MODULAR MARKET DEFINITION
Sean P. Sullivan
August 10, 2021
Surging interest in antitrust enforcement is exposing, once again, the dif-
ficulty of defining relevant markets. Past decades have witnessed the invention
of many tests for defining markets, but little progress has been made, or even
attempted, at reconciling these different tests. Modern market definition has
thus become a confused agglomeration of often conflicting ideas about what
relevant markets are and how they should be defined and used. The result—
unpredictable and unreliable market boundaries—is an unsure footing for the
complicated cases and policy questions now before us.
This Article responds to the problem of confused market definition with a
simple but powerful approach to dealing with multiple tests for defining mar-
kets. The basic insight is that different tests scope markets appropriate for serv-
ing different needs. Helpful market definition can thus proceed in two steps.
First, identify the substantive purposes for which markets are being defined in
a particular application. Second, select the test that defines markets most suited
to serving those purposes.
This modular approach to market definition offers several advantages over
the current conflation of different tests. First, the modular approach promises
University of Iowa College of Law: firstname.lastname@example.org. I am indebted to Jonathan
Baker, Daniel Francis, Giorgio Monti, Christopher Odinet, César Rosado, Stephen Ross, Steven
Salop, D. Daniel Sokol, Spencer Weber Waller, Kevin Washburn, Samuel Weinstein, and partic-
ipants of the 2020 Cambridge-Florida Virtual Antitrust Workshop and 2020 University of Iowa
Summer Workshop Series, 2021 University of Iowa Faculty Workshop Series, 2021 Thurman
Arnold Project Workshop Series, and 2021 Michigan-USC-Virginia Virtual Law & Economics
Workshop Series for thoughtful comments on earlier drafts of this article. Alexander Asawa,
Kassandra DiPietro, and Madison Tallant provided invaluable research assistance. Amy
Koopmann and the University of Iowa Law Library assisted in background research for this
greater predictability and reliability in market definition practice. Second, it
provides a more legally honest and economically coherent explanation of how
the various tests for defining markets fit together. Third, it contributes to on-
going policy discussions, clarifying how relevant markets work in antitrust law
and how they can be leveraged to empower more efficient and effective en-
Introduction ....................................................................................... 1
I. Process — The History of Market Definition ........................ 6
A. Tests based on commodity concepts ................................. 7
B. Tests based on popular perception ................................... 10
C. Tests based on joint market power .................................. 14
D. Tests based on individual market power ......................... 19
II. Purpose — The Functions of Market Definition ............... 26
A. Magnification ..................................................................... 27
B. Translation .......................................................................... 30
C. Exploration ......................................................................... 33
III. Practice — Modular Market Definition ............................. 37
A. Coordinated effects enabling tacit collusion ................... 38
B. Coordinated effects entrenching tacit collusion ............. 40
C. Concerted Conduct ............................................................ 42
D. Undifferentiated-product unilateral effects .................... 45
E. Differentiated-product unilateral effects ......................... 46
F. Monopolization ................................................................... 47
G. Structuralist Concerns ....................................................... 50
Conclusion ....................................................................................... 52
In 1945, the Aluminum Company of America (Alcoa) was like the Apple
or Amazon of today. Founded by a backyard inventor who had discovered a
low-cost way of smelting aluminum, the company had grown into a corporate
empire. From its humble start as a smelting operation, Alcoa expanded up-
stream into mining operations and power generation. It also expanded down-
stream into the fabrication of cookware, cables, and machine parts. Through
growth and acquisitions at every level, Alcoa’s operations soon blanketed the
nation and suffocated its rivals.
In the eyes of the public, Alcoa was a danger-
ous monopoly, the likes of which the antitrust laws were meant to condemn.
Unbeknownst to the public, Alcoa would soon become the catalyst to almost a
century of struggle with market definition.
The government sued Alcoa, accusing it of monopolizing the aluminum
ingot market, but lost badly at the district court.
Undeterred, it appealed to
the Supreme Court,
only to encounter recusals so numerous that a quorum
could not be reached. The stalemate was finally broken when the Court relin-
quished final review of the case to a panel of the Second Circuit.
By this pecu-
liar chain of events, Learned Hand came to write for the court of last resort.
Hand’s always clear-minded analysis did much to produce a reasoned and
See generally Spencer Weber Waller, The Story of Alcoa: The Enduring Questions of Market
Power, Conduct, and Remedy in Monopolization Cases, in ANTITRUST STORIES 121 (Eleanor M.
Fox & Daniel A. Crane eds., 2007) (providing detailed background on Alcoa).
Learned Hand observed as much in a preconference memo: “If we hold that [Alcoa] is not
a monopoly, deliberately planned and maintained, every-one who does not get entangled in the
legal niceties, and in the incredible nonsense that has emanated from the Supreme Court, will,
quite rightly I think, write us down as asses. Wherever the line of size should be drawn, it must
include such a company as this . . . .” Memorandum from Judge Learned Hand to Judges Augus-
tus N. Hand & Thomas W. Swan 13–14, United Sates v. Aluminum Co. of Am. (Alcoa), 148 F.2d
416 (2d Cir. 1945) (on file with author) [hereinafter Hand Memo].
United States v. Aluminum Co. of Am., 44 F. Supp. 97, <PIN> (S.D.N.Y. 1941), aff’d in
part and rev’d in part, 148 F.2d 416 (2d Cir. 1945).
Direct appeal was permitted under the Expediting Act. Ch. 544, 32 Stat. 823 (1903) (current
version at 15 U.S.C. § 29).
The proceeding was postponed indefinitely when recusals resulted in the absence of a
quorum. United States v. Aluminum Co. of Am., 320 U.S. 708, <PIN> (1943). Congress eventu-
ally intervened and the case was transferred to the Second Circuit. United States v. Aluminum
Co. of Am., 322 U.S. 716, <PIN> (1944).
persuasive resolution of the case. But Hand struggled, ultimately in vain, to
untie one defiant knot: he did not know what the “aluminum ingot market”
Because Alcoa produced everything from virgin ingot to pots, pans, and
machine parts, the market could potentially have included anything from min-
erals in the ground to the aluminum products scattered throughout kitchens
and garages around the country. That was a problem for Hand, because differ-
ent ways of slicing the aluminum ingot market pointed to different legal out-
comes. If the market included everything and the kitchen sink, then Alcoa’s
market share hovered around 60%—big, but not clearly monopoly.
If it in-
cluded just virgin ingot and the secondary metal produced from recycled scrap
aluminum, then Alcoa’s market share shrank to 30%.
But if the market was
defined as virgin ingot and its initial fabrication, then Alcoa’s share rocketed
up to an undoubtedly monopolistic 90%.
What Hand needed, to decide the case on a sound and persuasive basis,
was a predictable and reliable tool for choosing between these different plau-
sible definitions of the aluminum ingot market. Unfortunately, no such tool
existed. No prior opinion had grappled with this problem, at least not to the
point of producing useful precedent on defining markets.
Without such a
tool, Hand’s opinion was destined to disappoint. And it did. Alcoa lost the
case—Hand chose the 90% option—but the way the outcome of such an im-
portant case had come to teeter upon so unprincipled and unpredictable a line-
drawing exercise has haunted antitrust ever since.
The world has changed much since 1945 but, in many ways, it is still the
same. In place of massive metallurgy companies, public concern about
Hand Memo, supra note 2, at 12 (noting that “[t]he question whether Alcoa is a monopoly
under §2 of the Act depends in the first place upon its control of the production of virgin alumi-
num, and upon the extent to which through that control it has control of the market for alumi-
num ingot” and proceeding to consider uncertainties about what should be counted in the alu-
minum ingot market).
United States v. Aluminum Co. of Am. (Alcoa), 148 F.2d 416, 424 (2d Cir. 1945).
See infra notes 27–35 (describing market definition before Alcoa).
Alcoa, 148 F.2d at 422-26 (citing no authority on how to define a market). Hand’s market
analysis is weak by modern standards. See Waller, supra note 1, at 130-33. But his conclusions
were not necessarily wrong. See id. (providing other justifications for Hand’s choice of market).
See generally Peter J. Swan, The Influence of Recycling on Monopoly Power, 88 J. POL. ECON. 76
(1980) (defending important aspects of Hand’s market analysis).
monopoly now stacks against technology giants like Google, Amazon, Face-
book, and Apple. The products are different, but if we were trying to say
whether Google had monopolized the market for general search services,
would the difficulty of identifying the perimeter of that market be any different
today than it was for Judge Hand in 1945? Or take the claim that Facebook
dominates the market for personal social networking services.
Does it? To
answer this, we need a way of deciding what else is in this market. Would
video-based services like YouTube be included? Professional services like
LinkedIn? What about email and text messaging? Time marches forward, but
the challenges of market definition have not aged a day.
Another way that antitrust has developed much but changed little is in the
development of tests for defining markets. We are now officially overflowing
with these tests. We have tests that define markets based on interchangeability
of product uses and upon cross elasticity of demand.
We have tests based on
things like public perception and trade usage.
We have tests based on predic-
tions of how hypothetical monopolists would behave.
We have refinements
of these tests based on econometric and statistical techniques.
We have yet
more refinements based on observed behavior.
Just about the only thing we
See Complaint ¶ 92, United States v. Google, LLC, No. 1:20-CV-03010 (D.D.C. Oct. 20,
Complaint for Injunctive and Other Equitable Relief ¶¶ 61-62, Federal Trade Commission
v. Facebook, Inc., No. 1:20-CV-03590-JEB (D.D.C. Jan. 13, 2021).
Times-Picayune Publ’g Co. v. United States, 345 U.S. 594, 612 n.31 (1953); United States
v. E. I. du Pont de Nemours & Co. (Cellophane), 351 U.S. 377, 395 (1956).
Brown Shoe Co. v. United States, 370 U.S. 294, 325 (1962); see also United States v. E. I.
du Pont de Nemours & Co. (du Pont-General Motors), 353 U.S. 586, 593-94 (1957) (defining a
similar test around the identification of “peculiar characteristics and uses”).
E.g., LAWRENCE SULLIVAN, HANDBOOK OF THE LAW OF ANTITRUST 4 (1977) (defining mar-
kets by asking whether a price increase in a provisional market could be maintained for some
time); Rothery Storage & Van Co. v. Atlas Van Lines, Inc., 792 F.2d 210, 218 (D.C. Cir. 1986)
(characterizing Sullivan’s price-increase maintenance hypothetical as a well-known criterion for
defining markets); U.S. DEP’T OF JUSTICE & FED. TRADE COMM’N, HORIZONTAL MERGER GUIDE-
LINES § 4.1 (August 19, 2010) [hereinafter 2010 HORIZONTAL MERGER GUIDELINES], http://
www.justice.gov/sites/default/files/atr/legacy/2010/08/19/hmg-2010.pdf (describing the Hypo-
thetical Monopolist Tests for defining relevant markets).
E.g., Barry C. Harris & Joseph J. Simons, Focusing Market Definition: How Much Substi-
tution Is Necessary, 12 RES. L. & ECON. 207, 211-19 (1989); Joseph Farrell & Carl Shapiro, Anti-
trust Evaluation of Horizontal Mergers: An Economic Alternative to Market Definition, 10 B.E. J.
THEORETICAL ECON., Mar. 2010, art. 9, at 1, 4.
E.g., FTC v. Staples, Inc., 970 F. Supp. 1066, 1075-76 (D.D.C. 1997).
do not have is any way of balancing or reconciling this tangled mess of varied
and often conflicting approaches.
The result, of course, is that we still do not have a predictable and reliable
tool for defining markets. All that we have done is replace the need for judges
to make arbitrary and unprincipled decisions about the scope of markets with
the need for judges to make arbitrary and unprincipled decisions about the
tests that will be used to define the scope of markets. This has not made market
definition more certain or reliable. If anything, it has done the opposite.
Learned Hand at least had common sense and intuition on his side. Modern
market definition may not leave us even these.
Antitrust pays a heavy price for confused and unreliable market definition.
Enforcement failures and numbingly overcomplicated legal disputes are
among the visible consequences of market definition practices today.
ture consequences are even more concerning. A bill, cosponsored by Senator
Amy Klobuchar, seeks to strengthen antitrust enforcement by making mer-
gers, and certain other forms of conduct, presumptively illegal if undertaken
by firms holding more than a fifty percent share of a relevant market.
of the underlying objective aside, the problem with this plan is that it fails to
address the inextricable dependence of market-share thresholds on market
definition. The strongest presumption in the world means nothing if it is not
braced against a predictable and reliable system for defining markets. Without
that bracing, the effort could backfire. If defendants are able to exploit the slop-
piness of market definition to consistently minimize their apparent market
share, then the increased emphasis on market share statistics could hinder the
very enforcement that the bill seeks to strengthen.
To reduce a complicated situation down to simple terms, we can imagine
market definition as a carpentry tool. For the past several decades, judges and
antitrust experts have treated market definition like it was a hammer.
mers are simple tools. Sure, they come in different shapes, sizes, and colors.
But, for basic carpentry, any hammer will do. The same hammer fits every nail.
Those who have treated market definition like a hammer have tended to
See, e.g., FTC v. Facebook, Inc., No. 20-3590, at *2 (D.D.C. Jun. 20, 2020) (finding that the
FTC “failed to plead enough facts to plausibly establish . . . that Facebook has monopoly power
in the market for Personal Social Networking (PSN) Services”).
S. 225, 117th Cong. §§ 4, 26A (2021).
See, e.g., United States v. Aetna Inc., 240 F. Supp. 3d 1, 19–23 (D.D.C. 2017) (conflating at
least five separate tests under the common heading of “market definition”); FTC v. Sysco Corp.,
113 F. Supp. 3d 1, 25–27 (D.D.C. 2015) (similar).
assume that it shares these properties. Just as the same hammer can be carried
from one job to the next, the same test of market definition has been repur-
posed from one context to the next.
And just as every hammer leads to the
same result when used to pound a nail, every test of market definition has been
assumed to target and identify the exact same scope of trade.
The hammer analogy is old and in many ways appealing. But it has failed
antitrust. Suppose, instead, that we were to treat market definition like a power
drill. Unlike hammers, drills consist of separate parts. One part is the base,
which converts power into torque. The other part is the drill bit, which con-
verts torque into a hole of a specific size, depth, and shape. Using a drill re-
quires more thought than using a hammer. Even if the drill is only used for
making holes, the carpenter still must stop to select the appropriate drill bit for
every job. Treating market definition like a power drill would mean ascribing
it these properties. Just as the right drill bit must be chosen to produce the
desired hole, the right test of market definition would need to be chosen to
serve the purposes that the resulting market was meant to serve.
Power drills are examples of modular design: the ability to use different
drill bits for different applications is essential to the usefulness of the tool. My
thesis in this Article is that market definition is and always has been a modular
concept—a drill, not a hammer. This becomes clear when we stop to consider
how the process of defining markets, the tests we use to decide the scope of
markets in specific applications, relates to the purpose of market definition, the
substantive functions that relevant markets are meant to serve in specific ap-
plications. Once process and purpose are severed, it becomes natural to ask:
which of the available processes of market definition best serves the purposes
for which we are defining markets in this application? And once that question
is asked, it becomes easy to see answers throughout the history of market def-
inition. Put another way, this Article does not advance a new approach to mar-
ket definition. It simply reveals an approach that has been sitting under our
noses all along.
A modular understanding of market definition constitutes a predictable
and reliable tool for defining markets in antitrust cases. This Article extracts
that tool from antitrust history in Part I, observing how different tests of
See United States v. E. I. du Pont de Nemours & Co. (Cellophane), 351 U.S. 377, <PIN>
(1956) (“The ‘market’ which one must study . . . will vary with the part of commerce under con-
sideration. The tests are constant.”).
See infra notes 24–25 (citing examples of the modern everything soup statement of the
standard for market definition).
market definition were developed to meet different needs in response to
changes in the focus of substantive antitrust law. The Article describes that tool
in Part II, using three purposes that most relevant markets have traditionally
served to explore how the choice of market definition test relates to the fulfill-
ment of the expected purposes of relevant markets. Finally, the Article explains
how the tool can be used in Part III, enumerating various situations in which
relevant markets may be needed and discussing how to define helpful relevant
markets in each application. A brief conclusion offers four policy recommen-
dations that follow from this clarified understanding of market definition.
I. Process — The History of Market Definition
On what basis do we define markets? A glance at the market definition
section of any recent antitrust opinion will reveal several pages of potential
What it will not reveal is even the slightest effort to reconcile or balance
the—apparently simultaneous—application of all these varied and conflicting
approaches to the facts at hand.
The result is that summaries of the standard
for defining markets often read more like a whiplash tour of antitrust history
than they do a useful guide to market definition in a given case.
But, even setting aside the implementation problems raised by multiple
tests, on what basis have we come to believe that all these tests are simultane-
ously helpful in the first place? Antitrust law and policy evolved dramatically
over its history to date.
Might not different tests of market definition have
been designed to fit different substantive concerns?
E.g., FTC v. Qualcomm Inc., 411 F. Supp. 3d 658, 683-85 (N.D. Cal. 2019), rev’d and va-
cated on other grounds, 969 F.3d 974 (9th Cir. 2020) (citing, as bases for market definition, the
practical indicia factors, reasonable interchangeability and cross-elasticity of demand standards,
determination of available substitutes, and the hypothetical monopolist test); FTC v. Wilh. Wil-
helmsen Holding ASA, 341 F. Supp. 3d 27, 45-47 (D.D.C. 2018) (similar); United States v. An-
them, Inc., 236 F. Supp. 3d 171, 193-95 (D.D.C. 2017); Sysco, 113 F. Supp. 3d at 25-38 (similar).
