soft). The upper panel shows the cumulative residual returns from day !5 through day 5 around 23antitrust event dates involving Microsoft and marked by threshold declines in Microsoft's residualreturns. The sold line shows the cumulative returns for Microsoft, the long dashes for an equallyweighted portfolio of all other "rms in our sample, and the short dashes for a portfolio of equallyweighted segments (themselves equally weighted averages). The lower panel shows similar cumulat-ive returns for 24 antitrust event dates involving Microsoft and marked by threshold increases inMicrosoft's residual returns. The list of dates appears in Table 1. Enforcement actions accompaniedby Microsoft stock price (a) declines (n"23), (b) increases (n"24).
days 2 through 5 is not statistically signi"cant.) The average of all computersegments (and the average of all "rms) experience returns that are statisticallyindistinguishable from zero around these &positive' events. As seen in Fig. 2,cumulative industry returns move up about 1% in the four days prior to theevent dates.
G. Bittlingmayer, T.W. Hazlett / Journal of Financial Economics 55 (2000) 329}359 345
5.3. A xshing expedition
The results in Tables 3 and 4 o!er no support for the view that investorsexpect antitrust enforcement to improve e$ciency. Events implying an antitrustvictory for Microsoft do not harm the industry, and actions hurting Microsoftdo not help. Yet perhaps we have overlooked "nancial market evidence thatsupports the pro-consumer view of antitrust.
To investigate this possibility, we reverse our method. Instead of testingabnormal returns around event dates to determine whether the pattern suggeststhat investors expect antitrust policy to enhance e$ciency, we cull the data forwindows displaying the pattern consistent with pro-consumer policy expecta-tions } speci"cally, a negative correlation between Microsoft and computersector returns. What sorts of events generate these results, and, speci"cally, dopolicy interventions play a role?
We identify one-day and three-day periods over which either Microsoft'sabnormal return declines by at least 1.5 standard deviations while the abnormalreturn of the computer portfolio increases by at least one standard deviation, orthe inverse occurs (Microsoft up, computer industry down). Table 5 shows therelevant dates, one- and three-day returns, and possible causative events for theMicrosoft stock price movement and/or the large opposing movement in othercomputer "rm stocks. The #agged dates occur in 19 clusters, numbered at theleft. The bold percentages pass the respective hurdles (one- or three-day).Typically, the related news stories deal with earnings news, product announce-ments, or other non-antitrust litigation (e.g., Apple's &look-and-feel' suit, March7, 1991).
Of the 54 antitrust news events in the 1991}97 period, only three coincide withsubstantial capital losses for Microsoft and capital gains for the non-Microsoftcomputer sector, or the reverse. The "rst is the April 15, 1991 announcementthat the FTC was broadening its investigation. However, the signs of theone-day returns (Microsoft positive, industry negative) are opposite from thoseimplied by the pro-e$ciency theory of antitrust, while the three-day returns forboth Microsoft and the industry are negative. The second date is February 5,1993, when the FTC vote on issuing a preliminary injunction split 2-2. While theone-day price movements are consistent with the pro-e$ciency view of antitrust(Microsoft up, computer sector down), the three-day returns for both Microsoftand the industry are strongly negative. The third and "nal event involves theJuly 31, 1995 story that the DOJ was planning to examine Microsoft's bundlingpractices. This was preceded by a decline in Microsoft's shares and positivereturns for the rest of the industry on the three days centered on July 27.However, other events } such as the July 26 announcement that Microsoft hadsettled a California consumer protection lawsuit and a July 27 announcementthat Oracle, Apple, and IBM were teaming up against Microsoft } are moreclosely linked in time. Overall, the small number of instances in which Microsoft
346 G. Bittlingmayer, T.W. Hazlett / Journal of Financial Economics 55 (2000) 329}359
and computer industry shares react in substantial opposition around the an-nouncement of antitrust enforcement news concerning Microsoft, as well asexamination of the few episodes showing the appropriate correlation, lend littlesupport to the pro-e$ciency view of antitrust.
