United States of America et al v. American Express Company et al.
Filing
619
DECISION (Findings of Fact and Conclusions of Law). For the reasons set forth in the accompanying Decision, the court concludes that Plaintiffs have proven by a preponderance of the evidence that the challenged restraints constitute an unlawful restraint on trade under Section 1 of the Sherman Act. Following briefing by the parties in accordance with the Scheduling Order issued contemporaneously with this Decision, the court will separately issue a Remedial Order and a Judgment after it determines the appropriate remedy. Ordered by Judge Nicholas G. Garaufis on 2/19/2015. (Day, Lori)
UNITED STATES DISTRICT COURT
EASTERN DISTRICT OF NEW YORK
---------------------------------------------------------------------X
UNITED STATES OF AMERICA, et al.,
Plaintiffs,
DECISION
-against-
10-CV-4496 (NGG) (RER)
AMERICAN EXPRESS COMPANY and
AMERICAN EXPRESS TRAVEL RELATED
SERVICES COMPANY, INC.,
Defendants.
---------------------------------------------------------------------X
NICHOLAS G. GARAUFIS, United States District Judge.
TABLE OF CONTENTS
INTRODUCTION………. ........................................................................................................................... 2
FINDINGS OF FACT AND CONCLUSIONS OF LAW…. ....................................................................... 7
I.
BACKGROUND ............................................................................................................................... 7
A.
Overview of the GPCC Card Industry .............................................................................................. 8
B.
Competition and Pricing in the GPCC Card Industry ..................................................................... 14
C.
The Non-Discrimination Provisions ............................................................................................... 21
1.
Origins of Amex’s NDPs ............................................................................................................ 21
2.
The Challenged Restraints .......................................................................................................... 25
II.
LEGAL STANDARD ..................................................................................................................... 33
III.
MARKET DEFINITION................................................................................................................. 37
A.
GPCC Card Network Services Market ........................................................................................... 39
1.
2.
B.
IV.
The Relevant Product Is Network Services ................................................................................ 39
Debit Network Services Are Not Reasonably Interchangeable .................................................. 45
Plaintiffs’ Proposed T&E Submarket ............................................................................................. 61
MARKET POWER ......................................................................................................................... 65
A.
Market Share, Concentration, and Barriers to Entry ....................................................................... 67
B.
Cardholder Insistence...................................................................................................................... 71
1
C.
Pricing Practices.............................................................................................................................. 78
1.
2.
V.
Price Discrimination ................................................................................................................... 83
3.
D.
Value Recapture .......................................................................................................................... 79
Merchant Pricing Premium ......................................................................................................... 84
Amex’s Remaining Market Power Counterarguments ................................................................... 90
ADVERSE EFFECTS ON COMPETITION .................................................................................. 98
A.
The NDPs Impede Horizontal Interbrand Competition ................................................................ 100
B.
The NDPs Block Low-Cost Business Models .............................................................................. 107
C.
The NDPs Have Resulted in Higher Prices to Merchants and Consumers ................................... 111
D.
The NDPs Stifle Innovation .......................................................................................................... 115
E.
Removal of the NDPs Would Benefit Merchants and Consumers ............................................... 116
VI.
PRO-COMPETITIVE JUSTIFICATIONS ................................................................................... 127
A.
Defendants’ Ability To Drive Competition .................................................................................. 128
B.
Free-Riding ................................................................................................................................... 143
VII.
CONCLUSION ............................................................................................................................. 150
INTRODUCTION
The United States and the attorneys general of seventeen states 1 (collectively, “Plaintiffs”
or the “Government”) bring this antitrust enforcement action against Visa Inc. (“Visa”),
MasterCard International Incorporated (“MasterCard”), American Express Company, and
American Express Travel Related Services Company, challenging each network’s anti-steering
rules as anticompetitive restraints in violation of Section 1 of the Sherman Antitrust Act.
(Compl. (Dkt. 1).) Visa and MasterCard entered into consent decrees with the Government,
pursuant to which they voluntarily agreed to remove or revise the bulk of their challenged
restraints. (See Final J. as to Defs. MasterCard Int’l Inc. & Visa Inc. (Dkt. 143).) Defendants
American Express Company and American Express Travel Related Services Company
1
Originally, the United States and eighteen states brought this action, which was filed in 2010, but Hawaii
stipulated to the dismissal of its claims without prejudice before trial. (Stip. & Dismissal Without Prejudice of the
Claim of the State of Hawaii (Dkt. 104).)
2
(collectively, “Defendants,” “American Express,” or “Amex”) elected to litigate Plaintiffs’
challenge to their anti-steering rules, which they term American Express’s Non-Discrimination
Provisions (the “NDPs”). The NDPs, which are contained in both Defendants’ standard
acceptance agreement and also the more customized agreements they negotiate with a select
number of large merchants, prevent the roughly 3.4 million merchants who accept American
Express credit and charge cards from steering customers to alternative credit card brands, such as
Visa, MasterCard, and Discover.
Before turning to the contractual restraints at issue in this case, it is helpful to outline the
type of behavior that Defendants’ NDPs are intended to prevent. As a general matter, steering is
both pro-competitive and ubiquitous. Merchants routinely attempt to influence customers’
purchasing decisions, whether by placing a particular brand of cereal at eye level rather than on a
bottom shelf, discounting last year’s fashion inventory, or offering promotions such as “buy one,
get one free.” This dynamic, however, is absent in the credit card industry. Under American
Express’s NDPs, a merchant may not attempt to induce or “steer” a customer to use the
merchant’s preferred card network by, for example, offering a 10% discount for using a Visa
card, free shipping for using a Discover card, or a free night at a hotel for using an American
Express card.
Each time a customer uses a credit card, the merchant, in one way or another, pays a fee
to the network services provider that facilitates the customer’s purchase. Thus, when a customer
uses a Visa credit card, the merchant pays some combination of fees, commonly known as the
“discount rate” or the “merchant discount rate,” for the privilege of accepting that card. When a
customer uses an American Express card, the merchant similarly pays a fee. However, the
merchant’s cost of accepting American Express—one of the three largest network services
3
providers in the country—has tended to be greater than the cost of accepting other cards, such as
Visa or MasterCard. To speak in generalities that are perhaps unwarranted given the extensive
trial record in this case, all else being equal, a given merchant might prefer that a customer
carrying both a Visa card and an Amex card in her wallet use the Visa card, since the cost of the
transaction is likely to be lower for the merchant. But pursuant to Amex’s NPDs, merchants who
accept American Express are not permitted to encourage customers to pay for their transactions
with credit cards that cost the merchants less to accept.
As explained below, these NDPs create an environment in which there is nothing to
offset credit card networks’ incentives—including American Express’s incentive—to charge
merchants inflated prices for their services. This, in turn, results in higher costs to all consumers
who purchase goods and services from these merchants.
The court does not come to its decision in this case eagerly or easily. The credit card
industry is complex, and it is a critical component of commerce in the United States. General
purpose credit and charge (“GPCC”) card networks, including American Express, must balance
the demands of two sets of customers—merchants and cardholders—in a market that is highly
concentrated and distorted by a history of antitrust violations. The court recognizes that it does
not possess the experience or expertise necessary to advise, much less dictate to, the firms in this
industry how they must conduct their affairs as going concerns. For that reason, the court has
repeatedly urged the parties in this case to negotiate a mutually agreeable settlement that
appropriately balances American Express’s legitimate business interests with the public’s interest
in robust interbrand competition. However, the parties having failed to do so, the court is left
with no alternative but to discharge its duty by deciding the question before it: whether Plaintiffs
have shown by the preponderance of the evidence that Amex’s NDPs violate the U.S. antitrust
4
laws. Upon consideration of the case law in this circuit and the factual record developed at the
lengthy bench trial, which was held over a seven-week period during the summer of 2014 and
featured over thirty fact witnesses and four expert witnesses, the court finds that Plaintiffs have
made such a showing.
As noted, credit card networks cater to the needs of two distinct sets of consumers,
merchants and cardholders. Their very function is to bring these two sides together to
consummate value-generating transactions. Guided by the Second Circuit’s 2003 decision in
United States v. Visa, 344 F.3d 229 (2d Cir. 2003), which conducted an antitrust market analysis
in this industry to resolve a public enforcement action initiated by the Department of Justice
under Section 1, the court agrees with Plaintiffs that this two-sided platform comprises at least
two separate, yet deeply interrelated, markets: a market for card issuance, in which Amex and
Discover compete with thousands of Visa- and MasterCard-issuing banks; and a network
services market, in which Visa, MasterCard, Amex, and Discover compete to sell acceptance
services. For the reasons described herein, the court concludes that the relevant market for its
antitrust analysis in this case is the market for GPCC card network services. Notwithstanding
Defendants’ vigorous arguments to the contrary, the dramatic growth in customers’ use of debit
cards in the decade since Visa has not rendered obsolete the market definitions used in that case,
nor does it warrant debit’s inclusion in the relevant antitrust market. Indeed, both anecdotal
merchant testimony and the testimony of Plaintiffs’ economics expert presented at trial result in
the court’s determination that debit cards have not become reasonably interchangeable with
GPCC cards or network services in the eyes of credit-accepting merchants, who are the relevant
consumers in this case. Plaintiffs’ attempt to define a submarket for GPCC card network
services provided to merchants in travel and entertainment industries, however, is unavailing.
5
In reaching its decision, the court applies a full rule of reason analysis that takes stock of
the voluminous evidentiary record developed at trial, and also considers and accounts for the
interrelationships between the merchant and cardholder sides of the credit card platform. A few
of the court’s primary findings bear mentioning here. First, American Express possesses
sufficient market power in the network services market to harm competition, as evidenced by its
significant market share, the market’s highly concentrated nature and high barriers to entry, and
the insistence of Defendants’ cardholder base on using their American Express cards—insistence
that prevents most merchants from dropping acceptance of American Express when faced with
price increases or similar conduct. The record demonstrates, in fact, that Defendants have the
power to repeatedly and profitably raise their merchant prices without worrying about significant
merchant attrition. In addition, Plaintiffs have proven that American Express’s NDPs have
caused actual anticompetitive effects on interbrand competition. By preventing merchants from
steering additional charge volume to their least expensive network, for example, the NDPs shortcircuit the ordinary price-setting mechanism in the network services market by removing the
competitive “reward” for networks offering merchants a lower price for acceptance services.
The result is an absence of price competition among American Express and its rival networks. In
fact, the record shows that merchant prices have risen dramatically in the absence of merchant
steering. Defendants’ NDPs also have foreclosed the possibility of a current network or a new
entrant to the market differentiating itself from its competitors by pursuing a lowest-cost
provider strategy. Finally, the court has carefully considered American Express’s proffered procompetitive justifications and finds them to be insufficient to render the NDPs permissible under
Section 1.
6
FINDINGS OF FACT & CONCLUSIONS OF LAW
I.
BACKGROUND
This public enforcement action was tried without a jury before this court over a seven-
week period between July 7, 2014, and August 18, 2014. Over the course of the proceedings the
court received testimony from over thirty fact witnesses, including nearly twenty merchant
witnesses representing a selection of the nation’s largest retailers, airlines, and hotels;
representatives from each of the three major credit card networks that compete with American
Express; and an array of current and former American Express employees and executives,
including the company’s long-time Chairman and Chief Executive Officer, Kenneth Chenault.
Plaintiffs additionally presented expert testimony from Dr. Michael Katz, a Professor of
Economics at the Haas School of Business at the University of California at Berkley, and Dr.
Gary Ford, a Professor Emeritus at the Kogod School of Business at American University.
Professors Richard J. Gilbert, Ph.D., also of the University of California at Berkley, and
B. Douglas Bernheim, Ph.D., a Professor of Economics at Stanford University, provided expert
testimony on behalf of American Express. In addition to the testimony adduced at trial, which
amounts to nearly 7,000 transcript pages, the court received over 1,000 exhibits into the
evidentiary record. In reaching a determination as to Defendants’ liability under Section 1 of the
Sherman Act, the court has considered carefully the relevance of and appropriate weight to
afford this evidence, as well as the credibility of the parties’ respective witnesses and their
testimony. 2
2
On the final day of testimony, August 18, 2014, the court held the evidentiary record open to allow the parties to
resolve certain evidentiary disputes concerning summary evidence they wished to include in the evidentiary record
pursuant to Federal Rule of Evidence 1006. (Tr. at 6757:13-6758:1, 6768:13-18.) The parties were able to resolve
their disputes as to all but one of these additional exhibits, DX7828-63, which is a single slide from Dr. Bernheim’s
presentation that displays the corporate logos of twenty-six large companies and is captioned “Amex’s Rates Have
Declined for Many Merchants.” (See DX7828 at 63.) Plaintiffs object to the admission of DX7828-63 as a
violation of Rule 1006 and Federal Rule of Civil Procedure 26. (Aug. 27, 2014, Pls. Ltr. (Dkt. 594-1).) Because the
7
Based upon its measured consideration of the record just described, and upon the findings
of fact and conclusions of law set forth herein pursuant to Federal Rule of Civil Procedure 52(a),
the court concludes that Plaintiffs have proven by a preponderance of the evidence that the
challenged restraints violate Section 1 of the Sherman Act.
A.
Overview of the GPCC Card Industry
Since the advent of the modern payment card industry in the 1950s, general purpose
credit and charge cards, or “GPCC” cards, have become a principal means by which consumers
in the United States purchase goods and services from the nation’s millions of merchants. (See
PX2578A at ‘389-90 3; Tr. at 3581:10-12 (Silverman/Amex).) In 2013, for example, the four
dominant networks providing authorization and settlement services—Visa, American Express,
MasterCard, and Discover—facilitated roughly $2.399 trillion in credit and charge card spending
at participating merchants. (See DX6576 at 8.) The typical U.S. consumer carries multiple
forms of payment in his or her wallet. 4 Alongside general purpose credit and charge cards—the
focus of the court’s analysis in the this litigation—merchants also accept payment through some
combination of debit cards, proprietary or private label credit or charge cards issued by
individual merchants, direct Automated Clearing House or “ACH” transfers, checks, and cash,
court finds that DX7828-63 lacks any appreciable probative value, there is little cause not to admit the exhibit for
the purposes of a complete appellate record. Specifically, Dr. Bernheim’s determination that the discount rates
declined for the twenty-six merchants represented on DX7828-63 is premised on a methodology for calculating
merchants’ effective discount rates that the court finds to be unreliable and upon which it places little, if any, weight.
See infra Part IV.D. As a result, Plaintiffs suffer no prejudice from the admission of DX7828-63 to ensure a
complete record on appeal.
3
Throughout this Decision, where an exhibit’s pincite is preceded by an apostrophe, the number refers to the last
three digits of the control number of the particular page cited. Where the pincite is not preceded by an apostrophe,
the number refers to the ordinary pagination of the exhibit.
4
For the purposes of this Decision, the term “merchant” will be used to refer to the wide range of entities that need
to collect payments, including airlines, hotels, grocery stores, online retailers, and government agencies, among
myriad others. The terms “consumer” and “customer” are used herein to refer to those individuals and entities that
purchase goods and services from merchants; in other words, those who desire to make payments. However, it is
important to recognize that Amex-accepting merchants and Amex cardholders are both technically “consumers” of
the services provided by Defendants. Though nomenclature can be a source of unnecessary confusion in the
payments context, the court has endeavored to be as clear as possible.
8
among other means. While each of these methods competes to some degree for share of
consumers’ wallets, they have one essential characteristic in common: Each payment system
brings customers and merchants together in order to consummate a transaction that benefits both
participants.
An introduction to the credit and charge card industry is a suitable starting point for the
court’s inquiry into the competitive dynamics therein. Credit cards enable cardholders to make
purchases at participating merchants by accessing a line of credit extended to the cardholder by
the issuer of that card. (Joint Stmt. of Undisputed Facts (“Jt. Stmt.”) (Dkt. 447-1) ¶ 2.)
Cardholders are invoiced for purchases typically once per month and often have a grace period
during which payment may be made. (Id.) The delay between a purchase event and the
cardholder’s deadline for paying the bill on which that purchase appears is referred to as the
“float,” and it enables cardholders to temporarily defer payment on their purchases at no
additional cost (i.e., without paying interest). (Tr. at 6245:1-12, 6532:6-9 (Bernheim).) A
cardholder may either pay off the balance of his bill in full each month or pay it off over time
while accruing interest on the balance. (Jt. Stmt. ¶ 2.) Many credit card issuers impose a preset
spending limit on a cardholder’s outstanding credit amount, typically based on the issuer’s
determination of the cardholder’s creditworthiness. (Id.)
Charge cards similarly allow cardholders to make payments by accessing a line of credit
extended by the card issuer, but generally do not offer a revolving credit facility akin to that
offered on credit cards, and instead require that the cardholder pay the balance in full each
month. (Id. ¶ 3.) Some American Express cardholders, however, do have an ability to maintain
a balance, or “revolve,” on their Amex charge cards, blurring the distinction between
Defendants’ credit and charge offerings. (Tr. at 5161:5-24 (Gilbert).) Even though charge cards
9
typically are not paired with a line of credit, cardholders generally derive a benefit from the
ability to defer payment during the float period, depending on the point during the billing cycle
at which the purchase is made. (Jt. Stmt. ¶ 3; Tr. at 4065:15-19 (Katz).) Unlike credit cards,
charge cards typically do not have preset spending limits. (Jt. Stmt. ¶ 3.)
Two specific types of credit and charge cards bear brief mention here, and will be the
subject of greater discussion in later sections. First, general purpose credit cards may be issued
in partnership with a merchant pursuant to a co-brand agreement. These “co-brand” cards
typically bear the logos of the merchant, network, and issuing bank (where relevant), and enable
cardholders to earn rewards directly from the merchant partner when purchases are made on the
card. (See, e.g., Tr. at 1617:13-1618:22 (Brennan/Hilton), 3603:3-21 (Silverman/Amex).)
Prominent examples of co-brand cards include the Delta SkyMiles Credit Card issued by
American Express, which processes over the American Express network; the American Airlines
AAdvantage Card, which is issued by Citibank and runs on MasterCard’s network; and the
Marriott Rewards Premier Credit Card, which is issued by Chase on the Visa network. Second,
American Express also maintains a robust—indeed, the largest—corporate card business in the
industry. (Id. at 817:2-9 (Hochschild/Discover); PX2486 at ‘053.) American Express issues
corporate cards to individuals through a corporate account that has been established with their
employers, allowing employers to more easily monitor employees’ business expenditures and
streamline the accounting and reimbursement processes. (Tr. at 817:2-12
(Hochschild/Discover), 1226:18-1227:12 (Kimmet/Home Depot), 3963:11-24 (Katz).)
By facilitating transactions between merchants and their cardholding consumers, the
general purpose credit and charge card systems that are the subject of this litigation function as
“two-sided platforms.” (Id. at 3827:15-20, 3828:23-3829:3 (Katz), 5022:24-5023:22 (Gilbert).)
10
In a two-sided platform, a single firm or collection of firms sells different products or services to
two separate yet interrelated groups of customers who, in turn, rely on the platform to
intermediate some type of interaction between them. (Id. at 3828:23-3829:14 (Katz).) See
generally David S. Evans & Richard Schmalensee, Industrial Organization of Markets with TwoSided Platforms, 3 Competition Pol’y Int’l 150 (2007) [hereinafter Evans & Schmalensee
(2007)]; Jean-Charles Rochet & Jean Tirole, Two-Sided Markets: A Progress Report, 37 RAND
J. Econ. 645, 645-46 (2006) (describing basic contours of two-sided systems). Examples of such
two-sided models abound: Newspapers and other advertising-based forms of media sell distinct
products and services to subscribers and advertisers; shopping malls provide services jointly to
retailers and shoppers; computer operating systems provide a platform for bringing together
program developers and end users; and a seemingly endless array of Internet companies like
eBay, OpenTable, eHarmony, and Groupon exist to facilitate some form of value-generating
interaction between distinct sets of consumers. See David S. Evans & Michael Noel, Defining
Antitrust Markets When Firms Operate Two-Sided Platforms, 2005 Colum. Bus. L. Rev. 667,
672-79 (2005) [hereinafter Evans & Noel (2005)]. The fundamental function of a two-sided
platform is to reduce the transaction costs associated with the parties finding one another, and to
thereby enable their customers to realize gains from trade or other interactions that otherwise
might not occur. See Evans & Schmalensee (2007) at 151, 158.
Credit and charge cards, like all methods of payment, serve as two-sided intermediaries
between merchants and their cardholding customers. American Express, for example, provides
cardholders with card-payment services and merchants with card-acceptance services in order to
facilitate transactions between the two. Importantly, and unlike many two-sided platforms,
American Express provides these services simultaneously; for every unit of payment services
11
sold to the cardholder at the moment of purchase, a matching service is sold to the merchant in
order to execute the transaction, and vice versa. (Tr. at 6211:3-11, 6211:23-6212:13
(Bernheim).) Thus, credit and charge card networks are also referred to as two-sided
“transaction markets”—the two sides of the platform are brought together to consummate a
single, simultaneous transaction, and the products provided by the platform are consumed in
fixed proportions by the consumer and merchant. See Lapo Filistrucci et al., Market Definition
in Two-Sided Markets: Theory and Practice (Tilburg Law Sch. Legal Studies Research Paper
Series No. 09/2013) at 12, available at http://papers.ssrn.com/sol3/papers.cfm?abstract_
id=2240850.
The two-sided nature of the GPCC card industry necessarily affects the court’s antitrust
analysis in this case. While the nature and import of these effects are addressed in greater detail
where relevant, a number of observations relating to the symbiotic relationship between the two
sides of the credit card platform provide context both for American Express’s separate value
propositions for merchants and cardholders and the contractual restraints at issue in this
litigation. A key feature of the payment network services industry, like all two-sided platforms,
is that it is subject to indirect or cross-platform network effects, a phenomenon referred to in this
case as the “chicken and the egg problem.” (Tr. at 820:23-821:16 (Hochschild/Discover),
4296:11-4297:10 (Chenault/Amex).) See also ABA Section of Antitrust Law, Market Definition
in Antitrust: Theory and Case Studies 439-44 (2012) [hereinafter ABA, Market Definition].
Indirect network effects exist when the number of agents or the quantity of services bought on
one side of a two-sided platform affects the value that an agent on the other side of the platform
can realize. (Tr. at 3829:15-20 (Katz).) See ABA, Market Definition at 440-42; Evans &
Schmalensee (2007) at 151-52. In this case, for example, having a credit or charge card on a
12
particular network like Discover is more valuable to the cardholder when there are more
merchants willing to accept that card and, conversely, the value to merchants of accepting
Discover cards increases with the number of cards on that network in circulation. (Tr. at
3829:21-3830:7 (Katz), 4397:2-11 (Chenault/Amex); DX4184 at ‘856.)
The GPCC industry’s susceptibility to spillover effects is closely related. In the present
context, spillover refers to the phenomenon by which a cardholder’s experience at one merchant
when using a particular network’s card, here an American Express card, affects that cardholder’s
willingness to use the same card on the next transaction, whether at the same merchant or a
different merchant. (Tr. at 4169:5-17 (Katz), 4339:7-18 (Chenault/Amex).) As with indirect
network effects, spillover can be either positive or negative. For example, when a major
merchant is added to Amex’s network, cardholders are more likely to use their Amex cards at
other merchants in the same area. (See DX4007 at ‘928, ‘932-34 (ascribing this effect to inactive
cardholders becoming active cardholders, rather than to higher spending levels among active
cardholders).) American Express’s defense of the restraints in this case centers on the NDPs role
in avoiding negative spillover and preserving what the company calls “welcome acceptance.”
Defendants contend that if the NDPs are eliminated, a cardholder who is steered away from
American Express at one merchant will be less likely to use an Amex card at the next merchant,
even if that second merchant does not attempt to influence the card choice. (Tr. at 3066:43068:12 (Pojero/Amex), 6356:10-6357:24 (Bernheim).) See infra Part VI.A (discussing Amex’s
primary pro-competitive justification for the challenged restraints).
Therefore, in order to compete effectively, networks must account for the
interdependence between the demands of each side of the platform and strike a profitmaximizing balance between the two. As a result, even in a case such as this, where the court’s
13
analysis focuses on one side of the relevant platform (merchants), due consideration must be
given to the competitive dynamics on the other side (cardholders). This is particularly true here,
as merchants’ demand for payment card acceptance is largely derived from consumers’ demand
for payment card usage. (Jt. Stmt. ¶ 4; Tr. at 6628:24-6629:13 (Katz).) As explained by
Defendants’ economics expert, “[t]he only reason that a merchant wants to use a payment
product is that a customer wants to use the product” to purchase some good or service from the
merchant. (Tr. at 6217:5-12 (Bernheim).) Yet even though merchants may not have an
independent demand for American Express’s network services, the choice of which GPCC
network is used for any given transaction is a joint decision between the merchant and consumer.