E.g., Wilhelmsen, 341 F. Supp. 3d at 47 (citing both the Brown Shoe practical indicia and
the HMT as parallel sources of evidence upon which market definition would be based); United
States v. Aetna Inc., 240 F. Supp. 3d 1, 20-21 (D.D.C. 2017) (same); Sysco, 113 F. Supp. 3d at 27-
37 (same); United States v. H & R Block, Inc., 833 F. Supp. 2d 36, 50-52 (D.D.C. 2011) (same).
See William E. Kovacic & Carl Shapiro, Antitrust Policy: A Century of Economic and Legal
Thinking, 14 J. ECON. PERSP. 43, <PIN> (2000) (chronicling the historic convergence of legal and
economic analysis in antitrust).
A. Tests based on commodity concepts
Let us start as far back as possible. Long before the passage of the first U.S.
antitrust statutes, the common law provided that unreasonable restraints of
trade were void as against public policy.
Judges deciding restraint of trade
cases engaged in a simple form of implicit market definition. They assumed
that the scope of competitive effects followed commodity lines.
In the late 1800s, for example, courts in many states were called upon to
decide whether municipal market regulations were enforceable.
To take one
representative case, in City of Bloomington v. Wahl, a city ordinance provided
that fresh meat could only be sold within the designated space of the Bloom-
ington City Market; a meat vendor whose shop lay outside the perimeter of
this market objected to the rule as an unreasonable restraint of trade.
discussion of the case, the judge referred to the “market” as the physical space
in the city, but focused on the “business” of selling meat as the area of trade in
which the effect of the restraint might be felt.
Cases testing the enforceability
of voluntary or implied agreements not to compete tended to follow a similar
path. Here, judges focused their attention on the restriction of competition
within a popularly recognizable line of trade, such as “the business of boat-
or the “trade of a baker.”
This simple equation of the relevant scope of trade with commodity con-
cepts survived the passage of the Sherman Act. Thus, we see early antitrust
opinions address a company’s dominance over “the oil industry”
or the plight
of miners in “the coal industry.”
Also like the common law cases, early
See, e.g., Mitchel v. Reynolds (1711) 24 Eng. Rep. 347 (discussing the policy motivating
this doctrine as it applied to covenants not to compete).
See, e.g., City of Bloomington v. Wahl, 46 Ill. 489 (1868) ; City of Chicago v. Rumpff, 45
Ill. 90, 97-98 (1867) ; Caldwell v. City of Alton, 33 Ill. 416 (1864) ; Gale v. Vill. of Kalamazoo, 23
Mich. 344 (1871) ; Dunham v. Trustees of Rochester, 1826 WL 2016 (N.Y. Sup. Ct. 1826) ; Town
Council of Winnsboro v. Smart, 45 S.C.L. 551 (S.C. App. L. 1858).
City of Bloomington, 46 Ill. at 490-91.
Id. at 493 (“If this may be done, the business in this department would fall into the hands
of the few, and all competition would be destroyed, and the people oppressed.”).
Palmer v. Stebbins, 20 Mass. 188, 193 (1825); see also Bergamini v. Bastian, 35 La. Ann.
60, 66 (1883) (discussing the “line of business” of selling coffee and pastries).
Mitchel v. Reynolds (1711) 24 Eng. Rep. 347, <PIN>.
Standard Oil Co. of N.J. v. United States, 221 U.S. 1, 75 (1911).
Appalachian Coals v. United States, 288 U.S. 344, 361 (1933).
antitrust opinions used the term “market” in the sense of a physical location
in which a commodity was being traded.
Now, jump ahead to the state of antitrust law in the wake of Alcoa in 1945.
A systematic method for choosing between alternative views of market scope
was obviously needed, but no obvious methodology presented itself. In 1948,
the Supreme Court refereed a “sharp dispute” between the government and
some merging steel producers concerning the scope of “the market for rolled
Rather than use the opportunity to clarify market definition,
the Court recoiled from “the difficulty of laying down a rule as to what areas
or products are competitive,” contributing only a conclusory and fact-bound
declaration of the market’s scope in this particular case.
Finally, in the mid-1950s, the first tentative steps toward a modern test of
market definition began to appear. In Times-Picayune Publishing Company v.
United States, the Court counseled that the scope of the market “must be
drawn narrowly to exclude any other product to which, within reasonable var-
iations in price, only a limited number of buyers will turn; in technical terms,
products whose ‘cross-elasticities of demand’ are small.”
A few years later, in
the Cellophane case, it elaborated that the “market is composed of products
that have reasonable interchangeability for the purposes for which they are
produced—price, use and qualities considered.”
These were important steps in the history of market definition, and the
confident language of the Times Picayune and Cellophane opinions survives
today in nearly every statement of the standard for defining markets. In sub-
stance, however, these standards of market definition were like a fresh coat of
paint on an old car: nice new look, same old engine. The underlying goal was
still to define markets by implicit reference to commodity products. All that
the new language did was fussy up the prior practice.
One way to see this is to note the absence of a decision threshold in either
test. Are pens and pencils interchangeable writing instruments? This seems
See, e.g., id. (“Coal has been losing markets to oil, natural gas and water power and has
also been losing ground due to greater efficiency in the use of coal.”); Bd. of Trade of City of Chi.
v. United States, 246 U.S. 231, 239-40 (1918) (discussing the availability of grain markets in dif-
ferent Midwest states).
United States v. Columbia Steel Co., 334 U.S. 495, 508 (1948).
Id. at 511.
Times-Picayune Pub. Co. v. United States, 345 U.S. 594, 612 n.31 (1953). The opinion also
counseled that markets “cannot be measured by metes and bounds.” Id. at 611.
United States v. E. I. du Pont de Nemours & Co. (Cellophane), 351 U.S. 377, 404 (1956).
like something upon which people could disagree. But suppose for sake of ar-
gument that they are interchangeable; are they sufficiently interchangeable to
warrant placing them in a common market? How is a judge supposed to an-
swer this question except by personal intuition and reference to common
This was, of course, the prevailing practice before
these tests were introduced.
Another, more subtle, way that these tests reflect a focus on commodity
products is in their inattention to prevailing prices. In the imaginary world of
perfectly competitive trade in commodities, no firm has the power to influence
prices, so it is sensible to judge the closeness of products by their substitutabil-
ity at prevailing prices. This only works in that theoretical world, though. Even
a buyer who strongly prefers aluminum over steel may switch to buying steel
if the price of aluminum gets too high. If a company like Alcoa has the market
power to raise the price of aluminum, this means that the interchangeability
of steel becomes, in part, a matter of how high the price of aluminum is
To put it concretely, would we want to define markets in a way that
would let a company like Alcoa expand the market—shrinking its market
share and thereby escaping liability—simply by raising its prices? If that’s not
letting the fox guard the hen house, it is only because the fox has fewer moti-
vations to eat then hens than the monopolist has to raise its prices under such
The appeal of the Supreme Court’s early market definition tests is limited
to the commodity competition context in which they were devised. There is
nothing fundamentally wrong with scoping markets by commodity product
lines in this narrow context, but neither is there anything that recommends it.
No modern antitrust goal is advanced by defining markets according to the
Times Picayune and Cellophane tests today. Fortunately, the history of market
definition did not end here.
See David Glasner & Sean P. Sullivan, The Logic of Market Definition, 83 ANTITRUST L.J.
293, 305 (2020) (“[N]either [test] even attempts to articulate where the cutoff lies. How small
must be the cross-elasticity of demand, and how poor must be the interchangeability of use, be-
fore the edge of a relevant market has been reached?”).
In the antitrust literature, a closely related concern with this test of market definition is
discussed under the title of the “Cellophane fallacy” or the “Cellophane trap.” See, e.g., 2B PHILIP
E. AREEDA, HERBERT HOVENKAMP & JOHN L. SOLOW, ANTITRUST LAW ¶ 539 (4th ed. 2014); RICH-
ARD POSNER, ANTITRUST LAW 150–51 (2d ed. 2001).
B. Tests based on popular perception
A few years after the Times Picayune and Cellophane decisions, a spate of
Eisenhauer appointments reshaped the Supreme Court’s stance on antitrust.
From the late 1950s to the late 1960s, Warren Court antitrust emerged as a
machine bent on taking back political control of industry and commerce.
Looking at what antitrust was about during this unique epoch in its history,
we can quickly see why new and different tests of market definition would be
needed to achieve the Court’s ends.
One pillar of Warren Court antitrust was a certain type of concern with
economic efficiency. A strong sense of “structuralism” persuaded the Court—
and many economists at the time—that unconcentrated industries performed
better than concentrated industries,
and thus that economic efficiency could
be promoted by preventing increases in concentration whenever possible. In
United States v. Philadelphia National Bank, the Court balked at an attempted
merger of the second and third largest banks in a local geographic market.
Observing that “competition is likely to be greatest when there are many
sellers, none of which has any significant market share,”
the Court reasoned
that a merger that would give a single firm control over 30% of commercial
banking in a local area was “so inherently likely to lessen competition substan-
tially” that it should be enjoined with little further inquiry.
concern with competitor concentration undergirds many of the Warren
Court’s decisions. But even though economic efficiency was an important goal,
it was not the Court’s only focus at this time.
Another pillar was the related idea that small and local businesses needed
protection against competition from larger and more efficient rivals. In Brown
Shoe Company v. United States,
the Court placed this protectionist goal above
See Donald I. Baker & William Blumenthal, The 1982 Guidelines and Preexisting Law, 71
CALIF. L. REV. 311, 315 (1983) (“[M]erger policy during the 1960’s tended to flow from a simple
equation: increases in concentration lead to less efficient performance.”); Herbert Hovenkamp,
Markets in Merger Analysis, 50 ANTITRUST BULL. 887, 889 (2012) (observing that “highly influ-
ential in the economic literature of the 1960s, was structuralism, which found a close link be-
tween economic performance and market structure”).
United States v. Phila. Nat. Bank, 374 U.S. 321, 330 (1963).
Id. at 363 (internal markup omitted) (quoting Comment, ‘Substantially to Lessen Compe-
tition . . . ’: Current Problems of Horizontal Mergers, 68 YALE L.J. 1627, 1638-39 (1959)). The
Court cited both economists and Congress as supporting this proposition. Id. at 363, nn. 38-39.
Id. at 363.
Brown Shoe Co. v. United States, 370 U.S. 294 (1962).
efficiency concerns. In amending Section 7 of the Clayton Act, it said, “Con-
gress was desirous of preventing the formation of further oligopolies [because
of] their attendant adverse effects upon local control of industry and upon
The Court explicitly held that even if “higher costs and prices
might result” from the protection of small and local businesses, Congress had
“resolved these competing considerations in favor of decentralization.”
In short, Warren Court antitrust was about protecting little guys and fend-
ing off the creep of industrial concentration.
This was a time when even small
increases in concentration raised antitrust concerns.
It was a time when effi-
ciency advantages were often seen as problematic.
It was a time when cutting
prices to win customers could be attacked as anticompetitive.
It was a time
defined by a distinct shift in analytical focus and in need of a correspondingly
distinct shift in the approach to market definition.
Hints of that new approach can be glimpsed in the Supreme Court’s 1957
decision of the du Pont-General Motors case.
Studiously avoiding any men-
tion of the Times Picayune or Cellophane tests, the Court defined a narrow
market consisting of “automotive finishes and fabrics” by listing off a few
“characteristics and uses” that distinguished these products from the broader
field of “other finishes and fabrics.”
Alone, this pivot would not have been
that monumental, but it signaled bigger changes to come.
When, a few years
Id. at 333.
Id. at 344.
Cf. Thomas E. Kauper, The Warren Court and the Antitrust Laws: Of Economics, Populism,
and Cynicism, 67 MICH. L. REV. 325, 329 (1968) (discussing the “peculiar blend” of economic
theory and populism that motivated Warren Court antitrust).
See, e.g., United States v. Von’s Grocery Co., 384 U.S. 270, 272 (1966) (breaking up a mer-
ger that produced a firm with a total market share of about 7.5 percent).
See, e.g., Hovenkamp, supra note 42, at 895 (“The perceived injury in Brown Shoe was . . .
[that] Brown Shoe would acquire a competitive advantage over its competitors.”).
See, e.g., Utah Pie Co. v. Cont’l Baking Co., 386 U.S. 685, 690-98 (1967) (describing price
cutting that led to “a deteriorating price structure” as “lessening of competition”).
United States v. E. I. du Pont de Nemours & Co. (du Pont-General Motors), 353 U.S. 586,
Id. at 593-94, n.12.
See, e.g., Jesse W. Markham, The Du Pont-General Motors Decision, 43 VA. L. REV. 881,
884-88 (1957) (expressing concern about the future consequences of the du Pont-General Motors
test of market definition); Gregory J. Werden, The History of Antitrust Market Delineation, 76
MARQ. L. REV. 123, 143 (1992) (interpreting the du Pont-General Motors case as marking “a sig-
nificant shift in ideology on the Court, which was to prove decisive over the remainder of Chief
Justice Warren’s tenure”).
later, the Court decided Brown Shoe, the same opinion that pegged protection-
ism above efficiency also introduced the next major test of market definition.
The test that Brown Shoe announced defined markets
by reference to a
list of “practical indicia,” observational evidence of how businesspeople and
the public perceived industry boundaries:
[Market boundaries] may be determined by examining such
practical indicia as industry or public recognition of the [market]
as a separate economic entity, the product’s peculiar characteris-
tics and uses, unique production facilities, distinct customers,
distinct prices, sensitivity to price changes, and specialized ven-
The Court similarly scoped the geographic boundaries of markets by looking
to the “commercial realities of the industry.”
Put another way, what Brown
Shoe invited judges to do was define markets around such tactile landmarks as
an industry’s own self classification or a lay person’s everyday understanding
of a what constituted a market or an industry.
The invitation was eagerly accepted. Brown Shoe was a merger case, and a
review of subsequent merger cases from the 1960s and 1970s shows that the
practical indicia test quickly came to dominate market definition analysis.
also diffused into other areas of antitrust law. In 1966, the Court extended the
practical indicia test to monopolization cases.
Lower courts extended it fur-
ther to cover concerted action cases.
And when the DOJ released its first
To be precise, Brown Shoe described a test for defining “submarkets.” The difference be-
tween submarkets and relevant markets was unclear from the start and the two soon converged.
See Geneva Pharm. Tech. Corp. v. Barr Labs. Inc., 386 F.3d 485, 496 (2d Cir. 2004) (“The term
‘submarket’ is somewhat of a misnomer, since the ‘submarket’ analysis simply clarifies whether
two products are in fact ‘reasonable’ substitutes and are therefore part of the same market.”); 2B
AREEDA, HOVENKAMP & SOLOW, supra note 41, ¶ 533 (critiquing efforts to distinguish submar-
kets from relevant markets).
Brown Shoe Co. v. United States, 370 U.S. 294, 325 (1962).
Id. at 336.
Werden, supra note 55, at 172 (“In the two decades following the Supreme Court’s deci-
sion in Brown Shoe Co. v. United States, the submarket concept and the practical indicia domi-
nated thinking on market delineation in the lower courts.” (footnote omitted)).
United States v. Grinnell Corp., 384 U.S. 563, 572 (1966) (endorsing, in dicta, the use of
Brown Shoe’s practical indicia test in monopolization cases).
E.g., Columbia Metal Culvert Co. v. Kaiser Aluminum & Chem. Corp., 579 F.2d 20, 26-27
(3d Cir. 1978) (“[U]nder either s 1 or s 2 of the Sherman Act, judges . . . are adjured to follow the
merger guidelines in 1968, its own test of market definition drew obvious in-
spiration from both the du Pont-General Motors and Brown Shoe tests: markets
were defined as “[t]he sales of any product or service which is distinguishable
as a matter of commercial practice from other products or services.”
Like Times Picayune and Cellophane, judges still cite Brown Shoe as pri-
mary authority for defining markets today.
And, just like those earlier tests,
modern invocations of the practical indicia test are hard to love. Only a few of
the practical indicia have any plausible connection to antitrust’s current focus
on market power and constraints on that power.
Sure, some creative judges
and scholars have reinterpreted select indicia as evidentiary proxies for market
But this anachronism masks the point of the practical
indicia test, which was never about market power in the first place.
well-trodden trail illuminated by [Brown Shoe’s test of market definition].”); Heatransfer Corp.
v. Volkswagenwerk, A. G., 553 F.2d 964, 980 (5th Cir. 1977) (similarly applying the practical
indicia test to allegations of Section 1 and 2 violations).
U.S. DEP’T OF JUSTICE, MERGER GUIDELINES § 3(i) (1968).
E.g., FTC v. Sysco Corp., 113 F. Supp. 3d 1, 27 (D.D.C. 2015) (“Courts look to two main
types of evidence in defining the relevant product market: the ‘practical indicia’ set forth by the
Supreme Court in Brown Shoe and testimony from experts in the field of economics.”); FTC v.