5.4. Antitrust enforcement and the correlation between Microsoft and computerindustry returns
A "nal test is again generated by the negative correlation between Microsoft'sresidual return and the residual return of the computer sector that is implied bya pro-e$ciency antitrust regime. Successes and failures of such a regime shouldlower the typical positive correlation between Microsoft's residual returns andthe residual returns of the rest of the industry. In fact, the correlation increaseson the 54 event dates (r"0.3258) compared to the 1717 non-event days(r"0.2578). An analysis of the algebraic signs of residual returns yields a similarresult. Victories for e!ective antitrust should result in more negative Microsoftresidual returns accompanied by positive rest-of-industry residual returns. Infact, the frequency of that combination declines on the 54 event dates comparedto the rest of the sample, from 23.6% to 14.8%. That decline is statisticallysigni"cant at the 5% level (t"1.75). Similarly, setbacks for e!ective antitrustshould result in more positive Microsoft residual returns accompanied bynegative rest-of-industry returns. However, that pattern occurs at roughly thesame rate on the 54 event dates, 18.9% rather than 20.4% of the time.
5.5. Summary of empirical results
We have been unable to "nd evidence that antitrust initiatives against Micro-soft created expected gains for the rest of the computer industry. Policymeasures a!ect Microsoft share prices, which decline in response to pro-enforcement news and increase with the release of anti-enforcement news. Onaverage, however, the pro-enforcement (anti-enforcement) actions are accom-panied by declines (increases) in the rest of the computer industry. Even a delib-erate search for e!ects that would yield opposing price movements fails to "ndevidence that antitrust e!orts have helped the industry. Indeed, investors appearto believe that antitrust enforcement increases the link between the fortunes ofMicrosoft and other computer "rms.
6. The political economy of U.S. vs. Microsoft
If antitrust enforcement against Microsoft produces zero or negative returnsfor allegedly victimized computer "rms, why does it occur? We o!er threepossible explanations warranting further study.
G. Bittlingmayer, T.W. Hazlett / Journal of Financial Economics 55 (2000) 329}359 347
Tab
le5
One-
and
thre
e-da
ype
riod
sw
ith
larg
eneg
ativ
e(p
ositive
)M
icro
soft
retu
rns
and
larg
epo
sitive
(neg
ativ
e)re
turn
sfo
rth
ere
stofth
ein
dus
try
Theta
ble
show
sex
cess
retu
rnsfo
rday
tor
theco
mpoun
ded
retu
rnsfo
rda
yst!
1th
roug
ht#
1on
whic
hM
icro
soft's
exce
ssre
turn
exce
eded
1.5
stan
dar
dde
viat
ionsan
dth
ere
turn
forth
ein
dust
ryfe
llbel
ow1.
0st
anda
rdde
viat
ions,
orth
ere
vers
e.=
allS
tree
tJo
urna
lnew
sst
orie
sat
ornea
rea
chcl
ust
erof
dat
esei
ther
invo
lvean
titr
usto
rre
late
dle
gala
ctio
norot
her
wiseco
uld
expl
ain
thein
vers
em
ove
men
tofM
icro
soft
and
there
stof
thein
dust
ry.B
old
face
indi
cate
sre
turn
sth
atpas
sth
ehur
dle
for
inve
rse
move
men
ts.R
eturn
scl
ust
ered
bygr
oups
ofev
ent
dat
es.
Dat
esO
ne-
day
retu
rn(t)
Thre
e-day
retu
rn(t!
1,t,
t#1)
Sto
ryan
dD
ate
Mic
roso
ft(%
)In
dust
ry(%
)M
icro
soft
(%)
Indus
try
(%)
1.3/
6/91
!3.
71!
0.68
25.
591.
99A
fede
ralju
dge
boos
tsA
pple's
chan
ces
ofpr
o"ting
from
GU
Iby
striki
ng
dow
nke
yde
fense
argu
men
tsin
its
case
agai
nst
MS,
H-P
,3/7
3/7/
912
3.11
1.36
!7.
841.
272.
4/15
/91
5.69
!1.
60!
2.61
!1.
57FTC
bro
aden
ing
inve
stig
atio
n,4/
15;M
icro
soft
repor
tsin
crea
sed
earn
ings
,4/
173.
1/17
/92
!1.
230.
802
5.46
1.83
Net
inco
me"gu
res,
1/20
4.2/
5/93
5.08
22.
18!
3.37
!4.
65FT
Csp
lits
2!2
agai
nst
inju
nct
ion,a
dmin
istr
ativ
eac
tion
still
poss
ible
,2/8
5.2/
23/9
33.
562
1.71
3.06
!4.
04Pat
ent
o$ce
reje
cts
MS
righ
tsto`W
indo
wsa
trad
emar
k,2/
252/
24/9
33.
282
0.98
6.86
22.
056.