Steering, as Professor Katz correctly noted, describes “the interaction between the two sides in
order to make that joint decision.” (Tr. at 3834:7-24, 3831:1-21 (Katz).)
B.
Competition and Pricing in the GPCC Card Industry
American Express operates a business model that is materially different than that of Visa
and MasterCard, its primary competitors in the credit and charge card industry. However, to
understand American Express’s differentiated structure, it is helpful to understand how Visa and
MasterCard function. Visa and MasterCard sit at the center of a disaggregated platform that can
involve as many as five distinct actors: cardholders, issuers, networks, acquirers, and merchants.
Cardholders obtain their credit or charge cards from issuers, which are banks or other financial
institutions that issue cards with particular features (i.e., rewards, cash back, purchase
protection), set the financial terms on the cards (i.e., annual fees, interest rates, float periods),
extend cardholders credit where required, and issue cardholders their bills and collect required
payments. (Jt. Stmt. ¶ 5.) Similarly, to accept credit cards, a merchant must have a relationship
with an acquiring bank or financial institution. (Id. ¶ 6.) The acquirer is responsible both for
14
merchant acquisition (i.e., signing up merchants to accept particular brands of cards, providing
point-of-sale technology) and for accepting card transaction data from merchants for verification
and processing. (Id. ¶ 7.) The network sits as the platform’s middleman, bringing merchants and
their acquirers together with cardholders and their issuers. (Tr. at 3827:23-3828:13 (Katz).) The
network’s most fundamental function is to establish protocols and procedures by which issuers
and acquirers capture, authorize, and settle transactions; they also establish nearly all elements of
the price charged to merchants on each transaction (except for the fee charged by the
acquirer/processor), provide valuable fraud protection services, and operate the infrastructure
necessary to facilitate interactions between the two sides of the platform. (Id. at 3828:14-22.)
When a cardholder swipes his credit or charge card at a point-of-sale device in order to
make a purchase, the transaction information is immediately sent to the merchant’s acquirer.
The acquirer effectively obtains the receivable owed by the consumer arising from his purchase,
and therefore has a payable obligation to the merchant. (See Jt. Stmt. ¶ 12.) The acquirer
discharges this obligation by paying the merchant the funds owed on the transaction less the
“merchant discount fee,” which represents the merchant’s cost of accepting payment on the
credit or charge card used by the consumer. The merchant discount fee paid by the merchant
generally consists of an ad valorem element—i.e., a percentage discount rate multiplied by the
purchase price—but may include additional flat fees. (Id. ¶¶ 13, 15.) On the Visa and
MasterCard networks, which are frequently referred to as “4-party” or “5-party” systems to
reflect the number of agents involved, the merchant discount fee is primarily comprised of three
elements: a percentage interchange fee, an acquirer fee, and a network fee. (Id.) As the
terminology suggests, the acquirer fee is retained by the acquiring bank for services rendered to
the merchant, while the network fee is paid to Visa and MasterCard as the price of facilitating the
15
transaction. (Id.) Like the network fee, the interchange component, which represents the bulk of
the overall discount fee and is passed through to the issuing bank, is set by Visa or MasterCard.
(See Tr. at 2966:8-2967:12 (Pojero/Amex); DX7295 at 4 (MasterCard Interchange Rate
Programs).) Unlike American Express and Discover, as will be discussed shortly, the
interchange rate charged on the Visa and MasterCard network varies along two axes: (1) the
industry the merchant belongs to, and (2) the actual card product used by the cardholder.
MasterCard, for example, has more than 240 different interchange rate categories, and Visa has
more than 70 categories. (Jt. Stmt. ¶ 16.)
By contrast, American Express operates a partially integrated “3-party” or “closed-loop”
payment card system. In addition to operating its credit and charge card network, American
Express also acts as the card issuer and merchant acquirer for the vast majority of transactions
involving its cards. (See Tr. at 3791:4-15 (Silverman/Amex); see also Jt. Stmt. ¶ 9; Tr.
at 1069:18-1070:15 (Quagliata/Amex).) Thus, in most cases, American Express maintains direct
relationships with its cardholders and accepting merchants: It provides issuing services to
cardholders, acquiring and processing services to merchants, and network services to both sides
of the platform in order to facilitate the use and acceptance of its payment cards. The same
cannot be said of Visa and MasterCard, Defendants’ largest competitors. Rather, Visa and
MasterCard function exclusively as networks, providing certain core payment services but
relying on banks and other financial institutions to undertake the card issuance and merchant
acquisition and processing functions. (Jt. Stmt. ¶ 10; Tr. at 2116:4-10 (Berry/Amex).) Visa
cardholders therefore generally interact with the network through their issuing banks, the largest
of which include JPMorgan Chase, Bank of America, Citibank, and Capital One (see PX1560 at
10), and merchants that accept Visa cards interface with acquirer-processors like Chase
16
Paymentech and First Data Corporation, rather than with Visa itself (see Tr. at 2384:19-22
(Priebe/Southwest), 2965:2-7 (Pojero/Amex)).
American Express does not always interact directly with its cardholders and merchants,
however. Prior to the decision in United States v. Visa U.S.A., Inc., Visa and MasterCard
maintained bylaws preventing their member banks from issuing credit cards on competing
networks, like Discover and American Express. 5 163 F. Supp. 2d 322 (S.D.N.Y. 2001)
(“Visa I”), modified, 183 F. Supp. 2d 613 (S.D.N.Y. 2001), aff’d, 344 F.3d 229 (2d Cir. 2003)
(“Visa II”). After these so-called “exclusionary rules” were removed following the Department
of Justice’s successful antitrust enforcement action in Visa, Amex launched its Global Network
Services (“GNS”) business and began partnering with traditional issuing banks to disseminate
cards through non-proprietary channels. (Tr. at 4295:6-18 (Chenault/Amex), 2995:16-20
(Pojero/Amex).) There are currently nine GNS partners issuing Amex cards, which account for
roughly one percent of its total U.S. charge volume each year. (Id. at 4295:16-18, 4326:19-25
(Chenault/Amex.) Likewise, in its efforts to close its “merchant coverage gap”—American
Express is accepted in approximately three million fewer merchant locations than Visa,
MasterCard, and Discover—the company has increasingly relied on third-party merchant
acquirers to recruit small merchants to its network. (Id. at 2845:17-2850:2 (Pojero/Amex)
(discussing American Express’s External Sales Agent, OnePoint, and OptBlue initiatives).) See
also infra Part IV.D. Discover, the fourth and final significant competitor in this market, has
5
At the time of the Visa litigation, both Visa and MasterCard were organized as “open, joint venture associations
with members (primarily banks) that issue[d] payment cards, acquire[d] merchants who accept payment cards, or
both.” Visa I, 163 F. Supp. 2d at 332. For example, in return for the right to issue Visa cards and acquire Visa
transactions from merchants, member financial institutions agreed to follow Visa’s bylaws and operating
regulations, including its Exclusionary Rules. (Id.) Accordingly, the court occasionally refers to these ventures as
the “bank associations” throughout this Decision. Visa’s and MasterCard’s corporate structures changed in 2008
and 2006, respectively, when initial public offerings “converted each from a consortium of competitor banks into
single-entity, publicly traded companies with no bank governance.” See In re Payment Card Interchange Fee &
Merch. Disc. Antitrust Litig., 986 F. Supp. 2d 207, 212 (E.D.N.Y. 2013).
17
pursued a hybrid model; in addition to operating its network, Discover issues its own card
products but relies on third-party acquirers and processors to service the merchant side of the
platform. (See Tr. at 812:21-814:9, 815:12-24, 824:7-825:7 (Hochschild/Discover).)
American Express’s merchant pricing structure further differentiates its model from those
of Visa, MasterCard, and Discover. Whereas the discount rates applied to purchases on Visa or
MasterCard products vary depending on the type of card used—i.e., high-rewards cards are
subject to higher interchange rates and thus cost merchants more to accept—American Express
charges a single discount rate for all Amex credit and charge products, in addition to certain flat
fees charged on a per transaction basis. (See Jt. Stmt. ¶ 16; Tr. at 2132:14-19 (Berry/Amex),
2566:16-25 (Funda/Amex), 2978:6-2979:19 (Pojero/Amex); see also DX7295.) Accordingly,
the discount rate merchants are charged by American Express for purchases made on the highrewards Platinum Card, the “bedrock of [Amex’s] brand,” is the same as for purchases made on
its cards with less generous rewards, like the Green Card or EveryDay Credit Card. (Tr. at
2566:11-20 (Funda/Amex), 3112:9-20 (Pojero/Amex), 3602:5 (Silverman/Amex).) Like its
competitors, American Express strives to maintain “pricing integrity” within industry groups,
such as airlines, lodging, or gas stations. To that end, American Express sets a pricing table for
each merchant segment that contains a “headline” or “base” discount rate charged across the
entire industry, often with minor variations depending on the annual charge volume of each
individual merchant in that industry. (Tr. at 4684:3-4685:12, 4697:16-21 (Glenn/Amex).)
However, American Express will negotiate its acceptance agreements with certain large
merchants, and sometimes is required to provide monetary incentives such as signing bonuses or
cooperative marketing funds in order to ensure continued acceptance by the merchant. (See id.
at 4685:16-4687:3 (Glenn/Amex).) See also infra Part IV.D. These “side payments” are
18
credited against the merchant’s headline discount rate, often on a retrospective basis, in order to
calculate its “effective” discount rate.
Finally, the “spend-centric” nature of American Express’s business model is unique in
the industry. Unlike its competitors’ “lend-centric” models, which rely on the interest charged
on revolving balances to generate more than half of their revenue, the primary driver of
American Express’s revenue is the merchant discount fee. (See Tr. at 827:20-828:9
(Hochschild/Discover), 3534:19-3535:6 (Silverman/Amex), 4303:6-24 (Chenault/Amex).)
Together with its closed-loop framework, which also is a “key differentiator” for the network
vis-à-vis its competitors, Defendants’ spend-centric model is integral to the company’s value
proposition to its merchants. (Tr. 1069:20-1070:17 (Quagliata/Amex).) American Express
contends that due to its efforts to “encourage [cardholders] to maximize their spend on [Amex’s]
card products”—including, among other things, offering premium rewards programs, superior
customer service, and other ancillary benefits to cardholders—merchants that accept American
Express gain access to customers who are both “ready to spend” and who generally spend more
on an annual and per transaction basis than non-cardholders. (See id. at 1061:21-1062:5
(Quagliata/Amex), 2091:12-2092:12 (Glass/Amex), 3535:25-3536:4 (Silverman/Amex), 4304:813 (Chenault/Amex); see also id. at 3290:3-14 (Biornstad/MasterCard); DX7238 at ‘375.) See
also infra Part IV.C.3.
In addition to delivering higher spending customers, Amex’s merchant value proposition
relies on the network’s ability to leverage its closed-loop infrastructure to deliver marketing and
data analytics services to merchants that its competitors cannot match. (Tr. at 4305:7-4306:6
(Chenault/Amex), 4720:3-10 (Glenn/Amex).) By retaining end-to-end control of all spending
data on its network, American Express is able to sell its merchants information on and analysis
19
of its cardholders’ spending behaviors, allowing the merchant to engage in more effective
targeted marketing or identify new locations for geographic expansion, among other
applications. (Id. at 1072:5-1079:18 (Quagliata/Amex), 2117:2-2119:4, 2277:4-17
(Berry/Amex) (noting Amex can assist merchants in understanding their customers by providing
data on spending in the merchant’s own industry, as well as in other industries), 4307:8-4309:15
(Chenault/Amex), 4718:5-21, 4720:19-4721:6 (Glenn/Amex), 5530:10-5531:21
(Landau/DryBar); see also, e.g., DX7598 at ‘015.) The closed-loop system also allows
American Express to provide merchants and cardholders with advanced fraud management
services. (Tr. at 4306:7-17 (Chenault/Amex), 4933:22-4934:6 (Glenn/Amex).) Finally,
American Express provides dedicated client managers for its largest merchants (see id. at
625:12-626:24 (Quagliata/Amex)), and engages in a variety of efforts intended to promote
spending at small businesses—a merchant population where the network enjoys less widespread
acceptance than its competitors—including promotions like “Small Business Saturday” and
“Shop Small” (id. at 5704:2-5709:23 (Gilligan/Amex)).
On the other side of the platform, American Express’s cardholder value proposition
centers on the suite of rewards and other benefits the company provides to encourage cardholders
to use their cards for purchases at Amex-accepting merchants. These enticements commonly are
offered by the issuers of general purpose credit and charge cards and may include a combination
of per transaction benefits, such as “points,” cash back, or airline frequent flyer miles, as well as
other membership benefits, such as airport lounge access, purchase protection, or rental car
insurance. Cardholders enrolled in American Express’s Membership Rewards program, for
example, receive Membership Rewards points for purchases made on their Amex cards, and may
then redeem those points with Amex or one of its redemption partners for merchandise, gift
20
cards, frequent flyer miles, statement credits, or other goods and services. (Tr. at 3548:133549:22 (Silverman/Amex), 4298:20-4300:13 (Chenault/Amex).) When offered on Visa or
MasterCard products, these rewards are generally funded through the interchange fee paid by the
merchant that is passed through to the issuing bank. (See Tr. at 4040:21-24 (Katz).) As a
general matter, Visa and MasterCard products that tend to have rich rewards packages also tend
to carry higher interchange rates, which explains why it is more expensive for merchants to
accept high-rewards Visa and MasterCard cards when compared to more basic cards on the same
networks. See infra Part IV.D (discussing, among other things, Visa’s and MasterCard’s
introduction of premium interchange categories). In addition to its rewards programs, American
Express also offers its cardholders what it believes to be superior customer service, fraud
protection, and purchase and return protection, among other benefits. (Tr. at 3610:1-16
(Silverman/Amex), 4296:11-4297:10, 4309:20-4310:2 (Chenault/Amex).)
C.
The Non-Discrimination Provisions
The purpose and effect of American Express’s NDPs, as well as the vigor with which the
company defends them, cannot be fully appreciated without an understanding of their historical
context.
1.
Origins of Amex’s NDPs
American Express entered the payment cards industry in 1958, offering charge cards for
use primarily at travel and entertainment, or “T&E,” merchants. (Tr. at 4327:21-4328:6
(Chenault/Amex).) Intended to cater to the needs of business travelers, American Express’s
early charge cards competed with other niche payment systems offered by Diners Club and Carte
Blanche. (Id. at 4328:16-4329:10 (Chenault/Amex).) Following the entry of Visa and
MasterCard into this market in the mid-1960s, American Express undertook a concerted effort to
21
shift its payments business from a T&E-centric enterprise to a general purpose credit and charge
card network similar to those offered by the bank associations (Visa and MasterCard). For
instance, at the urging of its current Chief Executive Officer, Kenneth Chenault, and later under
his leadership, Amex endeavored to expand its acceptance network to include so-called
“everyday spend” merchants like gas stations, supermarkets, and pharmacies, with an aim toward
increasing its cards’ relevance to consumers’ everyday spending needs. (Id. at 4394:25-4395:5
(Chenault/Amex).) Amex additionally took steps during the late 1980s and early 1990s to
improve its value propositions to both merchants and cardholders by, for example, introducing
what would become the company’s touted Membership Rewards program and developing new
technology to better leverage its closed-loop network in service of its merchants. (Id. at
4333:18-4334:10, 4336:17-4337:18 (Chenault/Amex).) As a result of these efforts, Amex’s
share of credit and charge card spending in the United States rose to about 25% by 1990. (See
DX7828 at 50; Tr. at 5154:7-10 (Gilbert).)
Beginning in the late 1980s, however, Visa and MasterCard took a number of steps
intended to curtail American Express’s efforts to move into non-T&E merchant segments, which
had traditionally been the bankcards’ “bread and butter.” (See PX0132 at ‘867; Tr. at 3312:243313:6 (Morgan/Visa).) First, the bank associations adopted the so-called exclusionary rules,
which prevented member institutions from issuing card products on either American Express’s or
Discover’s networks. These rules were later found to violate the Sherman Act in Visa, and were
removed in 2004. (See Tr. at 859:21-860:4 (Hochschild/Discover)). Second, and more
importantly for the purpose of this litigation, Visa and MasterCard ran a number of marketing
campaigns that highlighted American Express’s perceived and actual competitive disadvantages
in the marketplace—specifically, Amex’s smaller merchant acceptance network, consumers’
22
resulting perceptions of the utility and value of Amex’s card products, and the network’s
significantly higher discount rates to merchants. (See id. at 3372:2-3373:12 (Morgan/Visa);
PX0132 at ‘930.) Visa, for example, sought to encourage consumer preference for its credit
cards and steer transactions away from its competitors through several advertising campaigns,
including the “It’s Everywhere You Want To Be” and “We Prefer Visa” initiatives. (Tr. at
3306:4-3307:11, 3321:21-3324:17, 3409:15-23 (Morgan/Visa); PX0082; PX0133 at ‘985-86; see
also Tr. at 3318:16-3320:24 (Morgan/Visa).) These efforts were remarkably effective. The “We
Prefer Visa” campaign, for example, appears to have contributed to a 25-45% shift in card
volume from American Express to Visa (see PX0133 at ‘986; see also Tr. at 4351:3-6
(Chenault/Amex)), and Amex’s overall share of GPCC charge volume dipped to approximately
20% by 1995. (Tr. at 5154:7-15 (Gilbert), 6305:18-6306:6 (Bernheim); DX7828 at 50.)
American Express responded to its competitors’ efforts to induce merchants to steer
volume away from its network by tightening the contractual restraints at issue in this litigation,
its so-called Non-Discrimination Provisions (“NDPs”). Formulated to control the manner in
which merchants treat Amex cardholders at the point of sale, limitations on merchant steering
have existed in Amex’s card acceptance agreements in one form or another since the 1950s.
(See PX1389 at 3 (1959 Agreement); DX0020 at ‘696 (1977 Agreement); Tr. at 4328:7-10
(Chenault/Amex).) In the late 1980s and early 1990s, however, Amex bolstered its NDPs to
ensure that merchants could not state a preference for any GPCC card network other than
American Express, and simultaneously intensified its efforts to enforce these provisions when it
detected merchant steering. (Compare PX1389 at ‘189 (sample 1992 agreement), ‘293 (sample
1998 agreement), with PX1389 at ‘155 (sample 1989 agreement containing less restrictive
NDPs); see also PX1103 at ‘353, ‘396-97 (discussing Amex’s response to “We Prefer Visa”
23
campaign); Tr. at 4490:13-4491:18 (Chenault/Amex).) Indeed, American Express’s CEO
acknowledged at trial that the NDPs were revised to preclude the use of preference language
favoring other issuers or networks after the company’s experience during the “We Prefer Visa”
campaign. (Tr. 4492:16-4493:14, 4531:17-4532:11 (Chenault/Amex).)
American Express’s response is hardly surprising. As a number of American Express
executives testified, a cardholder’s experience at the point of sale when using an Amex card is a
critical point of contact between American Express and the cardholder. (See Tr. at 3066:43067:13 (Pojero/Amex), 4953:8-23 (Hayes/Amex); see also id. at 884:11-20
(Hochschild/Discover), 6357:20-24 (Bernheim) (noting this is a critical touchpoint for the
network “that affects consumers’ perceptions of Amex and Amex isn’t there to control that at all,
the merchant is in charge of that”).) Since the entire purpose of carrying a payment card is to
enable the consumer to consummate transactions with merchants, the consumer’s decision to pull
an American Express card from his wallet at the point of sale represents a critical “moment of
truth” for the company. (Tr. at 3066:4-3067:13 (Pojero/Amex), 3573:11-15 (Silverman/Amex)
(“[T]here’s effectively two moments of truth for [Amex] customers . . . [o]ne is to get them to
buy the card . . . [t]he second is every time they make a purchase, to get them to use the card.”).)
Because the NDPs represent the Defendants’ attempt to control as much of that experience as
possible, purportedly to ensure its cardholders enjoy what the network calls “welcome
acceptance,” the network’s decision to tighten its restraints in response to the “We Prefer Visa”
campaign and similar initiatives is not tremendously surprising. (See DX0319 at ‘002; Tr. at
4372:8-4373:1 (Chenault/Amex).) In Amex’s view, merchant steering to less expensive card
networks—or “suppression,” as it is referred to within American Express—endangers the
cardholder’s purchasing experience and therefore endangers the network itself. (See Tr. at
24
4477:12-20 (Chenault/Amex).) Yet, as discussed in the remainder of this Decision, Amex’s
efforts to ensure welcome acceptance go too far in the view of the Sherman Act—the NDPs
unreasonably and unjustifiably suppress a critical avenue of interbrand competition in the
relevant market.
2.
The Challenged Restraints
The vast majority of American Express’s 6.4 million merchant locations are bound by the
company’s standard card acceptance agreement, and consequently, by its standard NDPs. (Tr. at
642:18-648:3 (Quagliata/Amex).) Unless a merchant has a customized or negotiated acceptance
agreement with American Express—and, notably, fewer than 1,000 merchants are in this
position—the terms and conditions of merchant acceptance of Amex’s credit and charge
products are set forth in a form card acceptance agreement with the network. (Tr. at 636:2639:13 (Quagliata/Amex), 2831:17-25 (Funda/Amex) (estimating Amex has fewer than 1,000
custom contracts); PX0003.) These form contracts incorporate by reference the company’s
Merchant Regulations and require Amex-accepting merchants to adhere to the policies and
procedures found therein. (Tr. at 642:15-643:5 (Quagliata/Amex); PX0003 at 13 (section 1.b.i).)
American Express’s standard NDPs are located in section 3.2 of Amex’s Merchant Regulations
(PX0003 at 13 (section 1.b.i)), and provide that a merchant who accepts American Express credit
or charge products may not:
•
indicate or imply that [it] prefer[s], directly or indirectly, any Other Payment
Products over [Amex’s] Card,
•
try to dissuade Cardmembers from using the Card,
•
criticize or mischaracterize the Card or any of our services or programs,
•
try to persuade or prompt Cardmembers to use any Other Payment Products or
any other method of payment (e.g., payment by check),
25
•
impose any restrictions, conditions, disadvantages or fees when the Card is
accepted that are not imposed equally on all Other Payment Products, except
for electronic funds transfer, or cash and check,
•
engage in activities that harm [Amex’s] business or the American Express
Brand (or both), or
•
promote any Other Payment Products (except [the merchant’s] own private
label card that [it] issue[s] for use solely at [the merchant’s] Establishments)
more actively than [it] promote[s] our Card.
(PX0002 at 16.)
Critically, section 3.2 includes two types of prohibitions that Plaintiffs expressly do not
challenge under the Sherman Act, which are therefore unaffected by the court’s resolution of this
case. First, Plaintiffs do not challenge section 3.2 insofar as it prohibits merchants from
imposing “fees” when accepting American Express cards that are not “imposed equally on all
Other Payment Products,” except for ACH, cash, or check. (PX0002 at 16.) Thus, the
Government is expressly not seeking to allow merchants to differentially surcharge American
Express cards vis-à-vis its competitors’ cards (i.e., charge a premium to consumers for using an
Amex card), though such steering is at issue in the multi-district litigation action against
American Express that is also before this court. (See Am. Compl. (Dkt. 57) ¶ 28; see also Tr. at
5828:8-5829:14 (Gilligan/Amex); PX2754 (demonstrative) (highlighting language in the NDPs
not challenged in this case).) See also In re Am. Express Anti-Steering Rules Antitrust Litig.
(No. II), No. 11-MD-2221 (NGG) (RER) (E.D.N.Y.). Second, Plaintiffs are not challenging the
NDPs to the extent they prohibit merchants from “mischaracteriz[ing]” the Card or “engag[ing]
in activities that harm [Amex’s] business or the American Express brand (or both).” (See Am.
Compl. ¶ 28; see also Tr. at 5828:8-5829:14 (Gilligan/Amex); PX2754 (demonstrative).) These
two carve-outs from Plaintiffs’ case are discussed in greater detail in connection with the court’s
analysis of Amex’s proposed pro-competitive justifications for the restraints. See infra Part
26
VI.A. For the sake of simplicity, any reference in this Decision to the NDPs or “challenged
restraints” refer only to those aspects of section 3.2, or parallel provisions in Amex’s customized
agreements, that Plaintiffs challenge as violations of the antitrust laws, unless otherwise noted.