Wilh. Wilhelmsen Holding ASA, 341 F. Supp. 3d 27, 47 (D.D.C. 2018) (similar); United States
v. Aetna Inc., 240 F. Supp. 3d 1, 21 (D.D.C. 2017) (citing Brown Shoe’s practical indicia as the
first of the “analytical tools at [the court’s] disposal” when defining markets).
See Jonathan Baker, Market Definition: An Analytical Overview, 74 ANTITRUST L.J. 129,
149 (2007) (commenting that not all of the practical indicia are related to substitution patterns
and noting “confusion and error” where use of the practical indicia has not focused on these
patterns); Robert Pitofsky, New Definitions of Relevant Market and the Assault on Antitrust, 90
COLUM. L. REV. 1805, 1815 (1990) (describing the distinct customers factor as “problematic” and
the industry or the public recognition factor as “decidedly marginal on the question of market
definition”); Werden, supra note 55, at 172-79 (criticizing the market power significance of sev-
eral of the practical indicia).
See, e.g., Rothery Storage & Van Co. v. Atlas Van Lines, Inc., 792 F.2d 210, 218-19 (D.C.
Cir. 1986) (Bork, C.J.) (reinterpreting the Court’s practical indicia as “evidentiary proxies for
direct proof of substitutability”); Jonathan Baker, Stepping Out in an Old Brown Shoe: In Quali-
fied Praise of Submarkets, 68 ANTITRUST L.J. 203, 205 (2000) (“Some of the seven practical indicia
appear related to the identification of buyer substitution patterns, the concern of market defini-
tion under the Merger Guidelines.”).
Hovenkamp, supra note 42, at 896-97 (“[T]he rationale for market definition in Brown
Shoe was very different from, and is fundamentally at odds with, the rationale for market defini-
tion . . . today.”).
Judges of Brown Shoe’s era rarely even thought to connect market defini-
tion with market power.
Across forty-four reporter pages, Brown Shoe in-
cludes one solitary reference to market power, buried in a footnote, without
contextual evidence that the Court meant what that term means today.
why should the Court have spent any time on market power? At that time,
antitrust was about preventing industrial concentration and harm to small
businesses. “Industry,” as the word is used by anyone other than today’s anti-
trust experts, has less to do with market power than it does with similarity of
production technologies and recognizable lines of trade—practical indicia fac-
To protect the small, local businesses in a market, one must have in
mind a group of businesses similar enough to each other and distinct enough
from others to constitute a recognizable market in which small, local compet-
itors can be identified and protected—again, practical indicial factors.
The practical indicia test is a process of defining markets appropriate for
addressing industrial concentration and harm to small businesses.
It is a test
designed for and around this particular set of antitrust concerns.
C. Tests based on joint market power
Jump ahead to the 1980s, and we find that antitrust law had changed again.
Warren Court antitrust lost momentum during the 1970s. As Donald Baker
and William Blumenthal observe: “the Supreme Court last sounded the key
See Werden, supra note 55, at 186 (noting the rarity of connecting market definition to
market power from the mid-1950s to the mid-1970s).
Brown Shoe Co. v. United States, 370 U.S. 294, 322 n.38 (1962).
See 2B AREEDA, HOVENKAMP & SOLOW, supra note 41, ¶530a, at 235 n.5 (commenting that
the lay term “market” often encompasses trading locations, like a farmers’ market, or a “category
of manufacture,” like the “motors and generators” market); Hovenkamp, supra note 42, at 891
(commenting that the term “line of commerce” describes, “in commercial law and other settings
. . . a set of products that a layperson might regard as in the same ‘line,’ such as clothing or
See, e.g., United States v. Von’s Grocery Co., 384 U.S. 270, 277 (1966) (“Congress sought
to preserve competition among many small businesses by arresting a trend toward concentration
in its incipiency before that trend developed to the point that a market was left in the grip of a
few big companies.”).
Cf. Lawrence A. Sullivan, The New Merger Guidelines: An Afterword, 71 CALIF. L. REV.
632, 639 (1983) (commenting that if one believed “that Congress wanted to maintain markets of
many small firms, regardless of effects on costs and prices,” then the Court’s approach to market
definition would be justified).
populist phrases, . . . retention of ‘local control’ and ‘protection of small busi-
nesses,’ . . . in 1974 in a dissenting opinion.”
But just as Warren Court anti-
trust was winding down, the Chicago-School philosophy of scholars like Rob-
ert Bork, Frank Easterbrook, and Richard Posner was gaining steam. Several
activist Reagan appointees
were all the additional encouragement that was
needed to crest the peak and descend screaming into a new regime. In a matter
of years, antitrust was “cut to a new pattern,” as one commentator put it.
meant changes in both antitrust policy and enforcement.
As far as policy, Chicago-School antitrust fiercely pursued one objective:
prevent certain exercises of market power and harm to consumer welfare as a
way of preserving economic efficiency. Populist goals which had nothing to do
with economic efficiency, like the protection of small and local businesses,
were unceremoniously jettisoned.
Thus, when the DOJ updated its merger
guidelines in 1982,
the only policy motivating merger enforcement was now
the promotion of economic efficiency.
Within a few years, federal judges
were extending this primacy of efficiency to all of antitrust law.
Baker & Blumenthal, supra note 42, at 320 n.41 (referring to Gulf Oil Corp. v. Copp Paving
Co., 419 U.S. 186, 207 (1974) (Douglas, J., dissenting)).
Cf. Phillip Areeda, Justice’s Merger Guidelines: The General Theory, 71 CALIF. L. REV. 303,
306-07 (1983) (commenting that DOJ officials had “given every indication of a mission to im-
prove and rectify antitrust law, a mission pursued through public statement, amicus briefs, and
Sullivan, supra note 71, at 632; see also William F. Baxter, Responding to the Reaction: The
Draftsman’s View, 71 CALIF. L. REV. 618, 618 (1983) (referencing a “trend in antitrust jurispru-
dence toward a focus on economic efficiency and consumer welfare”).
See Frank H. Easterbrook, Workable Antitrust Policy, 84 MICH. L. REV. 1696, 1703-04
(1986) (describing antitrust goals “other than efficiency (or its close proxy consumers’ welfare)”
as political questions of income redistribution without “any semblance of ‘legal’ criteria” upon
which judges could decide cases).
U.S. DEP’T OF JUSTICE, MERGER GUIDELINES (June 14, 1982) [hereinafter 1982 MERGER
See Baker & Blumenthal, supra note 42, at 317 (“Where economic, social, and political
considerations once received more or less equal billing as the basis for merger policy, economic
considerations have now achieved primacy.” (footnote omitted)); Robert G. Harris & Thomas
M. Jorde, Market Definition in the Merger Guidelines: Implications for Antitrust Enforcement, 71
CALIF. L. REV. 464, 465 (1983) (“[T]he thrust of the Merger Guidelines is that economic efficiency
is the only factor relevant to the enforcement of antitrust laws.”).
E.g., Morrison v. Murray Biscuit Co., 797 F.2d 1430, 1437 (7th Cir. 1986) (“The purpose
of antitrust law, at least as articulated in the modern cases, is to protect the competitive process
as a means of promoting economic efficiency.”); see also Westman Comm’n Co. v. Hobart Int’l,
As far as enforcement was concerned, Chicago-School antitrust focused on
a few specific ways that market power might be acquired. For instance, the
main injury contemplated by the 1982 Merger Guidelines was that a merger
would facilitate the exercise of joint market power by enabling explicit or tacit
collusion among competitors.
The basic idea was that, by eliminating a pre-
viously independent competitor, a merger could lift a constraint that had been
preventing—or at least frustrating—cooperation to jointly elevate prices.
Richard Posner captured this collusion-centric Chicago-School focus in Hos-
pital Corporation of America v. FTC: “When an economic approach is taken
in a [merger] case, the ultimate issue is whether the challenged acquisition is
likely to facilitate collusion.”
While judges still needed to define markets and measure market concen-
tration in order to decide merger cases under this new approach, their reasons
for doing so were far from what had motivated Warren Court judges. Market
boundaries were needed only to identify the groups of competitors that could
potentially collude on price elevation after a merger. Market concentration
mattered only because economic theory suggested that concentrated markets
would be more susceptible to collusion than unconcentrated markets.
short, the purposes for which judges were defining markets were now alien to
the purposes that had motivated market definition not twenty years before.
Inc., 796 F.2d 1216, 1220 (10th Cir. 1986) (“We adhere to the view that the antitrust laws should
not restrict the autonomy of independent businessmen when their activities have no adverse
impact on the price, quality, and quantity of goods and services offered to the consumer.”).
See Baker & Blumenthal, supra note 42, at 315 (“From among the many conceivable eco-
nomically based enforcement theories, the Department has plucked one of comparatively nar-
row (but hardly unanticipated) focus: mergers must not be permitted to enhance substantially
the risk of tacit collusion.”); id. (“[T]he principal risk associated with a merger is that it might
better enable firms in the industry to conspire tacitly to increase prices and restrain produc-
Pitofsky, supra note 64, at 1807 (“Merger enforcement . . . proceeds from the premise that
when a small group of firms occupies a large share of the relevant market, they can more easily
collude or coordinate sales policies in order to raise prices above competitive levels.”).
Hosp. Corp. of Am. v. F.T.C., 807 F.2d 1381, 1386 (7th Cir. 1986) (Posner, J.).
See infra notes 154–157 and accompanying text.
See Hospital Corporation of America, 807 F.2d. at 1386 (“[T]he economic concept of com-
petition, rather than any desire to preserve rivals as such, is the lodestar that shall guide the con-
temporary application of the antitrust laws . . . .”); Baker & Blumenthal, supra note 42, at 316
(“Unlike the 1960’s cases, however, the Guidelines view concentration as mattering not for its
own sake, but because it increases the likelihood of collusion.”).
A creature of that earlier time, Brown Shoe’s practical indicia test was use-
less for these new purposes. Influential economists like Janusz Ordover and
Robert Willig criticized the practical indicia, and earlier tests of market defini-
tion, as “inadequate substitute[s] for, and a diversion from, sound direct as-
sessment of a merger’s effects.”
George Stigler called previous market defini-
tion “an almost impudent exercise in economic gerrymandering.”
Blumenthal castigated Warren Court market definition as “ad hoc evidentiary
selection, hand-waving, or result orientation.”
The common theme was ob-
vious: a new process of market definition was needed.
The 1982 Merger Guidelines responded to that need.
Monopolist Test (“HMT”), promulgated by the guidelines, delineated markets
not by reference to commodity concepts, or by popular perceptions of market
boundaries, but by analytically identifying a scope of trade in which collusion
among competitors could lead to higher prices.
The approach of the HMT
was to start by taking a group of producers to be a small provisional market
and to ask whether the firms in that market would, if they were hypothetically
joined together to act as a monopolist not constrained by price regulation or
the entry of new firms, choose to implement at least a small but substantial
price increase. If the answer to this question was “yes,” then that provisional
market was validated as a relevant market for antitrust analysis. If “no,” then
more producers would be added to the provisional market and the process was
repeated until a price increase would be imposed. At base, the HMT defined a
Janusz A. Ordover & Robert D. Willig, The 1982 Department of Justice Merger Guidelines:
An Economic Assessment, 71 CALIF. L. REV. 535, 536 (1983).
George J. Stigler, The Economists and the Problem of Monopoly, 72 AM. ECON. REV., May
1982, at 1, 8.
Baker & Blumenthal, supra note 42, at 324.
Though the 1982 Merger Guidelines are often credited as introducing this test, the basic
idea seems to have occurred to various authors at about the same time. See, e.g., SULLIVAN, supra
note 16, at 4 (defining markets by whether a price increase in a provisional market could be
maintained for some time); 2 PHILLIP AREEDA & DONALD F. TURNER, ANTITRUST LAW 347 (1978)
(defining markets as groups of firms that would have market power if acting in unison); Gregory
J. Werden, The Use and Misuse of Shipments Data in Defining Geographic Markets, 26 ANTI-
TRUST BULL. 719, 721 (1981) (defining markets by whether a merger of producers would result
in a price increase); Kenneth D. Boyer, Is There a Principle for Defining Industries?, 50 S. ECON.
J. 761, 763 (1984) (defining markets as ideal collusive groups).
The HMT has been revised over the years but has retained its core structure. Compare
1982 MERGER GUIDELINES, supra note 76, § II.A, with 2010 HORIZONTAL MERGER GUIDELINES,
supra note 16, § 4.1.
market as a group of competitors who could, at least under ideal circum-
stances, collude to jointly raise their prices.
The HMT was the darling of 1980s antitrust. Ordover and Willig called it
a “noteworthy intellectual feat” that focused “much of the best available eco-
nomic learning” on the task of “appropriate economic analysis” in merger
Robert Pitofsky called it a “formidable achievement”
and credited its
“orderly, intellectual approach” with making market definition “a more coher-
ent exercise during the 1980s than in previous decades.”
promulgated by the DOJ, the FTC adopted the HMT internally,
joined the DOJ in advocating the test.
Lower courts similarly adopted the
HMT when defining markets in merger cases.
Enthusiasm for the HMT was no lucky accident. The HMT was a test of
market definition designed for and around the substantive policies of Chicago-
School antitrust. Like Chicago-School antitrust generally, the HMT was about
market power, its enhancement, and its exercise.
Like Chicago-School mer-
ger enforcement specifically, the HMT looked for groups of competitors that
could collude to raise prices.
The HMT was a process of market definition
specifically tailored to the substantive law it was helping to apply.
See SULLIVAN, supra note 16 (proposing a similarly relaxed version of the test).
Ordover & Willig, supra note 84, at 539. See also id. at 537 (describing the HMT as “con-
sistent with economic learning and helpful for logically resolving otherwise difficult [market de-
Pitofsky, supra note 64, at 1822.
Id. at 1808.
David Scheffman, Malcolm Coate & Louis Silvia, Twenty Years of Merger Guidelines En-
forcement at the FTC: An Economic Perspective, 71 ANTITRUST L.J. 277, 281 (2003) (“[A]lmost
from the beginning, FTC legal staff embraced the DOJ Guidelines as the analytical framework
for merger analysis.”).
U.S. DEP’T OF JUSTICE & FED. TRADE COMM’N, HORIZONTAL MERGER GUIDELINES § 1.11
(April 2, 1992) [hereinafter 1992 HORIZONTAL MERGER GUIDELINES].
See Gregory J. Werden, The 1982 Merger Guidelines and the Ascent of the Hypothetical
Monopolist Paradigm, 71 ANTITRUST L.J. 253, 270-75 (2003) (cataloging, by circuit, lower court
opinions adopting all or part of the HMT).
See Pitofsky, supra note 64, at 1822 (“[B]y focusing on the capacity for the future exercise
of market power, [the HMT asked] a central question that often had been inadequately treated
in the past.”).
See Areeda, supra note 73, at 307 (“[The HMT] correctly relate[s] market definition to the
ultimate legal issue—the prospect that the merging firms will achieve price-raising power or that
the merger will facilitate price coordination among oligopolists.”).
Chicago-School antitrust extended beyond merger control, and even
within merger control it at least recognized the possibility of other types of
But the triumph of the HMT was in connecting mar-
ket definition to the type of joint market power at issue in collusion-facilitation
concerns. When the HMT was extended to other antitrust concerns, it needed
refinements to fit them.
D. Tests based on individual market power
Jump ahead another twenty years, however, and the HMT had morphed
from darling to demon. So devastating was its fall from grace that some com-
mentators wondered aloud whether not just the HMT but all of market defi-
nition was soon to find itself upon the chopping block.
What had changed?
A shift in substantive law had pushed the purposes of market definition out of
alignment with the HMT, at least in certain important applications.
The seeds of this disruption were planted in the 1980s. The simple models
of competition behind Chicago-School antitrust were known from the start to
be oversimplified in some important respects,
and efforts to enrich them had
From the early 1990s to the 2010s, economists revisiting
then derelict antitrust concerns—like predatory pricing and vertical restraints
1982 MERGER GUIDELINES, supra note 76, § III.A.2 (devoting a single paragraph to a dif-
ferent theory of harm under the heading of the “leading firm proviso”).
See Glasner & Sullivan, supra note 40, at 312-24 (discussing the need to customize the
HMT to meet other theories of anticompetitive harm).
Cf. Daniel A. Crane, Market Power Without Market Definition, 90 NOTRE DAME L. REV.
31, 33 (2014) (“[T]he handwriting is on the wall for market definition.”).
See, e.g., Herbert Hovenkamp, Antitrust Policy After Chicago, 84 MICH. L. REV. 213, 256-
64 (1985) (critiquing Chicago-School antitrust as relying too heavily on static models of compe-
tition without strategic considerations); Richard Schmalensee, Another Look at Market Power,
95 HARV. L. REV. 1789, 1793-98 (1982) (illustrating how one influential Chicago-School model’s
implications changed when restrictive assumptions were relaxed or varied).
E.g., Steven C. Salop & David T. Scheffman, Raising Rivals’ Costs, AM. ECON. REV., May
1983, at 267 (illustrating how a dominant firm might profit by strategically raising the produc-
tion costs of its rivals); Louis Kaplow, Extension of Monopoly Power Through Leverage, 85
COLUM. L. REV. 515, <PIN> (1985) (critiquing the persuasiveness of Chicago-School arguments
against antitrust intervention in some leveraging cases); Robert D. Willig, Merger Analysis, In-
dustrial Organization Theory, and Merger Guidelines, 1991 BROOKINGS PAPERS ON ECON. ACTIV-
ITY: MICROECONOMICS 281, 299-305 (1991) (describing a modern unilateral effects model for a
merger of competitors in a differentiated product space).
of trade—would time and again find them more worrying than Chicago-
School antitrust had supposed.