2/8/
942
2.65
1.09
!3.
442.
49N
ost
ories
7.4/
18/9
4!
0.99
!0.
446.
312
1.75
Ear
nings
repor
t,4/
194/
19/9
46.
542
1.60
7.73
24.
30N
ewO
S`C
hica
goa,
4/25
4/20
/94
2.12
!2.
318.
672
2.68
8.4/
18/9
5!
0.42
!0.
546.
442
2.33
Ear
nings
repor
t,4/
14;re
port
son
pen
din
gIn
tuit
dea
l,4/
219.
7/27
/95
!0.
621.
272
5.04
2.10
MS
sett
led
Cal
if.la
wsu
it,7
/26;
Ora
cle,
App
le,a
nd
IBM
team
up
agai
nst
MS,
7/27
;DO
Jto
look
atbund
ling
ofw
ebso
ftw
are,
7/31
348 G. Bittlingmayer, T.W. Hazlett / Journal of Financial Economics 55 (2000) 329}359
10.
12/6
/95
4.98
21.
272.
76!
2.95
MS
anno
unc
esN
BC
pac
t,12
/7;M
San
noun
ces
new#urr
yof
new
inte
rnet
pro
duct
s,12
/711
.12
/20/
952
3.43
1.04
1.42
3.81
No
stories
12.
1/16
/96
3.05
21.
24!
1.77
!4.
28M
icro
soft
tobu
yV
emee
r(w
ebso
ftw
are
co.),
1/16
13.
9/25
/96
!1.
251.
382
4.97
2.24
Bin
gam
anre
cuse
she
rsel
ffrom
MS
case
,9/
2314
.12
/17/
963.
292
1.73
2.82
!2.
21Sto
ryon
MS
issu
eof
pref
erre
dst
ock,po
ssib
lera
idon
com
pet
itors
,12/
1612
/18/
962.
070.
545.
612
3.49
15.
4/4/
972
2.77
1.32
1.02
2.24
MS,In
tel,
and
Com
paq
initia
tive
on
dig
ital
TV
stan
dar
ds,
4/4;
MS
tobuy
Web
TV
,4/
716
.4/
17/9
7!
0.05
0.68
8.68
22.
894/
18/9
78.
942
1.54
10.4
02
2.13
Ear
nings
repor
t,4/
184/
21/9
71.
39!
1.27
11.0
82
4.95
4/22
/97
0.57
!2.
226.
072
3.95
17.
5/2/
972
3.28
0.98
27.
523.
13N
ost
ories
5/5/
97!
3.99
0.86
28.
872.
505/
6/97
!1.
860.
642
5.63
3.17
Borlan
dsu
esM
S,5/
818
.7/
18/9
72
3.95
1.08
!5.
120.
87Ja
vaan
noun
cem
ent,
7/18
19.
7/22
/97
2.92
21.
23!
2.23
!1.
15M
Sto
acqu
ire
Pro
gres
sive
Net
wor
ks,7
/22;
MS
toin
crea
sesa
les,
mar
ket
ing
e!ort
s,7/
24
G. Bittlingmayer, T.W. Hazlett / Journal of Financial Economics 55 (2000) 329}359 349
2 &Culture clash', National Journal, May 16, 1998, p. 1102; &Book Bork, browser Bork', SlateMagazine, Dec. 10, 1998; &Browser Bork replies', Slate Magazine, Dec. 29, 1998; &What antitrust is allabout', New York Times, May 4, 1998, op-ed page; &The most misunderstood antitrust case', WallStreet Journal, May 22, 1998, op-ed page; &Full text of Bork's &White Paper' on DOJ vs. MS, ZDNN,July 29, 1998, online at www.zdnet.com.
6.1. Private use of antitrust
Clearly, some of Microsoft's competitors have an economic interest in govern-ment action. This possibility is recognized in the &raising rivals' costs' literature(Krattenmaker and Salop, 1986) and in work that emphasizes the use ofantitrust to constrain competitors' strategies (Baumol and Ordover, 1985;Baron, 1998). Attacks on Microsoft can be pro"table even for "rms that bene"tfrom Microsoft products generally. Netscape's browser competes with Micro-soft's Internet Explorer, for instance, but is complementary with (&dependent on'in the antitrust allegation) Microsoft's operating systems MS-DOS, Windows95, Windows 98, and NT.