As suggested above, American Express does negotiate its card acceptance agreements
with certain large merchants. (See, e.g., Tr. at 684:12-16, 685:5-686:5 (Quagliata/Amex),
2831:17-2832:2 (Funda/Amex).) However, the anti-steering provisions underlying this action,
which are intended to dictate merchants’ treatment of Defendants’ cards, are only rarely subject
to negotiation. In fact, Amex identified only 139 merchants with agreements that contained nonstandard NDPs. (See Stip. Regarding Non-Standard Card Acceptance Agreements (“Schmitt
Stip.”) (Dkt. 590) ¶ 1.) And even where a merchant has negotiated non-standard NDPs into its
contract with American Express, the rules still restrict nearly all forms of point-of-sale steering,
including merchants’ ability to express a preference for a particular card brand. Although in the
course of negotiations many of these larger merchants have requested that Amex remove its
NDPs from their agreements entirely (see, e.g., Tr. at 1613:2-6 (Brennan/Hilton), 1257:251258:19 (Kimmet/Home Depot), 1697:13-1698:13 (Dale/Sprint)), Defendants generally grant
merchants limited exceptions to the NDPs in only two contexts.
First, certain merchants, such as Sears, Crate & Barrel, Home Depot, and Hilton Hotels &
Resorts, have negotiated the right to steer toward their private label or co-brand cards. (See, e.g.,
PX1270 at ‘152-53 (Crate & Barrel); PX1915 at ‘805-06 (Home Depot); DX7530 at ‘649
(Hilton); Tr. at 1327:3-18 (Kimmet/Home Depot), 1653:18-1655:12 (Brennan/Hilton), 2338:182339:22 (Bruno/Crate & Barrel).) Under its custom card acceptance agreement and nonstandard NDPs, for example, Sears is permitted to steer customers to its private label card or to
its co-brand card with MasterCard by having its cashiers ask every customer as they approach
27
the point of sale if they “could put [their purchase] on their Sears card.” (Tr. at 577:17-578:9
(Bouchard/Sears).) Southwest Airlines is similarly able to steer customers to its Rapid Rewards
co-brand card with Visa by offering “$50 off, $100 off on the first purchase” or by offering “up
to 50,000 rewards points.” (PX0332 at ‘750-51; Tr. at 2426:25-2427:11 (Priebe/Southwest).)
While this type of steering is effective at shifting share to the merchants’ private label or cobrand cards (see, e.g., Tr. at 578:20-22, 2427:9-11 (Bouchard/Sears)), those merchants who have
negotiated this exception nonetheless remain subject to significant restrictions on whether and
how they can influence the customer’s payment choice as it relates to standard GPCC cards.
Hilton, for example, is permitted under its card acceptance agreement to steer toward its cobranded cards, but it still may not disclose to customers truthful information about how its cobrand cards compare to Amex’s credit and charge products. (Id. at 1613:12-19 (Brennan/Hilton),
2426:1-3, 2426:16-24 (Priebe/Southwest).)
Second, American Express occasionally has granted certain large merchants an exception
to the NDPs permitting them to engage in limited, short-term promotions with other credit card
issuers or networks. (See, e.g., DX2770 at ‘009 (Dell); PX0612 at ‘403-04 (Sprint); Tr. at
1699:5-19 (Dale/Sprint); PX0650 at ‘229-30 (Enterprise); Tr. at 525:9-13, 527:12-19
(Satkowski/Enterprise).) While this exception, when granted, allows merchants to incentivize
customers to shift their transactions to a less expensive network for a limited time period—such
clauses typically limit promotions to three or six months in duration—certain modes of steering
remain blocked, including merchants’ ability to express a preference for any brand other than
American Express. (See DX2770 at ‘009 (Dell); PX0612 at ‘403-04 (Sprint).) For example,
Sprint has been permitted to conduct a number of short-term promotions with Visa and
MasterCard whereby it offers customers a statement credit if they sign up for Sprint’s recurring
28
billing program with a particular card brand; however, Sprint remains unable to tell its customers
“Sprint prefers Visa” or to offer them discounts, incentives, or other benefits for paying their bill
with a non-Amex card. (Tr. at 1699:17-1702:6, 1703:18-1704:5, 1705:6-9 (Dale/Sprint).)
Further, a number of merchants whose card agreements contain this short-term promotion
exception dispute the efficacy of such programs, given the strict time limitations imposed by
American Express. (See Tr. at 579:15-580:3 (Bouchard/Sears) (testifying that the three-month
limitation in Sears’s contract affects the company’s willingness to engage in such promotions as
“short term promotions generally change behavior for that period, and then [it] goes back to
normal”); see also id. at 405:16-24 (Robinson/Ikea), 1702:25-1703:11 (Dale/Sprint) (testifying to
difference in effectiveness of long-term and short-term promotions).)
In practice, the NDPs operate to block Amex-accepting merchants from encouraging their
customers to use any credit or charge card other than an American Express card, even where that
card is less expensive for the merchant to accept. Steering among the various card brands could
be accomplished by offering discounts or other monetary incentives to customers who pay with a
particular type of card, offering non-monetary benefits for using a lower-cost card, displaying the
logo of one brand more prominently than others, expressing the merchants’ preference as to
which type of card it would rather accept, or posting each card’s cost of acceptance and letting
customers make their own decisions as to which mode of payment they prefer. Under
Defendants’ standard NDPs, however, a merchant can do none of these things. The NDPs
disable merchants from attempting to influence their customers’ card choices by, among other
things:
•
Offering a 10% discount off the posted purchase price, free shipping, free checked
bags, gift cards, or any other monetary incentive for using their Discover card;
29
•
Providing customers a designated checkout lane, priority boarding on an airline, or
any other non-monetary incentive if using a MasterCard;
•
Posting a sign saying “We Prefer Discover” at the point of sale, or otherwise
signaling a preference for a non-Amex payment card;
•
Answering the phone by saying “Thank you for calling us, we proudly accept the
Discover card” or posting a sign that says “Thank You For Using Discover”;
•
Posting a sign that discloses the merchant’s actual cost of accepting each network’s
cards or that compares the relative costs of acceptance across card brands, even if
such information is accurate and truthful;
•
Asking customers to “please keep in mind that credit and charge expenses are some
of our highest costs”;
•
Informing customers that it costs more for the merchant to accept American Express
than it does other card brands, even if the statement is true; or
•
Inviting customers to inquire or answering a customer’s inquiry into its credit card
costs, or in any way signaling that the merchant’s retail prices might be lower if it
were better able to control its credit card costs.
(See Tr. at 648:24-658:9, 667:17-691:15, 785:11-19, 792:2-796:19 (Quagliata/Amex); PX26202631 (demonstrative exhibits showing signage that would be prohibited by NDPs).) Importantly,
the NDPs inhibit steering even when American Express is not mentioned, resulting in the
restraints’ effects being inflicted across the GPCC industry. For instance, a merchant may not
post a sign saying “Please Use Your Discover Card, Visa Is More Expensive To Accept” or offer
a discount to MasterCard cardholders if they use a Visa card instead. (Tr. at 671:7-672:21,
792:2-5 (Quagliata/Amex); PX2626; PX2627.)
American Express’s NDPs do not, however, restrict merchants from steering customers
to cash, check, or ACH transfers. (PX0002 at 16.) For example, gas stations are able to offer
customers a lower price per gallon of gasoline if the customer pays in cash as opposed to using a
credit or charge card. Merchants may also steer their customers to debit cards to an extent. This
freedom is a product of the Durbin Amendment to the Dodd-Frank Wall Street Reform and
30
Consumer Protection Act of 2010, Pub. L. No. 111-203, § 1075, 124 Stat. 1376, 2068 (codified
in relevant part at 15 U.S.C. § 1693o-2), and its enabling regulations, 12 C.F.R. § 235, which
provide that merchants are legally entitled to offer discounts or other in-kind incentives to
encourage their customers to use a debit card, provided that the merchant does not differentiate
between card issuers or the various debit networks. See 15 U.S.C. § 1693o-2(b). Such steering
has been referred to as “untargeted” in the course of this litigation, and will be revisited in
greater detail in connection with the parties’ arguments concerning market definition and the
efficacy of merchant steering generally. (See Tr. at 3866:1-23 (Katz).)
American Express actively monitors for non-compliance with the NDPs and vigorously
enforces the restraints where steering or “suppression” is found to have occurred. Among its
large merchants, such monitoring is accomplished through the oversight of the merchant’s client
manager at American Express, random on-site visits, and reports from cardholders. (Tr. at
4786:11-4787:1 (Glenn/Amex).) Similarly, Amex monitors smaller, unmanaged merchants’
compliance with the NDPs through cardholder complaints and by monitoring the merchants’
charge volume. (Id. at 4789:1-16 (Glenn/Amex) (noting that when a small merchant’s volume
falls precipitously, the company believes the merchant has gone out of business, has stopped
accepting Amex, or is suppressing its card).) When the company identifies a situation in which a
merchant is believed to be steering, it will raise the issue through the merchant’s assigned client
manager or through its small merchant team, and, if the merchant does not voluntarily agree to
cease encouraging customers to pay with a product other than American Express, the company
will take remedial action up to and including termination of the merchant’s card acceptance
agreement with the network. (Id. at 4787:2-11, 4789:17-4790:5 (Glenn/Amex), 4491:6-16,
4514:14-19 (Chenault/Amex) (confirming that Amex terminated merchants for steering under
31
NDPs); see also PX1103 at ‘398-412 (guidance document concerning official responses to
steering).)
For example, in the early 2000s, Travelocity entered into a promotion agreement with
MasterCard whereby in exchange for significant financial remuneration from the network, the
online travel agency agreed to communicate to customers on its website that MasterCard was its
preferred form of payment. (See Tr. at 3245:14-3246:4 (Biornstad/MasterCard).) When
American Express learned of this preference relationship in or about June 2003, the company
immediately sought to enforce the NDPs and compel Travelocity to remove the offending
preference language, going so far as to send Travelocity a notice that it intended to terminate the
travel agency’s ability to accept American Express cards in December 2003. (See id. at 2887:412, 2891:8-2900:24 (Pojero/Amex); see also PX1085; PX0466.) Amex also tried to enforce its
restraints against MasterCard; the company’s Chief Litigation Counsel sent a letter to
MasterCard stating that Travelocity’s use of preference language pursuant to its promotion
agreement with MasterCard violated American Express’s card acceptance agreements and
“demand[ing] that MasterCard immediately cease and desist all such advertising and activities.”
(PX0385; Tr. at 2897:1-10 (Pojero/Amex).) Travelocity ultimately capitulated, agreeing to
change its website to indicate that MasterCard was the “official card” of Travelocity, rather than
the preferred card. (PX0449; Tr. at 3246:16-3247:20 (Biornstad/MasterCard).) Defendants have
enforced their NDPs in a similar manner at a number of other merchants to end the merchants’
use of preference language favoring Amex’s competitors. 6
6
See Tr. at 2901:2-2904:18 (Pojero/Amex) (NDPs used to end a “We Prefer MasterCard” campaign by Chelsea
Piers), 3332:22-3334:13 (Morgan/Visa) (Amex terminated its relationship with Steamboat Ski Area after the
merchant ran a “Steamboat Prefers Visa” promotion in early 1990s); see also id. at 2906:3-2908:18 (Ravina
Festival), 3334:21-3335:5 (Laura Ashley Holdings plc), 2909:13-2910:20, 4822:9-4827:2 (discussing Amex’s
efforts to end preference campaigns at Zagat, CheapTickets.com, Regal Cinemas, and Liberty Travel, among
others).
32
II.
LEGAL STANDARD
Plaintiffs challenge the legality of American Express’s NDPs under Section 1 of the
Sherman Act, which prohibits “every contract . . . in restraint of trade or commerce among the
several States.” 15 U.S.C. § 1. While this statutory text might be read literally to staggeringly
broad effect, as nearly every contract restrains some type of trade to some degree, the Supreme
Court has repeatedly held that Section 1 is intended to “outlaw only unreasonable restraints.”
State Oil Co. v. Khan, 522 U.S. 3, 10 (1997); see also Leegin Creative Leather Prods., Inc. v.
PSKS, Inc., 551 U.S. 877, 885 (2007). As the NDPs are contained in American Express’s card
acceptance agreements with its merchants—satisfying the “concerted action” element of a
Section 1 violation—the court is left to determine whether the challenged contractual provisions
qualify as unreasonable restraints on competition. See Geneva Pharm. Tech. Corp. v. Barr Labs.
Inc., 386 F.3d 485, 506 (2d Cir. 2004) (“To prove a § 1 violation, a plaintiff must demonstrate:
(1) a combination or some form of concerted action between at least two legally distinct
economic entities that (2) unreasonably restrains trade.”).
As non-price vertical restraints between firms at different levels of production—namely,
between the network and its merchant-consumers—American Express’s NDPs are properly
analyzed under the rule of reason. 7 See Continental T.V., Inc. v. GTE Sylvania Inc., 433 U.S.
7
Defendants’ contention that the Government seeks to impose liability under a truncated rule of reason, or “quick
look” analysis, is without merit. (Defs. Post-Trial Br. (Dkt. 605) at 8-10.) American Express argues that the “actual
adverse effects” test relied upon by Plaintiffs is inapposite here, as it was taken from a portion of the Supreme
Court’s opinion in F.T.C. v. Indiana Federation of Dentists, 476 U.S. 447, 461-62 (1986), which applied a “quick
look” analysis to a horizontal agreement. (Defs. Post-Trial Br. at 8.) See also Major League Baseball Props., Inc. v.
Salvino, Inc., 542 F.3d 290, 317 (2d Cir. 2008) (discussing state of Supreme Court precedent on “quick look”
analyses). Yet as the Second Circuit clarified in Todd v. Exxon Corp, 275 F.3d 191, 206-07 (2d Cir. 2001), and as
the court clarifies herein, “[t]he use of anticompetitive effects to demonstrate market power . . . is not limited to
‘quick look’ or ‘truncated’ rule of reason cases,” and may be used as a supplement to or in lieu of a market share
approach to proving market power. Id. at 207 (2d Cir. 2001) (rejecting antitrust defendants’ attempt to equate the
direct means of proving anticompetitive effect under the rule of reason with a “quick look” or “truncated” rule of
reason inquiry). Plaintiffs have unequivocally stated their position that the NDPs should be assessed using a full
rule of reason inquiry (see Tr. at 15:18-16:2 (Opening Statement), 6783:8-13 (Closing Argument)), and the court
agrees that the various complexities in this case preclude a finding that the anticompetitive effects flowing from the
33
36, 57-59 (1977); see also Leegin, 551 U.S. at 885-86. Yet, as Plaintiffs correctly note, the
challenged restraints do not fit neatly into the standard taxonomy of federal antitrust law. Unlike
most vertical distribution agreements between manufacturers/suppliers or dealers/distributors,
the NDPs do not purport to restrain intrabrand competition in favor of greater interbrand
competition. See, e.g., Leegin, 551 U.S. at 890-91; GTE Sylvania, 433 U.S. at 54. See also infra
Part VI.A. Rather, more akin to exclusive dealing or tying arrangements 8—though the NDPs
cannot be fairly characterized as either—Amex’s anti-steering rules admittedly have the primary
effect of restraining one form of interbrand competition among the GPCC card networks in favor
of alternative forms of interbrand competition. This effect is limited to horizontal competition at
the network level, however, as the NDPs do not purport to affect competition among the millions
of Amex-accepting merchants. Given the nature of the NDPs’ effects on competition, and
recognizing that “the primary purpose of the antitrust laws is to protect interbrand competition,”
State Oil, 522 U.S. at 15, the court accordingly approaches its rule of reason analysis in this case
with due caution and care.
The rule of reason, which is the most searching form of antitrust analysis, requires the
court in its capacity as factfinder to “weigh[] all of the circumstances of a case in deciding
whether a restrictive practice should be prohibited as imposing an unreasonable restraint on
competition.” See GTE Sylvania, 433 U.S. at 49; see also Major League Baseball Props., Inc. v.
Salvino, Inc., 542 F.3d 290, 316-17 (2d Cir. 2008). This context-specific inquiry enables the
court to “form a judgment about the competitive significance of the restraint,” i.e., to determine
“whether the challenged agreement is one that promotes competition or one that suppresses
NDPs are “obvious.” See Cal. Dental Ass’n v. F.T.C., 526 U.S. 756, 770-71 (1999) (noting the Supreme Court has
generally applied a truncated analysis only where “an observer with even a rudimentary understanding of economics
could conclude that the arrangements in question would have an anticompetitive effect on customers and markets”).
8
See Tr. at 4187:4-4189:18 (Katz).
34
competition” on the whole. Nat’l Soc’y of Prof’l Eng’rs v. United States, 435 U.S. 679, 691-92
(1978) (“‘The true test of legality is whether the restraint imposed is such as merely regulates
and perhaps thereby promotes competition or whether it is such as may suppress or even destroy
competition.’” (quoting Chi. Bd. of Trade v. United States, 246 U.S. 231, 238 (1918))); see
Leegin, 551 U.S. at 886 (“In its design and function the rule [of reason] distinguishes between
restraints with anticompetitive effect that are harmful to the consumer and restraints stimulating
competition that are in the consumer’s best interest.”); Major League, 542 F.3d at 316-17.
Factors appropriate for consideration in the course of such analysis include “specific information
about the relevant business, its condition before and after the restraint was imposed, and the
restraint’s history, nature, and effect.” State Oil, 522 U.S. at 10; see also Leegin, 551 U.S. at
885.
In applying a rule of reason analysis, the court utilizes a three-step burden shifting
framework. Plaintiffs bear an initial burden of demonstrating that the challenged restraints have
had an “adverse effect on competition as a whole in the relevant market.” Geneva Pharm., 386
F.3d at 506-07 (emphasis removed); United States v. Am. Express Co., No. 10-CV-4496 (NGG)
(RER), 2014 WL 1817427 (E.D.N.Y. May 7, 2014) (“Amex I”). Two independent avenues exist
by which this burden may be discharged: directly, by “show[ing] an actual adverse effect on
competition” caused by the restraint in the relevant market, such as increased prices or a
reduction in output; or indirectly, by “establishing that [the defendant] had sufficient market
power to cause an adverse effect on competition,” Tops Mkts., Inc. v. Quality Mkts., Inc., 142
F.3d 90, 96 (1998), and that there are “other grounds to believe that the defendant’s behavior will
harm competition market-wide,” K.M.B. Warehouse Distribs., Inc. v. Walker Mfg. Co., 61 F.3d
123, 129 (2d Cir. 1995) (“This court has not made a showing of market power a prerequisite for
35
recovery in all § 1 cases.”). See also F.T.C. v. Ind. Fed’n of Dentists, 476 U.S. 447, 460-61
(1986); Geneva Pharm., 386 F.3d at 509 (“If plaintiff can demonstrate an actual adverse effect on
competition, such as reduced output, . . . there is no need to show market power in addition.”
(citations omitted)); Todd v. Exxon Corp., 275 F.3d 191, 206-07 (2d Cir. 2001); Amex I, 2014
WL 1817427, at *6-7. Yet the Second Circuit has also recognized that, in many ways, these dual
paths are two sides of the same coin. Just as an indirect market analysis represents a court’s
effort to discern whether a defendant firm has the capacity to harm competition and, relatedly,
whether the challenged restraint is likely to have anticompetitive effect in the relevant market,
see K.M.B. Warehouse, 61 F.3d at 128-30; Tops Mkts., 142 F.3d at 96 (“Market power is but a
‘surrogate for detrimental effects.’” (citation omitted)), proof of actual adverse effects on
competition is compelling evidence that the defendant firm does, in fact, possess sufficient
power to profitably restrain competition in the relevant market, see Todd, 275 F.3d at 206 (thenJudge Sotomayor observing that actual detrimental effect “arguably is more direct evidence of
market power than calculations of elusive market share figures,” and that “[i]f a plaintiff can
show that a defendant’s conduct exerted an actual adverse effect on competition, this is a strong
indicator of market power”); Tops Mkts., 142 F.3d at 98. 9
In the event Plaintiffs are able to discharge their initial burden, the burden then shifts to
Defendants “to offer evidence of the pro-competitive effects of their agreement.” See Geneva
Pharm., 386 F.3d at 507; Capital Imaging Assocs., P.C. v. Mohawk Valley Med. Assocs., Inc.,
9
American Express maintains its position, first articulated in its summary judgment briefing, that Plaintiffs must
prove that it possessed market power in the relevant market in order to prevail under Section 1. For the same
reasons set forth in Amex I, the court rejects this position as inconsistent with clear and binding precedent in this
Circuit. See Amex I, 2014 WL 1817427, at *6-7. In any event, as suggested by the Second Circuit in Todd and
Tops Markets, when a plaintiff has discharged his initial burden in a Section 1 case by proving that the challenged
restraint caused actual detrimental effects on competition, the plaintiff implicitly has also proven that the defendant
possessed sufficient antitrust market power to cause such competitive harms. See Todd, 275 F.3d at 206; Tops
Mkts., 142 F.3d at 98. Additionally, as discussed later in this Decision, the court concludes that American Express
does in fact possess antitrust market power in the relevant market. See infra Part IV.
36
996 F.2d 537, 543 (2d Cir. 1993) (noting defendant must “offer evidence of the pro-competitive
‘redeeming virtues’” of the challenged conduct). If Defendants are able to demonstrate such
justification, the burden ultimately shifts back to Plaintiffs to prove that any “legitimate
competitive benefits” proffered by Defendants could have been achieved through less restrictive
means. Geneva Pharm., 386 F.3d at 507.
III.
MARKET DEFINITION
In order to determine whether Amex’s NDPs violate the Sherman Act, the court first
must determine the contours of the relevant antitrust market and thereby define an appropriate
context for the remainder of its analysis. See Geneva Pharm., 386 F.3d at 496 (noting market
definition is relevant regardless of whether a plaintiff intends to prove actual adverse effect or
market power); Visa I, 163 F. Supp. 2d at 334-35; Carell v. Scubert Org., Inc., 104 F. Supp. 2d
236, 264 (S.D.N.Y. 2000). Antitrust markets are defined by reference to both a “product
market” and a “geographic market.” AD/SAT, Div. of Skylight, Inc. v. Associated Press, 181
F.3d 216, 226 (2d Cir. 1999). Here, as the parties have agreed that the relevant geographic
market is the territorial United States (see Tr. at 3859:1-6 (Katz)), the court need only determine
the relevant product market.
Under the federal antitrust laws, a product market is “composed of products that have
reasonable interchangeability” from the perspective of the relevant consumer with the product
sold by the defendant firm. See United States v. E.I. du Pont de Nemours & Co., 351 U.S. 377,
404 (1956); Geneva Pharm., 386 F.3d at 496 (“The relevant market is defined as all products
‘reasonably interchangeable by consumers for the same purposes,’ because the ability of
consumers to switch to a substitute restrains a firm’s ability to raise prices above the competitive
level.” (citation omitted)); Xerox Corp. v. Media Scis., Inc., 660 F. Supp. 2d 535, 543 (S.D.N.Y.
37
2009) (“[P]roducts constitute part of a single product market if they are ‘reasonably
interchangeable by consumers for the same purposes,’ such that there is high cross-elasticity of
demand for the products.” (citation omitted)). This factual determination requires the court to be
cognizant of the “commercial realities” faced by a defendant’s consumers, see Eastman Kodak
Co. v. Image Technical Servs., Inc., 504 U.S. 451, 482 (1992); Brown Shoe Co. v. United States,
370 U.S. 294, 336 (1962), and to consider the various factors that might influence consumers’
choice to switch to a substitute product, including functional interchangeability, price, and
quality, see E.I. du Pont de Nemours, 351 U.S. at 401-04; Visa I, 163 F. Supp. 2d at 335. By
identifying the range of reasonably interchangeable substitute products, the court is able to
“identify the market participants and competitive pressures that restrain an individual firm’s
ability to raise prices or restrict output” and better assess the competitive dynamics in which the
defendant firm and challenged restraint operate. Geneva Pharm., 386 F.3d at 496; Balaklaw v.
Lovell, 14 F.3d 793, 799 (2d Cir. 1994) (“The basic principle is that the relevant market
definition must encompass the realities of competition.” (internal citation and quotation marks
omitted)).
For the reasons set forth below, the court adopts the general market definition advanced
by Plaintiffs’ economics expert, Dr. Michael Katz, and accordingly finds that the relevant market
for the purpose of the court’s antitrust analysis in this case is the market for general purpose
credit and charge card network services. Importantly, notwithstanding Defendants’ vigorous
arguments to the contrary, the court declines to conclude that debit network services should be
included in the relevant market. However, the court agrees with American Express that Plaintiffs
have not adequately proven the existence of a cognizable submarket for GPCC card network
services provided to merchants in T&E industries.
38
A.