This research also led to the development of
entirely new antitrust concerns.
One such concern was about the unilateral effects of mergers. Recall that
in the coordinated effects focus of Chicago-School antitrust, the problem with
mergers was that they could facilitate joint exercises of market power—help
competitors collude on raising prices. In a unilateral effects focus, the worry is
instead that the elimination of all competition between the merging parties
would directly enable the merged company to individually raise its prices—
even without any collusion or cooperation from its competitors.
A prototypical unilateral effects concern was presented in FTC v. Swedish
Match, a case involving the attempted merger of loose-leaf tobacco sellers Swe-
dish Match and National Tobacco.
In the differentiated product space of
loose-leaf tobacco, comparable prices, flavor profiles, and brand messages
made the tobacco products of Swedish Match and National the best and next-
best options for many consumers. If Swedish Match tried to raise its prices,
many of these consumers would switch to National, and vice versa if National
tried to raise its prices.
The concern presented by the merger of these com-
panies was that the termination of their special rivalry would give the merged
company an individual incentive to raise its prices. Since customers could no
longer punish a price increase by running to the arms of the now-merged rival,
the merged firm now had the freedom to raise its prices a bit.
The incentive to raise prices following a merger of close competitors in a
differentiated product space is intuitive enough, but the magic of the theory is
that it can be mathematically modeled with some basic economic assumptions
about the competitive process.
An economist with adequate data and an
See, e.g., United States v. AMR Corp., 335 F.3d 1109, 1114-15 (10th Cir. 2003) (noting
that “[r]ecent scholarship has challenged the notion that predatory pricing schemes are implau-
sible and irrational” and that “[p]ost–Chicago economists have theorized that price predation is
not only plausible, but profitable, especially in a multi-market context,” thus “we do not [ap-
proach that theory] with the incredulity that once prevailed”); Herbert Hovenkamp, Post-Chi-
cago Antitrust: A Review and Critique, 2001 COLUM. BUS. L. REV. 257, 258 (attributing to Post-
Chicago antitrust a less permissive view of the conduct of dominant firms, a more serious con-
cern for the potential effects of mergers, and a greater willingness to consider the anticompetitive
potential of vertical restraints).
FTC v. Swedish Match, 131 F. Supp. 2d 151, 153-54 (D.D.C. 2000).
Id. at 169.
See generally Margaret Slade, Merger-Simulations of Unilateral Effects: What Can We
Learn from the UK Brewing Industry?, in CASES IN EUROPEAN COMPETITION POLICY: THE
appropriate model of competition can even produce a numeric prediction of
what the unilateral price effects of a merger will be. In Swedish Match, the
FTC’s expert economist testified that “the merger will result in a price increase
of Swedish Match’s loose leaf brands of approximately eleven percent and a
price increase for National’s brands of approximately twenty-one percent.”
The attractiveness of simple numeric predictions in antitrust litigation really
cannot be overstated.
As a consequence, unilateral effects exploded onto the scene.
compiled from FTC investigations, Malcolm Coate reports that unilateral ef-
fects rose from being the primary focus of less than 20 percent of merger in-
vestigations at the start of the 1990s—and presumably something closer to zero
before that—to well over 75 percent of merger investigations by 2010.
same change in focus is reflected in the Merger Guidelines. Where the 1982
Merger Guidelines devoted barely a paragraph to a simple precursor of unilat-
eral effects, the 1992 revisions treated unilateral effects and coordinated effects
in roughly the same detail, and the 2010 revisions now devote twice as much
ECONOMIC ANALYSIS 312, 313-21 (Bruce Lyons ed., 2009) (providing intuition and technical de-
tails); Gregory J. Werden & Luke M. Froeb, Unilateral Effects of Horizontal Mergers, in HAND-
BOOK OF ANTITRUST ECONOMICS 43 (Paolo Buccirossi, ed., 2008) (same); Gregory J. Werden,
Unilateral Competitive Effects of Horizontal Mergers I: Basic Concepts and Models, in 2 ISSUES IN
COMPETITION LAW AND POLICY 1319 (Wayne Dale Collins, ed., 2008) (same); Willig, supra note
102 (providing an early and clear articulation of this approach); Jonathan B. Baker & Timothy
F. Bresnahan, The Gains from Merger or Collusion in Product-Differentiated Industries, 33 J. IN-
DUS. ECON. 427 (1885) (providing what appears to be the first demonstration of this empirical
Swedish Match, 131 F. Supp. 2d at 169.
See Jonathan B. Baker, Why Did the Antitrust Agencies Embrace Unilateral Effects?, 12
GEO. MASON L. REV. 31, 33-36 (2003) (attributing the popularity of unilateral effects theories to
the availability of empirical methods of measuring market power).
Malcolm B. Coate, The Merger Review Process at the Federal Trade Commission from 1989
to 2016, Table 4 (SSRN Working Paper No. 2955987, February 28, 2018), https://ssrn.com/ab-
stract=2955987 (excluding merger-to-monopoly cases in calculating these figures, and so possi-
bly undercounting the rate at which unilateral effects are the primary concern); see also Carl
Shapiro, The 2010 Horizontal Merger Guidelines: From Hedgehog to Fox in Forty Years, 77 AN-
TITRUST L.J. 49, 60 (2010) (“The biggest shift in merger enforcement between 1992 and 2010 has
been the ascendency of unilateral effects as the theory of adverse competitive effects most often
pursued by the Agencies.”).
space to unilateral effects as they do to a breezier account of coordinated ef-
Returning to how this relates to the HMT, one input that is not required in
unilateral effects analysis is the definition of an HMT market. A product of a
time when antitrust was about joint market power—not individual market
power—the HMT simply focused on different issues than unilateral effects.
For economists and practitioners looking for unilateral effects in mergers, time
spent on the HMT was time wasted.
One complaint was that the HMT placed competitors either inside or out-
side a market, with no accounting for degrees of competitive closeness.
had always been true, but its visibility was accentuated by the differentiated
products focus of the new unilateral effects concern. Economists like Joseph
Farrell and Carl Shapiro warned that, in the differentiated-products context,
efforts to delineate markets via the HMT risked allowing outcomes to turn on
“an inevitably artificial line-drawing exercise.”
Another complaint was that the HMT’s indirect path to inferring the im-
plications of a merger was obviated by “direct” estimation of market power in
unilateral effects analysis.
Economists like Dennis Carlton criticized use of
the HMT as a “crude” way of predicting market power.
Farrell and Shapiro
Compare 1982 MERGER GUIDELINES, supra note 76, § III.A.2, with 1992 HORIZONTAL
MERGER GUIDELINES, supra note 94, §§ 2.1–2.2, with 2010 HORIZONTAL MERGER GUIDELINES,
supra note 16, §§ 6–7.
E.g., 2B AREEDA, HOVENKAMP & SOLOW, supra note 41, ¶ 530, at 238 (“This ‘either-or’
nature of market definition can readily be criticized to the extent that compromises between full
inclusion or full exclusion are typically not available.); Mark A. Lemley & Mark P. McKenna, Is
Pepsi Really a Substitute for Coke? Market Definition in Antitrust and IP, 100 GEO. L. REV. 2055,
2098 (2012) (commenting that market definition “draws an arbitrary line when what we need is
a continuum that reflects the partial differentiation of products”).
Joseph Farrell & Carl Shapiro, Antitrust Evaluation of Horizontal Mergers: An Economic
Alternative to Market Definition, 10 B.E. J. THEORETICAL ECON., March 2010, art. 9, at 1, 4.
See, e.g., id. at 2, 5 (suggesting that inferences derived from HMT markets are less direct
than inferences derived from unilateral effects models); see also Malcolm B. Coate & Jeffrey H.
Fischer, Is Market Definition Still Needed After All These Years, 2 J. ANTITRUST ENFORCEMENT
422, 448 (2014) (describing an analytical choice between market definition and direct estimation
of the likely effects of a merger).
Dennis W. Carlton, Market Definition: Use and Abuse, 3 COMPETITION POL’Y INT’L 3, 3
called the HMT “clumsy.”
Louis Kaplow called it “counterproductive,”
and some less flattering things as well.
At previous inflection points in the history of market definition, we saw
frustration with existing tests of market definition herald new methodologies.
Here, dissatisfaction with the HMT arose from its poor performance in iden-
tifying mergers likely to bring about unilateral market power. The price pre-
dictions of unilateral effects models were a ready-made solution for identifying
this type of harm. We might, therefore, guess that this made the prediction of
a price increase by an appropriate unilateral-effects model the new test of mar-
ket definition for this concern.
Unilateral effects predictions did slot into this role, but a rhetorical wrinkle
complicated things. Early proponents of unilateral effects models introduced
this methodology not as a form of market definition, but as a replacement for
Thus, the current merger guidelines declare: “[s]ome of the analytical
tools used . . . to assess competitive effects do not rely on market definition;”
the “[diagnosis of] unilateral price effects based on the value of diverted sales
need not rely on market definition;”
and “[unilateral effect] merger simula-
tion methods need not rely on market definition.”
What justification could
possibly explain the surprising move of treating unilateral effects predictions
as not market definition?
One possibility is that the two-competitor scope of trade bounded by uni-
lateral effects concerns is too narrow to be called a market.
The problem with
this idea is that “relevant market” has long been a term of art in antitrust.
Farrell & Shapiro, supra note 112, at 1.
Louis Kaplow, Market Definition and the Merger Guidelines, 39 REV. IND. ORGAN. 107,
Louis Kaplow, Why (Ever) Define Markets?, 12 HARV. L. REV. 437, 442 (2010) (“useless”);
Louis Kaplow, Market Definition: Impossible and Counter-Productive, 79 ANTITRUST L.J. 361,
367 (2013) (“pointless”); Louis Kaplow, Market Definition Alchemy, 57 ANTITRUST BULL. 915,
926 (2012) (“perverse”).
See Farrell & Shapiro, supra note 112, at 1-2 (proposing one unilateral effects model as
an alternative to market definition); see also supra note 113 (citing sources for the implicit claim
that market definition is not needed merger effects can be directly estimated).
2010 HORIZONTAL MERGER GUIDELINES, supra note 16, § 4, para. 2.
Id. § 6.1, para. 6.
Id. § 6.1, para. 7.
See Hovenkamp, supra note 42, at 908 (commenting that unilateral harm in a differenti-
ated product space “does not fit well into our conception of market definition”).
See United States v. H & R Block, Inc., 833 F. Supp. 2d 36, 50 (D.D.C. 2011).
The HMT, universally understood to be a process of market definition, does
not identify broad and intuitive markets:
it scopes an area of trade in which
a certain type of market power could be exercised. So does a unilateral effects
prediction. In fact, if we take the merging parties as the provisional market in
the HMT, and if we take the unilateral effects prediction as evidence that the
hypothetical monopolist would increase its prices by a small but substantial
amount, then the unilateral effects prediction would validate the merging par-
ties as a relevant market under the very methodology of the HMT. How could
one of these be market definition if the other is not?
Well, maybe precision is the difference. Perhaps the ability of some unilat-
eral effects models to predict specific price effects differentiates this analysis
from the more qualitative market-based inferences supported by something
like the HMT. The problem with this idea is that experienced antitrust practi-
tioners never simply accept the predictions of unilateral effects models.
everything in economics, these models depend on assumptions about human
behavior and the competitive process.
The predictions can be sensitive to
even slight changes in these assumptions.
Of course, the accuracy of assump-
tions can be bolstered by proof that they qualitatively match observed
See 2010 HORIZONTAL MERGER GUIDELINES, supra note 16, § 4, para. 8 (“Relevant anti-
trust markets defined according to the hypothetical monopolist test are not always intuitive and
may not align with how industry members use the term ‘market.’”).
See, e.g., Duncan Cameron, Mark Glick, & David Mangum, Good Riddance to Market
Definition?, 57 ANTITRUST BULL. 719, 734 (2012) (“[O]ne should not confuse the apparent pre-
cision of these models . . . with a tool that will generate accurate and reliable measures of market
power when applied in the complexity of the real world.”).
See supra note 102 (listing references to model parameters and assumptions).
See Philip Crooke, Luke Froeb, Steven Tschantz, & Gregory J. Werden, Effects of Assumed
Demand Form on Simulated Postmerger Equilibria, 15 REV. INDUS. ORGAN. 205, 206-08 (1999)
(observing how demand curvature can substantially affect model predictions); Roy J. Epstein &
Daniel L. Rubinfeld, Understanding UPP, 10 B.E. J. THEORETICAL ECON., May 2010, art. 21, at 1,
8 (observing that “the accuracy and reliability of the [a unilateral effects price-pressure index]
depends crucially on the accuracy of the diversion ratio [parameter]”); Luke Froeb, Steven
Tschantz, & Gregory J. Werden, Pass-Through Rates and the Price Effects of Mergers, 23 INT'L J.
INDUS. ORG. 703, 710-11 (2005) (noting demand curvature sensitivity); Slade, supra note 106, at
331-38 (illustrating the sensitivity of costs, demand systems, and unilateral effects predictions to
various possible modeling assumptions); Gregory J Werden & Luke M Froeb, Choosing Among
Tools for Assessing Unilateral Merger Effects, 7 EUR. COMPETITION J. 155, 158 (2011) (comment-
ing that unilateral effects predictions are valid “only if the model actually captures the essence of
competition in a particular industry, and only if the merger itself does not fundamentally change
how competitors interact”).
behavior, and sensitivity can be addressed by proof that different assumptions
lead to qualitatively similar predictions.
But shuffling the qualitative parts of
the inference around the table doesn’t make them disappear. If the HMT’s re-
liance on qualitative inferences is what earns it the title of market definition,
then unilateral effects predictions deserve that honor as well.
As a final stab, we might consider whether unilateral effects predictions
were strategically carved off from market definition as strategy for directing
generalist judges away from the HMT and other market definition tests when
looking at unilateral effect concerns. Well, if that was the plan, it was not the
result. Judges—who do not see antitrust cases every day—have proven unsur-
prisingly uncomfortable with the idea of simply skipping a step in rule of rea-
son analysis as venerable as defining relevant markets.
And, with unilateral
effects predictions professedly not market definition, they have reached for
tests like the HMT to define these markets—precisely the wrong result.
cannot look at this situation and help but speculate that market definition
might be clearer and more accurate today if unilateral effects predictions had
only been labeled a process of market definition from the start.
Stripped of all the math and rhetoric, the only real difference between the
HMT and unilateral effects predictions is the type of market power at issue.
The HMT defines markets around potential exercises of joint market power.
Unilateral effects predictions define markets around potential exercises of in-
dividual market power. The processes are different, but they are both tests for
defining markets. Each test seeks a scope of trade in which a particular type of
market power might be exercised.
Cf. 2010 HORIZONTAL MERGER GUIDELINES, supra note 16, § 6.1, para. 7 (“The Agencies
. . . place more weight on whether their merger simulations consistently predict substantial price
increases than on the precise prediction of any single simulation.”).
E.g., FTC v. Whole Foods Mkt., Inc., 548 F.3d 1028, 1036 (D.C. Cir. 2008) (objecting that
“Inexplicably, the FTC now asserts a market definition is not necessary . . . in contravention of
the statute itself”); see also City of New York v. Grp. Health Inc., 649 F.3d 151, 155 (2d Cir. 2011)
(making failure to allege a plausible relevant market grounds for dismissal); Queen City Pizza,
Inc. v. Domino's Pizza, Inc., 124 F.3d 430, 436 (3d Cir. 1997) (same).
E.g., United States v. Aetna Inc., 240 F. Supp. 3d 1, 19-21 (D.D.C. 2017); FTC v. Sysco
Corp., 113 F. Supp. 3d 1, 25-38 (D.D.C. 2015); FTC. v. Swedish Match, 131 F. Supp. 2d 151, 159-
60 (D.D.C. 2000).
II. Purpose — The Functions of Market Definition
The history of market definition is like the history of western architecture.
Victorian architectural norms dominated one period; Brutalist aesthetics an-
other; the small-house movement is the most striking fad today. These are all
ways to build a house. But imagine a design document that reads “The façade
should reflect the elegance of the time, but try to stick with just concrete, and
if it doesn’t fit on a trailer then don’t even bother.” Equally absurd mashups
now appear in every description of the standard for defining markets.
return to carpentry tools, this is the hammer analogy at work.
History resists the hammer analogy at every turn. Different tests of market
definition were developed to address different issues. New tests were devel-
oped not to reproduce the same results as earlier tests but to identify different
scopes of trade more helpful to the analytical questions then at issue. In car-
pentry terms, the better analogy is a power drill. One market definition module
after another has been added to the antitrust toolbox, each one designed to
meet a distinct and specific set of needs.
The history of market definition offers several glimpses into how specific
processes of market definition have been developed to meet specific analytical
With both process and purpose changing simultaneously, how-
ever, it can be difficult to make out the underlying patterns. We can see more
if we hold fixed a few common purposes of market definition and compare
different processes of market definition against them. The following picks out
three traditional purposes of relevant markets and explores how different tests
define relevant markets appropriate for serving these purposes in different
contexts. The point of this discussion is not to classify the universe of potential
purposes for defining markets in antitrust analysis.