In fact, managers of several computer "rms have actively promoted the caseagainst Microsoft. Netscape cooperated with the 1998 Department of Justicecase, and retained former Judge Robert Bork to write a &white paper' for publicdissemination and to argue the case in television debates and op-ed pagearticles.2 Sun Microsystems and Oracle executives have also endorsed the caseagainst Microsoft. Sun CEO Scott McNealy comments, &I think the governmentis doing all the right things. Government has to come in and discipline (Micro-soft) until the rest of the world catches up' (&Microsoft antitrust case splitsValley', San Jose Mercury News, October 12, 1998). Similar sentiments have beenexpressed by Oracle CEO Larry Ellison (&In parts of Silicon Valley, mutedapplause', New York Times online edition, May 19, 1998). Novell prevailed uponOrrin Hatch (R-UT), its home state senator and powerful head of the JudiciaryCommittee, to hold hearings on the problem of monopoly in the computerindustry, and to pressure the Department of Justice to take sterner enforcementmeasures against Microsoft. A trade association has been established to pres-sure policymakers to increase regulatory scrutiny of Microsoft. The Project toPromote Competition and Innovation in the Digital Age (ProComp) is &fundedby Microsoft's competitors including Sun Microsystems and Netscape'. (&Micro-soft, foes square o! over Windows 98', San Jose Mercury News, May 6, 1998).
We examine the returns of individual "rms and "nd no systematic evidencethat the likely bene"ciaries (e.g., Netscape, Sun, Novell, Apple) realize higherreturns when antitrust enforcement measures are taken against Microsoft. Inspeci"c instances, however, antitrust enforcement actions did seem to bene"tNetscape shareholders. For example, the December 12, 1997 court order thatkept Microsoft from bundling its software was accompanied by negative returns
350 G. Bittlingmayer, T.W. Hazlett / Journal of Financial Economics 55 (2000) 329}359
for both Microsoft and the industry as a whole but a 6.2% increase in Netscape'sprice. (This gain was quickly reversed.) If the expected bene"ts to competing"rms are in fact scant, managerial support for antitrust in this instance couldre#ect a principal-agent problem. The specter of Microsoft could providea rationalization for managerial shortcomings. As shown, such distractions donot come cheaply. Companies producing highly complementary goods toMicrosoft stand to incur the largest capital losses from ine$cient antitrustinterventions. Consistent with shareholder interests revealed in the "nancialmarket data examined above, they have opposed the case. This anti-antitrustgroup includes the major computer makers (Compaq, Dell, Hewlett-Packard),the major chip maker (Intel), and the major computer retailers (CompUSA andVanstar). In fact, the CEO's of 26 major computer industry "rms signed a lettersent to the Department of Justice in May 1998 requesting that the governmentrefrain from "ling additional antitrust charges against Microsoft (&Microsoftseeks allies to halt antitrust move', Wall Street Journal Interactive Edition, May 1,1998; &PC Makers, despite Microsoft suit, aren't clamoring to control screen',Wall Street Journal Interactive Edition, May 19, 1998).
The behavior of individual states also sheds light on the political support foraction against Microsoft. The May 1998 suit "led by the DOJ was accompaniedby a suit "led by 20 states. As has been widely observed in the press, these statesappear to have been the subject of intense lobbying pressure from locally basedcomputer companies. California, home to important Microsoft competitorssuch as Sun, Oracle, and Netscape, "led; Texas, home to important Microsoftcomplement suppliers such as Compaq, Dell, and CompUSA, did not, despitethe fact that the Texas attorney general had originally begun the states' invest-igation of Microsoft (&Microsuits', Slate Magazine online, May 22, 1998; &Politicsplay a role in states' status in antitrust action against Microsoft', Wall StreetJournal Interactive Edition, May 28, 1998).
6.2. Bureaucratic self-interest
Government agents may themselves gain from legal action. Top-level anti-trust o$cials typically enjoy short tenure and receive much of their compensa-tion in human capital (Wilson, 1980). At the state level, attorneys general oftenuse high-visibility litigation to enhance their political stature. The Microsoftcase is &the type of case attorneys general dream about, regardless of how deeplyit a!ects } or fails to directly a!ect } their states' consumers or businesses2 It'salso true that a number of the AGs (the letters are sometimes said to stand for&aspiring governors') involved in the suit are seeking higher o$ce' (&Politics playa role in states' status in antitrust action against Microsoft', Wall Street JournalInteractive Edition, May 28, 1998). In the case of Microsoft, antitrust o$cialshave also received substantial favorable publicity. Tellingly, rival agencies havefought over federal jurisdiction, and the Federal Trade Commission initiated
G. Bittlingmayer, T.W. Hazlett / Journal of Financial Economics 55 (2000) 329}359 351
The rent-seeking literature points out that policy makers exploit the competi-tion between interest groups for government favors. Policy makers may alsoextract rents from private parties. In this framework, political o$ceholders arenot entirely passive. They achieve some degree of market power and exploit thisincumbency to demand payments from various economic actors with rents atrisk (McChesney, 1997). Transfers can take the form of campaign contributions,non-monetary political support, or public cooperation on a policy issue (includ-ing a consent decree in an antitrust case).