GPCC Card Network Services Market
Plaintiffs propose a relevant product market limited to general purpose credit and charge
card network services. (See Am. Compl. ¶ 33; Tr. at 3858:14-25 (Katz).) These services include
the core enabling functions provided by networks, which allow merchants to capture, authorize,
and settle transactions for customers who elect to pay with their credit or charge card. American
Express disputes this definition, arguing alternatively (1) that the market should be defined by
reference to “transactions” so as to account for both sides of the credit card platform, and/or (2)
that debit services are a reasonably interchangeable substitute for general purpose credit and
charge card services and therefore should be included in the market definition. Each argument is
considered, and ultimately rejected, in turn. The court finds that Plaintiffs have appropriately
accounted for the two-sided features and competitive realities that affect the four major firms
operating in the GPCC card network services market—as distinguished from the card issuance
market—and that a practical and nuanced application of the standard tools for defining product
markets establishes that Plaintiffs’ proposed definition is an appropriate underpinning for the
court’s analysis in this case.
1.
The Relevant Product Is Network Services
This court is not the first to have been called upon to conduct a market analysis in the
credit and charge card industry, and the court does not expect that it will be the last. Nor have
the two-sided features of the GPCC industry escaped judicial inquiry in prior decisions. See
Visa II, 344 F.3d at 238-40. Yet, as American Express correctly notes, product markets must be
defined by reference to current competitive realities. See Eastman Kodak, 504 U.S. at 482
(“[M]arket definition . . . can be determined only after a factual inquiry into the ‘commercial
realities’ faced by consumers.” (citation omitted)); Todd, 275 F.3d at 199 (stating “market
39
definition is a deeply fact-intensive inquiry”). Thus, while prior judicial experience in this
industry necessarily informs the court’s market definition analysis, it must nonetheless remain
sensitive to the fact that market conditions may well have shifted during the intervening years.
In United States v. Visa, the Department of Justice brought an antitrust enforcement
action against Visa and MasterCard, alleging that the bank associations’ so-called “exclusionary
rules,” pursuant to which member banks were prohibited from issuing credit cards on either the
American Express or Discover networks, violated federal antitrust laws. Visa I, 163 F. Supp. 2d
at 327-30. 10 En route to concluding that the exclusionary rules did, in fact, violate Section 1, the
district court adopted the market definitions proposed by the Government’s expert economist,
Dr. Michael Katz—the same expert who testified on Plaintiffs’ behalf in this case—and
determined that the GPCC card industry included two separate yet complementary product
markets: (1) a “general purpose card network services market” in which Visa, MasterCard,
American Express, and Discover competed to sell network services to merchants, and in the
associations’ case, to issuing banks; and (2) a “general purpose card market” that covered the
issuance of card products by issuing banks, as well as American Express and Discover, to
cardholders. Id. at 331, 334-39. With regard to the network services market, which is
effectively the same product market proposed by Plaintiffs here, 11 the court determined that
10
The Government additionally, albeit unsuccessfully, challenged the dual governance structures of Visa and
MasterCard, which permitted members of each association to sit on the Board of Directors of either Visa or
MasterCard. See Visa I, 163 F. Supp. 2d at 378-79.
11
Dr. Katz’s characterization of the relevant market at trial differed from both the definition alleged by Plaintiffs in
the Amended Complaint and the definition adopted in Visa, insofar as Dr. Katz added the notation “provided to
merchants” after the core product definition “general purpose credit and charge card network services.” (Compare
Tr. at 3858:14-25 (Katz), with Am. Compl. ¶ 33.) Having reviewed the whole of Dr. Katz’s testimony, the court
cannot conclude that by including this additional language Dr. Katz intended to depart from the product markets
adopted by the district court and Second Circuit in Visa. Rather, it seems far more likely that the addition of the
“provided to merchants” qualification was included by Dr. Katz as a foil against which he could define a submarket
of network services provided to only T&E merchants, as opposed to all merchants. The court accordingly concludes
that Plaintiffs and Dr. Katz propose that it adopt a substantively identical market as the network services market
used in Visa. As the Second Circuit explained in Visa, in the network services market the four major networks
40
“general purpose card network services . . . constitute a product market because merchant
consumers exhibit little price sensitivity and the networks provide core services that cannot
reasonably be replaced by other sources.” Id. at 338-39. The Second Circuit agreed, confirming
that the GPCC platform encompassed “two interrelated, but separate product markets”—namely,
a card issuance market and a network services market. Visa II, 344 F.3d at 238-40 (affirming the
district court’s determination that “there are no products reasonably interchangeable . . . with the
network services provided by the four major brands”).
At trial in this case, American Express urged the court to depart from the decisions in
Visa and to define the relevant product market in terms of “transactions,” rather than network
services. (See Tr. at 5015:7-5016:16 (Gilbert), 6211:12-6216:25 (Bernheim).) As explained by
Dr. Richard Gilbert, one of Defendants’ expert witnesses, a transactions-based market effectively
would collapse all services provided to merchants and cardholders in the context of the GPCC
card platform into a single antitrust market. (Id. at 5015:7-5016:16 (noting a market defined in
terms of transactions would capture “the flow around this whole loop between consumers and
merchants”).) But this takes the concept of two-sidedness too far. The goal in defining a
relevant product market is not to obfuscate or confuse market realities, but rather to “recognize
competition where, in fact, competition exists.” United States v. Cont’l Can Co., 378 U.S. 441,
452 (1978) (internal quotation marks omitted); Geneva Pharm., 386 F.3d at 496. Competition in
the GPCC card industry occurs on at least two distinct yet interrelated levels: (1) at the card
issuance level, where American Express and Discover compete against each other and against
compete with one another to sell services to both the issuing banks and merchants. See Visa II, 344 F.3d at 239
(“[I]n the market for general purpose card network services, the four networks themselves are the sellers, and the
issuers of cards and merchants are the buyers.” (emphasis removed)). In other words, like the overall GPCC
platform, the network services market itself is two-sided; each network competes for the business of two separate
sets of consumers. Of course, due to its large proprietary issuing business, American Express does not need to focus
on the demands of the issuing banks to the same degree as Visa and MasterCard, yet it cannot wholly ignore these
demands either, given its third-party issuing or GNS business.
41
the thousands of Visa- and MasterCard-issuing banks; and (2) at the network services level,
where Visa, MasterCard, American Express, and Discover compete. 12 Visa I, 163 F. Supp. 2d at
333; see also Visa II, 344 F.3d at 237, 239. American Express itself recognizes this reality. As
one executive explained at trial, Amex competes in “three businesses: we’re an issuing bank,
we’re a merchant acquirer, and we’re a network.” (Tr. at 3791:8-15 (Silverman/Amex).) To
conflate these separate avenues of competition into a single product market for “transactions”
that is coextensive with the platform itself, as Defendants encourage, would impermissibly and
unnecessarily frustrate the court’s analysis in this case. The court believes it is both necessary
and appropriate to define separate product markets that reflect the competitive realities in the
GPCC industry, although recognizing that these markets are inextricably linked with one
another, and appreciating and accounting for the effects that flow from such a relationship.
Defendants have provided the court with no reason to depart from the approach of the Second
Circuit in Visa II, and the court declines to do so on its own accord by collapsing the issuance
and network services markets into a single platform-wide market for transactions.
Nor is it necessary that the relevant product market be defined by reference to how
American Express chooses to compete in the industry. (See Tr. at 2020:24-2021:15, 2022:5-23
(Gilbert) (noting Amex does not offer network services in isolation nor charge a “network fee”).)
As previously noted, Defendants’ “closed-loop” model entails a significant degree of vertical
12
As the Second Circuit recognized in Visa II, a third avenue of competition occurs at the “acquiring” level among
American Express and the various acquirers/processors that provide services to Visa-, MasterCard-, and Discoveraccepting merchants. 344 F.3d at 237. This level of competition, however, has little bearing on the court’s analysis
of Amex’s NDPs, except insofar as it affects each network’s respective merchant coverage. American Express
generally does not rely on acquirers to act as intermediaries between the network and its merchants, while those
competitors that do rely on acquiring banks to deal with merchants still dictate nearly all key terms of the networkmerchant relationship, including interchange rates (which form the bulk of the merchant discount fee), anti-steering
rules, and other conditions of acceptance. (See Tr. at 3828:13-22, 3833:17-24 (Katz).) Accordingly, as in Visa, the
court finds that it is able to conduct a thorough and comprehensive analysis of the challenged restraints without
extensive consideration of the acquiring market; therefore, the role and functions of acquiring banks will be
addressed only where relevant, and the court will otherwise assume a direct relationship between network and
merchant on the merchant side of the GPCC platform.
42
integration into each level of competition comprising the GPCC card platform—card issuance,
network services, and merchant acquisition. Yet each constituent product market in this industry
is distinct, involving different sets of rivals and the sale of separate, though interrelated, products
and services to separate groups of consumers. This is evident, for example, from the fact that
Visa and MasterCard offer only network services but do not issue cards or acquire merchants,
and from the function in this industry of Citibank and Bank of America, which act as card issuers
but not networks. That American Express has elected to compete at each of these levels by
partially integrating into the issuing and acquiring businesses does not compel the court to
collapse these distinct markets into a single “transactions” market to more closely resemble
Amex’s chosen business strategy. See, e.g., United States v. Eastman Kodak Co., 63 F.3d 95,
98-99, 104 (2d Cir. 1995) (defining a product market for “amateur color negative photographic
film,” but not considering the camera market in which Kodak also competed). Instead, the
network services market is a distinct product market for purposes of antitrust analysis, and a
firm’s conduct therein may be separately scrutinized under the Sherman Act, provided the court
recognizes and accounts for the fact that such conduct may indirectly affect competition at
another level within the GPCC platform.
Additionally, the court is aware of no authority—and Defendants have supplied none—
that requires the court to define the relevant product market to encompass the entire multi-sided
platform. 13 See Times-Picayune Publ’g Co. v. United States, 345 U.S. 598, 610 (1953)
13
The court in National Bancard Corp. v. Visa U.S.A., Inc., 596 F. Supp. 1231 (S.D. Fla. 1984) (“NaBanco”), aff’d,
799 F.2d 592 (11th Cir. 1986), however, did define a relevant product market for all “payment systems,” finding that
Visa’s GPCC cards were reasonably interchangeable with other credit cards, ATM cards, travelers checks, personal
checks, and cash. Id. at 1257-58. In reaching this determination, the NaBanco court appears to have recognized
some of the two-sided features of the credit card industry, though it did not expressly consider how such a structure
might affect its definitional analysis. Id. Importantly, however, the broad market definition endorsed by NaBanco
was rejected by the district court in Visa. As Judge Jones noted, “although it is literally true that . . . cash and checks
compete with general purpose cards as an option for payment by consumers . . . cash and checks do not drive many
of the means of competition” in the GPCC industry. Visa I, 163 F. Supp. 2d at 335-38. Relying on the qualitative
43
(recognizing “every newspaper is a dual trader in separate though interdependent markets” for
advertisers and readers, but defining the relevant market by reference only to advertisers and
noting “[t]his case concerns solely one of these markets”); Visa II, 344 F.3d at 238-40. Amex’s
position that the relevant market should be defined in terms of “transactions,” rather than by
reference to the various levels of competition occurring in the credit card industry, is plainly
inconsistent with the Second Circuit’s holding in Visa II as well as subsequent case law in this
district. See id.; see also In re Payment Card Interchange Fee & Merch. Disc. Antitrust Litig.,
562 F. Supp. 2d 392, 396-97 (E.D.N.Y. 2008) (noting, in the course of denying a motion to
dismiss, that “[t]he commodity in each product market is ‘Network Services’”); In re Visa
Check/MasterMoney Antitrust Litig., No. 96-CV-5238 (JG), 2003 WL 1712568, at *2 (E.D.N.Y.
2003) (implicitly defining a product market around network services, rather than transactions,
noting “[o]verwhelming evidence establishes that merchant demand for credit card services is
distinct from merchant demand for debit card services”). Absent a compelling reason why the
court in this case should depart from established precedent and instead endorse a single, unified
market for “transactions” in the credit card industry, which Defendants have not provided, the
court concludes that the relevant product market for purposes of its analysis of Amex’s NDPs is
the market for general purpose credit and charge card network services.
Nonetheless, American Express is correct that the court must account for the two-sided
features of the credit card industry in its market definition inquiry, as well as elsewhere in its
antitrust analysis. As previously noted, the purpose of defining a relevant market is to identify
differences between GPCC cards and other forms of payment, the fact that Visa and MasterCard did not consider the
costs of non-GPCC systems in pricing their own services, and evidence of consumer payment preferences, the Visa
court ultimately concluded that “neither consumers nor the defendants view debit, cash and checks as reasonably
interchangeable with credit cards.” Id. The Second Circuit agreed, Visa II, 344 F.3d at 238-40, and, given both the
passage of time since the NaBanco decision and the imprecise reasoning applied by the district court in that case,
this court finds little justification in NaBanco for the “transactions” definition advocated by American Express.
44
the competitive constraints on Defendants’ business behavior, i.e., the substitute products,
competitors, or other market forces that could potentially limit the anticompetitive effect of the
competitive practice at issue. In the present context, the functional reality of a multi-sided
platform must be taken into account, since the antitrust significance of a restraint that nominally
affects conduct on only one side of the platform cannot be assessed without considering its
impact on the other side of the platform. For example, a price increase imposed by a firm on
only one set of consumers in a two-sided platform may appear profitable when one considers
only the direct effect of that practice on demand among the targeted consumers. Yet in reality,
the suppression of demand on one side of the market may, by virtue of indirect network effects,
trigger a response among consumers on the other side of the platform that, on the whole, renders
the practice unprofitable. See generally ABA, Market Definition at 445-52; Evans &
Schmalensee (2007) at 173-75; Evans & Noel (2005) at 696-700.
As a result, rote application of the standard mechanical market definition exercises—
which were developed for single-sided markets—risks significantly overstating or understating
the breadth of the relevant market. While these tools remain helpful to the court’s definitional
analysis, they must be applied in a manner that carefully accounts for the competitive realities in
multi-sided platforms. Notwithstanding the two-sidedness of the credit card industry, however,
the court finds inadequate cause to depart from the approach of the Second Circuit in Visa and
accordingly defines the relevant market by reference to network services, rather than
transactions.
2.
Debit Network Services Are Not Reasonably Interchangeable
The evidence presented at trial also established that the relevant product market includes
only general purpose credit and charge card network services, and does not extend to debit card
45
network services. Contrary to American Express’s position at trial, the significant rise in
spending among U.S. consumers on debit cards over the past decade has not rendered obsolete
the determination in Visa that debit cards and GPCC cards, as well as their associated acceptance
services, belong to separate antitrust markets. See Visa I, 163 F. Supp. 2d at 335-39; Visa II, 344
F.3d at 269 (noting “there are no products reasonably interchangeable . . . with the network
services provided by the four major brands”). Exclusion of debit from the market is similarly
consistent with American Express’s public statements and litigation positions prior to initiation
of this lawsuit. 14
Together with the thorough and convincing price sensitivity analysis conducted by Dr.
Katz, the record developed at trial regarding the competitive realities in this marketplace—
including the limited substitutability of credit and debit services from the merchant’s perspective,
the various functional differences between the products, and empirical evidence suggesting low
cross-elasticity of demand between the products—demonstrates that credit and debit have not
become reasonably interchangeable in the decade since Visa. Both categories of evidence are
discussed in turn below. Ultimately, there is no indication that merchants—the “relevant
consumer” for defining the relevant product market in this case—historically have been or would
be inclined to switch to debit network services (i.e., drop acceptance of credit cards) in response
to rising prices in the GPCC card network services market, or that such substitution, if it did
occur, would be sufficient to temper an exercise of market power therein.
14
For example, prior to 2010, American Express consistently and repeatedly represented to courts, federal agencies,
and its investors that it did not compete with debit card networks because of the limited substitutability between
credit and debit. (See PX1408 at 22 (Amex 2009 Form 10-K) (noting “[t]he ability to substitute debit cards for
credit and charge cards is limited); PX0106 ¶¶ 72, 152 (Compl., Am. Express Travel Related Servs. Co. v. Visa
U.S.A., Inc., No. 04-CV-8967 (BSG) (DFE) (S.D.N.Y. Nov. 15, 2004) (ECF No. 1)); PX0254 at ‘641 (2005
presentation to Federal Reserve); PX0004 at ‘046 (2007 presentation to General Accountability Office noting “[w]e
consider our market to be [GPCC] cards; debit is a different market”); see also PX2072 ¶ 200 (Defendants’ expert,
Dr. Gilbert, endorsing a market of “general-purpose credit and charge cards” in 2008); Tr. at 5160:14-5164:2
(Gilbert).)
46
a.
Dr. Katz’s Price Sensitivity Analysis
In determining that the relevant market includes only general purpose credit and charge
card network services, Dr. Katz applied the price sensitivity inquiry outlined in the Department
of Justice and Federal Trade Commission Horizontal Merger Guidelines, which has been used
routinely by courts in the Second Circuit as a means of applying the reasonable
interchangeability standard. See U.S. Dep’t of Justice & Fed. Trade Comm’n, Horizontal
Merger Guidelines § 4.11 (2010) (“Merger Guidelines”); see also Todd, 275 F.3d at 202;
AD/SAT, 181 F.3d at 228-29; Emigra Grp., LLC v. Fragomen, Del Rey, Bernsen & Loewy,
LLP, 612 F. Supp. 2d 330, 352 (S.D.N.Y. 2009) (recognizing the Merger Guidelines as a “tool
used to define a relevant market”). Under this test, a relevant product market is properly defined
if a hypothetical profit-maximizing monopolist that is the only seller of the product(s) included
in the proposed market could profitably impose a small but significant and non-transitory price
increase (“SSNIP”)—i.e., “without losing so many sales to other products that its price became
unprofitable.” Visa I, 163 F. Supp. 2d at 335; see also Emigra Grp., 612 F. Supp. 2d at 352. By
contrast, if buyers are able and inclined to switch away from the product(s) in numbers sufficient
to render the SSNIP unprofitable, the proposed market definition likely needs to be expanded.
Recognizing that a substantial portion of the discount fees charged to merchants is passed
through to card issuers in the form of interchange—or, in the case of American Express’s
proprietary channel, to the issuing side of its business—Dr. Katz first tested the viability of his
proposed network services market using a SSNIP applied to the network or “switch” fee. (Tr. at
3914:25-3915:11 (Katz); PX2702 at 56, 59.) The network fee, which is the price charged by the
network for clearing a given transaction, may be thought of as a two-sided price; that is, by
offsetting the discount fees collected from merchants (via acquirers) with the amount passed
47
through to issuers (and ultimately to cardholders), an analyst may approximate the amount
retained by the monopolist network as the price for intermediating each transaction. Using an
analysis performed by American Express that compared its network business to those of Visa
and MasterCard, Dr. Katz calculated that a 10% increase in the network fee charged by a
monopolist with a 100% price-cost margin would have to experience a reduction in charge
volume by over 9.1% in order to be unprofitable. (Tr. at 3915:12-3917:11, 3922:9-3923:23
(Katz); DX2702 at 59.) Both the size of the SSNIP and extreme margin assumed by Dr. Katz,
which maximizes the “pain” to the network associated with merchant substitution away from
GPCC network services, ensured a conservative application of this test. (Tr. at 3922:14-20
(Katz).)
Dr. Katz additionally considered the effect of a SSNIP on the average merchant discount
rate, appreciating that merchants’ acceptance decisions are driven by the full price paid for
accepting credit cards and not just the relatively small network component of that price. (Tr. at
3924:3-24 (Katz) (describing his “belt and suspenders approach” to market definition); PX2702
at 61.) Using the difference between Amex’s average discount rate and the pass-through rate it
pays to third-party issuers as a conservative estimate for Amex’s gross margin, Dr. Katz
calculated that a 5% SSNIP applied to the overall discount rate charged to merchants by a
monopolist would be profitable unless such an increase resulted in a 23% decrease in credit and
charge card volume across all merchant segments. (Tr. at 3924:3-24 (Katz).) As this inquiry
focused on only the merchant side of the GPCC platform, Dr. Katz accounted for the two-sided
features at play in this system by (1) assuming that the price charged to cardholders (or the
rewards given) via the issuing bank remained unchanged and (2) allowing for the possibility that
the SSNIP might result in cross-platform feedback effects. (Id. at 3904:2-10 (Katz) (noting his
48
analysis is “sensitive to the fact that we’re dealing with two-sided platforms here”); see also id.
at 5177:8-5178:12 (Gilbert) (testifying that one way to adapt the hypothetical monopolist test to
the two-sided context is to raise price to one side and hold the price to the other side constant).)
The profitability of a price increase in the network services market is ultimately
determined by how much charge volume would be lost to other payment systems in response.
Though, as a general matter, merchants and their customers jointly make the decision of which
method of payment is used for any given transaction, the customer neither sees nor pays the
additional cost when networks increase the price of network services to merchants (other than in
the form of higher retail prices, which are paid by all consumers); thus, the customer cannot be
expected to initiate substitution in the first instance. Instead, the “relevant consumer” for
purposes of assessing price sensitivity in the proposed market and for identifying reasonably
interchangeable substitute products is the merchant. (Tr. at 3864:21-3865:6 (Katz).) The
merchant has two choices when faced with an increase in the price of network services: (1) It
can attempt to steer its customers to other forms of payment to the extent permissible by law,
including untargeted discounting to debit cards under the Durbin Amendment; or, and more
importantly with respect to the court’s market definition analysis in this case, (2) it can defect
from the network and cease accepting credit cards entirely. (Tr. at 3864:21-3865:25 (Katz).)
The court concurs in Dr. Katz’s determination that it is implausible that merchants
accounting for either 9.1% or 23% of all credit and charge card volume in the United States
would cease to accept credit cards in the face of either a 3 basis point (10% SSNIP to the
network fee) or 13 basis point (5% SSNIP to the average discount rate) increase in the total price
charged by the network, respectively. (Id. at 3922:22-3924:24 (Katz); PX2702 at 60-61; see also
Tr. at 3917:23-3919:15 (Katz).) Dr. Katz further determined, and the court agrees, that
49
untargeted discounting to debit is unlikely to serve as a meaningful constraint on a monopolist’s
prices. (Tr. at 3866:3-3867:22 (Katz) (noting untargeted discounting is unlikely to be a costeffective response to increased prices for GPCC network services due to merchants’ inability to
distinguish those customers who would have used debit without the offered incentive).)
However, because merchant demand for network services is derived from cardholders’ desire to
pay with credit and charge products, the decision to stop accepting GPCC cards is strongly
influenced by the payment preferences of the merchant’s customers, and their willingness to
switch to other products. (See, e.g., id. at 3875:12-18 (Katz).)
In weighing whether to defect from credit and charge card acceptance in response to
higher discount fees, each merchant will likely conduct its own cost-benefit analysis. A rational
economic actor would be willing to drop credit card acceptance only if the incremental profit
earned on transactions that could be successfully switched to other methods of payment (i.e., the
cost difference between GPCC cards and the less expensive form of payment) would be greater
than the losses associated with those customers who would shop elsewhere or spend less if they
could not pay with a credit or charge card. (Tr. at 3867:23-3871:24 (Katz).) Because the
merchant’s profit margin is likely to significantly exceed the price differential between credit and
other forms of payment, however, the foregone profits associated with each transaction lost to a
competitor because the merchant no longer accepts credit will significantly exceed the per
transaction savings realized from successfully shifting a customer to a less expensive payment
method. (Id. at 3870:23-3872:5 (Katz); see also PX2702 at 35 (demonstrative); PX0111 at ‘826
(Amex using similar logic in convincing merchants to accept its premium rates).) As a result,
and as Dr. Katz noted in his testimony, a merchant would need to lose relatively little business
associated with those customers who insist on using their credit cards in order to conclude that
50
“the economically rational thing to d[o] is to keep accepting credit and charge cards.” (Tr. at
3871:20-24 (Katz).)
The merchant’s decision to switch away from GPCC network services, therefore, is
largely driven by its estimate of its credit-insistent customers. (Id. at 3867:23-3871:24, 3875:143876:1, 3899:18-3901:19 (Katz); see also PX0111 at ‘826; Tr. at 3872:8-3874:12 (Katz) (Amex
employs similar logic in negotiating with its merchants).) Based on American Express’s own
estimates of insistence levels among its cardholder base—and the reasonable assumption that the
percentage of credit-loyal customers would significantly exceed the percentage of Amex-loyal
customers (see Tr. at 3896:10-3898:10 (Katz) (noting Amex cardholders may be prompted to use
another GPCC card, which they likely carry given the ubiquity of Visa and MasterCard, but may
not be as easily switched from GPCC cards to another form of payment entirely); PX1240 at
‘103; PX2702 at 48)—Dr. Katz concluded that it was extremely unlikely that merchant attrition
would reach the thresholds necessary to render either SSNIP unprofitable (see Tr. at 3917:203919:15, 3922:22-3924:24 (Katz)). Similarly, data of individual customers’ spending patterns
(correlated using loyalty card numbers) collected from five of the nation’s largest supermarkets
and drugstores indicates that approximately 25% to 35% of customers shopping at these
merchants exclusively use a GPCC card. (See DX7828 at 31; see also Tr. at 3893:2-3894:24
(Katz) (describing these customers as the “set of people that are loyal to credit, and that the
merchant would be concerned about losing if it were to decide not to accept credit cards”).) This
credit-insistent business would be put at risk by dropping acceptance of GPCC cards altogether.