For the limited goals of
exploring important patterns and providing context for Part III, however, a
discussion centered on these few and overlapping traditional purposes will suf-
See supra notes 24–25.
See supra Part I.
See Glasner & Sullivan, supra note 40, at 296-98 (citing number purposes that have been
attributed to market definition in antitrust analysis).
One traditional purpose of market definition has been to act like a micro-
scope trained upon a specific area of concern. The full, interconnected web of
commerce—of all possible products and technologies and consumptive uses
and trading partners—is simply too big and too overwhelming to provide use-
ful context for antitrust analysis. Market definition responds to this problem
by zooming in on one or more relevant strands of the web, focusing attention
where it is needed and cropping out those peripheral details that would only
end up distracting from the necessary analysis.
The Supreme Court has at times come close to equating market definition
with this magnification purpose. The Court seemed to be alluding to it in FTC
v. Indiana Federation of Dentists when it said that the only purpose of market
definition is to help “determine whether an arrangement has the potential for
genuine adverse effects on competition.”
It had previously come closer in
Philadelphia National Bank, saying that market definition identified the part
of trade where “the effect [of the challenged act] on competition will be direct
These statements are a good start at explaining the micro-
scope purpose of market definition, but they are incomplete in one important
respect: they neglect to explain what strength of magnification we seek.
Is the point to fully bound the range of potential harm? If so, we should
draw markets broadly and magnify weakly, capturing every part of the web
where harm is possible but scooping up lots of irrelevant stuff, too. Or is the
point to clearly identify one area of concern at a time? If that is the case, then
we would draw markets narrowly and magnify strongly, zooming in tight on
one area of concern without necessarily capturing the full range of harm—at
least, not all at once. We need not speculate on this point. From a quick glance
at the market definition modules discussed so far, the typical objective is the
second option: narrow markets and strong magnification.
Of all market definition tests to date, only the early commodity concept
tests followed the weak magnification path. There are reasons why this may
have made sense at the time. Rule of reason analysis was once a vague and
FTC v. Ind. Fed’n of Dentists, 476 U.S. 447, 460 (1986) (according this definition to “the
inquiries into market definition and market power”); see also FTC v. Actavis, Inc., 570 U.S. 136,
153-56 (2013) (returning to the “potential for genuine adverse effects on competition” as a gating
question in determining whether antitrust remedies are available).
United States v. Phila. Nat. Bank, 374 U.S. 321, 357 (1963) (describing definition of the
geographic market in a merger case, but not obviously limiting the principle to this context).
Before reliance on economics began to sharpen focus on in-
dividual actors and market behavior,
there would have been some benefit in
matching the wide prowl of the rule of reason with equally expansive markets.
The Times Picayune and Cellophane tests pursued wide markets, and little
But the early commodity-concept tests are outliers in every sense. In Brown
Shoe, the Supreme Court tellingly dismissed the Times Picayune and Cello-
phane tests as merely identifying the “outer boundaries of a product mar-
And in the decades since Brown Shoe, no test of market definition has
shown the slightest interest in these outer boundaries. Modern tests instead
focus strong lenses on the areas of trade relevant to specific concerns.
Start with Brown Shoe itself. The practical indicia test was a response to
concerns about increasing concentration and the plight of small and local busi-
To guard against rising concentration in industries as popularly un-
derstood, judges needed to be able to identify industries as popularly under-
stood. To protect small and local businesses, judges needed to be able to pick
out groups of businesses whose common interests could be evaluated and pro-
The practical indicia test served up the type of narrow and intuitive
markets needed to address these concerns.
Of course, strong magnification tends to crop out peripheral details. And
in the case of the practical indicia test, this meant that the markets defined by
the test could not purport to scope the full range of potential harm. The public
could recognize a national shoe market while also recognizing city-level mar-
The Supreme Court anticipated this issue in Brown Shoe and addressed
See, e.g., Bd. of Trade of City of Chicago v. United States, 246 U.S. 231, 238-39 (1918)
(prescribing, as the “true test of legality,” a broad factual inquiry without clear standards for
deciding when a restraint of trade was reasonable or unreasonable).
Cf. Kovacic & Shapiro, supra note 26, at 49-52 (describing the late introduction of market
structure inferences in rule of reason analysis).
See supra notes 40–41 and accompanying text.
Brown Shoe Co. v. United States, 370 U.S. 294, 325 (1962).
See supra notes 42–48 and accompanying text.
Brown Shoe observed that Congress had intended “the protection of competition, not
competitors.” Brown Shoe, 370 U.S. at 320. This position is reconcilable with the protection of
small businesses if that protection is required to accrue to the benefit of all small and local busi-
nesses in a recognizable market, not just to specific competitors.
See, e.g., id.. at 325 (identifying a national shoe market for one aspect of a merger and a
series of local shoe markets defined around cities with populations exceeding 10,000 for another
aspect of the merger).
it with an invitation to define multiple relevant markets where necessary.
This has not been necessary as often as one might think, however, since harm
in any single market usually suffices to establish illegality.
The HMT is a very different test, but it closely follows the strong magnifi-
cation pattern of the practical indicia test. Here, the concern is that a merger
could facilitate something like tacit collusion on joint price elevation.
HMT scopes markets around this potential injury.
In so doing, it too adopts
the narrow market and strong magnification approach. Suppose that a merger
of two steel manufacturer-and-fabricator companies risks facilitating collusion
among other manufacturers, other fabrications, or both. The HMT could be
used to validate either or both of these potential collusive rings as relevant
markets. Nested markets are also possible: if joint market power could be ex-
ercised by the competitors in a given market, then it could usually be exercised
by the competitors in any arbitrary expansion of that market as well.
fore, the HMT zooms in on one potential area of harm without prejudice to
other potential areas of harm. Nothing logically prohibits the use of multiple
HMT markets where doing so would be helpful.
Unilateral effects predictions follow the same pattern again. There is no
need to belabor the details. Here, the concern is about individual exercises of
market power. Predicted price increases following the merger of two compet-
itors identify these competitors as capable of exercising that market power.
Again, a tight zoom means that the scope of the market does not exhaust the
scope of potential harm. A merger could, for example, threaten both unilateral
Id. at 325, 336 (commenting that within any broader market there may exist submarkets
that are also appropriate markets for antitrust scrutiny).
See id. at 325 (commenting that “it is necessary to examine the effects of a merger in each
[economically significant market]” because if a probable lessening of competition is found in
any such market, “the merger is proscribed”); United States v. E. I. du Pont de Nemours & Co.
(du Pont-General Motors), 353 U.S. 586, 595 (1957) (similar).
See supra notes 79–81 and accompanying text.
See 2B AREEDA, HOVENKAMP & SOLOW, supra note 41, ¶ 533e, at 275 (“The function of
defining a market [by the HMT] is to determine that grouping of sales that, if controlled by a
single firm or a cartel, could charge noncompetitive prices.”).
See Glasner & Sullivan, supra note 40, at 332-33 (elaborating on this point and providing
an illustrative example); Baker, supra note 64, at 148 and n.68 (similarly observing that relevant
markets may nest or overlap).
See Glasner & Sullivan, supra note 40, at 326-36 (providing an extended defense of the
delineation of multiple relevant markets using something like the HMT).
See supra note 106 (summarizing unilateral effects models).
and coordinated effects. And the pursuit of these different concerns would
benefit from scoping different relevant markets for each concern.
The dominant pattern that emerges from this discussion of the magnifying
role of market definition is one of tight magnification on a specific area of an-
titrust concern. Tests that follow this pattern—like the practical indicia test,
the HMT, and unilateral effects predictions—do not purport to exhaust the
full range of potential harm. Each identifies at least one area of potential harm
without prejudice to the possibility of other areas of potential harm. The point
of these relevant markets is not to identify the single “right” market for a given
case. The point is merely to help courts and litigants focus their attention on
one area of concern at a time.
Another function of market definition is to translate concerns arising from
abstract political and economic theory into statements about the actions of
producers, consumers, and competitors in the world. Competitors emerge,
victims step forward, the heroes and villains of an antitrust narrative come to
life, all within the universe of a relevant market. Resist the urge to write this
off as poetic and trivial. It is not.
Take the markets implicitly defined by unilateral effects predictions. So-
phisticated economic models may be used to produce these predictions. The
jargon alone can be an obstacle to a layperson’s understanding, to say nothing
of the calculus and statistics used to arrive at model predictions, or the body
of assumed principles and prior results upon which the models are based. As
theory goes, this is not a gentle and approachable variety.
But now consider how a unilateral effect prediction translates that theory
into statements about actors in the world. These two companies are close com-
petitors. If they merge, they might raise their prices.
Here are the consumers
See Gregory J. Werden, Why (Ever) Define Markets? An Answer to Professor Kaplow, 78
ANTITRUST L.J. 729, 740–43 (2013) (discussing the narrative role that market definition plays in
antitrust litigation); Steven C. Salop, The First Principles Approach to Antitrust, Kodak, and An-
titrust at the Millennium, 68 ANTITRUST L.J. 187, 191-92 (2000) (observing how market defini-
tion identifies key actors in a theory of harm).
See 2010 HORIZONTAL MERGER GUIDELINES, supra note 16, § 6.1, para. 3 (“Unilateral price
effects are greater, the more the buyers of products sold by one merging firm consider products
sold by the other merging firm to be their next choice.”).
who would feel the pinch of that price increase.
And here are the competitors
best positioned to stop the price increase.
This concrete, tractable narrative
is not a simplistic dumbing down of economic theory. It is a precise translation
of that theory into statement about the interactions of actors in the world. And
that is precisely where attention should be placed. It is the future interactions
of these actors, not abstract model parameters, that will determine whether
competitive injuries arise.
As another example, consider the translations performed by the HMT.
Economic theory paints collusion as an enticing but unstable form of cooper-
ation among competitors.
Every member of a collusive group stands to profit
by joining the collusive scheme but, soon as collusion takes off, every member
of the collusive group also stands to profit by defecting on the arrangement.
The threat of collusion therefore turns on how likely the members of a collu-
sive group are to successfully stabilize their arrangement. Who are the poten-
tial collaborators? Do they have like incentives?
How many of them would
need to cooperate for collusion to succeed?
Evaluating the threat of collusion
involves answering questions like these.
See id., para. 4 (identifying the subset of consumers most likely affected by such a merger).
See id., para. 8 (discussing the potential responses of non-merging competitors). Whether
we call this market structure or entry analysis, the substance is the same. See Coate & Fischer,
supra note 113, at 433 (suggesting that market structure analysis, entry analysis, and reposition-
ing considerations all address common concerns).
See, e.g., Louis Kaplow & Carl Shapiro, Antitrust, in 2 HANDBOOK OF LAW AND ECONOM-
ICS 1073, § 3.2.1 (A. Mitchell Polinsky & Steven Shavell eds., 2007) (summarizing elements of
successful collusion in economic models of oligopoly).
Participation in a joint ten percent price increase sounds great until compared with the
option of poaching customers while others blindly raise their prices.
See generally Jonathan B. Baker, Mavericks, Mergers, and Exclusion: Proving Coordinated
Competitive Effects Under the Antitrust Laws, 77 N.Y. U. L. REV. 135 (2002) (discussing incen-
tive-heterogeneity considerations at length).
See id. at 1112 (“Collusive outcomes are less likely to occur in industries with more firms
because greater numbers make it more difficult to satisfy the . . . conditions necessary for suc-
cessful collusion.”); Ordover & Willig, supra note 84, at 555 (“The view that a reduction in the
number of firms facilitates coordinated use of assets among the incumbent firms is a rock upon
which much of industrial economics has been built.”); George J. Stigler, A Theory of Oligopoly,
72 J. POL. ECON. 44, 55 (1964) (modeling cartel stability as a function of market concentration,
itself a function of the number of important competitors in a market).
Cf. 2010 HORIZONTAL MERGER GUIDELINES, supra note 16, § 7.2 (listing these and many
related factual considerations).
In that vein, recall what the HTM does. It defines a market as a potential
collusive group. The competitors in an HMT market are, by construction, a
group of competitors with the joint market power to collude. To evaluate the
threat of collusion among these competitors, we can count their number and
try to decide which of them would need to cooperate for collusion to suc-
We can also inspect the operations of individual competitors in trying
to look for differences or similarities in incentives to collude. Here again, mar-
ket definition translates abstract economic theory into factual questions about
actual actors in the world. And market definition really is where this transla-
tion takes place. If the relevant market had been defined by anything other
than the HMT, then the number and incentives of competitors in the relevant
market would not generally correspond to the threat of potential collusion
among these competitors.
The previous examples concern antitrust evaluations of potential changes
in market power, but the translation function of market definition extends to
other sociopolitical objectives as well. In the era of Warren Court antitrust, for
example, changes in market concentration were not opposed because they
might affect specific forms of market power. The inference was broader and
stronger than that.
To politicians and economists of the time, and to many
who still subscribe to their viewpoint today, increasing concentration was not
evidence of a problem—increasing concentration was the problem.
Brown Shoe’s practical indicia test translates an abstract political prefer-
ence for deconcentrated industries into factual terms administrable by courts.
Scoping markets around recognizable industries facilitates the measurement
HMT markets may encompass the smallest group of competitors needed to bring about
a joint price increase, but this is not a guaranteed property of these markets.
E.g., George J. Stigler, Mergers and Preventive Antitrust Policy, 104 U. PA. L. REV. 176,
181-82 (1955) (describing a tight connection between industrial concentration and effectiveness
of competition); see Baker, supra note 156, at 138 (The dominant and largely unquestioned view
among economists and antitrust commentators [at this time] was that when only a few firms
competed in an industry, they readily would find a way to reduce rivalry, collude tacitly, and
raise prices above the competitive level); cf. CARL KAYSEN & DONALD F. TURNER, ANTITRUST
POLICY: AN ECONOMIC AND LEGAL ANALYSIS 132-36 (1959) (suggesting the presumptive illegal-
ity of any merger resulting in a firm with more than a twenty percent share of the market).
See, e.g., Brown Shoe Co. v. United States, 370 U.S. 294, 315 (1962) (noting “fear of what
was considered to be a rising tide of economic concentration in the American economy”); id. at
346 (“We cannot avoid the mandate . . . that tendencies toward concentration in industry are to
be curbed in their incipiency . . . .”); id. at 345 n.72 (concluding that Congress sought “to prevent
even small mergers that added to concentration in an industry”).
of concentration in terms familiar to legislators and lay people. If the objective
is to satisfy the public’s interest in controlling concentration as the public per-
ceives it, then there can be no substitute for defining relevant markets accord-
ing to public perception.
In each of these examples, market definition translates abstract concerns
into concrete statements about actors in the world. While the details differ
from one context to the next, the basic function of every translation is the
same: the relevant market is defined so that it is populated with the actors
whose interactions are most important to evaluating the threat of a given in-
jury. Antitrust cases are in every sense about the characters and controversies
that emerge in these translations.
Another purpose of market definition is to provide a context for exploring
counterfactual aspects of antitrust concerns.
Take entry analysis, as an ex-
ample. The possibility of future entry is nearly always an issue in antitrust lit-
igation. Without barriers to entry, potential market power gains are ephem-
eral, since any attempt to exercise that power would attract others to compete
for the newly profitable transactions, eventually unwinding any market power
gains through increased competition. Fact finders must therefore ask whether
and how quickly new competitors would enter a relevant market in response
to an exercise of market power—a question which presumes the context of a
In this and other ways, market definition facilitates the ex-
ploration of antitrust concerns.
See supra notes 69–70 and accompanying text.
This is often described as an “organizing” function of relevant markets. See SULLIVAN,
supra note 87, at 64 (“[T]he only purpose for defining a market is to organize available data in a
way which facilitates judgment about the extent of that power.”); Franklin M. Fisher, Economic
Analysis and “Bright-Line” Tests, 4 J. COMPETITION L. & ECON. 129, 130 (2008) (“Market defini-
tion can be a useful tool, a way to begin organizing the material that must be studied.”).
See Fisher, supra note 163, at 131 (“Ease of entry must also be considered, and one might
reasonably say that such a consideration requires one to know what it is that is being entered.”);
Werden, supra note 150, at 729 (“Even if antitrust analysis never used market shares, the relevant
market would remain essential for examining entry prospects and the durability of market
power.”); see also Crane, supra note 100, at 48 (questioning how entry can be assessed “in a ‘di-
rect’ market power analysis since entry barriers require identification of a market into which
entry is difficult”).
What other types of exploration does it facilitate? Fully evaluating the
threat of post-merger tacit collusion requires courts to engage a long list of
factual inquiries. Philip Areeda once summarized the challenge as follows:
[Market structure inferences] do not purport to be determinative
but are to be considered along with ease of entry, degree of prod-
uct homogeneity, next closest products or producers excluded
from the market definition, buyer concentration, information
availability or exchanges, economic performance, prior disrup-
tiveness of a merging firm, and such practices as price protection
clauses, product standardization, delivered pricing, past collu-
sion, and other matters affecting the ease of tacit price coordina-
He could have kept going. Historic market stability can be an important con-
as can things like the frequency of contact between competitors
across different markets
and the way that a merger changes the incentives of
It is hard to miss the connection between these factors and the surrounding
context of a relevant market defined by the HMT. Many of these inquiries are
intuitive and sensible only within the context of such a market. Moreover,
these inquiries would not be facilitated by relevant markets defined according
to other tests. For purposes of the assessing the “degree of product homogene-
ity” or the “prior disruptiveness of a merging firm,” the narrow relevant
Areeda, supra note 73, at 309.