According to some accounts, the government took action against Microsoftat least in part because it lacked a &Washington presence'. Compared to otherlarge "rms, Microsoft historically contributed only small sums to politicalcampaigns. It only recently opened a Washington o$ce to represent its interests.Policy makers might have been attempting to &correct' this &underinvestment' inpolitical goodwill with heightened antitrust scrutiny (&Culture clash', NationalJournal, May 16, 1998, p. 1102).
7. Conclusion
In what The Economist identi"es as &the biggest antitrust case in a generation'(&At war with Microsoft', The Economist, May 23}29, 1998, online version),a large number of "rms have products that are tied to the success of the allegedlymonopolized product, desktop operating systems. These "rms will prosper ifactions are taken } by Microsoft or the DOJ } to make operating systems costless, function better, or provide a more convenient platform for popular prod-ucts. Policy actions that are expected to e!ectively constrain Microsoft's marketpower should simultaneously increase economic e$ciency and improve pro"t-ability for "rms throughout the sector.
This study turns to the stock market evidence. Have repeated antitrustinitiatives against Microsoft increased the expected earnings of (non-Microsoft)"rms in the computer industry? The answer is a decisive &no'. In fact, govern-ment action against Microsoft appears to in#ict capital losses on the computersector as a whole. Retreats in antitrust enforcement o!er symmetric con"rma-tion: withdrawals from policy enforcement have been accompanied by positiveshareholder returns throughout the computer sector.
352 G. Bittlingmayer, T.W. Hazlett / Journal of Financial Economics 55 (2000) 329}359
These results deserve attention for three reasons. First, the burden of proofought to be on policy intervention. A case of this magnitude generates clear coststo the government (i.e., taxpayers) and to Microsoft. In fact, each enforcementaction lowered Microsoft's stock by 1.2%, roughly $3 billion at May 1998 shareprices. It should produce visible gains in the form of positive returns to the largenumber of other "rms in the computer sector. That such gains fail to be in evidenceis the key "nding of this study. Moreover, rather than generating o!setting gains,each enforcement action decreased a broad index of other computer stocks by0.7%, equivalent to an additional loss of $5 billion dollars in May 1998.
The second reason these results demand attention is that they suggest thatantitrust policy has lowered returns to investments in the computer sector anddiscouraged capital in#ows. Policy risk increases the cost of capital across anentire industry when intervention occurs at a key point within that sector.Pindyck (1991) and Dixit and Pindyck (1994) model the investment-reducinge!ect of public policy uncertainty.
Third, stock market data produce a bottom-line evaluation of public policy.Rather than narrowly focusing on alleged abuses of market power, "nancialmarkets also consider likely policy outcomes. When actual markets are com-pared to hypothetical alternatives in traditional antitrust analysis, policymakers run the risk of &fanciful reasoning [wherein] much mischief and littleuseful purpose is served by pronouncing failure if no clearly superior feasiblealternative can be described and implemented with expected net gains' (Gilbertand Williamson, 1998, p. 7). Financial markets set asset prices by utilizing thebest available predictions as to what the full spectrum of antitrust enforcementactions will achieve. The verdicts rendered by &courts of investor opinion'constitute reliable economic evidence compared to the available alternatives.
Appendix A
Perm numbers and names of "rms used in this study, from Hoover's Guide toComputer Companies 1995. Firms are assigned to an industry segment on thebasis of the company description. Not all "rms are publicly listed during theentire 1991}1997 period.