Moreover, the determination that merchant attrition is unlikely to reach levels necessary to
render the SSNIPs unprofitable is confirmed by the quantitative evidence showing the response
among merchants to the actual price increases imposed by American Express during its Value
51
Recapture initiative, see infra Part IV.C.1, as well as the effect—or rather, lack thereof—on
GPCC card network services prices following the decline of debit network prices under the
Durbin Amendment, see infra Part III.A.2.b.
As in Visa, however, there is limited direct quantitative evidence in the record from
which the court might make a definitive calculation of either merchants’ or cardholders’
sensitivity to pricing changes in the network services market. 15 Notwithstanding this difficulty,
the court is persuaded by Dr. Katz’s logical and thorough analyses of likely merchant responses
to a monopolist’s imposition of a SSNIP in the proposed product market, and finds it highly
unlikely that merchant attrition resulting from such a move would be sufficient to render it
unprofitable, given the high rates of credit-insistent spend merchants would place at risk by
switching away from credit card acceptance. See Emigra Grp., 612 F. Supp. 2d at 352
(acknowledging that “[d]irect evidence of cross-elasticity of demand is rare,” and requiring
courts instead “to look to a number of ‘criteria designed to focus, directly or indirectly, on crosselasticity’” (citation omitted)). Yet the court’s determination that the applicable product market
does not include debit network services is not premised solely on a formalistic implementation of
the reasonable interchangeability standard, particularly in light of the complexities introduced
into this analysis by the two-sided nature of the GPCC platform, and the varying levels of
vertical integration that may tend to obscure market realities. See Evans & Noel (2005) at 696
15
Defendants dispute the probative value of Dr. Katz’s application of the SSNIP test in this case due to a lack of
concrete quantitative data on price sensitivity among cardholders and merchants, noting instead that Dr. Katz’s
analysis is merely “another way of expressing one’s judgment about the qualitative evidence.” (Tr. at 6219:186221:4 (Bernheim).) The court disagrees. While there is limited direct quantitative data on price sensitivity among
cardholders or merchants, the court finds Dr. Katz’s analysis to be both carefully executed and persuasive. Emigra
Grp., 612 F. Supp. 2d at 352 (“Direct evidence of cross-elasticity of demand is rare.”). Moreover, the more
pragmatic, factual analysis Defendants favor also supports Dr. Katz’s proposed market definition. See infra Part
III.A.2.b. Nor does the court find fault in Dr. Katz’s decision to begin his SSNIP analysis by assuming a test market
equivalent to that adopted by the district court and Second Circuit in Visa, namely, the market for general purpose
credit and charge card network services, rather than beginning with a single-brand market for American Express
cards alone. (See Tr. at 4057:10-23, 4059:11-24 (Katz).)
52
(urging caution when defining antitrust markets in two-sided platforms, noting “mechanical
market definition exercises are particularly likely to obscure market realities” given the
complexity of such systems).
b.
Competitive Realities in the Network Services Market
The “pragmatic, factual approach” to market definition advocated by American Express,
which relies on evidence of the “actual dynamics of the market” rather than concrete data
regarding cross-elasticity, confirms Dr. Katz’s analysis and provides further support for the
court’s determination that debit cards and their related acceptance services are not reasonably
interchangeable with credit and charge products. See Brown Shoe, 370 U.S. at 336 (noting
courts should apply “a pragmatic, factual approach to the definition of the relevant market and
not a formalistic, legalistic one”); Geneva Pharm., 386 F.3d at 495-500 (noting that in defining
product markets, “[t]he emphasis always is on the actual dynamics of the market rather than rote
application of any formula”).
In the decade since Visa, the payment systems market has undergone a significant
evolution as consumers are increasingly using their debit cards to purchase goods and services at
the point of sale. The number of households carrying debit cards has more than quadrupled since
1995, and today nearly as much purchase volume flows across debit networks as across GPCC
networks. (Tr. at 6227:2-6228:10 (Bernheim); DX7828 at 10-11.) To American Express, debit’s
dramatic growth over the past decade, together with new data that purport to show both an
increased willingness among customers to use their debit and GPCC cards interchangeably and
also greater alignment between cardholders’ perceptions of these two platforms, provides
evidence that the product markets used in Visa are no longer viable and that debit must be
included in the relevant product market in this case. But the market in this case cannot be
53
defined solely by reference to cardholders’ views on substitutability between debit and GPCC
cards at the point of sale, as customers’ preferences are relevant only insofar as they provide
insight into how merchants might respond to a price increase as to credit and charge cards.
Instead, the critical question for the court is whether debit card network services are reasonably
interchangeable with GPCC network services from the merchant’s perspective such that they
impose some form of competitive discipline on the suppliers of the latter, such as American
Express, Visa, and MasterCard. 16 While merchants’ acceptance decisions are primarily driven
by customer preferences, it is essential to recognize that other characteristics of credit and debit
acceptance services are also material to that calculus, including training costs, speed of pay, and
customer service. (See Tr. at 3831:22-3832:8 (Katz).) For the following reasons, the court finds
that the actual dynamics in the marketplace for network services confirm that debit is properly
excluded from the relevant product market in this case.
First, the record suggests a prevailing view among the GPCC card networks themselves
that debit cards and acceptance services do not meaningfully affect the prices they charge to their
merchants. See Brown Shoe, 370 U.S. at 325 (noting that among the “practical indicia” of
reasonable interchangeability are “industry or public recognition of the [proposed market or]
submarket as a separate economic entity” and whether “distinct prices” are charged); Geneva
16
Dr. Bernheim’s conclusions, to the contrary, focused almost entirely on cardholders’ inclination to substitute
between credit and debit products, which, while certainly relevant to the reasonable interchangeability of the
associated acceptance services from a merchant’s perspective, does not tell the whole story. For example, Dr.
Bernheim analyzed data regarding individual customers’ spending patterns at specific supermarkets and drugstores,
which showed that some customers were switching between debit and credit products, even when shopping at the
same merchant. (Tr. at 6261:12-6262:15, 6267:4-6269:5 (Bernheim); DX7828 at 31-33, 35-36.) But Dr. Bernheim
made no attempt to link that substitution to changes in the price for network services (see Tr. at 6635:22-6636:23
(Katz)), and did not explain why between two-thirds and three-quarters of the customers whose spending patterns
were analyzed used only credit or debit and did not switch between the two forms of payment. (Id.) Defendants’
expert also cited new data from a survey conducted by the Boston Federal Reserve to illustrate the changing
perceptions of credit and debit products (id. at 6237:21-6244:3, 6248:22-6249:17 (Bernheim); DX7828 at 26), but
his analysis appears to have overstated the significance of the shift in consumer attitudes (see id. at 6641:4-6643:15
(Katz)). For the reasons set forth here and in the remainder of this section, the court finds Dr. Bernheim’s and
Defendants’ arguments concerning debit’s inclusion in the relevant market to be unpersuasive.
54
Pharm., 386 F.3d at 496; Emigra Grp., 612 F. Supp. 2d at 356 (“‘[E]vidence of industry or public
recognition of [a proposed market or] submarket as a separate economic unit’ is important in
determining its relevance for antitrust purposes ‘because we assume that economic actors usually
have accurate perceptions of economic realities.’” (internal citation and quotation marks omitted)
(alterations in original)). American Express, for example, sets its pricing tables for each
merchant segment by reference to Visa’s and MasterCard’s all-in credit card rates, not their debit
card rates or blended credit/debit rates. (Tr. at 2562:16-2563:3, 2564:17-2565:4 (Funda/Amex);
PX1240 at ‘091 (illustrative pricing methodology).) As explained by the American Express
executive responsible for global merchant pricing, the company’s pricing methodology compares
Amex’s discount rates to the all-in credit rates for Visa and MasterCard, rather than their blended
credit/debit rates, because credit possesses “a sufficiently different feature set” when compared
to debit and “a sufficiently different cost structure . . . [such] that it should be priced on its own
merits and not combined with debit.” (Tr. 2730:17-23 (Funda/Amex).) Indeed, there is very
little indication in the record that American Express views its merchant pricing as being
constrained by debit rates. Discover, which offers both debit and credit products, similarly does
not consider the price of debit services when setting merchant pricing for its credit card network
services. (See id. at 818:16-23 (Hochschild/Discover).) American Express additionally rebuffs
attempts by merchants to compare its higher discount rates to blended credit/debit rates charged
by Visa and MasterCard, and instead recommends that merchants compare Amex’s pricing
against its competitors only “on a credit to credit basis.” (See PX0010 (“[Amex] made it clear
we do not compete with debit so we didn’t include it in [the rate] analysis.”); PX1110 at ‘311
(recommending that merchants consider Amex’s pricing premium only against what they “pay
for similar Visa/MasterCard credit cards”); see also PX0068 at ‘514 (in a meeting with one large
55
retailer, Amex representatives explained that comparing Amex to Visa and MasterCard “on a
credit to credit basis” is “a more accurate comparison” than a comparison including debit).)
Second, with regard to the cardholder side of the platform, product substitution at the
point of sale is limited by the core functional differences between debit and GPCC cards. See
Brown Shoe, 370 U.S. at 325 (considering indicia such as the “product’s peculiar characteristics
and uses” and where the products have “distinct customers”). Debit cards are “pay now”
products, allowing cardholders to make purchases using the funds deposited in their linked
demand deposit accounts. (See PX1408 at 22 (Amex 2009 Form 10-K).) Credit cards, by
contrast, are “pay later” products by virtue of the attached credit facility; cardholders may pay
for their purchases in the future and carry a balance beyond the month in which a purchase was
made. Similarly, charge cards enable customers to defer payment by virtue of the float, and
increasingly allow cardholders to carry a balance forward to the next billing cycle. See supra
Part I.A. Put simply, debit cards enable consumers to deduct from their existing funds—similar
to checks—while spending on credit and charge cards results in an accruing balance that can be
paid in the future. GPCC cards also offer an array of ancillary benefits to cardholders that
typically are not offered on debit cards, including robust rewards programs. (Tr. at 3880:193883:6 (Katz).)
While debit and GPCC network services need not be perfect substitutes to be included in
the same product market, see AD/SAT, 181 F.3d at 227-28, American Express has itself
acknowledged that the functional differences between these two payment systems limit their
interchangeability at the point of sale 17 and affect customers’ spending behaviors. For example,
17
Indeed, in the last Form 10-K filed by American Express before initiation of this lawsuit, the company publicly
acknowledged that “[t]he ability to substitute debit cards for credit and charge cards is limited because there is no
credit extended and the consumer must have sufficient funds in his or her demand deposit account to pay for the
56
many consumers compartmentalize or tailor their spending to some degree on their GPCC or
debit cards depending on the type or size of a transaction. (Tr. at 754:9-24 (Quagliata/Amex)
(recognizing that “many people compartmentalize their spend . . . us[ing] different products for
different reasons”), 3701:1-3703:5, 3744:7-3745:5 (Silverman/Amex); see also id. at 819:17820:5 (Hochschild/Discover) (testifying to the compartmentalization phenomenon in the
payment card industry).) According to one study performed by American Express, consumers
tend to use their debit cards “when a purchase feels mundane or ‘everyday’” or when a “purchase
falls below a personal threshold (anywhere from $20-$200).” (See PX2543 at ‘624, ‘634, ‘643
(indicating approximately 40% of Amex cardholders compartmentalize between credit and debit
to some degree); see also Tr. at 3885:19-3889:10 (Katz) (tailoring is affected by the
characteristics of both the consumer and the transaction); PX2702 at 42-47; Tr. at 6122:10-15,
6151:23-6152:6 (Mitchell/Official Payments) (noting Official Payments’s customers tend to use
debit on purchases below $300, and credit or charge on purchases above that amount); DX7828
at 30 (median transaction size on debit is approximately 25% lower than on GPCC cards).)
Conversely, cardholders are more likely to put luxury or big-ticket purchases on their credit or
charge cards. (See, e.g., Tr. at 563:22-564:24 (Bouchard/Sears), 1231:25-1232:12
(Kimmet/Home Depot); PX2702 at 43.) Business travelers likewise tend to use GPCC products
when making purchases at T&E merchants in order to benefit from the deferred payment options
while awaiting reimbursement by their employers. (See id. at 251:10-22 (Thiel/Alaska Airlines),
5908:2-18 (Flueck/Starwood Hotels).)
Though consumers in the United States are undeniably turning to debit cards with greater
frequency than in the past—which, in the court’s view, may be attributable in part to shifting
purchase at the time of the transaction, as opposed to charge cards where payment is due at the end of the month or
credit cards where payment can be extended over a period of time.” (PX1408 at 22 (Amex 2009 Form 10-K).)
57
attitudes toward credit in the wake of the Great Recession—the functional differences between
debit and credit cards have not changed since the decisions in Visa. Merchants recognize that
these qualitative differences influence customer payment behavior, and that the resulting
spending patterns limit their ability to switch away from GPCC acceptance. (See, e.g., Tr. at
387:8-388:7 (Robinson/Ikea) (recognizing that Ikea’s “experience has shown us different
customers[] have different payment preferences, and we have to offer those choices,” and
discussing various rationales customers might use when compartmentalizing spend), 564:13565:9 (Bouchard/Sears) (testifying that Sears would be at a competitive disadvantage if it shifted
to debit-only acceptance given customer payment preferences), 1232:13-24 (Kimmet/Home
Depot), 1525:9-21 (O’Malley/Best Buy), 2402:10-18, 2404:7-16 (Priebe/Southwest); see also id.
at 5908:15-18 (Flueck/Starwood Hotels) (Starwood must continue to accept credit cards or risk
losing business travelers).)
Third, the evidence adduced at trial indicates that merchants do not view debit and GPCC
acceptance services as reasonably interchangeable. Even if debit’s increasing market share and
shifting consumer perceptions as compared to GPCC products were taken to signal a greater
inclination toward substitutability among cardholders, American Express has presented no
evidence to suggest that merchants—the relevant consumers in this analysis—are willing to
switch away from GPCC cards in favor of debit. To the contrary, the testimony elicited at trial
suggests that merchants are sensitive to the spending preferences and credit-insistence of their
customer base and, as a result, do not view debit network services as an economically viable
substitute for GPCC networks services given the revenue that presumably would be lost to their
credit-accepting competitors. 18 Other merchants strongly prefer payment by credit and charge
18
See Tr. at 252:25-253:2 (Thiel/Alaska Airlines) (“Q: Can Alaska [A]irlines as an institution substitute the
acceptance of debit cards for the acceptance of credit cards? A: No.”), 387:8-388:7 (Robinson/Ikea) (noting that
58
cards for reasons independent of cardholder demand, particularly when the merchant requires
some form of security for a purchase. (See Tr. at 482:5-484:8, 485:7-23 (Satkowski/Enterprise)
(Enterprise strongly prefers customers use credit because such cards provide superior identity
verification and security for car rentals), 1610:2-1611:21 (Brennan/Hilton) (Hilton would
actively discourage guests from using a debit card at check-in because the hotel must take an
“overage” out of the linked checking account to ensure there are sufficient funds to cover
incidental charges, and it may take up to six weeks for the funds to be refunded), 5906:175907:20 (Flueck/Starwood Hotels). American Express’s effort to undercut the merchant
testimony presented at trial by pointing to the fact that some merchants use its blended
debit/credit rate, which would be lower than a credit-only rate, in attempting to negotiate a lower
discount rate with Amex is unpersuasive. 19 (See, e.g., id. at 2284:11-23 (Berry/Amex).) Such
use is, in the court’s view, more reasonably understood as a negotiating tactic employed by
merchants in the hopes of negotiating a lower effective rate.
Fourth, and finally, the reaction—or, more accurately, the lack thereof—among
merchants and the credit card networks to the significant decline in debit interchange rates that
occurred as a result of the Durbin Amendment is further evidence that debit does not constrain
Ikea’s “experience has shown us that different customers [] have different preferences, and we have to offer those
choices”); 564:20-565:9 (Bouchard/Sears) (testifying that Sears would be at a competitive disadvantage were it not
to accept GPCC cards), 1231:1-1232:24 (Kimmet/Home Depot) (given Home Depot’s large average ticket size, it is
“almost required to accept credit cards” and noting that if it dropped GPCC cards “a segment of the customer base . .
. would clearly move to our competitors”), 1525:9-21 (O’Malley/Best Buy) (testifying that Best Buy has never
considered accepting only debit cards because “consumers are expecting to pay with credit cards”), 5908:2-18
(Flueck/Starwood Hotels) (noting that Starwood must accept GPCC cards because business travelers generally book
travel on their credit cards to allow time for reimbursement); see also id. at 1233:10-25 (Kimmet/Home Depot)
(testifying that Home Depot would likely continue accepting GPCC cards in the face of a 10% increase in the
discount rate), 1683:24-1684:5 (Dale/Sprint) (same).
19
Only one merchant, Jetro, appears to have successfully negotiated a lower discount rate based on the lower debit
rates that took effect after the Durbin Amendment. (Tr. at 2282:17-2284:10 (Berry/Amex).) Yet even in that case,
internal Amex documents made it clear that “Jetro will need to be educated on why debit and credit are separate
products and why AXP should only be compared to credit.” (PX0011 at ‘627.) Accordingly, the court finds little
probative weight in this example, and certainly not enough to outweigh the prevailing evidence presented at trial that
credit and debit are viewed by networks and merchants alike as distinct product groups.
59
credit card network pricing and thus does not belong in the relevant market. See Brown Shoe,
370 U.S. at 325 (advising courts to consider evidence of “sensitivity to price changes”). The
Durbin Amendment authorized the Federal Reserve to regulate interchange rates on debit cards
issued by certain financial institutions, see Dodd-Frank Act § 1075, 15 U.S.C. § 1693o-2, which
was accomplished by regulation in October 2011, see Regulation II, Debit Card Interchange Fees
and Routing, 12 C.F.R. § 235.3. According to data from the Federal Reserve, following
implementation of the Durbin Amendment the average debit interchange rate in the United States
fell by 37% (see Tr. at 3926:10-24 (Katz); PX2702 at 63 (showing decline in the debit
interchange rate from 1.24% to 0.78% in the three months following implementation)), and Dr.
Katz calculated that all-in debit rates actually paid by merchants (including interchange, network,
and acquirer rates) fell nearly 30% (Tr. at 3927:4-3928:10 (Katz); see also id. at 3929:5-25
(Katz) (noting two-sided price declined as well)). 20 Were debit in the relevant market, the court
would expect to have seen the falling debit rates place downward pressure on GPCC prices as
merchants switched away from credit network services to debit, given the increased price
differential between the two forms of payment. (Id. at 3925:1-3926:3 (Katz).)
Yet the market’s reaction was quite different. Following implementation of the Durbin
Amendment in 2011, there was no significant merchant attrition from the major credit card
networks in favor of debit services, notwithstanding the increased savings that could be
recognized by switching consumers from credit to debit cards. 21 (See id. at 819:14-16
20
The court is not persuaded by Dr. Bernheim’s argument that the proper measure of the Durbin Amendment’s
impact in the marketplace is an unweighted average of rates paid by merchants, as opposed to the figures used by
Dr. Katz and the Federal Reserve to measure the Durbin Amendment’s effect on debit rates, which are weighted by
charge volume. (Tr. at 6519:15-6520:24 (Bernheim).) While the raw measure of how many merchants did or did
not see a decline in their debit rates is interesting, the more relevant metric for determining interchangeability from
the merchant’s perspective, including how much incremental profit a merchant might realize by dropping GPCC
cards entirely, cannot be calculated or even approximated on an unweighted basis.
21
The anecdotal evidence provided by those merchants that testified at trial supports the observations of Dr. Katz.
Despite the decline in debit rates after the Durbin Amendment, none of these merchants saw any decline in their
60
(Hochschild/Discover), 2723:2-19 (Funda/Amex), 3931:3-3932:21 (Katz) (noting “merchants
did not respond by dropping credit cards en masse,” and that in fact, merchant acceptance grew
during this period).) And, relatedly, lower debit rates did not result in any appreciable decrease
in the price of credit card network services charged to merchants by American Express and its
fellow GPCC networks. (See id. at 820:6-8 (Hochschild/Discover) (Discover did not reduce
rates in response to Durbin), 3930:1-7 (Katz) (noting “very little, if anything” happened to GPCC
prices); see also PX0920 (expressing concern about potential merchant responses to falling debit
rates, but noting “fact [is] that debit and credit are not substitutes in the consumer’s (or Durbin’s)
mind”); PX0089 at ‘343 (Amex analysis of likelihood merchants would switch to debit in
response to the Durbin Amendment).) Thus, the natural experiment afforded by implementation
of the Durbin Amendment for observing the degree of cross-elasticity between GPCC and debit
acceptance services, or, for that matter, the degree to which debit fees constrain pricing by and
the business behavior of GPCC card networks, provides additional support for excluding debit
from the relevant market. 22
B.
Plaintiffs’ Proposed T&E Submarket
Within the boundaries of the broader network services market, Plaintiffs additionally
propose that the court recognize a submarket consisting of GPCC card network services provided
to T&E merchants. (See Tr. at 3912:15-19, 6645:16-19 (Katz); Pls. Post-Trial Br. (Dkt. 606) at
credit card pricing. (Tr. at 251:23-252:8 (Thiel/Alaska Airlines), 407:13-408:6 (Robinson/Ikea), 1525:24-1526:6
(O’Malley/Best Buy), 1611:22-1612:5 (Brennan/Hilton).) Nor did any of the merchants switch away from GPCC
acceptance after Durbin. (See id. at 1234:12-1235:5 (Kimmet/Home Depot), 1525:9-1526:3 (O’Malley/Best Buy),
2321:21-2322:7 (Bruno/Crate & Barrel), 5297:16-5298:5 (Gutierrez/Strictly Bicycles); see also id. at 2723:5-8
(Funda/Amex) (testifying that he was unaware of any merchant that ceased accepting GPCC cards and relied on
debit).)
22
Of course, declining interchange fees on debit cards cannot be equated with lower prices for debit network
services, as these are two separate elements of the overall debit discount fee. Nonetheless, the absence of any
meaningful reaction among merchants to both lower interchange rates and lower all-in rates for debit acceptance
supports the court’s determination that debit and credit network services are not reasonably interchangeable from the
merchants’ perspective.
61
34-35.) See also Geneva Pharm., 386 F.3d at 496 (“Reasonable interchangeability sketches the
boundaries of a market, but there may also be cognizable submarkets which themselves
constitute the appropriate market for antitrust analysis.” (citing Brown Shoe, 370 U.S. 325)).
Relying primarily on American Express’s practice of segmenting its merchant base by industry,
and setting different pricing tables for each merchant segment, Plaintiffs attempt to establish the
existence of their submarket by proving a hypothetical monopolist would similarly be able to
segment and target T&E merchants for higher discount rates. (See Tr. at 3906:20-3912:19,
6645:16-6646:16 (Katz).) See also Merger Guidelines § 4.1.4 (recognizing that markets may be
defined around “targeted consumers” in limited circumstances). Courts have recognized that
these so-called “price discrimination markets” may be defined around a set of consumers that are
particularly vulnerable to such practices. See, e.g., E.I. du Pont de Nemours & Co. v. Kolon
Indus., Inc., 637 F.3d 435, 448 (4th Cir. 2011); Spirit Airlines, Inc. v. Nw. Airlines, Inc., 431
F.3d 917, 935 (6th Cir. 2005); see also 2B Phillip E. Areeda, Herbert Hovenkamp & John L.
Solow, Antitrust Law ¶ 534d(1), at 269 (3d ed. 2007) [hereinafter Areeda & Hovenkamp]
(“Successful price discrimination means that the disfavored geographic or product class is
insulated from the favored class and, if the discrimination is of sufficient magnitude, should be
counted as a separate relevant market.”). (See also Tr. at 3911:2-7 (Katz), 6296:25-6297:10
(Bernheim).)