See Edward J. Green & Robert H. Porter, Noncooperative Collusion under Imperfect Price
Information, 52 ECONOMETRICA 87, 90-91 (1984) (relating this and other aspects of competitive
structure to the feasibility of self-enforcing collusion); see also Subhasish M. Chowdhury & Car-
sten J. Crede, Post-Cartel Tacit Collusion: Determinants, Consequences, and Prevention, 70 INT’L
J. OF INDUS. ORG., May 2020, at 1 (discussing experimental evidence on how prior success at
collusion may similarly facilitate coordination).
See B. Douglas Bernheim and Michael D. Whinston, Multimarket Contact and Collusive
Behavior, 21 RAND J. ECON. 1 (1990) (suggesting how multimarket contact may facilitate collu-
sion); Federico Ciliberto & Jonathan W. Williams, Does Multimarket Contact Facilitate Tacit
Collusion? Inference on Conduct Parameters in the Airline Industry, 45 RAND J. ECON. 764 (2014)
(providing empirical evidence on this relationship).
See Baker, supra note 156, at 166-77 (describing different ways that a merger may facili-
tate collusion by changing the incentives of one of the merging parties).
markets defined by unilateral effects predictions would be just as inapt as the
industry concepts scoped by the practical indicia test.
On that point, note that Brown Shoe’s practical indica test serves different
but still recognizable exploration purposes. The structuralist objectives that
motivated antitrust in Brown Shoe’s time were linked to the understanding of
industries as permanent fixtures in the world. Locating analysis within a spe-
cific industry allowed courts to consider the surrounding context of that in-
dustry when evaluating the structural effects of mergers. The apparent trajec-
tories of industries were matters of special importance: “only . . . examination
of the particular market—its structure, history and probable future—can pro-
vide the appropriate setting for judging the probable anticompetitive effect of
But other features of the industry mattered as well.
The practical indicia test responded to “Congress’ express intent” that mer-
gers be assessed “within an industry framework almost inevitably unique in
by defining relevant markets to align with recognizable and per-
sistent industry concepts. Again, the test is unique in its satisfaction of this
purpose. Few of the relevant contextual inquiries make much sense outside of
the industry concepts sought by the practical indicia test. A potential collusive
group, defined as a relevant market by the HMT, would be little help in trying
to assess the trajectory of an industry, and the narrow market scoped by a uni-
lateral effects prediction would be simply worthless.
Exploration functions are, however, performed by unilateral effects predic-
tions in the different context of evaluating unilateral effects concerns. The need
for exploration might come as a bit of a surprise in view of the apparent
See 2010 HORIZONTAL MERGER GUIDELINES, supra note 16, § 7.2 (describing the many
market-based factors that are used to assess vulnerability to coordinated conduct); Philip
Areeda, Market Definition and Horizontal Restraints, 52 ANTITRUST L.J. 553, 562-63 (1983)
(drawing a similar distinction between the scope and purpose of relevant markets defined for
assessing coordinated conduct and relevant markets defined for assessing unilateral or monop-
oly market power).
Brown Shoe Co. v. United States, 370 U.S. 294, 322 n.38.
See, e.g., United States v. Cont’l Can Co., 378 U.S. 441, 458-66 (1964) (discussing various
features of the metal can and glass industries and how a merger would change the nature of
competition in these industries); see also United States v. Bethlehem Steel Corp, 168 F. Supp.
576, 583 (S.D.N.Y. 1958) (“The contending positions of the parties can be understood only
against the background and general pattern of the iron and steel industry, the making and dis-
tribution of steel and steel products, the nature, size and location of the companies in the indus-
try, the nature of competition in the industry generally . . . .”).
Brown Shoe, 370 U.S. at 322 n.38.
precision of these models in directly predicting price effects.
But the literal
accuracy of unilateral effects predictions is always limited to the toy models of
competition they assume.
The more that actual patterns of competition di-
verge from these models, the more unilateral effects predictions work like ed-
To be blunt, in empirical studies to date, unilateral effects
models have not had great success in predicting observed behavior.
This does not undermine the usefulness of unilateral effects predictions. It
simply means that these predictions are the beginning of competitive effects
analysis, not the end of it. Further exploration is always needed. As Shapiro
explains, “measuring upward pricing pressure, or even performing a full mer-
ger simulation, typically is not the end of the story” because “[r]epositioning,
entry, innovation, and efficiencies must also be considered.”
When the po-
tential price responses of other firms are not reflected in the underlying model,
these too must be considered in evaluating the potential for unilateral market
See supra note 113 and accompanying text.
See Cameron, Glick, & Mangum, supra note 125, at 734 (warning not to confuse the pre-
dictions of a restricted economic model with accurate statements about complex, real-world
markets); see also Franklin M. Fisher, Games Economists Play: A Noncooperative View, 20 RAND
J. ECON. 113, 115 (1989) (decrying economic testimony “that one should analyze real markets by
using [simple models of competition]” as “theory run riot”).
See supra notes 126–128 and accompanying text.
See Craig Peters, Evaluating the Performance of Merger Simulation: Evidence from the U.S.
Airline Industry, 49 J. L. & ECON. 627, 627 (2006) (reporting that “standard simulation methods,
which measure the effect of the change in ownership on unilateral pricing incentives, do not
generally provide an accurate forecast”); Dennis W. Carlton & Mark Israel, Will the New Guide-
lines Clarify or Obscure Antitrust Policy?, ANTITRUST SOURCE, Oct. 2010, 1, at 4 (“[T]here is only
weak empirical evidence establishing the usefulness of merger simulation as a tool to predict
anticompetitive mergers.”). See generally Jonas Björnerstedt & Frank Verboven, Does Merger
Simulation Work? Evidence from the Swedish Analgesics Market, 8 AM. ECON. J. 125 (2016) (re-
porting some successes, but also several respects in which merger simulation failed to adequately
explain the apparent price and share effects of an observed merger); Lars Mathiesen, Øivind Anti
Nilsen, & Lars Sørgard, A Note on Upward Pricing Pressure: The Possibility of False Positives, 8 J.
COMPETITION L. & ECON. 881 (2012) (illustrating false positives in UPP analysis); Matthew C.
Weinberg, More Evidence on the Performance of Merger Simulations, 101 AM. ECON. REV., May
2011, at 51 (reporting a retrospective study in which merger simulations substantially underpre-
dicted the actual estimated price effects of a merger); see also Douglas D. Davis & Bart J. Wilson,
Differentiated Product Competition and the Antitrust Logit Model: An Experimental Analysis, 57
J. ECON. BEHAV. & ORG. 89, 91 (2005) (describing uninspiring experimental results.).
Carl Shapiro, Update from the Antitrust Division, Remarks as Prepared for the American
Bar Association Section of Antitrust Law Fall Forum 26 (Nov. 18, 2010), https://www.jus-
power to be gained and exercised. The appropriate scope of trade for exploring
these questions is precisely the implicit market defined by the predicted exer-
cise of market power. Like all other tests of market definition, unilateral effects
predictions expose a context in which relevant analysis can be helpfully per-
Across each of these different applications, market definition performs the
same basic function. It marks out a context in which the various issues relating
to a concern may be explored and addressed. This exploration function could
be trivial—in a tautological sense, any relevant market provides a context for
exploring every inquiry—but the right connection between the choice of test
and the application at hand elevates it to something far more valuable.
Helpful relevant markets facilitate exploration by matching the market to
the needs of the inquiry. A market defined by the HMT is helpful context for
exploring joint market power concerns. A market defined by a unilateral price
effect prediction is helpful context for exploring individual market power con-
cerns. Neither market is typically a helpful context for the other’s concern. Just
like the functions of market definition discussed above, the utility of a relevant
market for exploration purposes derives from using the right test for the right
III. Practice — Modular Market Definition
Everything discussed so far points toward a simple, two-step solution for
choosing the appropriate market definition test in any application. First, iden-
tify the substantive purposes for which relevant markets are being defined.
Second, identify the test that delineates markets best suited to serving those
purposes. This modular approach is undoubtedly different from current prac-
tice, but it is not a departure from binding precedent. If anything, the modular
approach to market definition is more faithful in its adherence to precedent
than any recent cases have been.
The modern approach—treating every test of market definition as some-
how interchangeable—is a false account of antitrust history. As illustrated in
Part I, even a brief review of the different contexts in which the different tests
were developed reveals the unremarkable fact that different tests of market
definition were developed to serve different purposes. It is true that several
market concepts have served some broadly similar purposes in their respective
contexts. Examples of this, like the magnification function of relevant markets,
were discussed in Part II. But as Part II also illustrated, the ability of a market
to serve these purposes derives from the connection between a specific test and
its intended application.
Modular market definition accepts and exploits the differences between
different tests of market definition. It anticipates that the most helpful test will
be selected for every application. How does this work in practice? That is easier
to show than tell. The following illustrates how a modular approach to market
definition quickly identifies the details of appropriate market definition tests
across a variety of situations.
A. Coordinated effects enabling tacit collusion
In a standard coordinated effects theory, the concern is that a merger will
allow the exercise of joint market power through tacit collusion. Put another
way, the worry is that the members of a possible collusive group will, because
of the merger, become capable of colluding on things like joint price increases.
As discussed before, the HMT was made to fit this concern.
The HMT, im-
plemented by hypothesizing the joint exercise of market power as a small but
substantial price increase over prevailing prices, is the right test for defining
markets in this context.
With that said, many of the fussier details of the HMT as described in the
2010 Horizontal Merger Guidelines can be omitted. The Guidelines talk about
expanding markets to fill perceived gaps in the set of products in the market,
a heuristic sometimes called the “circle principle.”
They also talk about
shrinking markets to match the smallest group of competitors that could
Note, in passing, that the modular approach to market definition tailors tests to concerns,
not cases. Were a case to raise more than one concern, it would typically be important to define
different relevant markets for each concern. This approach of defining different relevant markets
for different concerns was employed by the Supreme Court in Brown Shoe. See Brown Shoe Co.
v. United States, 370 U.S. 294, 324-28, 335-39 (1962) (conducting separate market definition for
the vertical and horizontal concerns raised by the same merger). It enjoys theoretical as well as
practical justifications. See Glasner & Sullivan, supra note 40, § III (providing an extended justi-
fication for a multiple market paradigm).
See supra notes 96–97 and accompanying text.
2010 HORIZONTAL MERGER GUIDELINES, supra note 16, § 4.1.1, para. 4 (suggesting the
inclusion of products in a market when those products are perceived to be interstitial to other
products already included in the market).
E.g., United States v. Aetna Inc., 240 F. Supp. 3d 1, 37 (D.D.C. 2017).
a heuristic sometimes called the “smallest market princi-
As descriptions of how the agencies choose to structure their internal
analysis, these heuristics are questionable judgement calls but nothing more.
That tolerance does not extend to treating the heuristics as immutable parts of
the market definition process.
A stark illustration of this point is provided by the recent case of FTC v.
The FTC complained, in this case, that a merger of two hydro-
gen peroxide producers risked facilitating tacit collusion among other produc-
ers of commodity hydrogen peroxide.
This is a simple theory, and it begs a
simple market. Do producers of commodity hydrogen peroxide have the joint
market power to raise prices if they cooperate to do so? If so, then that is a
relevant market under the HMT.
True, the HMT might validate other rele-
vant markets as well, but the availability of alternative relevant markets does
not and cannot preclude consideration of any other market validated by the
HMT. Non-exclusivity is a property of all strong-magnification tests.
The district court missed this point in Rag-Stiftung, in part because it con-
fused some of the Merger Guidelines’ heuristics with the test of market defini-
Treating the smallest market principle and some language about
supply-side substitution as immutable principles of market definition,
court erroneously rejected the simple relevant market that fit the FTC’s com-
plaint, and instead substituted relevant markets defined so narrowly that data
were apparently unavailable for them.
This prevented the court from
2010 HORIZONTAL MERGER GUIDELINES, supra note 16, § 4.1.1, para. 5 (“[W]hen the
Agencies rely on market shares and concentration, they usually do so in the smallest relevant
market satisfying the hypothetical monopolist test.”).
E.g., United States v. H & R Block, Inc., 833 F. Supp. 2d 36, 59 (D.D.C. 2011).
FTC v. Rag-Stiftung, 436 F. Supp. 3d 278 (D.D.C. 2020).
Administrative Part 3 Complaint at 2-3, Evonik/PeroxyChem, No. 191 0029 (2020),
See supra notes 87–99 and accompanying text.
See supra Part II.A.
See RAG-Stiftung, 436 F. Supp. 3d at 292–300.
See id. at 292-93 (treating the smallest market principle as a rule of market definition); id.
at 293-300 (plucking language from explanatory footnotes and asides to create a novel frame-
work for deciding when it would be permissible to accept a broad market if narrower markets
Id. at 303, 310.
reaching the merits of the FTC’s complaint.
Wrongly defined markets re-
pelled, rather than facilitated, the substantive analysis that the case required.
All that is needed to avoid this type of error is to start market definition
with a clear understanding why markets are being defined. The concern, in a
standard coordinated effects theory, is that a group of competitors could co-
ordinate to raise prices. The HMT validates markets in which competitors have
the joint market power to do just that. If the members of a candidate market
appear not to satisfy the test—because they are too few or too easily replaced
to exercise joint market power—then expansion of the candidate market is
necessary. But, if the candidate market satisfies the HMT, it simply makes no
difference that some other broader or narrower group of competitors could
also exercise joint market power.
If broader or narrower markets could also
satisfy the HMT criteria, then all that we have learned is that these, too, may
constitute scopes of trade in which the threatened harm is theoretically possi-
ble. Alternative HMT markets can add to the list of potential scopes of con-
cern; they can never subtract from it.
B. Coordinated effects entrenching tacit collusion
In a different version of the coordinated effects theory, tacit collusion has
already taken hold among a group of competitors and the concern is that a
merger will entrench the ongoing exercise of joint market power. Put another
way, the threatened harm is not the exercise of new joint market power, but
Id. at 310 (describing lack of evidence as dispositive). The court did comment briefly, in
dicta, on the feasibility of coordinated effects, but this analysis, too, was obstructed by the court’s
market definition error. See id. at 312 (admitting that “the Court’s review is necessarily limited”
by the unavailability of evidence related to the narrow relevant markets that the court chose to
See supra notes 185–187 and accompanying text.
See supra note 147 and accompanying text.
the stabilization of old joint market power.
Vetting entrenchment theories
requires market analysis.
What test should be used to define those markets?
Because it scopes markets around joint market power concerns, the HMT
is again the place to start. But applying the HMT by hypothesizing a price in-
crease over prevailing prices now leads to strange results. Suppose that ongo-
ing collusion has already raised prices as high as they profitably can go. Ap-
plying the HMT to these elevated prices reveals that the collusive group lacks
the joint market power to raise prices further (which is true) but would then
move on to expand the market on the assumption that the collusive group
lacks joint market power (which is false).
Worse than wrong, this brings us
back to the fox and henhouse problem:
the higher the collusive group man-
ages to raise its prices, the wider the market becomes, and the more freedom
the members of this group have to lock in their successful collusion through
These problems trace to a mismatch between the usual articulation of the
HMT and the substantive concern in entrenchment theories. The usual artic-
ulation identifies a scope of trade in which new joint market power could be
exercised, but entrenchment is about old joint market power.
is that ongoing collusion may be set in stone by a merger involving members
of an already collusive group. This can happen without any new market power
One way to define markets around the entrenchment concern would be to
run the HMT against an estimate of the competitive price, asking whether the
See Areeda, supra note 169, at 564 (commenting that “[m]erger precedents have been
concerned not only with combinations creating new power but also with those reinforcing pre-
sent power” and that “a merger which reinforces pre-existing monopoly or oligopoly pricing”
may be anticompetitive); 2010 HORIZONTAL MERGER GUIDELINES, supra note 16, § 1, para. 5
(“[M]ergers should not be permitted to create, enhance, or entrench market power or to facilitate
See generally Sean P. Sullivan, Anticompetitive Entrenchment, 68 U. KAN. L. REV. 1133
(2020) (discussing entrenchment theories of harm in merger enforcement).
See Salop, supra note 150, at 194 (describing similar interpreting errors under the labels
of the “Cellophane trap” and the “Price-Up trap”).
See text following note 41.
See Sullivan, supra note 195, at 1136 (contrasting “the usual claim that the merger or
acquisition would make things worse in the relevant market” with the entrenchment concern
that market power is already being exercised in the relevant market and the “corresponding pub-
lic interest in preserving opportunities for the grind of competitive frictions to make things bet-
ter in the future”).
members of the apparently collusive group would have the joint market power
to raise prices above a reasonable guess at what the price would be but for their
More directly, we could try to run the HMT in reverse,
seeking a scope of trade in which the defection of one of the merging parties
could destabilize ongoing coordination. These are complex inquiries, but they
are not insuperable.