Permd Name
Components Thin-"lm magnetic heads, printed circuit boards, disk drives, harddrives, SCSI hardware.77039 READ RITE CORPORATION57808 VISHAY INTERTECHNOLOGY INC68161 S C I SYSTEMS INC66384 WESTERN DIGITAL CORP
G. Bittlingmayer, T.W. Hazlett / Journal of Financial Economics 55 (2000) 329}359 353
Permd Name
89894 MAXTOR CORP69607 SEAGATE TECHNOLOGY88795 HUTCHINSON TECHNOLOGY INC48267 L S I LOGIC CORP65024 QUANTUM CORP11365 KOMAG INC11970 AMERICAN POWER CONVERSION CORP77183 SYQUEST TECHNOLOGY INC10353 ADAPTEC INC75831 DIGI INTERNATIONAL INC79251 BELL MICROPRODUCTS INC
Computers: PC's, workstations, servers.12490 INTERNATIONAL BUSINESS MACHS CORP27828 HEWLETT PACKARD CO43916 DIGITAL EQUIPMENT CORP14593 APPLE COMPUTER INC68347 COMPAQ COMPUTER CORP10078 SUN MICROSYSTEMS INC44792 INTERGRAPH CORP74617 TANDEM COMPUTERS INC85041 A S T RESEARCH INC11081 DELL COMPUTER CORP57592 DATA GENERAL CORP79973 GATEWAY 2000 INC11283 SEQUENT COMPUTER SYSTEMS INC
Corporate: Systems integration, information systems consulting.10890 UNISYS CORP40125 COMPUTER SCIENCES CORP80057 WANG LABORATORIES INC NEW77882 CONTROL DATA SYSTEMS INC73075 STRATUS COMPUTER INC76723 FUTURE NOW INC76846 TECHNOLOGY SOLUTIONS CO81639 PSINET INC
354 G. Bittlingmayer, T.W. Hazlett / Journal of Financial Economics 55 (2000) 329}359
Permd Name
29209 ARROW ELECTRONICS INC12260 MERISEL INC11964 EGGHEAD INC78180 MICRO WAREHOUSE INC10182 TECH DATA CORP61524 COMDISCO INC43781 INACOMP COMPUTER CTRS INC89763 COMPUCOM SYSTEMS INC11484 MICROAGE INC11475 INTELLIGENT ELECTRONICS INC79483 ELEK TEK INC76790 LIUSKI INTERNATIONAL INC75779 RANDOM ACCESS INC11233 TIGER DIRECT INC
Network: LAN hubs, modems, network "le servers.75625 CABLETRON SYSTEMS INC76754 BAY NETWORKS INC76076 CISCO SYSTEMS INC76129 3COM CORP51393 B B N CORP77049 U S ROBOTICS CORP76654 CHIPCOM CORP72486 STANDARD MICROSYSTEMS CORP79024 WALL DATA INC77793 STRATACOM INC76842 ARTISOFT INC86917 COMPUTER NETWORK TECHNOLOGY CP79152 AUSPEX SYSTEMS INC77451 XIRCOM INC77740 NETFRAME SYSTEMS INC76683 PROTEON INC79016 TRICORD SYSTEMS INC79940 ASANTE TECHNOLOGIES INC81106 SHIVA CORP
PC Software: Database programs, spreadsheets, fax software, entertainment,educational, word processing.10104 ORACLE CORP90609 NOVELL INC
G. Bittlingmayer, T.W. Hazlett / Journal of Financial Economics 55 (2000) 329}359 355
Permd Name
50156 LOTUS DEVELOPMENT CORP76792 SYBASE INC75902 BORLAND INTERNATIONAL INC85631 AUTODESK INC75510 ADOBE SYSTEMS INC75607 SYMANTEC CORP77837 BANYAN SYSTEMS INC78975 INTUIT INC12259 SIERRA ON LINE INC77331 FRAME TECHNOLOGY CORP77094 BRODERBUND SOFTWARE INC11917 PHOENIX TECHNOLOGY LTD10114 ACCLAIM ENTERTAINMENT INC91871 SOFTWARE PUBLISHING CORP80252 MAPINFO CORP76698 QUARTERDECK CORP77971 ELECTRONICS FOR IMAGING INC80337 MINNESOTA EDUCATIONAL COMP CORP81248 GENERAL MAGIC INC79991 MACROMEDIA INC
Peripherals: Scanners, printers, storage and retrieval systems, video monitors.27983 XEROX CORP58464 STORAGE TECHNOLOGY CORP60425 CONNER CORP40061 TEKTRONIX INC10147 E M C CORP MA76566 ZILOG INC45891 IOMEGA CORP77781 AMPEX CORP DEL76281 RADIUS INC78070 MEDIA VISION TECHNOLOGY INC20897 CAMBEX CORP80244 GLOBAL VILLAGE COMMUNICATION79413 PINNACLE MICRO INC