Here, however, Plaintiffs have failed to establish that a hypothetical monopolist’s ability
to charge a higher price in T&E industries constitutes price discrimination such that it would be
appropriate to define a submarket around these “targeted customers.” See Merger Guidelines
§§ 3, 4.1.4. All four GPCC networks pursue pricing strategies that require the ability to segment
their merchant bases by industry and to charge different rates to different segments. (See Tr. at
62
733:14-22 (Quagliata/Amex) (describing industry segmentation as the “guiding criteria” for
Amex’s pricing strategy), 3906:20-3907:23, 3909:18-3910:7 (Katz).) But differential pricing is
discriminatory, and thus relevant to the definition of the T&E submarket, only if the different
prices being charged do not reflect actual differences in the network’s costs of providing services
to the different merchant segments. (Tr. at 4093:10-14 (Katz) (agreeing that “price
discrimination is a supplier’s charging different prices to two customers after accounting for
differences in the supplier’s incremental costs of serving those two customers”).) See also
Eastman Kodak, 63 F.3d at 106-07 (“[E]vidence that Kodak film sells for different prices in
different parts of the world is insufficient to establish price discrimination without proof that
Kodak’s costs are uniform throughout the world.” (emphasis in original)); 2B Areeda &
Hovenkamp § 517, at 150-55.
Looking, as Plaintiffs do, to evidence of American Express’s pricing practices in T&E
industries, the court finds insufficient evidence in the record to conclude that the higher prices
charged therein are discriminatory. (See Tr. at 6297:19-6301:9 (Bernheim).) In attempting to
show that Defendants’ higher discount rates in T&E industries are not driven by higher costs,
and are therefore discriminatory, Dr. Katz’s analysis compared two proposed measures of
American Express’s price-cost margin on network services across multiple merchant segments.
Each measure, however, is flawed. Dr. Katz first used internal data from American Express
concerning the company’s “contribution margin” across various industries, which is a measure of
Amex’s revenue less its variable costs on each dollar of charge volume. (Tr. at 3984:1-20
(Katz); PX2702 at 87.) Not only does contribution margin fail to account for the network’s fixed
costs, resulting in an inexact measure of margin depending on how those costs are allocated by
industry, but the internal database from which these figures were drawn also does not distinguish
63
between Amex’s various business lines. As a result, the margin calculations include revenues
and costs associated with Defendants’ issuing business, not just the network services business.
(Tr. at 4102:10-4103:20 (Katz).)
Alternatively, Dr. Katz attempted to approximate American Express’s margins by
beginning with the network’s average discount rate in each industry, and then subtracting
Defendants’ third-party “issuer rate,” which is analogous to the interchange rates charged on
Visa’s and MasterCard’s systems. (Id. at 3982:11-3983:23; PX2702 at 86.) The portions of the
discount rate retained by Amex vary significantly by industry segment, and, according to Dr.
Katz, “are larger than any plausible . . . cost differences” that may exist in servicing each set of
customers. (Tr. at 3983:13-17 (Katz).) Yet this approach also fails to disaggregate Amex’s
various lines of business and provides, at best, only an indirect approximation of Amex’s
margins on network services in each industry. Similarly, the court lacks data regarding the other
networks’ costs of providing network services in the proposed T&E submarket.
Accordingly, Plaintiffs have not established that American Express, or any of its
competitors, are able to charge discriminatory prices in the industries that would appear to
comprise Plaintiffs’ proposed T&E submarket. Though a hypothetical monopolist likely would
be able to isolate and impose higher prices in such industries, the court lacks a reliable basis for
inferring that those prices would not be driven by cost differences associated with serving T&E
merchants. (See id. at 6347:19-6348:3 (Bernheim).) Additionally, Plaintiffs have not clearly
defined the contours of their proposed submarket. It is not apparent which merchant industries
would be included in a “travel and entertainment” submarket and which would not, and Plaintiffs
have not provided the court with adequate grounds for making those determinations itself. For
64
these reasons, the court finds that Plaintiffs have failed to establish a cognizable price
discrimination market around merchants in T&E industries.
*
*
*
As a result, the court concludes that Plaintiffs have established that the relevant market
for the court’s antitrust analysis in this case is the market for general purpose credit and charge
card network services in the United States. Plaintiffs have not proven the existence of a
cognizable submarket for network services sold to T&E merchants. That Plaintiffs have failed to
establish a T&E submarket is ultimately of no consequence, however, as the discussion below
holds that Plaintiffs have successfully demonstrated that Defendants possess market power in the
GPCC card network services market.
IV.
MARKET POWER
As explained in Parts IV and V of this Decision, the court concludes that Plaintiffs have
successfully discharged their initial burden under the rule of reason using both the direct and
indirect methods of proof: Amex’s NDPs have adversely affected competition in the network
services market, and American Express possesses sufficient market power to cause such effects.
In the interests of continuity and clarity, the court begins by completing the market analysis
begun in the prior section before considering Plaintiffs’ efforts to prove that the NDPs have
caused actual detrimental effects on competition.
To prevail on a market power theory of liability, Plaintiffs must prove that American
Express possesses market power in the GPCC card network services market and that there are
“other grounds to believe that the defendant’s behavior will harm competition market-wide.”
K.M.B. Warehouse, 61 F.3d at 129. Defined by the Supreme Court as the “power to control
prices or exclude competition,” E.I. du Pont, 351 U.S. at 391; Kodak, 504 U.S. at 481 (“the
65
ability of a single seller to raise price and restrict output”); see also K.M.B. Warehouse, 61 F.3d
at 129 (market power is “the ability to raise price significantly above the competitive level
without losing all of one’s business”), market power may be proven directly through evidence of
“specific conduct indicating the defendant’s power to control prices or exclude competition,” or
it may be inferred based on the defendant firm’s large share of the relevant market when viewed
in the context of the competitive dynamics therein, see Kodak, 504 U.S. at 464; K.M.B.
Warehouse, 61 F.3d at 129; Todd, 275 F.3d at 206; Commercial Data Servers, Inc. v. Int’l Bus.
Machs. Corp., 262 F. Supp. 2d 50, 73 (S.D.N.Y. 2003). The Government presents evidence on
both points. Although an assessment of whether American Express possesses antitrust market
power is a question of fact to be decided on the basis of the record developed at trial, the Visa
decisions nonetheless provide a helpful roadmap for the court’s analysis in this case. Both the
district court and Second Circuit in Visa relied on three categories of evidence in finding that
Visa and MasterCard, both jointly and separately, possessed market power in the GPCC card
network services market: (1) defendants’ market shares and the structural characteristics of the
market; (2) cardholder insistence; and (3) the networks’ pricing practices and merchants’
continued acceptance despite price increases. See Visa I, 163 F. Supp. 2d at 340-42; Visa II, 344
F.3d at 239-40.
Considering many of the same types of evidence introduced into the factual record in this
case, and for the additional reasons outlined below, the court concludes that American Express
does possess antitrust market power in the GPCC card network services market sufficient to
cause an adverse effect on competition. Specifically, the court finds that Defendants enjoy
significant market share in a highly concentrated market with high barriers to entry, and are able
to exercise uncommon leverage over their merchant-consumers due to the amplifying effect of
66
cardholder insistence and derived demand. In addition, American Express’s ability to impose
significant price increases during its Value Recapture initiatives between 2005 and 2010 without
any meaningful merchant attrition is compelling evidence of Defendants’ power in the network
service market. Plaintiffs’ other pricing arguments are less persuasive, and are ultimately
unnecessary to the court’s finding that American Express possesses market power.
A.
Market Share, Concentration, and Barriers to Entry
American Express’s percentage share of the network services market is compelling
evidence of market power. In reaching this determination, the court remains mindful that data
regarding a firm’s raw share of the relevant market is probative of market power only after “full
consideration of the relationship between market share and other relevant market
characteristics,” including the “strength of the competition, the probable development of the
industry, the barriers to entry, the nature of the anticompetitive conduct[,] and the elasticity of
consumer demand” that characterize this particular market. Tops Mkts., 142 F.3d at 98; see also
In re Payment Card, 562 F. Supp. 2d at 402. In Visa, the court inferred market power on the part
of Visa and MasterCard—which at the time accounted for 47% and 26% of total general purpose
credit card spending, respectively—based upon their “large market shares in a highly
concentrated network market with only four significant competitors.” Visa I, 163 F. Supp. 2d at
341-42; Visa II, 344 F.3d at 240. Few of the structural elements of the network services market
noted in Visa have changed in the intervening years.
Today, American Express is the second largest GPCC card network when measured by
charge volume. As of 2013, Amex accounted for 26.4% of general purpose credit and charge
card purchase volume in the United States. (Jt. Stmt. ¶ 20.) It trails only Visa’s 45% market
share, and is larger than both MasterCard (23.3%) and Discover (5.3%). (Id.) Despite Amex’s
67
protestations to the contrary, the proper metric for assigning market shares among the four
GPCC networks is the dollar value of the transactions facilitated on those networks. See Visa I,
163 F. Supp. 2d at 341; Visa II, 344 F.3d at 240; In re Payment Card, 562 F. Supp. 2d at 400; see
also Merger Guidelines § 5.2 (“In most contexts, the Agencies measure each firm’s market share
based on its . . . revenues in the relevant market.”). Although other measures of a network’s size,
such as the number of cards in circulation, the breadth of its merchant acceptance network
(whether actual or perceived), and the total number of transactions, will affect that firm’s ability
to compete in a market characterized by network effects, charge volume is the most direct
measure of output in this particular market, and is also the primary determinant of the
remuneration networks receive from merchants in exchange for network services. See Merger
Guidelines § 5.2 (“In most contexts, each firm’s market share is based on its actual or projected
revenues from the targeted customers.”). As a result, in terms of raw percentage share of the
relevant market, American Express is larger today than MasterCard was at the time of the Visa
litigation, when the Second Circuit held that MasterCard possessed market power. 23
23
Defendants attempt to undercut Plaintiffs’ market share argument by arguing that firms with under 30% market
share presumptively lack market power in the Second Circuit. (Tr. at 6932:16-23 (Closing Argument).) Defendants
find support for this market power threshold in Commercial Data Servers, Inc. v. International Business Machines
Corp., in which Judge McMahon of the Southern District of New York noted that “[c]ourts have consistently held
that firms with market shares of less than 30% are presumptively incapable of exercising market power.” 262 F.
Supp. 2d 50, 74 (S.D.N.Y. 2003) (citation and internal quotation marks omitted). Yet absent clear precedent from
the Second Circuit establishing the 30% threshold advocated by American Express—and there is none—the court is
unwilling to endorse Defendants’ unduly formalistic and arbitrary approach to market power. Market share is but
one factor considered when attempting to approximate a defendant firm’s power in a relevant market, and that a
firm’s share falls below some arbitrary threshold cannot disprove allegations of market power without reference to
the other competitive dynamics at play. See United States v. Columbia Steel Co., 334 U.S. 495, 528 (1948) (“The
relative effect of percentage command of a market varies with the setting in which that factor is placed.”); AllenMyland, Inc. v. Int’l Bus. Machs. Corp., 33 F.3d 194, 209 (3d Cir. 1994) (“‘Market share is just a way of estimating
market power, which is the ultimate consideration. When there are better ways to estimate market power, the court
should use them.’” (quoting Ball Mem’l Hosp., Inc. v. Mut. Hosp. Ins. Co., 784 F.2d 1325, 1336 (7th Cir. 1986)
(Easterbrook, J.))). Moreover, Amex’s proposed threshold is inconsistent with the Second Circuit’s finding in Visa
that MasterCard, which at the time possessed a 26% share of the network services market, possessed market power
sufficient to establish a Section 1 violation. Visa II, 344 F.3d at 240.
68
Furthermore, the network services market remains highly concentrated and constrained
by high barriers to entry, just as it was in Visa. American Express is one of only four major
suppliers of GPCC card network services, and three of the competitors in this market (Visa,
American Express, and MasterCard) are significantly larger than the fourth (Discover).
(Tr. at 3826:9-3827:9 (Katz); see also id. at 3939:5-3941:22 (Katz) (finding the HerfindahlHirschman Index for GPCC card network services is “well above” the “threshold for being
highly concentrated”).) The structural susceptibility of this market to an exercise of market
power is exacerbated by its inherently high barriers to entry, which further reduce the likelihood
that an attempt at anticompetitive conduct would be defeated by new suppliers entering the
market. See Visa I, 163 F. Supp. 2d at 342 (“The higher the barriers to entry, and the longer the
lags before new entry, the less likely it is that potential entrants would be able to enter the market
in a timely, likely, and sufficient scale to deter or counteract any anticompetitive restraints.”
(citing Merger Guidelines § 3.0)); see also In re Payment Card, 562 F. Supp. 2d at 402 (noting
that “whereas a seller in a market with low entry barriers could not raise its prices without the
risk that a new seller would enter the market and offer the same product for a lower price, a
competitor in a market with high entry barriers could raise its prices unfettered by the prospect of
a new entrant into the market who would undercut prices”). In addition to the sizable setup costs
associated with developing the infrastructure and branding necessary to compete in the network
services market, any new entrant would also need to overcome what executives from Amex and
Discover have termed the “chicken and the egg problem.” That is, due to the aforementioned
network effects inherent in this platform, a firm attempting entry into the GPCC network market
would struggle to convince merchants to join a network without a significant population of
cardholders and, in turn, would also struggle to convince cardholders to carry a card associated
69
with a network that is accepted at few merchants. (Tr. at 820:23-821:16 (Hochschild/Discover),
3942:16-3943:2 (Katz), 4296:12-24 (Chenault/Amex).) See also Visa I, 163 F. Supp. 2d at 34142. Accordingly, the network services market is not only highly concentrated, it is also
remarkably static; no firm has entered the GPCC card network services market in the United
States since Discover launched its network in 1985. 24
The rise of new digital payment options like PayPal, Square, and Google Wallet do not
pose an entry threat to American Express or the other GPCC card networks at this time. Rather
than establish their own payment networks, these services piggyback on existing methods of
payment—including credit and charge, debit, and ACH—in order to facilitate their use at both
online and brick-and-mortar merchants. (See Tr. at 3714:6-18 (Silverman/Amex), 3945:63946:13 (Katz).) PayPal, for example, functions as an electronic wallet, enabling consumers to
load multiple payment methods into a single online account and then choose which payment
option to use at the moment of purchase. (See id. at 1271:16-1272:9 (Kimmet/Home Depot),
2428:8-15 (Priebe/Southwest).) These companies are viewed by American Express and its
network rivals as GPCC-accepting merchants, not as competitors in the network services market.
(See id. at 823:19-23 (Hochschild/Discover), 3714:5-3715:1 (Silverman/Amex).) Although
electronic wallets like PayPal and Square are recognized by American Express to present unique
competitive challenges to its business, these difficulties are traceable to the fact that their
services have proven effective at steering customers to debit and ACH and that they interrupt the
24
Discover’s successful entry in 1985 may be attributed to two unique factors that are unlikely to be replicated in
today’s market. First, Discover offered a “breakthrough value proposition” for cardholders, offering cash back
rewards at no annual fee—in fact, Discover was “the first card to have any form of rewards.” (Tr. at 820:23-821:16
(Hochschild/Discover).) Second, and perhaps more importantly for overcoming the network effects just described,
Discover was initially owned and operated by one of the nation’s largest retailers, Sears, which marketed Discover’s
cards to its already significant population of private label cardholders. (Id. at 823:3-16 (Hochschild/Discover).)
Similar entry today would be “impossible,” according to Discover. (Id.)
70
typically direct relationship between Amex and its cardholders. (Id. at 3711:20-3713:23, 3715:214 (Silverman/Amex); DX7524; see also Tr. at 1270:3-18 (Kimmet/Home Depot).)
Consequently, American Express’s 26.4% share of a highly concentrated market with
significant barriers to entry suggests that the firm possesses market power. See, e.g., Visa II, 344
F.3d at 239-40; see also Toys “R” Us, Inc. v. F.T.C., 221 F.3d 928, 937 (7th Cir. 2000) (finding
firm exercised market power notwithstanding only 20% share of national wholesale market).
Yet, Amex’s market share alone likely would not suffice to prove market power by a
preponderance of the evidence were it not for the amplifying effect of cardholder insistence.
B.
Cardholder Insistence
American Express’s highly insistent or loyal cardholder base is critical to the court’s
finding of market power in this case. The ability of merchants to resist potential anticompetitive
behavior by Amex, including significant price increases, by shifting customers to less expensive
credit card networks or other forms of payment is severely impeded by the segment of Amex’s
cardholder base who insist on paying with their Amex cards and who would shop elsewhere or
spend less if unable to use their cards of choice. In Visa, both the district court and Second
Circuit recognized the amplifying effect of cardholder loyalty on Visa’s and MasterCard’s
positions in the market, noting that insistence effectively precluded merchants from dropping
acceptance of either Visa or MasterCard credit cards and supported a finding of market power as
to both networks. See Visa I, 163 F. Supp. 2d at 340-41 (noting merchants “cannot refuse to
accept Visa and MasterCard even in the face of significant price increases because the cards are
such preferred payment methods that customers would choose not to shop at merchants who do
not accept them”); Visa II, 344 F. 3d at 240. Here, the record developed at trial illustrates a
similar dynamic among Defendants’ cardholders and merchants, supported not only by merchant
71
testimony on the effect of cardholder insistence, but also by American Express itself, which
expressly recognizes, quantifies, and leverages the loyalty of its cardholders in its business
dealings with merchants.
Cardholder insistence is derived from a variety of sources. First, and perhaps most
importantly, cardholders are incentivized to use their Amex cards by the robust rewards
programs offered by the network. 25 Enrollees in American Express’s Membership Rewards
program, for example, receive points for purchases made with their Amex cards, and may then
redeem those points with Amex or one of its redemption partners for merchandise, gift cards,
frequent flyer miles, statement credits, or other goods and services. (Tr. at 3548:13-3549:22
(Silverman/Amex), 4298:20-4300:13 (Chenault/Amex).) Cardholders who value the ability to
earn points, miles, or cash rebates often centralize their spending on their Amex cards to
maximize these benefits. (See PX0426 at ‘649.) Similar “single-homing” behavior is also
observed among the approximately 10-20% of Amex cardholders who own or regularly carry
only their Amex cards (PX0815 at ‘290; DX7249 at ‘207; DX7828 at 85-86), as well as among
those cardholders who consolidate their credit card spending on their American Express cards for
other reasons. Amex’s industry-leading corporate card program, for instance, drives a significant
degree of insistent spending, particularly at those T&E merchants that cater to the needs of
business travelers. (Tr. at 2569:13-2570:9 (Funda/Amex), 3962:3-3964:19 (Katz), 6378:166379:20 (Bernheim); PX2486 at ‘053 (stating Amex captured 64.3% of corporate card spend in
first half of 2013).) Indeed, according to one study by American Express, approximately 70% of
25
Tr. at 3962:3-24 (Katz) (noting Amex’s “very attractive rewards program” to be “the big source of insistence” for
most Amex cardholders), 4759:15-29 (Glenn/Amex); PX0426 at ‘649 (Amex presentation noting “research indicates
strong loyalty to American Express Cards” “[d]riven by [the] ability to earn points, miles, or cash rebates” and that
many cardholders “use American Express exclusively to consolidate rewards”); PX0111 at ‘812 (asserting
“American Express rewards programs drive greater loyalty” in pitch to Alaska Airlines); see also PX0815 at ‘284
(study noting 84% of Amex cardholders are enrolled in one of Amex’s rewards programs).
72
Corporate Card consumers are subject to some form of “mandation” policy, by which employers
require the employee-cardholders to use Amex cards for business expenses. (PX0634 at ‘112.)
As in Visa, Plaintiffs also have presented merchant testimony illustrating the manner in
which cardholder insistence effectively prevents merchants from dropping American Express.
While a number of merchant witnesses testified that their companies had never considered
terminating acceptance of Amex due to the network’s share of the merchants’ receipts or a
generalized concern that their customers would shop elsewhere if unable to use their American
Express cards, 26 others have analyzed the issue in detail and arrived at the same conclusion: The
foregone profits associated with losing Amex-insistent customers rendered dropping Amex
commercially impractical. 27 Though American Express may be fairly characterized as a
discretionary card for consumers when compared to the ubiquity enjoyed by Visa and
MasterCard (see Tr. at 4312:10-22 (Chenault/Amex)), the degree to which its cardholders insist
on using their Amex cards affords the network significant power over merchants, particularly in
a market in which merchants’ primary recourse when faced with a price increase or similar
conduct is an “all-or-nothing” acceptance decision. (See id. at 3974:18-3976:12 (Katz).)
Defendants’ efforts to minimize the significance of cardholder insistence by recasting it as mere
26
E.g., Tr. at 247:25-248:13 (Thiel/Alaska Airlines), 573:6-574:5 (Bouchard/Sears) (testifying that Sears had never
considered dropping American Express because “we would lose an unacceptable amount of sales”), 1262:23-25
(Kimmet/Home Depot), 2322:8-25 (Bruno/Crate & Barrel) (stating that he would be “extremely nervous” about
adverse effects on sales associated with dropping Amex because of the “large percentage of spend coming through
[Amex]”), 1606:4-18 (Brennan/Hilton) (stating Hilton would likely lose about two-thirds of its current Amex charge
volume if it no longer accepted American Express), 3146:2-6 (Gibson/Sinclair) (noting it would “be crazy not to
take [American Express]” at Sinclair’s hotel properties because it constitutes “34 percent of the business”).
27
E.g., id. at 389:10-390:10 (Robinson/Ikea) (noting there had been “internal discussions about whether or not it
would be feasible to drop American Express,” but that after conducting internal surveys among its customers, Ikea
concluded the company could not drop Amex without “suffering a loss in sales” attributable to Amex-insistent
cardholders shopping elsewhere), 491:1-494:8 (Satkowski/Enterprise) (testifying that Enterprise determined it could
not drop Amex because its “corporate customers were not interested in paying fo[r] their rental with a different
method of payment”), 1529:6-1536:22 (O’Malley/Best Buy) (stating that Best Buy had conducted a “war game” to
evaluate feasibility of dropping Amex, but after estimating levels of insistence among various cardholder segments
it concluded “the numbers [were] pretty stark” that Best Buy should continue acceptance), 1687:12-1690:12,
1759:13-23 (Dale/Sprint) (Sprint twice considered dropping Amex, but did not do so because “[t]here was a concern
that we would lose customers . . . if we made that decision,” particularly among corporate cardholders).
73
“brand loyalty” are unavailing. Amex I, 21 F. Supp. 3d at 200-01 (finding Defendants’ brand
loyalty argument unconvincing).
Nor may insistence be dismissed as a mere marketing ploy. American Express itself uses
insistence-based calculations to inform its pricing strategy and to persuade merchants of the
importance of accepting its cards. When evaluating its industry-specific rate tables, for example,
Amex’s pricing team begins with a baseline rate equal to the Visa/MasterCard all-in credit card
rate (i.e., the average rate charged across Visa’s and MasterCard’s various card offerings) and
then adds the value of “the incremental business that [Amex] bring[s] to [its] merchants,” which
the company equates with the amount of spend attributable to its insistent cardholders. (Tr. at
2563:5-22 (Funda/Amex) (acknowledging that “[t]he metric that we use to back into how much
incremental volume we bring to our merchants is an insistence-based metric”).) The company
tracks and applies three measures of insistence for this purpose: (1) “walk away” insistence,
which captures those cardholders who would shop elsewhere if unable to use their Amex cards,
(2) “spend less” insistence, which captures those cardholders who would shop less frequently at
a merchant or spend less per visit, and (3) corporate insistence, which captures the effect of
employer mandation policies among Amex’s corporate cardholders. (Id. at 2567:23-2570:9
(Funda/Amex); see also 2571:9-2573:15 (Funda/Amex); PX1240 at ‘091, 102-03 (Amex
calculations of “total insistence” for use in its pricing strategy).) Amex regularly surveys its
cardholders to estimate the degree of insistence in each industry segment to ensure its
calculations reflect actual market realities. (Tr. at 2570:10-2571:11 (Funda/Amex).) Defendants
use these figures to approximate the amount of charge volume and associated profit realized by
participants in each industry segment that it believes is traceable to insistent spend—i.e., the
incremental value of Amex acceptance or, alternatively, the business put at risk by defecting
74
from the network—and, in turn, to inform its decision-making when identifying opportunities to
raise merchant prices relative to the value it believes it delivers. 28 (Id. at 2639:14-22, 2819:192820:16 (Funda/Amex); PX1240 at ‘104; see also Tr. at 3957:20-3961:16 (Katz) (testifying that,
based on his review of Amex’s pricing methodology, Defendants “recognize[] insistence is one
of the things that gives them the ability to charge merchants higher prices”).)