In Valspar Corp. v. E.I. Du Pont De Nemours & Co., the Third Circuit re-
cently affirmed summary judgement against a plaintiff’s price fixing claim,
reasoning that the members of a titanium dioxide oligopoly were tacitly col-
luding without express agreement to do so.
The Seventh Circuit reached a
similar conclusion in In re Text Messaging Antitrust Litigation, reasoning that
the small number of competitors offering text messaging services could easily
engage in tacit collusion,
and even reasoning that an email offered as proof
of collusion failed to do so because it “rather clearly refers to tacit collusion.”
The same evidence that persuades courts of tacit collusion in these cases
should suffice to prove narrow relevant markets around the tacitly colluding
groups in an entrenchment theories of coordination.
C. Concerted Conduct
The seemingly concerted conduct of competitors often raises joint market
power concerns. In some cases, this conduct trips a per se rule of illegality.
See Glasner & Sullivan, supra note 40, 319-24 (discussing this point as an example of a
broader need to match the HMT to the specifics of alleged collusion); Salop, supra note 150, at
197 (“Using the lower price that would prevail in the absence of the alleged anticompetitive con-
duct as the competitive benchmark is appropriate whenever it is alleged that a restraint will pre-
vent price from falling to a lower level.”).
Valspar Corp. v. E.I. Du Pont De Nemours & Co., 873 F.3d 185, 200 (3d Cir. 2017) (“It
makes sense that each firm would implement [parallel pricing] strategies, since conscious paral-
lelism allows firms in an oligopoly to in effect share monopoly power and maintain prices at a
profit-maximizing, supracompetitive level.” (internal quotation marks omitted)).
In re Text Messaging Antitrust Litig., 782 F.3d 867, 871 (7th Cir. 2015) (“[T]he fewer the
firms, the easier it is for them to engage in “follow the leader” pricing (“conscious parallelism,”
as lawyers call it, “tacit collusion” as economists prefer to call it) . . . .”).
Id. at 872.
United States v. Socony-Vacuum Oil Co., 310 U.S. 150, 228 (1940) (“[P]rice-fixing com-
binations which lack Congressional sanction are illegal per se; they are not evaluated in terms of
their purpose, aim or effect in the elimination of so-called competitive evils.”); Nat’l Soc. of Pro.
Engineers v. United States, 435 U.S. 679, 692 (1978) (identifying per se illegality as applying to
In all other cases, market definition may be an important step in evaluating
the anticompetitive potential of the arrangement.
Because the focus is on
exercises of joint market power, there are overlaps between market definition
in this setting and in coordinated effects analysis. There are also differences,
tracing to how alleged conspiracies direct analysis.
As a concrete example, consider Todd v. Exxon Corp., in which the plaintiff
alleged that a group of oil companies was using an information exchange to
suppress employee wages.
In describing how the plaintiff might seek to
prove market power in her case, then-Judge Sotomayor identified two alterna-
tive paths that could be pursued. One path involved defining a relevant market
and establishing that the defendants held a large share of this market. The
other involved producing evidence of observable anticompetitive effects.
These are well-supported alternatives in antitrust law,
but their apparent dis-
tinction is curious. If market definition is done right, what would be the dif-
ference between each path?
Start with the market definition option. The plaintiff complained that
Exxon, Mobil, B.P., and others were exchanging information in order to sup-
press wages. If wages were already suppressed as a result, then a helpful market
definition test would ask whether the alleged group of companies would have
the joint market power to suppress wages relative to where they would be but
for the ongoing information exchange.
This is the conduct-analogue of the
agreements “so plainly anticompetitive that no elaborate study of the industry is needed to es-
tablish their illegality”).
See Cal. Dental Ass’n v. FTC, 526 U.S. 756, 779 (1999) (commenting that even when a
per se rule would ultimately be applied “considerable inquiry into market conditions” may be
required to justify the per se rule).
Todd v. Exxon Corp., 275 F.3d 191, 195 (2d Cir. 2001).
Id. at 206 (“If a plaintiff can show that a defendant's conduct exerted an actual adverse
effect on competition, this is a strong indicator of market power. In fact, this arguably is more
direct evidence of market power than calculations of elusive market share figures.”).
See, e.g., FTC v. Ind. Fed’n of Dentists, 476 U.S. 447, 460-61 (1986) (noting that reduced
output “can obviate the need for an inquiry into market power” through market definition).
See Salop, supra note 150, at 197 (noting that if it is alleged “that certain conduct has
already permitted a firm to raise its price,” then “the proper competitive benchmark is not the
current price [but] the lower price that would have prevailed absent the alleged restraint”). Put
another way, the Cellophane fallacy, widely recognized as complicating market definition in the
monopolization context, applies equally in the concerted action context. Cf. supra note 41 (dis-
cussing the Cellophane fallacy).
HMT as applied to an entrenchment theory of coordination.
could be answered by circumstantial evidence of market power, but the most
compelling proof would be evidence of actual wage suppression.
Now consider the direct proof of effects option. Here, the plaintiff presents
evidence that the defendants’ conduct was resulting in observable wage sup-
Unless this ends the inquiry, it still may be helpful to identify one
or more relevant markets in which further analysis could be conducted—entry
barriers, business justifications, and similar inquiries being difficult to study
in a vacuum.
But, as we have just discussed, evidence of actual effects is itself
compelling proof that those engaged in the challenged conduct have the joint
market power to constitute a relevant market under the HMT.
Put another way, when market definition is properly conducted, the mar-
ket definition and direct proof inquiries ask the same questions and lead to the
same results. When proof of effects is unavailable, market definition is the only
path forward. But when proof of effects is available, the two coincide. Proof of
actual anticompetitive effects does not obviate market definition; proof of ac-
tual effects subsumes market definition.
Now, what happens when the challenged conduct has not yet had time to
cause actual effects? Suppose the oil companies in Todd had only just begun to
swap information, for example. Here, market definition asks whether the co-
conspirators would have the joint market power to bring about future wage
suppression. This is the conduct-analog of the HMT as applied to standard
Prevailing prices—here, wages—are the appropriate
baseline for the tests, and the feasibility of the exercise of joint market power
can only be tested by circumstantial evidence.
See supra Part III.A.2.
E.g., Todd, 275 F.3d at 214 (alleging ongoing salary suppression and specific reductions
in wage competition over prior years).
See supra Part II.C (discussing markets as test labs).
See generally Crane, supra note 100 (discussing direct inferences of market power from
conduct); John B. Kirkwood, Market Power and Antitrust Enforcement, 98 B. U. L. REV. 1169,
1196 (2018) (elaborating on how and when market power can be identified from conduct).
See supra notes 118–130 (similarly interpreting unilateral effects predictions).
See supra Part III.A.1.
D. Undifferentiated-product unilateral effects
Returning to merger analysis, one type of unilateral effects concern strad-
dles the line between individual and joint market power. The setting is an un-
differentiated product space—a context in which competing products are al-
most perfectly interchangeable—and the concern is that a merger of
competitors will create a firm with a large enough market share to be able to
profit by curtailing its own production. The artificial scarcity that this curtail-
ment creates would drive up the price of all producers.
Hence, this is an ex-
ercise of joint market power. But the merged firm collects enough of the profits
to find curtailment worthwhile even without the cooperation of its competi-
tors. Hence, this is unilateral conduct.
In economic models of this type of unilateral effect, features of a market
like the relative shares of the merging parties and the likely supply responses
of non-merging rivals inform the seriousness of concern.
How should mar-
kets be defined when trying to identify these features?
While this is a theory of unilateral conduct, the effect is still an exercise of
joint market power, and so the HMT is again the appropriate test for the job.
In fact, this is one setting in which some of the fussier details of market defi-
nition as described in the 2010 Horizontal Merger Guidelines may be applica-
Since the theory of harm is an automatic price increase propagated
across all producers by a supply retraction, scoping the impact of the price
increase, and how much of the profit would return to the merged firm, may
well require filling in gaps and identifying sufficiently narrow markets in this
See 2010 HORIZONTAL MERGER GUIDELINES, supra note 16, § 6.3, para. 1 (describing con-
cern that “the merged firm will find it profitable unilaterally to suppress output and elevate the
market price” (emphasis added)).
See 2010 HORIZONTAL MERGER GUIDELINES, supra note 16, § 6.3 (discussing this concern
in greater detail).
See id., para. 2 (listing relevant factors). See generally Joseph Farrell & Carl Shapiro, Hor-
izontal Mergers: An Equilibrium Analysis, 80 AM. ECON. REV. 107 (1990) (describing price and
welfare effects of merger in this context); R. Preston McAfee & Michael A. Williams, Horizontal
Mergers and Antitrust Policy, 40 J. INDUS. ECON. 181 (1992) (similar); Luke M. Froeb & Gregory
J. Werden, A Robust Test for Consumer Welfare Enhancing Mergers Among Sellers of a Homoge-
neous Product, 58 ECON. LETTERS 367 (1998) (similar).
If the merger is still unconsummated, then the prevailing price is the appropriate baseline.
See, e.g., supra notes 180–183.
This form of unilateral effect concern is not often pursued today, but recent
cases like FTC v. Tronox confirm that it is still an important part of the en-
When undifferentiated-product unilateral effects theo-
ries are pursued, the HMT is the appropriate test for defining relevant markets.
E. Differentiated-product unilateral effects
Today, the far more common unilateral effects concern involves a unilat-
eral price increase arising from a merger of close competitors in a differenti-
ated product space. Think, for example, of a merger of tobacco companies or
soft-drink manufacturers. Here, the concern is not that a merger will facilitate
the exercise of joint market power by all competitors but that it will facilitate
the exercise of individual market power by the merging parties.
priate test of market definition in this setting is the credible prediction of a
unilateral price increase in an appropriate model of the competitive process.
The appropriate test is not the HMT or Brown Shoe’s practical indicia test.
True, unilateral effects cases have been successfully argued despite their reli-
ance on relevant markets defined by these tests.
But this is cold comfort if it
merely signals that cases with unilateral effects concerns are only being
brought when they happen to exhibit impressive concentration in relevant
markets defined by these inappropriate tests. When the relevant concern is
about a potential exercise of individual market power, defining relevant mar-
kets around hypothesized exercises of joint market power offers at best a blurry
view of the relevant scope of trade. And the contribution of the practical indi-
cia test is next to nothing.
The one important exception to this rule is a special case of unilateral ef-
fects analysis in which market shares in a broader relevant market substitute
Fed. Trade Comm'n v. Tronox Ltd., 332 F. Supp. 3d 187, 208-12 (D.D.C. 2018) (finding
the post-merger firm would have “meaningful market incentives to manage prices by withhold-
ing TiO2 supply”); see also Administrative Part 3 Complaint at 2, Superior/Canexus, No. 161
0020 (2016), https://www.ftc.gov/system/files/documents/cases/160627superiorcanex-
uscmpt.pdf (alleging that the merger “would increase the likelihood of future anticompetitive
output reductions to increase price”).
See supra notes 104–107, 118–130 and accompanying text.
See supra notes <NOTE #s> and accompanying text.
E.g., F.T.C. v. Swedish Match, 131 F. Supp. 2d 151, 160 (D.D.C. 2000) (applying both the
practical indicia and HMT tests to define the relevant market for a differentiated-product uni-
lateral effects concern); Fed. Trade Comm'n v. Sysco Corp., 113 F. Supp. 3d 1, 26-37 (D.D.C.
2015) (same); United States v. Aetna Inc., 240 F. Supp. 3d 1, 20-21 (D.D.C. 2017) (same).
for more precise measures of the closeness of competitors.
There are serious
theoretical limitations to this approach, and these typically limit share-based
unilateral effects predictions to the shallow waters of data scarcity and time
But, when this path is taken, the HMT is a reasonable process
for scoping a market in which shares could at least potentially identify fre-
quency of head-to-head competition. The practical indicia test claims no such
Monopolization is a vast subject. The different versions of offense—mo-
nopolize, attempt to monopolize, conspiracy to monopolize—are distinct
enough to warrant separate treatment.
Atop that division, the subject further
divides by type of concern: acquisition of monopoly power and maintenance
of existing monopoly power are different enough to warrant separate treat-
And atop that subdivision the additional complexity emerges that ex-
clusionary conduct encompasses various different acts and consequences. Rel-
evant markets and market shares may connect in different ways to the
anticompetitive harm threatened by exclusive dealing, predatory pricing, re-
fusals to deal, and the other ways in which a dominant firm might seek to ex-
clude rivals or raise their costs.
The analysis may also depend on questions
of timing and focus, as where concerns relate to the exclusion of potential com-
petitors, or where concerns involve changes in innovation or research inten-
In short, monopolization is not one antitrust concern, but really a factorial
expansion encompassing many related but different concerns. Rather than
See, e.g., Willig, supra note 102, at 299-305 (explaining this approach).
See Jerry Hausman, 2010 Merger Guidelines: Empirical Analysis, ANTITRUST SOURCE 1
(Oct. 2010) (noting that this approach requires the IIA property, which is “unrealistic in many
situations”); Willig, supra note 102, at 301 (commenting that the assumptions behind this ap-
proach “are unlikely to be valid in many areas of application”).
See 15 U.S.C. § 2 (1994).
See United States v. Grinnell Corp., 384 U.S. 563, 570–71 (1966) (affirming that the of-
fense of monopolization applies to both the acquisition of monopoly power and the maintenance
of monopoly power through exclusionary conduct).
See generally 3 PHILIP E. AREEDA & HERBERT HOVENKAMP, ANTITRUST LAW ch. 7 (2015).
See JONATHAN B. BAKER, THE ANTITRUST PARADIGM: RESTORING A COMPETITIVE ECON-
OMY 152-175 (2019) (discussing innovation harms arising from mergers and exclusionary con-
attempt what would necessarily be a cursory treatment of the individual ele-
ments of this expansion,
we can consider three overarching questions that
inform any use of market definition in a monopolization context.
First, what are relevant markets being used for? A firm’s possession of a
large share of a relevant market is commonly interpreted as evidence that the
firm possesses monopoly power in that market.
And a firm’s market share
in a relevant market is also commonly used to infer something about the firm’s
incentive to engage in exclusionary conduct—certain forms of exclusion only
making sense when undertaken by firms of sufficiently large relative size.
Note, however, that nothing logically requires these two inferences to use the
same relevant market. It may often be helpful to delineate separate relevant
markets for each inference.
Second, is the relevant concern about the acquisition, possession, or
maintenance of monopoly power? Markets defined for purposes of assessing a
firm’s possession or maintenance of monopoly power face the challenge of the
Cellophane fallacy: current substitution patterns will overstate the significance
of competitive constraints—and thus understate the firm’s actual market
power—whenever a firm is already exercising the market power it possesses.
For initial sketches of a comprehensive treatment, see Sean P. Sullivan, Market Definition,
in 1 RESEARCH HANDBOOK ON ABUSE OF DOMINANCE AND MONOPOLIZATION (Pinar Akman,
Konstantinos Stylianou, & Or Brook eds., Edward Elgar forthcoming); Lawrence J. White, Mar-
ket Power and Market Definition in Monopolization Cases: A Paradigm Is Missing, in 2 ABA
SECTION OF ANTITRUST LAW, ISSUES IN COMPETITION LAW AND POLICY 913 (W. Dale Collins ed.,
2008). For closely related discussion of the meaning of exclusion and monopoly power, see
Thomas G. Krattenmaker, Robert H. Lande & Steven C. Salop, Monopoly Power and Market
Power in Antitrust Law, 76 GEO. L.J. 241, <PIN> (1987); Thomas G. Krattenmaker & Steven C.
Salop, Anticompetitive Exclusion: Raising Rivals’ Costs to Achieve Power over Price, 96 YALE L.J.
209, <PIN> (1986).
See, e.g., supra notes 7–11 and accompanying text (describing the identification of mo-
nopoly power by market share in Alcoa).
See HERBERT HOVENKAMP, FEDERAL ANTITRUST POLICY: THE LAW OF COMPETITION AND
ITS PRACTICE 110–11 (5th ed. 2016) (citing predatory pricing, foreclosure, and tying as examples
of exclusionary conduct offenses “that are plausible only [when] the defendant occupies a large
portion of the relevant market in question”).
See Sullivan, supra note 230, § 3.1 (discussing the use of separate relevant markets to
assess existing market power and conduct threatening to increase that market power).
See supra notes 41 and 196 and accompanying text (discussing the interpretive error of
evaluating competition at prices already reflecting the exercise of market power); Sullivan supra
note 230, § 2.2 (observing that all substitution-based tests of market definition are subject to this
This is the same problem encountered in defining HMT markets for entrench-
ment theories of coordinated conduct.
Just as in the entrenchment context,
a correction is needed to define markets relative to something like substitution
patterns at competitive baseline prices when using relevant markets for these
purposes. Importantly, that correction is not needed when relevant markets
are being used to assess potential increases in a firm’s market power.
the Cellophane fallacy applies to a relevant market defined for purposes of as-
sessing an alleged monopolist’s current market power but does not apply to
relevant markets defined for purposes of assessing the same firm’s ability to
increase its market power through exclusionary conduct.
Third, what understanding of monopoly power will the relevant markets
be used to evaluate? One might suppose that this had long ago been settled but
consider the current state of the law. The most common definition of monop-
oly power, today, is “something greater than [normal] market power.”
it is also said that monopoly power is “the power to control prices or exclude
Recent outcries against the threat of technological monopolies
echo something of past interpretations of monopoly power as defined by in-
sufficient availability of alternatives
or as companies having attained sizes
“great enough to cause just anxiety on the part of those who love their country
more than money.”