356 G. Bittlingmayer, T.W. Hazlett / Journal of Financial Economics 55 (2000) 329}359
Permd Name
22779 MOTOROLA INC61241 ADVANCED MICRO DEVICES INC14702 APPLIED MATERIALS INC53613 MICRON TECHNOLOGY INC79848 V L S I TECHNOLOGY75603 CIRRUS LOGIC INC12067 NOVELLUS SYSTEMS INC
Non-PC Software: Mainframe data processing software, CAD/CAM, electricaldesign software, executive information systems.38914 CERIDIAN CORP25778 COMPUTER ASSOCIATES INTL INC10419 CREATIVE TECHNOLOGIES CORP65074 STERLING SOFTWARE INC30884 LEGENT CORP11403 CADENCE DESIGN SYSTEMS INC77859 COMPUTERVISION CORP NEW10691 INFORMIX CORP78711 NEWBRIDGE NETWORKS CORP52813 MENTOR GRAPHICS CORP11103 SYSTEMS SOFTWARE ASSOC INC75912 PARAMETRIC TECHNOLOGY CORP79190 SANTA CRUZ OPERATION INC THE75828 ELECTRONIC ARTS INC77357 SYNOPSYS INC11976 B M C SOFTWARE INC76752 PROGRESS SOFTWARE INC11531 FILENET CORP10398 INTERLEAF INC10765 CONVEX COMPUTER CORP13777 AMERICAN SOFTWARE INC76620 PLATINUM TECHNOLOGY INC78083 PEOPLESOFT INC77030 HYPERION SOFTWARE CORP79081 DAVIDSON and ASSOCIATES INC79829 F T P SOFTWARE INC53621 LEARNING COMPANY INC86597 CHEYENNE SOFTWARE INC79385 FOURTH SHIFT CORP77663 NETWORK COMPUTING DEVICES INC
G. Bittlingmayer, T.W. Hazlett / Journal of Financial Economics 55 (2000) 329}359 357
Permd Name
78890 CENTURA SOFTWARE CORP79157 C D W COMPUTER CENTERS INC73294 SULCUS COMPUTER CORP75873 CAERE CORP79379 DATAWARE TECHNOLOGIES INC77581 STAC INC77538 LEARNING COMPANY80488 XCELLENET INC80262 PARCPLACE DIGITALK INC80272 SOFTDESK INC79577 CORNERSTONE IMAGING INC81169 NETCOM ON LINE COMM SVCS INC77976 MCAFEEASSOCIATES INC
References
Alexander, J., Spivey, M., 1994. CEBA of 1987 and the security returns and market risk of savingsand loan institutions: a note. Journal of Banking and Finance 18, 1205}1215.
Banerjee, A., Eckard, W., 1998. Are mega-mergers anticompetitive? Evidence from the "rst greatmerger wave. Rand Journal of Economics 29, 803}827.
Baron, D., 1998. Integrated market and nonmarket strategies in client and interest group politics.Paper presented at the Positive Political Theory of Business Strategy Conference, Haas Schoolof Business, University of California at Berkeley, October 15}17.
Baumol, W., Ordover, J., 1985. Use of antitrust to subvert competition. Journal of Law andEconomics 28, 247}266.
Binder, J., 1985. On the use of the multivariate regression model in event studies. Journal ofAccounting Research 23, 370}383.
Binder, J., 1988. The Sherman Antitrust Act and the railroad cartels. Journal of Law and Economics31, 443}468.
Bizjak, J., Coles, J., 1995. The e!ect of private antitrust litigation on the stock-market valuation ofthe "rm. American Economic Review 85, 436}461.
Bork, R., 1978. The Antitrust Paradox. Basic Books, New York, NY.Buchanan, J., Lee, D., 1992. Private interest group support for e$ciency enhancing antitrust policies.
Economic Inquiry 30, 218}224.Cornett, M., Tehranian, H., 1990. An examination of the impact of the Garn-St. Germain Deposi-
tory Institutions Act of 1982 on commercial banks and savings and loans. Journal of Finance 45,95}111.
Cutler, D., Summers, L., 1988. The costs of con#ict resolution and "nancial distress: evidence fromthe Texaco-Pennzoil litigation. Rand Journal of Economics 19, 157}172.