When negotiating with merchants, Defendants also rely on the restraining effects of
cardholder insistence to explain why ceasing to accept American Express cards would be
unprofitable. In 2008, for example, when justifying a price increase imposed as part of a
repricing initiative termed “Value Recapture,” see infra Part IV.C.1, American Express reminded
merchants in the airlines group that the network’s “highly insistent cardholders” or “loyalists”
were responsible for hundreds of millions in charge volume that would be put “at risk” by not
accepting the price increase, rendering it “essential [for the airlines] to accept American
Express.” (See, e.g., PX0111 at ‘806, ‘814 (Alaska Airlines); PX1601 at ‘263, ‘271 (Southwest);
PX0517 at ‘026 (American Airlines).) Similarly, in a standardized presentation used by Amex
client managers when justifying a series of Value Recapture price increases to restaurants in
2010, Amex paired its standard value proposition to merchants with a warning that, according to
its data, nearly 50% of Amex cardholders “[w]ould no longer dine, [w]ould dine less often or
would spend less” if the restaurants chose not to accept American Express. (PX0957 at ‘900,
‘914, ‘916-17, ‘921, ‘923-24.)
28
Though Amex’s value-based pricing methodology is only one of several inputs the network uses when setting its
headline discount rates (see Tr. at 2564:18-2565:4 (Funda/Amex)), the conceptual framework used by Amex’s
pricing team nonetheless demonstrates Defendants’ own reliance on estimates of cardholder insistence in the
ordinary course of business, as well as a recognition by the network that the uncommon loyalty of its cardholders
amplifies its leverage over merchants. This is, in effect, the same dynamic observed in Visa. See Visa I, 163 F.
Supp. 2d at 340-41.
75
The existence and practical effect of cardholder insistence on merchant choice are not
merely theoretical, as demonstrated by the various “real world” examples cited by Plaintiffs. For
example, when Murphy Oil, a chain of gas stations located primarily in Wal-Mart parking lots,
terminated its acceptance of American Express cards during the Great Recession in 2008, the
network tracked the response among its cardholders and found that the actual insistence rate it
observed among Murphy Oil’s customers was twice what the company’s research had previously
estimated. (See PX0031 at ‘668, ‘671.) As noted by Jack Funda, the head of Amex’s pricing
unit, in an email to his colleagues, the data illustrated that Murphy Oil’s decision to terminate
was “irrational” and that “this case example suggests that [cardholder] insistence in Oil is real
and strong -- we should be able to make use of this data in our merchant negotiations.” (Id.
at ‘668.) Murphy Oil eventually resumed its acceptance of American Express. (Tr. at 2703:2325 (Funda/Amex).)
Even the nation’s largest merchants are not immune to the effects of cardholder
insistence. Walgreen, which was at the time the ninth largest retailer in the United States, was
forced to retreat from a 2004 decision to terminate acceptance of American Express in the face of
public outcry from its customers, who told the drugstore that they would take their business to a
competitor if unable to use their Amex cards. (See id. at 1343:2-4, 1352:1-1399:7
(Rein/Walgreen).) Like Murphy Oil, Walgreen ultimately “capitulated” to American Express
and agreed to a new acceptance agreement containing the pricing terms that were substantially
similar to those the retailer had previously deemed unacceptable and which had led to its
decision to drop Amex. (Id. at 1363:11-1365:19, 1398:18-1399:4, 1517:2-11 (Rein/Walgreen)
(testifying that Walgreen believed Amex’s offer of a 10 basis point decrease to its discount rate
was unacceptable given the magnitude of American Express’s premium over competitor rates,
76
but that Walgreen ultimately agreed to a similar decrease—albeit on a different timeline);
PX1966; PX1969 at ‘367; PX1965; DX2143 at ‘943 (noting Amex’s proposal left it with a 50
basis point premium over Visa/MasterCard).)
This is not to suggest, however, that merchant defection is something Amex takes lightly;
losing a merchant hurts the network’s bottom line as the merchant presumably also has loyal
customers and, perhaps more importantly, risks spillover effects at other acceptance locations.
(See Tr. at 5958:3-5959:5 (McNeal/Amex); see also id. at 1626:18-1628:6 (Brennan/Hilton).)
Nonetheless, the experiences of Murphy Oil and Walgreen, together with the other “natural
experiments” cited by Plaintiffs, illustrate the manner in which cardholder loyalty, and the
prospect of losing insistent charge volume by terminating acceptance, constrains merchants’
ability to resist anticompetitive behavior by American Express. (See PX0031 at ‘671 (noting
cost to Murphy Oil of defection far exceeded the cost to Amex).)
Finally, the court is unconvinced by Defendants’ argument that cardholder insistence
cannot be a source of durable market power. (See Defs. Post-Trial Br. (Dkt. 605) at 44-45; see
also Tr. at 5067:6-5068:9 (Gilbert), 6350:6-6351:25 (Bernheim).) Though Defendants are
correct that transitory market power is not of particular concern under the federal antitrust laws,
the requirement that market power be “durable” speaks to whether a new entrant or other market
forces could quickly bring the defendant’s exercise of power to an end. See Geneva Pharm., 386
F.3d at 509 (“[A] transitory advantage does not significantly harm competition and therefore
should not violate § 1 . . . .”); AD/SAT, 181 F.3d at 229; see also 2B Areeda & Hovenkamp
¶ 510, at 110 (“[T]ransitory power may safely be ignored by antitrust law. The social costs of
antitrust intervention (including its error potential) are likely to exceed the gains when market
forces themselves would bring the defendant’s power to an end fairly quickly.”). The court is
77
aware of no authority that supports Defendants’ position that market power is not durable if its
maintenance requires continual and replicable investment by the defendant firm. Put simply,
American Express cannot avert a finding of market power premised on cardholder insistence
merely because that loyalty and its current market share would dissipate if the company were to
stop investing in those programs that make its product valuable to cardholders. Of course it
would, as would the share of any company that abandoned a core element of a successful
business model. Here, the durability of Defendants’ power is ensured by the sustained high
barriers to entry in the network services market, see supra Part IV.A, as evidenced by the lack of
any meaningful entry into the market since 1985, and the decades-long persistence of the
restraints at issue in this case. See United States v. Microsoft, 253 F.3d 34, 82 (D.C. Cir. 2001);
Colo. Interstate Gas Co. v. Natural Gas Pipeline Co. of Am., 885 F.2d 683, 695 n.21 (10th
Cir. 1989).
C.
Pricing Practices
Certain of Amex’s pricing practices provide direct evidence of the company’s market
power in the network services market, albeit to varying degrees. As discussed below, the record
shows that between 2005 and 2010, American Express repeatedly and profitably raised its
discount rates to millions of merchants across the United States as part of its Value Recapture
(“VR”) initiative without losing a single large merchant and losing relatively few small
merchants as a result. Similar evidence of low defection rates among merchants following
repeated network price increases was viewed by the district court in Visa as strong evidence of
Visa and MasterCard’s market power. See Visa, 163 F. Supp. 2d at 340 (in finding defendants
possessed market power, noting “both Visa and MasterCard have recently raised interchange
78
rates charged to merchants a number of times, without losing a single merchant customer as a
result”). The court finds the same is true here.
In addition, Plaintiffs aver that American Express’s ability to price discriminate between
various industry segments, and its stated policy of maintaining a pricing premium over the mixadjusted rates of Visa and MasterCard, are additional evidence of market power. While
probative of Amex’s ability to control prices in the network services market, this evidence is less
persuasive than that regarding Value Recapture and, in the court’s view, is not necessary to its
finding that American Express possesses market power.
1.
Value Recapture
Faced with a declining premium over the all-in rates charged by Visa and MasterCard in
the early 2000s, see infra at Part IV.C.3, American Express executed a series of targeted price
increases in certain industry segments between 2005 and 2010, with the stated purpose of better
aligning its prices with the value it perceived as being delivered to both cardholders and
merchants. (Tr. at 1983:2-15 (Berry/Amex), 4399:3-4400:2 (Chenault/Amex), 6038:23-6039:7
(McNeal/Amex).) Because these Value Recapture initiatives were not paired with offsetting
adjustments on the cardholder side of the platform, the resulting increases in merchant pricing
are properly viewed as changes to the net price charged across Amex’s integrated platform. (Tr.
at 3985:5-24 (Katz) (noting that without “offsetting changes on the cardholders’ side, . . . the
two-sided price [across Amex’s platform], increased”).) Given the low rates of merchant
defection observed in response to this initiative, which increased prices that were already at or
above the competitive level, Value Recapture illustrates Amex’s “successful exercise of market
power.” (Id. at 3985:3-24, 3989:13-3991:15 (Katz) (noting “American Express was starting
79
from prices that were not below competitive levels” and that these “two-sided price increases
were profitable. . . . [as] [t]hey both raised revenues and they raised profits”).)
Taken as a whole, Amex’s Value Recapture initiatives comprised at least twenty separate
price increases accomplished through a combination of increased discount rates, new or
increased per transaction fees, and reduced side payments to merchants. (PX0121 at ‘459.) The
increases were imposed on an industry-specific basis, with several merchant segments—typically
those with relatively high rates of cardholder insistence—targeted for multiple rounds of price
hikes. (PX1240 at ‘100-04 (overview of VR as of 2010); see, e.g., PX0056-A at ‘237-38 (VR in
airlines industry); PX0778 (VR in lodging).) For example, between 2007 and 2010, American
Express moved its airline merchants to a higher discount rate table, resulting in the headline
discount rates for many airlines rising between 7% and 15% over the four year period, and
driving over $90 million in additional pre-tax income to the network. (PX0056-A at ‘237.) The
restaurant industry was also subject to three separate rounds of Value Recapture increases
between 2007 and 2010, with individual merchants subject to some combination of the following
pricing “levers”: a 5-15 basis point increase in the discount rate, a new 30 basis point “Card Not
Present Fee,” and/or a $0.05 fixed fee per transaction, as well as subsequent expansions of the
same. (PX0062 at ‘392-93, ‘395-96.) In whole, the Value Recapture programs targeting
restaurants increased prices for at least 280,000 restaurants in the United States. (Tr. at 717:2224 (Quagliata/Amex); PX0062 at ‘393, ‘395.) In 2009 alone, Amex aimed to increase prices for
over one million small merchants and more than 3,000 larger, managed merchants in the United
States. (PX0121 at ‘461.) Value Recapture was not limited to T&E industries; Amex also
targeted so-called “everyday spend” merchants like supermarkets and retailers. (See, e.g.,
PX1201 (Giant Eagle); PX0121 at ‘461-62.) On the whole, these price increases affected a
80
“substantial portion” of Amex’s acceptance network, with merchants accounting for 65% of
American Express’s annual charge volume paying higher prices for its network services as a
result of Value Recapture. (PX1240 at ‘100; see also PX0121 at ‘459; PX1240 at ‘088
(estimating Value Recapture raised Amex’s average discount rate in the United States by nearly
9 basis points).)
Although American Express’s decision to adjust its pricing in response to perceived
increases in its costs on either side of its integrated platform is not itself evidence of market
power (see Tr. at 6337:9-17 (Bernheim)), the company’s ability to profitably impose such price
increases across a broad swath of its merchant base with little or no meaningful buyer attrition is
compelling proof of such power (see id. at 3985:3-3990:12 (Katz)). Notwithstanding the breadth
of the Value Recapture program, the court is unaware of any large merchant in the United States
that elected to cancel its acceptance of Amex cards in response to the Value Recapture price
increases. (Tr. at 2079:6-2080:10 (Berry/Amex), 2675:23-2676:4 (Funda/Amex); PX1099 at
‘555 (noting 100% retention among largest global merchants); see also id. at 4739:25-4740:3
(Glenn/Amex) (testifying that “no large merchants canceled”).) Similarly, in Amex’s Regional
Client Group (or its organizational predecessor), which was responsible for managing the
accounts of the approximately 9,000 merchants with annual Amex charge volume between $3
million and $100 million, the network was able to retain over 99.9% of merchants following
Value Recapture increases in both 2009 and 2010. (See PX0706 at ‘676, ‘685; PX1000 at ‘929;
Tr. at 622:24-625:5 (Quagliata/Amex).) In 2010, for example, only three merchants in this group
voluntarily stopped accepting Amex, and it is not clear whether those decisions were related to
Value Recapture. (Tr. at 742:22-746:1.) Finally, among the millions of small merchants without
designated Amex client managers, American Express appears to have concluded that Value
81
Recapture was profitable on the whole, even though the network observed higher rates of
cancellation and card suppression among this population when compared to its larger merchants.
(PX1753-A at ‘033; Tr. at 2677:18-2679:20 (Funda/Amex).)
American Express, for its part, contests the Government’s assertion that Value Recapture
may be viewed as a profitable endeavor. It asserts that the price increases were the result of
hotly contested negotiations, and notes that the increases were at least partly offset by significant
concessions the network was forced to make to certain large merchants to ensure their continued
acceptance of American Express cards, such as deferrals of the rate increases, additional
marketing funds, and similar financial consideration. (Tr. at 4401:16-23 (Chenault/Amex),
4740:4-4741:12 (Glenn/Amex); see also PX0121 at ‘463 (noting usage of such “relief valves”
was below projections).) Yet the record concerning the profitability of Value Recapture is clear.
Internal Amex documents show that the Value Recapture initiatives resulted in $1.3 billion in
incremental pre-tax income for Amex over the five-year period from 2006 to 2010, and nearly a
9 basis point improvement to Amex’s weighted average discount rate in the United States.
(PX0357 at ‘949.) Even after accounting for the concessions Amex paid to retain certain
merchant accounts and the forgone profits associated with the minimal merchant cancellations
that did occur during the period—whether or not those cancellations are traceable to the price
increase—the evidence plainly shows that Value Recapture was profitable on a return-oninvestment basis. (PX1753-A at ‘032-33; PX0008 at ‘487-89; Tr. at 2688:12-2689:11
(Funda/Amex) (VR for small merchants was profitable when considering merchant attrition,
though that determination did not include intangible effects like changing perceptions of
coverage among cardholders).) While the court recognizes that Value Recapture likely had
certain secondary effects, including implications for the network’s ability to acquire new
82
merchants and cardholders’ perception of coverage, Defendants have not shown that these
“speculative” or intangible costs rendered the increases unprofitable on the whole. (Tr. at
2683:3-21 (Funda/Amex).) 29
Value Recapture was ultimately ended in 2010 by American Express due to merchant
dissatisfaction with the price increases given the economic climate at the time and not, it would
appear, due to any competitive pressures imposed by Amex’s competitors in the GPCC network
services market. (See Tr. at 2690:1-24 (Funda/Amex), 5717:11-5718:2 (Gilligan/Amex) (VR
ended because “the pain that it was causing to many of our relationships with merchants was not
worth the gain”).)
2.
Price Discrimination
Next, Plaintiffs rely on Amex’s ability to price discriminate between various industry
groups as evidence of the network’s market power. See Visa I, 163 F. Supp. 2d at 340-41
(noting Visa and MasterCard’s ability to “charge differing interchange fees based, in part, on the
degree to which a given merchant category needs to accept general purpose cards,” in other
29
Specifically, a number of American Express witnesses cited the harm done to the network’s relationship with
Continental Airlines as evidence of the financial damage caused by Value Recapture, which is not captured in the
profitability analyses just described. As part of its Value Recapture initiative in the airline industry, American
Express raised Continental’s headline discount rate in January 2008 by approximately 10%, which equated to
several million dollars in additional discount fees paid by the airline annually. (See PX0056-A at ‘237; PX0211 at
‘046; DX3821 at ‘533, ‘560-01; PX1033 at ‘151.) More than two years later, in June 2010, Continental decided not
to renew its participation in Amex’s Membership Rewards program as a redemption partner, and withdrew from
Amex’s airport lounge access program, pursuant to which Platinum and Centurion cardholders enjoyed free access
to a number of airlines’ lounges. (PX1033 at ‘151; Tr. at 5720:16-5722:8 (Gilligan/Amex).) While these moves
were undoubtedly detrimental to American Express’s bottom line and the utility of its cards to cardholders (see Tr.
at 5721:22-5723:19 (Gilligan/Amex)), Defendants’ efforts to establish causation between the VR price increases and
the airline’s decision to alter its relationship with Amex is not convincing. To the contrary, the contemporaneous
record evidence plainly shows that while Continental was unhappy with the magnitude of the price increase imposed
years earlier, the primary reason for its withdrawal from Membership Rewards and the lounge access program was
to avoid heavy penalties under its new co-brand contract with Chase. (See PX1033 at ‘151 (“As we’ve discussed
with several members of your team over the past year, any extension of M[embership] R[ewards] is very difficult
because of our co-brand agreement with Chase.”); see also Tr. at 5471:4-14 (Codispoti/Amex) (testifying that the
network was “fairly certain that Continental would exit the [Membership Rewards] program” because of
Continental’s co-brand relationship with Chase); PX1203-A at ‘245 (“Re-signing these Amex partnerships would
result in significant penalties to Continental from their co-brand issuer (Chase).”).)
83
words, cardholders’ credit-insistence, “illustrates their market power”). Yet the court finds such
evidence to be of limited probative value in this case.
While it is beyond dispute that American Express sets different prices for merchants
depending on the industry segment to which they belong, this fact alone does not prove its
market power. As the court previously discussed in connection with Plaintiff’s proposed T&E
submarket, American Express’s practice of charging different prices to different merchant groups
is discriminatory only if those prices are unrelated to differences in the costs associated with
providing network services to each group. See supra Part III.B. Yet Plaintiffs have not provided
a reliable measure of American Express’s per transaction margins across its industry groups, and
the court accordingly determines there to be inadequate grounds to find price discrimination
sufficient to establish antitrust market power. (Tr. at 6347:19-6348:3 (Bernheim).)
Nonetheless, like the district court in Visa I, this court does take note of the ease with
which American Express is able to identify and target merchant segments for differential pricing
based on its estimates of merchant demand in each industry—demand which, as previously
discussed, and as recognized in Visa, is largely a reflection of the degree to which Amex
cardholders insist on using their cards.
3.
Merchant Pricing Premium
Defendants’ express pricing strategy of charging merchants a premium over its
competitors’ rates presents similar difficulties to the court when proffered as evidence of market
power. That American Express may charge a higher price to merchants than Visa and
MasterCard, two firms previously found to possess market power in the relevant market, is not
necessarily proof that such prices are supracompetitive; merchants may be receiving
commensurate value for the higher price, similar to the manner in which Lamborghini and
84
Toyota both sell cars, but the former can charge a higher price because it offers a differentiated,
and ostensibly superior, product. (See Tr. at 5063:7-13, 5086:6-11 (Gilbert).) See also Xerox
Corp. v. Media Scis., Inc., 660 F. Supp. 2d 535, 549 (S.D.N.Y. 2009) (noting that
“‘[c]ompetitive markets are characterized by both price and quality competition, and a firm’s
comparatively high price may simply reflect a superior product’” (quoting Harrison Aire, Inc. v.
Aerostar Int’l, Inc., 423 F.3d 374, 381 (3d Cir. 2005))). Here, as discussed below, there is
conflicting evidence as to whether American Express still maintains a premium price over Visa
and MasterCard and whether that premium, to the extent it exists, represents a supracompetitive
price or due compensation for Amex’s higher quality product. Given this ambiguity, evidence
relating to Defendants’ premium pricing strategy provides only limited support for a finding of
market power 30 and is ultimately cumulative support for the court’s conclusion that American
Express has had and continues to have market power in the network services market.
The success of American Express’s differentiated business model, according to the
network, is largely predicated on its ability to charge a premium over its competitors’ all-in
prices to merchants. (See Tr. at 709:15-710:3 (Quagliata/Amex), 2643:15-2644:9 (Funda/Amex)
(testifying that the absence of a premium would “stall[] our closed-loop engine . . . of making
sure we had enough revenue to invest and differentiate our products”), 3544:15-22
(Silverman/Amex); PX0051 at ‘677 (“Maintaining our rate premium is an integral part of driving
positive results to our P&L.”); PX0357 at ‘948, ‘952, ‘960-61 (charging a “reasonable premium”
is one of Amex’s core “pricing principles”), ‘953 (describing need for premium to offset scale
30
The probative value of Plaintiffs’ evidence concerning Amex’s pricing premium also is undercut by the fact that
it speaks only to one component of the net or two-sided price charged across Defendants’ partially integrated
platform—i.e., the merchant discount rate. (See Tr. at 6340:2-7 (Bernheim).) As the court later explains, the
evidentiary record does not include a reliable measure of the two-sided price charged by American Express that
correctly or appropriately accounts for the network’s expenses on the cardholder side of the platform, from which
the court might draw comparisons to Visa and MasterCard’s pricing. See infra Part.IV.D.
85
disadvantages and invest in differentiated product); see also Tr. at 3979:5-7 (Katz) (stating that
“[o]ne element of American Express’[s] pricing strategy has been to have premium pricing”).)
Indeed, among Amex’s primary justifications for the Value Recapture rate increases between
2005 and 2010 was the erosion of its premium over Visa and MasterCard during the preceding
years, which, according to the network, endangered its ability to deliver on its distinct value
propositions on both sides of the platform. (See DX7537 at ‘489 (noting Value Recapture was in
part “[t]riggered by the on-going decline of our Discount Rate premium”); PX0051 at ‘677.)
Nonetheless, American Express has successfully pursued a premium pricing strategy for
decades, most recently in 2013—the last year for which data was provided to the court—
maintaining an 8 basis point and 3 basis point premium over Visa and MasterCard, respectively,
on a mix-adjusted basis. 31 (See PX2702 at 85.) Historically this premium has been even higher.
(See PX0357 at ‘959 (showing premiums between 28 and 89 basis points between 1997 and
2009); Tr. at 3599:8-12 (Silverman/Amex) (noting we “have traditionally earned a premium
discount rate”).) Not surprisingly, the magnitude of Amex’s premium varies by industry, with
larger premiums maintained in those industries where American Express delivers more
insistence-driven value to merchants, such as airlines, rental cars, lodging, and other T&E
segments. (See Tr. at 2705:14-2707:24 (Funda/Amex); PX0357 at ‘960; see also PX1240 at
‘091 (illustrating that Amex’s premium pricing is informed in part by the “surplus value” it
31
American Express’s merchant discount rates are typically compared to those of its competitors on a mix-adjusted
basis, meaning Defendants’ headline rates—which are the same for all Amex card products—are compared on a
weighted basis to a comparable “mix” of Visa and MasterCard card products in order to ensure an apples-to-apples
comparison. (See Tr. at 919:5-20 (Hochschild/Discover) (agreeing that to fairly compare merchant discount rates
one should do an “apples to apples” comparison that compares fees on like card products), 2567:1-22
(Funda/Amex).) Yet the court recognizes that merchants’ acceptance decisions often are not so nuanced; from the
merchant’s perspective, the relevant price when comparing the network services provided by the four GPCC
networks likely is the total per transaction cost, regardless of what type of Amex, Visa, MasterCard, or Discover
card is used. (See, e.g., id. at 385:10-386:5 (Robinson/Ikea), 474:18-477:19 (Satkowski/Enterprise).) Without
adjusting for mix, Amex’s premiums appear significantly larger. (See PX2702 at 85.)
86
delivers to merchants, which the network measures using internal estimates of cardholder
insistence).)
American Express rejects Plaintiffs’ use of its higher merchant prices as evidence of
market power, arguing instead that its premium rates have been and are justified by the
differentiated value it delivers to merchants. (See Tr. 3599:8-16 (Silverman/Amex).) American
Express does, in fact, deliver on its differentiated value proposition to merchants in many
respects. See supra Part I.B. (See also PX1408 at 10 (Amex 2009 Form 10-K).) For instance,
Amex delivers, on average, more affluent cardholders who are “ready to spend” at participating
merchants. 32 In particular, Amex cardholders tend to spend more on average per transaction, 33
spend more on an annual basis per card, and spend more often 34 than cardholders on competitor
networks. (See DX6576 at 10.) Furthermore, Amex leverages its “closed-loop” model to
provide merchants with data analytics, targeted marketing solutions, fraud protection, and other
business-building benefits not provided to the same degree by its competitors. (See, e.g., Tr.
at 4305:7-4306:17 (Chenault/Amex), 4720:19-4723:15 (Glenn/Amex), 2117:2-2119:4, 2277:425 (Berry/Amex); DX7598 at ‘015-17.)
Yet a number of merchant witnesses at trial disputed the actual value of these additional
services, noting, among other things, that American Express can and does sell that same closedloop data to their competitors as well. (See, e.g., Tr. at 575:8-576:8 (Bouchard/Sears), 2437:3-
32
See PX0111 at ‘809 (noting average household income for Amex cardholders is 21% higher than that of noncardholders’ households); DX6576 at 10.
33
Numerous merchant witnesses called by Plaintiffs testified that American Express had the largest average ticket
size for any GPCC network. (See, e.g., Tr. at 433:10-12 (Robinson/Ikea), 588:14-18 (Bouchard/Sears), 1297:241300:4 (Kimmet/Home Depot), 2347:3-8 (Bruno/Crate & Barrel).) Amex additionally delivers incremental value to
airlines and other T&E merchants due in large part to its strong corporate card business. (See PX1601 at ‘273
(Amex presentation showing its cardholders have “1.6 times more air tickets purchases v. non-Cardholders” and
“3.2 times higher spend on air travel”); PX0111 at ‘809 (Amex presentation stating that its cardholders purchase
first or business class tickets 2.3 times as often as non-cardholders). )
34
See DX7238 at ‘375; Tr. at 3290:5-14 (Biornstad/MasterCard).