Each of these understandings of monopoly power focuses on something
different, and a test of market definition appropriate to one understanding of
monopoly power may be ill-suited to another. For the significant market
power interpretation, a version of the HMT could suffice. That test would not
be much help, however, if monopoly power was defined as having a dominant
share of a recognizable industry or as the mere ability to exclude competi-
For these, the practical indicia test would seem to offer the better
See Part III.B.
See Sullivan, supra note 230, § 3.1 (elaborating on this point in the additional context of
attempted monopolization claims).
Eastman Kodak Co. v. Image Tech. Servs., Inc., 504 U.S. 451, 481 (1992).
United States v. E. I. du Pont de Nemours & Co. (Cellophane), 351 U.S. 377, 391 (1956).
United States v. Aluminum Co. of Am., 148 F.2d 416, 427 (2d Cir. 1945) (describing the
illegality of monopolization as reflecting a preference for “a system of small producers . . . [over]
one in which the great mass of those engaged must accept the direction of a few”).
N. Sec. Co. v. United States, 193 U.S. 197, 407 (1904) (Holmes, J., dissenting).
Here, mere injury to competitors contrasts with an expectation that injury to competitors
would translates into harm to consumers. See generally Thomas G. Krattenmaker & Steven C.
approach. And if monopoly power was defined by the absolute size or asset
value a company alone, then market definition would be entirely unnecessary.
Establishing the possession of monopoly power would be an accounting exer-
These questions only begin to address the appropriate use of market defi-
nition in monopolization cases. If this seems complicated, it is because the
substantive law is complicated. Modular market definition selects the test of
market definition by looking to the analytical needs of the substantive law. And
monopolization currently encompasses a sprawling and under-theorized col-
lection of different concerns.
If market definition is to be reliably helpful in
evaluating these concerns, then it must start from first principles in every
case—asking what the underlying concern is and what needs to be addressed
to evaluate that concern.
G. Structuralist Concerns
We have yet to address the type of structuralism that was prominent in
1960s antitrust, and for good reason. It has been a long time since antitrust last
fought for the retention of unconcentrated industries and the protection of
But while these concerns have not moved courts in several
decades, the history of antitrust law is one of shifting norms and policy goals,
and recent events may signal the return of structuralism, perhaps even protec-
tionism, to the antitrust stage.
Bearing the short title “Consolidation Prevention and Competition Pro-
motion Act of 2019,”
a bill introduced by Senator Klobuchar in 2019 recited
that “unprecedented consolidation is reducing competition and threatens to
Salop, Anticompetitive Exclusion: Raising Rivals’ Costs to Gain Power over Price, 96 YALE L.J. 209
(1986) (discussing in detail the different possible standards of exclusion).
Cf. Thomas E. Kauper, Section Two of the Sherman Act: The Search for Standards, 93 GEO.
L.J. 1623 (2005) (describing, with exasperation, more than a century of uncertainty about the
standard of illegality in monopolization cases).
See supra note 72 and accompanying text.
See William E. Kovacic, The Modern Evolution of U.S. Competition Policy Enforcement
Norms, 71 ANTITRUST L.J. 377, 400 (2003) (commenting on the inevitability of evolution and
change in competition policy and enforcement norms); Robert Pitofsky, The Political Content of
Antitrust, 127 U. PA. L. REV. 1051, 1052 (noting attention to “political” concerns throughout
most of antitrust history). See generally Kovacic & Shapiro, supra note 26 (discussing the historic
evolution of antitrust norms).
S. 307, 116th Cong. § 1 (2019).
place the American dream further out of reach for many consumers in the
As discussed previously, another bill recently introduced by
the Senator would erect presumptions of illegality for mergers and exclusion-
ary conduct undertaken by firms with more than a fifty percent share of any
Meanwhile, a recent House subcommittee report urges
greater congressional attention to “increased market concentration in our
and references with apparent approval a recommendation that
Congress should “investigate factors which tend to . . . injure small business . .
. or promote undue concentration of economic power . . . .”
Calls to break
up and restructure large businesses now come from commentators in positions
of authority in the administration and federal agencies.
The result of these initiatives remains to be seen. If the outcome is indeed
the return to antitrust of a strong form of structuralism concerned with con-
or a type of the protectionism that seeks to defend small
businesses against larger rivals even if doing so means lower economic effi-
then this will have follow-on effects in market definition.
Simply put, tests like the HMT are inappropriate for pursuing concerns
about industry-level concentration or the protection of small businesses.
Scopes of trade designed to focus on possible exercises of joint market power
do not exhibit systematic relationships with either of these concerns. Instead,
the return of this type of structuralism to antitrust would necessitate a corre-
sponding return to relevant markets defined by something like Brown Shoe’s
Id. § 2.
S. 225, 117th Cong. §§ 4, 26A (2021).
STAFF OF H. SUBCOMM. ON ANTITRUST, COM. & ADMIN. L. OF THE COMM. ON THE JUDICI-
ARY, 116TH CONG., INVESTIGATION OF COMPETITION IN DIGITAL MARKETS 7 (Comm. Print 2020)
[hereinafter DIGITAL MARKETS REPORT].
Id. at 7–8.
E.g., TIM WU, THE CURSE OF BIGNESS: ANTITRUST IN THE NEW GILDED AGE 132-33 (2018)
(suggesting the use of structural breakups as a means of increasing competition in certain areas
of trade); Elizabeth Warren, Here’s How We Can Break Up Big Tech, MEDIUM (Mar. 8, 2019),
ilar); Lina M. Khan, The Separation of Platforms and Commerce, 119 COLUM. L. REV. 973 (2019)
(proposing to restrict platform hosts from vertically integrating into the sale of products, so that
separate entities would need to engage in each activity).
See Herbert Hovenkamp, The Looming Crisis in Antitrust Economics, B.U. L. REV., 43-44
(forthcoming 2021) (critiquing at least one bill as “focused far too much on increased concen-
tration or absolute size for their own sake”).
See supra notes 46–49 and accompanying text.
practical indicia test.
This is not to claim that the practical indicia test of the
1960s perfectly fits these concerns. The point is simply that markets scoped by
popular perception are the appropriate setting for addressing concerns about
concentration as popularly perceived and for pursuing related protectionist
It bears emphasis that nothing in this Article is a proposal for changing
antitrust law. No legislation is required to implement any part of this ap-
proach. Indeed, nothing stops any court from adopting the modular approach
to market definition today. True, modular market definition differs from cur-
rent practice, and its adoption would mark a departure from recent precedent.
But the only innovation of this Article is to give due weight to the differences
between different tests of market definition—to honestly acknowledge that
different tests were intentionally developed to address different concerns. Eve-
rything else flows from that. To the extent that this Article proposes a depar-
ture from recent precedent, it is only because recent precedent has itself de-
parted from faithful interpretation and application of earlier cases and
The details of the modular approach to market definition have been illus-
trated in enough detail above to obviate yet another summary here. Instead, I
prefer to close with four broad suggestions for antitrust practice that follow
from this modular approach to market definition. These suggestions are closer
to proposals for changing antitrust law.
First, because proper market definition is an inherently flexible exercise,
courts should take care not to inject Hobson’s choices into its practice in the form
of rules that restrict markets to fit one specific mold or none at all.
One example to avoid is the claim of Newcal Industries v. Ikon Office Solu-
tion that markets “must encompass the product at issue as well as all economic
substitutes for the product.”
Judges in the Ninth Circuit have spent over a
decade declaring defective, under this rule, any market that arguably omits
See supra notes 67–71 and accompanying text.
Newcal Indus., Inc. v. Ikon Office Sol., 513 F.3d 1038, 1045 (9th Cir. 2008).
Whether the inclusion of these omitted substitutes
would actually change anything about the evaluation of dispositive issues in
the case appears to be an academic question under this rule.
Another example is the Hobson’s choice that antitrust plaintiffs often face
in merger litigation. When seeking to enjoin mergers, the government now
defines markets by the HMT even when the complaint turns on nothing but
unilateral effects predictions,
and even when those HMT markets have noth-
ing to do with its substantive basis for seeking an injunction.
ment does this for the simple reason that many judges expect markets to be
defined this way.
Here, no less than in the Ninth Circuit, we see a single
process of market definition being forced upon every application.
The errors in the Newcal rule and current merger enforcement are plain to
see when looked at from the perspective of modular market definition.
ful relevant markets are defined by matching the process of market definition
with the purposes that relevant markets are meant to serve. Rules rigidly dic-
tating process without regard to purpose are guaranteed to get things wrong.
Second, because different relevant markets relate to their associated con-
cerns in different ways, care should be taken not to assume that the correspond-
ence between features of one market and its concern apply to other markets and
See, e.g., AFMS LLC v. United Parcel Serv. Co., 105 F. Supp. 3d 1061, 1077-78 (C.D. Cal.
2015), aff'd sub nom. AFMS LLC v. United Parcel Serv., Inc., 696 F. App’x 293 (9th Cir. 2017) ;
Payment Logistics Ltd. v. Lighthouse Network, LLC, No. 3:18-CV-00786-L-AGS, 2018 WL
5311907, at *3 (S.D. Cal. Oct. 24, 2018).
See, e.g., Shapiro, supra note 177 (reassuring listeners that DOJ “recognizes the necessity
of defining a relevant market as part of any merger challenge we bring”).
Compare 2010 HORIZONTAL MERGER GUIDELINES, supra note 16, § 6.1, ¶¶ 6-7 (disclaim-
ing the need to define markets in unilateral effects analysis), with Shapiro, supra note 109, at 56
(disclaiming the ability to skip market definition “when going to court”).
See supra notes 129–130 and accompanying text.
The Newcal error is especially pronounced as it appears to intend that the expansive rel-
evant markets sought by old commodity concepts tests be used in every context. Amazingly, the
lonely authority that Newcal produces for its rule is the very page of Brown Shoe—the full page
is cited—in which the Court first excused markets from matching the wide breadth of the early
commodity concept tests. Newcal, 513 F.3d at 1045 (citing Brown Shoe v. United States, 370 U.S.
294, 325 (1962)).
A concrete illustration of this point is provided by the tortured history of
market concentration inferences in antitrust law.
From about the 1950s to
the 1960s, industrial organization economists reported what seemed to be
strong relationships between industrial concentration and various measures of
market power. This research was soon attacked as suffering from measure-
ment problems (market power is hard to measure) and causality issues (con-
centration can actually result from market power).
Decades later, only ech-
oes of the initial claims remain. Few today would deny that there is a positive
but hard-to-generalize relationship between concentration and price in many
but the typical soundbite is that economists have not established
more than a “weak” empirical relationship between concentration and market
Looked at from the perspective of modular market definition, this whole
project is a curiosity. Modular market definition acknowledges that there are
different ways of defining markets.
It also acknowledges that there are dif-
ferent types of market power.
The unstated assumption in the research and
antitrust conversation seems to be that market concentration and market
power must admit some one-size-fits-all relationship. But if different tests of
market definition connect different versions of market concentration to dif-
ferent types of market power, then what exactly are we averaging?
this perspective, weakness in the observed relationship between concentration
See generally Richard Schmalensee, Inter-industry Studies of Structure and Performance,
in 2 HANDBOOK OF INDUSTRIAL ORGANIZATION 951 (Richard Schmalensee & Robert Willig eds.,
1989) (providing a critical review of this literature).
See Steven Berry, Martin Gaynor & Fiona Scott Morton, Do Increasing Markups Matter?
Lessons from Empirical Industrial Organization, 33 J. ECON. PERSP. 44, 46-47 (2019) (summariz-
ing similar concerns); Jonathan B. Baker & Timothy F. Bresnahan, Economic Evidence in Anti-
trust: Defining Markets and Measuring Market Power, in HANDBOOK OF ANTITRUST ECONOMICS
1, 24-25 (Paolo Buccirossi ed., 2008) (same).
See Schmalensee, supra note 260, at 988 (summarizing the literature as supporting the
stylized fact: “In cross-section comparisons involving markets in the same industry, seller con-
centration is positively related to the level of price”).
See, e.g., Carlton, supra note 114, at 4 (“Unfortunately, there is only a weak link between
change in market share and change in competitive performance . . . .”).
See supra Part I.
See supra Part II.
Cf. Berry, Gaynor & Scott Morton, supra note 261, at 45, 47-48 (noting that there is no
single causal relationship between concentration and price, but actually a series of different re-
lationships, a subset of which may be applicable in any given context).
and market power could owe as much to muddled market thinking as it does
to any actual absence of economic relationships in the data.
Third, safe harbors based on market structure thresholds should be limited
to specific concerns or not used at all.
Ever since 1982, the merger guidelines have set aside certain mergers as
unlikely to require serious analysis. These unproblematic mergers are identi-
fied by their small size or effect on concentration in a market defined by the
Mergers not meeting minimum concentration thresholds are rarely
investigated at all—the thresholds are thus safe harbors or quasi-safe harbors
depending on who you ask.
Looked at from the perspective of modular market definition, it is aston-
ishing that this practice has remained undisturbed for so long. Back when the
minimum concentration threshold was first put in place, merger review was
essentially concerned with only coordinated effects analysis.
At that time, a
merger unlikely to lead to significant concentration in any HMT market prob-
ably did not raise competitive concerns.
Today, however, unilateral effects
concerns are by far the bigger emphasis in government enforcement efforts.
And a merger may create unilateral market power without creating any risk of
So, why is concentration in a market designed only to
assess coordinated conduct still being used as a plenary screen for enforce-
See 1982 MERGER GUIDELINES, supra note 76, § III.A.1.a; 1992 HORIZONTAL MERGER
GUIDELINES, supra note 94, §§ 1.0, 1.51; 2010 HORIZONTAL MERGER GUIDELINES, supra note 16,
§ 5.3, para. 6.
Cf. FED. TRADE COMM’N & U.S. DEP’T OF JUSTICE, HART-SCOTT-RODINO ANNUAL REPORT
5–6 (FY 2019), https:// www.ftc.gov/system/files/documents/reports/federal-trade-commis-
p110014hsrannualreportfy2019_0.pdf (reporting that only about two to three percent of merger
notifications were subject to requests for additional information and documentary material in
See supra notes 79–81 and accompanying text.
See 1982 MERGER GUIDELINES, supra note 76, § III.A.1.a (commenting that “implicit co-
ordination among firms is likely to be difficult [in unconcentrated markets]”).
See supra notes 111–117 and accompanying text.
Cf. Shapiro, supra note 109, at 69 (commenting that despite the increased emphasis on
unilateral effects concerns, the “DOJ continues to apply the HHI thresholds to all horizontal
None of the obvious justifications are satisfying. This practice could be de-
fended as economizing on government budgets and as increasing the predict-
ability of merger enforcement.
But the same could be said of any underin-
clusive test. Should we turn all patients away at the hospital doors if they do
not present with high blood pressure?
This practice could also be defended
on the theory that narrowly drawn HMT markets may approximate unilateral
effects predictions in many cases. But why torture HMT markets to serve this
role when more appropriate tests are already available for unilateral effects
In the end, there is no salvaging a rule that ignores one antitrust concern
on the basis of evidence against another, different concern. Screens based on
market concentration in HMT markets are a principled way of disposing co-
ordination concerns relating to those relevant markets. But screens based on
market concentration in one HMT market are not a principled way of dispos-
ing concerns relating to any other relevant market. At a time when many are
looking for opportunities to strengthen merger enforcement,
tion of this strange and unprincipled practice should be a high priority.
Fourth, market definition should be expected and allowed to continue to
evolve with changes in the substantive law.
Market definition could not have survived for so many decades, and en-
dured so many changes in antitrust policy, if it did not have the ability to
change and grow with the underlying law.
That flexibility is especially im-
portant today. Giants of the tech world now face scrutiny and challenge on
Matters of long-term innovation and privacy competition
See John Kwoka, The Structural Presumption and the Safe Harbor in Merger Review: False
Positives or Unwarranted Concerns?, 81 ANTITRUST L.J. 837, 844-45 (2017) (summarizing the
history and policy goals of safe harbor provisions in merger analysis).
See Salop, supra note 150, at 191-98 (critiquing of the idea that market definition acts as
a preliminary “filter” for ruling out competitive effects).
See supra notes 118–130 and accompanying text.
E.g., BAKER, supra note 229, at 3-7, 209; Carl Shapiro, Antitrust: What Went Wrong and
How to Fix It, ANTITRUST MAGAZINE (forthcoming).
See supra Part I.
See, e.g., Complaint for Injunctive and Other Equitable Relief, Federal Trade Commission
v. Facebook, Inc., No. 1:20-CV-03590-JEB (D.D.C. Jan. 13, 2021) ; Complaint, United States v.
Google, LLC, No. 1:20-CV-03010 (D.D.C. Oct. 20, 2020). See generally DIGITAL MARKETS RE-
PORT, supra note 248 (outlining cases against Amazon and Apple as well).
increasingly sound in antitrust.
Opposing philosophies are entering into a
battle over control of the next few decades of antitrust policy.
clouds portend little certainty in this area of law and the need for flexibility in
all tools of analysis.
Modular market definition responds to that need. What are the boundaries
of the general search