Dixit, A., Pindyck, R., 1994. Investment Under Uncertainty. Princeton University Press, Princeton,NJ.
Department of Justice (DOJ), 1998. U.S. Department of Justice Complaint in United States v.Microsoft (May 18), online at http://interactive.wsj.com/edition/resources/documents/mssuit0518-1.htm.
358 G. Bittlingmayer, T.W. Hazlett / Journal of Financial Economics 55 (2000) 329}359
Eckbo, B., 1983. Horizontal mergers, collusion, and stockholder wealth. Journal of FinancialEconomics 11, 241}273.
Gibbons, M., 1980. Econometric models for testing a class of "nancial models } an application of thenonlinear multivariate regression model. Ph.D. dissertation, University of Chicago, IL.
Gilbert, R., Williamson, O., 1998. Antitrust policy. In: Newman, P. (Ed.), New Palgrave Dictionaryof Economics and the Law, 3}9. Stockton Press, New York.
Gilligan, T., Marshall, W., Weingast, B., 1989. Regulation and the theory of legislative choice: theinterstate commerce act of 1887. Journal of Law and Economics 32, 35}62.
Hertzel, M., Smith, J., 1993. Industry e!ects of inter"rm lawsuits: evidence from Pennzoil v. Texaco.Journal of Law, Economics and Organization 9, 425}444.
Jarrell, G., Peltzman, S., 1985. The impact of product recalls on the wealth of sellers. Journal ofPolitical Economy 93, 512}536.
Kara"ath, I., Glascock, J., 1989. Intra-industry e!ects of a regulatory shift: capital market evidencefrom Penn Square. Financial Review 24, 123}134.
Krattenmaker, T., Salop, S., 1986. Anticompetitive exclusion: raising rivals' costs to achieve powerover price. Yale Law Journal 96, 209}293.
Liebowitz, S., Margolis, S., 1999. Winners, Losers and Microsoft: Competition and Antitrust in HighTechnology. Independent Institute, Oakland, CA.
McChesney, F., 1997. Money for Nothing: Politicians, Rent Extraction, and Political Extortion.Harvard University Press, Cambridge, MA.
Mitchell, M., Netter, J., 1989. Triggering the 1987 stock market crash: antitakeover provisions in theproposed house ways and means tax bill. Journal of Financial Economics 24, 37}68.
Mullin, G., Mullin, J., Mullin, W., 1995. The competitive e!ects of mergers: stock market evidencefrom the U.S. Steel dissolution suit. Rand Journal of Economics 26, 314}330.
Nash, K., 1996. Netscape warns investors of threat from Microsoft. Computerworld, Nov. 11 Onlinevia Computer Magazine Archive (http://cma.zdnet.com/texis/cma/cma/).
Pindyck, R., 1991. Irreversibility, uncertainty and investment. Journal of Economic Literature 29,1110}1152.
Prager, R., 1992. The e!ects of horizontal mergers on competition: the case of the northern securitiescompany. Rand Journal of Economics 23, 123}133.
Quittner, J., Slatalla, M., 1998. Speeding the Net: The Inside Story of Netscape and How itChallenged Microsoft. Atlantic Monthly Press, New York, NY.
Rose, N., 1985. The incidence of regulatory rents in the motor carrier industry. Rand Journal ofEconomics 16, 299}318.
Salinger, M., 1992. Standard errors in event studies. Journal of Financial and Quantitative Analysis27, 39}53.
Schipper, K., Thompson, R., 1983. The impact of merger-related regulations on the shareholders ofacquiring "rms. Journal of Accounting Research 21, 184}221.
Schwert, G., 1981. Using "nancial data to measure e!ects of regulation. Journal of Law andEconomics 24, 121}158.
Smith, R., Bradley, M., Jarrell, G., 1986. Studying "rm-speci"c e!ects of regulation with stockmarket data: an application to oil price regulation. Rand Journal of Economics 17, 467}489.
Stillman, R., 1983. Examining antitrust policy towards horizontal mergers. Journal of FinancialEconomics 11, 225}240.
Wilson, J., 1980. The politics of regulation. Basic Books, New York, NY.Wolf, C., 1993. Markets or Governments? Choosing Between Imperfect Alternatives, 2nd Edition.
MIT Press, Cambridge, MA.
G. Bittlingmayer, T.W. Hazlett / Journal of Financial Economics 55 (2000) 329}359 359