87
2438:10 (Priebe/Southwest), 5375:15-5376:23, 5395:24-5396:21 (Miller/Delta).) In addition,
American Express plainly lags its competitors in other aspects of its merchant services, including
the speed with which merchants receive payment from the network. (See Tr. at 2581:24-2582:9
(Funda/Amex).) Amex’s own biannual survey data on merchant satisfaction indicate that Amexaccepting merchants do not believe they receive commensurate value from the network in return
for its higher discount rates. When a representative sample of merchants were asked in 2010
how they would “rate the value” they receive from the three primary networks given the prices
they pay, the company found that the “[p]erceived value is significantly higher for Visa/MC than
for Amex” among large and small merchants alike. (See PX0043 at ‘965-66 (showing a
“Performance Gap” between Amex and Visa/MasterCard on the cost-value question averaging
nearly 20% among managed, unmanaged, and OnePoint merchants); Tr. at 1795:11-1796:13,
1803:11-1804:15 (Ford); see also id. at 1804:21-1807:12, 1811:23-1814:20 (Ford) (finding
similar results on “value” in 2006 and 2012 surveys supported the reliability of 2010 results).)
Several American Express executives testified at trial that the network no longer
maintains any premium over Visa or MasterCard on a mix-adjusted basis. (See Tr. at 702:7-25
(Quagliata/Amex), 2666:24-2667:2 (Funda/Amex), 4403:4-15 (Chenault/Amex).) However, no
data or expert analysis was proffered to substantiate these assertions, and the court is hesitant to
rely on the self-interested statements of Defendants’ executives absent some form of
documentation. (See id. at 6343:19-6344:1 (Bernheim) (relying on prior testimony of Amex
executives to support absence of premium, not independent evidence).) Nonetheless, Amex’s
premium plainly has eroded over time. (See PX0357 at ‘959.) While this decay can be traced in
part to Amex’s declining net effective discount rate, which is primarily the result of intentional
business decisions by American Express as discussed in the following section, see infra
88
Part IV.D, the erosion of Defendants’ premium is in large part a result of rising prices on Visa’s
and MasterCard’s networks. (PX0357 at ‘952, ‘959 (showing more than 80% of erosion due to
price increases by Visa and MasterCard); PX0028 at ‘398; see Tr. at 2667:3-6 (Funda/Amex),
4426:5-4427:22 (Chenault/Amex).) Beginning around 2006, for example, both Visa and
MasterCard introduced new premium card categories with higher interchange rates intended to
enable issuers to more effectively compete with Amex’s high-rewards products. (See PX0028 at
‘396 (Amex presentation noting “MasterCard and Visa are targeting our premium economics by
introducing higher interchange product categories”); DX0447 at ‘445 (explaining development
of Visa Signature products); DX7116 at ‘788; DX4961 at ‘306; DX7237 at ‘015.) Issuing banks
promptly and aggressively began converting cardholders to these new premium offerings,
shifting Visa’s and MasterCard’s overall card mixes toward higher interchange categories and—
together with new fees charged by the networks—driving up the networks’ all-in discount rates.
(PX0357 at ‘939-43; see also id. at ‘959 (showing Visa’s effective all-in rate rose nearly 15 basis
points from 2006 to 2009 as issuers moved cardholders to Visa’s higher interchange categories).)
As Visa’s and MasterCard’s rates rose, Amex’s premium shrunk.
Even if, as Defendants contend, Amex no longer maintains a premium in the network
services market, evidence of its historical pricing is nonetheless material to the court’s market
power analysis. (See Tr. at 4151:17-4152:15, 4256:7-4257:3 (Katz).) In addition, given the
manner in which the premium gap was narrowed by Visa and MasterCard over the years, the
court cannot credit Amex’s argument that the erosion of its premium is compelling evidence of a
lack of market power. See also infra Part IV.D. Quite to the contrary, the virtual impunity with
which Visa and MasterCard were able to raise their merchant pricing suggests an absence of
inter-network competition on the basis of price attributable to rules prohibiting merchant
89
steering, which is a condition Amex has been able to perpetuate even after Visa and MasterCard
abandoned their anti-steering rules as a result of this litigation. Nonetheless, given the absence
of clarity with respect to whether Amex maintains a premium in today’s market and whether
such premium is or has been justified by the network’s differentiated value propositions, the
court finds Plaintiffs’ evidence of Amex’s pricing premium to be of limited utility in the present
market power analysis.
D.
Amex’s Remaining Market Power Counterarguments
American Express’s remaining arguments regarding whether it possesses antitrust market
power are unavailing.
First, to the extent American Express’s average effective discount rate has declined over
time, that decrease does not show a lack of market power. To the contrary, the record indicates
that any reduction in Amex’s average effective rate is primarily the result of the network’s
successful efforts to increase its share of spending at so-called “everyday spend” merchants. (Tr.
at 2649:17-2654:19 (Funda/Amex).) These industry segments, which include supermarkets, gas
stations, and pharmacies, generally pay significantly lower discount rates to American Express
than merchants in the types of T&E industries that traditionally had formed the core of Amex’s
acceptance network. (Tr. at 2650:9-2651:1 (Funda/Amex); PX0357 at ‘960.) As the overall mix
of merchants at which Amex cardholders were spending shifted toward lower-priced industry
groups, the network’s average discount rate across all industries fell accordingly. 35 (Tr.
at 2657:19-2663:1 (Funda/Amex); PX0791 at ‘144-45 (quantifying negative impact of various
types of uncontrollable mix effects on Amex’s 2008 and 2009 average discount rates); PX1753-
35
Additionally, it appears that while Amex’s average effective discount rate continued to decline during Value
Recapture (see Tr. at 5718:3-16 (Gilligan/Amex)), those price increases more than offset all sources of downward
pressure on Amex’s overall rates except those attributable to changes in mix. (PX1753-A at ‘029, ‘034-35 (noting
“Value Recapture initiatives continue to offset controllable rate investments globally”).)
90
A at ‘029.) This was an “intentional move” by American Express, which recognized that
changes to its inter-industry mix were the “primary” reason its average discount rate had been
declining since the late 1990s. (PX0890 at ‘338 (“This shift from predominantly T&E to a more
balanced industry mix has decreased our overall rate.”); PX0254 at ‘647; PX0004 at ‘055.)
Amex’s belief that its rate was falling as a result of mix effects was confirmed by Dr. Katz’s
analysis. When Plaintiffs’ economics expert controlled for the changing composition of Amex’s
merchant base, he found that the network’s average effective discount rate had, in fact, increased
slightly over time. (Tr. at 6654:11-6656:2 (Katz); PX2778 at 5.)
Nor is the court is swayed by Dr. Bernheim’s “two-sided price” calculations, which are
intended to capture the all-in price charged to merchants and consumers across Defendants’
entire platform on a per transaction basis. 36 Defendants’ expert calculates Amex’s two-sided
price—which, in Amex’s view, is the proper metric for analyzing price effects in this case—by
offsetting the headline discount rates charged to merchants with (1) any payments made to
merchants, including payments made pursuant to agreements other than card acceptance
agreements, and (2) payments made to cardholders in the form of rewards. (Tr. at 6311:106312:13 (Bernheim); DX7828 at 54.) According to Dr. Bernheim’s estimates, American
Express’s two-sided price has fallen precipitously since 2002, indicating that in his opinion the
network does not possess the power to control price in the relevant market. (Tr. at 6314:10-21
(Bernheim); DX7828 at 56-57.)
These calculations are flawed in a number of respects, however. On the merchant side of
the platform, Dr. Bernheim improperly applied billions of dollars in remuneration paid to a
36
Cardholders effectively pay a “negative” price for acceptance services in Amex’s GPCC platform in the form of
rewards earned on a per transaction basis. (Tr. at 3852:11-3853:2, 3905:17-3906:5, 4021:16-24 (Katz).) See also
Evans & Schmalensee (2007) at 151 (“[P]rofit-maximizing prices may entail below-cost pricing to one set of
customers over the long run and, as a matter of fact, many two-sided platforms charge one side prices that are below
marginal cost and are in some cases negative.”).
91
handful of merchants by American Express in connection with its co-brand agreements to reduce
the network’s average effective discount rate. 37 (Tr. at 6657:8-6658:3, 6661:11-24, 6662:36663:3 (Katz); PX2778 at 6; see also Tr. at 6556:18-6557:1 (Bernheim).) Payments made to
obtain or retain co-brand partnerships—which benefit the issuing side of Amex’s business by
opening new channels for acquiring cardholders (DX5561 at ‘128; Tr. at 5420:12-5421:9
(Codispoti/Amex))—should not, in the court’s view, be used to offset the price paid by those
companies for network services in their capacity as Amex-accepting merchants. (See Tr. at
6658:15-6659:1 (Katz); see also id. at 2717:7-18 (Funda/Amex); PX0842 at ‘284; PX0999 at
‘041 (“An issuer co-brand relationship is independent from the card acceptance relationship and
needs to support its own value proposition.”).) This determination is consistent with American
Express’s internal business practice of not including the co-brand remuneration identified by Dr.
Bernheim when estimating its own effective discount rate. (See, e.g., PX0791 at ‘144.) Dr.
Bernheim’s adjustments to Amex’s average effective discount rate is enough to render his entire
two-sided price calculation unreliable. In addition, however, there also is a sharp disagreement
between Drs. Katz and Bernheim regarding the proper measure of payments made to
cardholders—i.e., the negative price charged in the form of rewards and other benefits—which
further obscures any effort to credibly determine a two-sided price in this market. (See Tr. at
6663:24-6675:17 (Katz), 6312:14-6313:11, 6314:25-6316:9 (Bernheim).) Given these findings,
37
For example, Dr. Bernheim’s two-sided price calculations include a one-time $1 billion pre-purchase of SkyMiles
paid by American Express to Delta Airlines—which was effectively an “interest-free loan” intended to provide
increased liquidity for the airline—as well as foregone interest on these funds, as an offset to discount revenue. (Tr.
at 6656:9-6659:1 (Katz), 5359:16-19 (Miller/Delta), 5658:16-23 (Codispoti/Amex).) Ordinarily, American Express
purchases SkyMiles earned on its co-brand card with Delta when the cardholder makes a purchase. Delta itself
recognized that payments made pursuant to the co-brand agreement, including the $1 billion pre-purchase of
SkyMiles that was critical to the airline’s decision to renew its co-brand partnership with American Express, were
not related to the discount rate or Delta’s status as an Amex-accepting merchant, but were instead to compensate the
airline for marketing, issuing, and other efforts undertaken to promote the co-brand relationship. (See id. at
5330:19-5331:8, 5340:13-5341:4, 5342:19-25, 5345:5-5346:23, 5376:11-20 (Miller/Delta).)
92
the court cannot conclude that Dr. Bernheim’s analysis provides a reliable basis for finding that
Amex’s two-sided price has declined during any relevant period.
Next, Amex endeavors to distinguish its present position in the network services market
from that of MasterCard during the Visa litigation by emphasizing the degree to which it trails its
competitors in metrics other than charge volume. Of particular relevance here, American
Express contends that its smaller acceptance network—Amex is accepted by roughly 3.4 million
merchants at 6.4 million different merchant locations, approximately 3 million fewer locations
than Visa, MasterCard, and Discover—belies a showing of market power. (See Jt. Stmt. ¶ 20;
DX6576 at 10; PX1985; see also Tr. at 6932:24-6934:2 (Closing Argument).) According to
Amex, its merchant coverage gap drives a lower perception of coverage among current and
potential cardholders, one which generally trails the network’s actual coverage level, and in turn,
affects consumers’ willingness to acquire and/or use an Amex card. In sum, this deficit
represents a significant competitive challenge for American Express in the card issuance market
and, indirectly by virtue of the intertwined nature of the two sides of the GPCC platform, affects
its ability to compete for share of charge volume in the network services market. 38 (See Tr. at
2948:8-2951:5, 2955:13-17 (Pojero/Amex), 4435:24-4437:17, 4395:18-4396:5
(Chenault/Amex); see also id. at 4125:2-9 (Katz) (Dr. Katz agreed that he testified in Visa that “a
38
Similarly, Defendants note that American Express has the fewest cardholders and fewest issuing banks of any
major network. (Defs. Post-Trial Br. at 4; see also Jt. Stmt. § 18; Tr. at 4295:16-21 (Chenault/Amex).) These
metrics concern Amex’s position in the card issuance market, which was at issue in Visa, but is not directly relevant
to the court’s analysis in this case. Nonetheless, the court recognizes that Amex’s performance in the issuing market
undoubtedly, though indirectly, affects its ability to compete against Visa, MasterCard, and Discover in the network
services market. The more Amex-branded cards merchants see come through their doors, the more important it is
for them to accept American Express; and the breadth of Amex’s merchant network, or consumers’ perception
thereof, influences cardholders’ views on the utility of, and thus their willingness to adopt or use, an Amex card.
(Tr. at 2960:7-23 (Pojero/Amex); DX7575 at ‘137; DX6791 at ‘566.) Additionally, the amount of merchant demand
for Amex acceptance is also somewhat affected by the fact that Defendants process fewer transactions than Visa and
MasterCard (see DX6576 at 8); however, as discussed previously, the more relevant metric for merchants
considering whether to begin or stop accepting American Express is how much Amex cardholders are spending at
the point of sale. While these additional metrics present a variety of competitive challenges for Amex, the
network’s robust share of the network services market when measured by charge volume suggests that these
disadvantages have not precluded the network from assuming a dominant position in the relevant market.
93
system with limited acceptance is of limited value to potential cardholders because of the
network effect”).) See Visa I, 163 F. Supp. 2d at 387-88 (noting “[m]erchant acceptance, and the
consumer perception of merchant acceptance, is vital to a network,” and that increases in both
metrics “can lead to an increase in card issuance and transaction volume”). Multiple witnesses,
for example, testified to the fact that Amex’s coverage gap in the United States and its effect on
the actual or perceived utility of the network’s GPCC cards significantly impedes American
Express’s ability to compete with the other networks and/or issuing banks for co-brand,
corporate card, and third-party or “GNS” issuing agreements. (See, e.g., Tr. at 2995:21-2997:5
(Pojero/Amex), 4441:21-4442:6, 4438:14-4439:8 (Chenault/Amex).) Yet Amex’s smaller
merchant network, while undoubtedly a competitive disadvantage, does not preclude a finding of
market power.
In fact, the trial record indicates that American Express’s smaller acceptance network is
largely a product of its own business decisions. The network realizes, for instance, that its
premium pricing strategy is likely incompatible with 100% merchant coverage, and that this
element of its business model remains a “formidable obstacle” in signing new merchants and
expanding coverage. (See PX0013 at ‘237; PX1611 at ‘317; DX6791 at ‘565 (noting “[h]igher
pricing” is a driver of Amex’s coverage gap); Tr. at 1153:23-1154:21 (Quagliata/Amex), 4810:16 (Glenn/Amex).) However, Amex affirmatively has elected not to reduce prices in order to
expand merchant coverage due to a concern that existing merchants might demand a lower price
if they learn Amex is reducing its price to improve coverage (known as “price spillover”), and a
firm belief that the company would be unable to fuel its differentiated business model at a lower
price point. (Tr. at 3043:4-3044:1 (Pojero/Amex), 4703:6-4704:3 (Glenn/Amex).)
94
Other elements of Amex’s business also make it less attractive to merchants, and likely
contribute to the network’s coverage gap. 39 For instance, American Express’s strategy of having
direct contractual relationships with merchants has frustrated its efforts at expanding coverage
among small merchants, as merchants traditionally had to deal separately with American Express
both in order to join the network and also during the duration of their relationships with Amex.
(See DX3750 at ‘818; Tr. at 3002:13-3004:2, 3096:15-3097:18, 3103:1-4, 3108:2-12
(Pojero/Amex), 4607:12-4609:22 (Chenault/Amex).) Amex was a late adopter of the third-party
acquirer model, and only began to work with such acquirers to sign new merchants in 2007—and
even then on a limited basis as part of its OnePoint program. (Tr. at 2847:16-2848:19, 3096:83098:12 (Pojero/Amex).) The OnePoint program, which is an Amex program aimed at acquiring
new merchants, was effective at expanding Amex’s merchant coverage (id. at 2847:16-2848:16,
3002:13-3004:2, 3097:11-24 (Pojero/Amex), 4810:10-16 (Glenn/Amex)), and Amex’s
documents signal internal optimism that the network’s recent initiatives with acquiring banks—
including its OptBlue program, which affords acquirers greater pricing flexibility—will yield
similar results and make significant headway in expanding merchant coverage. (PX2745; Tr.
at 5756:15-5759:10, 5765:16-5766:22 (Gilligan/Amex).) Indeed, Discover made a similar move
around 2005 after the Visa decisions, and subsequently it has been able to close an almost
identical coverage gap to near parity with Visa and MasterCard. (Tr. at 813:22-814:5, 824:7825:10 (Hochschild/Discover), 3931:3-3932:21 (Katz); DX6576 at 10; PX2702 at 65.)
Moreover, the record shows that at present, Amex’s current acceptance network is able to satisfy
39
As the court previously noted, American Express is also slower to pay funds to merchants on any given
transaction as compared to its competitors. (See, e.g., Tr. at 188:6-18, 189:9-12 (Thiel/Alaska Airlines), 558:9559:22 (Bouchard/Sears), 2391:3-2392:6 (Priebe/Southwest); see also id. at 2582:1-3 (Funda/Amex) (“We recognize
here there are parts of our business that are actually less attractive to merchants than Visa and MasterCard and speed
of pay is one.”).)
95
a substantial portion—94%—of its cardholders’ GPCC spending needs. 40 (Tr. at 4440:214441:7 (Chenault/Amex); PX0924 at ‘809; see also PX1412 at 8 (Amex 2013 Form 10-K).)
Defendants similarly dispute a finding of market power on the grounds that the
acquisition and retention of larger merchants is often the product of extended and intense
negotiations. However, the fact that some merchants have some degree of leverage when
negotiating with American Express does not disprove that the network possesses antitrust market
power. Even monopolists are sometimes required to negotiate with their consumers. 41 (See also
Tr. at 4268:14-4269:12 (Katz).) American Express nonetheless places significant emphasis on
the fact that it negotiates “every term” of its acceptance agreements with certain large merchants,
see supra Part I.C (discussing certain negotiated exceptions to Amex’s NDPs), and that the
network is sometimes compelled to make monetary concessions in the form of signing bonuses,
cooperative marketing funds, and volume incentives that yield lower effective discount rates in
order to ensure the merchants’ continued acceptance of Amex cards. (Tr. at 5951:15-5954:15
(McNeal/Amex); see also id. at 2292:7-2293:11 (Berry/Amex).) Yet these pricing concessions
are often quite small when compared to the discount fees paid to American Express under the
new agreements (see id. at 215:10-22 (Thiel/Alaska Airlines), 1667:10-15 (Brennan/Hilton),
40
When merchant acceptance is evaluated by the percentage of merchants that accept GPCC cards that also accept
Amex, which is referred to as “Locations in Force Coverage,” or “LIF Coverage,” American Express’s coverage
among GPCC-accepting merchants is below 80%. (See Tr. at 2945:25-2946:20 (Pojero/Amex).) However, LIF
Coverage is not weighted by charge volume, which distorts the relevance of this statistic; when calculating LIF
Coverage, for example, Delta Airlines and the corner florist are given equal weight. (See id. at 2852:24-2853:7
(Pojero/Amex); see also id. at 4802:13-4803:2 (Glenn/Amex) (discussing concerns with integrity of LIF data).)
Spend coverage, by contrast, reflects Amex’s estimate of how much of its cardholders’ credit card spending could be
accommodated at merchants that accept American Express relative to the cardholders’ overall GPCC spending.
(See PX1412 at 8; Tr. at 1159:7-1160:8 (Quagliata/Amex), 4440:21-4441:7 (Chenault/Amex).)
41
See Toys “R” Us, 221 F.3d at 930, 932, 936-37 (affirming FTC’s finding of market power notwithstanding the
fact that the defendant firm was forced to negotiate with manufacturers in order to impose the challenged restraint);
United States v. Grinnell Corp., 236 F. Supp. 244, 254, 257 (D.R.I. 1964), aff’d in relevant part, 384 U.S. 563, 576
(1966) (finding the defendants monopolized their industry even though they had “not always been able to receive the
standard they [had] set for themselves, the so-called ‘Minimum Basic Rates’, . . . or annual service charges” due to
“fringe” competition); see also Microsoft, 253 F.3d at 51-56 (finding Microsoft had monopoly power), and Deiter v.
Microsoft Corp., 436 F.3d 461, 468 (4th Cir. 2006) (noting that the “prices that [Microsoft’s] customers paid were
negotiated and, as a consequence, were both discounted and unique to each transaction”).
96
2386:21-2390:18 (Priebe/Southwest); PX2661 (demonstrative)), and they frequently come with
strings attached that limit their value to a merchant (see, e.g., Tr. at 2385:5-2386:20
(Priebe/Southwest), 5911:5-11 (Flueck/Starwood Hotels); DX7278 at ‘889 (detailing restrictions
on the use of marketing funds)). Though certain large merchants do receive lower effective
discount rates as a result of such negotiations, when viewed against the breadth of Amex’s
merchant base and the relative infrequency with which Amex makes meaningful pricing
concessions, the court does not view evidence of Amex’s willingness to negotiate certain terms
in their acceptance agreements to bar a finding of market power.
Lastly, the Government’s admission that Discover lacks market power does not compel a
similar finding as to American Express, even if both networks share certain characteristics. With
only a 5.3% share of the network services market, for example, Discover cannot leverage its
loyal cardholder base into an ability to control prices or restrict output. Merchants can profitably
drop Discover if the network overplays its hand. (See, e.g., Tr. at 457:2-8 (Robinson/Ikea),
504:18-23 (Satkowski/Enterprise) (“[W]e’re fully comfortable that if we were to eliminate the
Discover brand that we will not see any loss in business.”), 1608:7-13 (Brennan/Hilton)
(testifying that it would be easier to drop Discover due to the lower volume of spending on
Discover cards than on Amex).) As previously discussed, that is not the case for the majority of
merchants with regard to American Express, given its share of spending. Similarly, Discover’s
ability to raise its discount rates in a number of industries to levels that match or exceed the
prices charged by Visa and MasterCard is not necessarily proof of that network’s market power.
If anything, Discover’s rationale for raising its prices and its ability to do so illustrate an absence
of price competition in the network services market. See infra Part V.B.
*
*
97
*
In sum, the court concludes that American Express possesses sufficient market power in
the general purpose credit and charge card network services market to satisfy Plaintiffs’ initial
burden under the rule of reason. Even if this were not the case, however, Plaintiffs alternatively
may (and do) discharge their burden under the rule of reason by proving actual adverse effects on
competition that are attributable to the NDPs, to which the court now turns.
V.
ADVERSE EFFECTS ON COMPETITION
Plaintiffs additionally have proven that the NDPs have caused and continue to cause
actual harm to competition in the network services market. As described below, American
Express’s merchant restraints sever the essential link between the price and sales of network
services by denying merchants the opportunity to influence their customers’ payment decisions
and thereby shift spending to less expensive cards. With the NDPs in place, merchants lack any
meaningful means of controlling their consumption of network services in response to changes in
price, short of dropping acceptance altogether. Thus, by disrupting the price-setting mechanism
ordinarily present in competitive markets, the NDPs reduce American Express’s incentive—as
well as those of Visa, MasterCard, and Discover—to offer merchants lower discount rates and,
as a result, they impede a significant avenue of horizontal interbrand competition in the network
services market. On the basis of the record developed at trial, the court finds that the challenged
restraints have impaired the competitive process in the network services market, rendering lowprice business models untenable, stunting innovation, and resulting in higher prices for
merchants and their consumers.
Proof of anticompetitive harm to merchants, the primary consumers of American
Express’s network services, is sufficient to discharge Plaintiffs’ burden in this case. Cf. F.T.C. v.
H.J. Heinz Co., 246 F.3d 708, 719 (D.C. Cir. 2001) (“[N]o court has ever held that a reduction in
98
competition for wholesale purchasers is not relevant unless the plaintiff can prove impact at the
consumer level.”); United States v. Dentsply Int’l, Inc., 399 F.3d 181, 190 (3d Cir. 2005); Visa I,
163 F. Supp. 2d at 340-42 (finding plaintiffs satisfied their initial burden with regard to the
network services market by showing a likelihood of harm to merchants). In this case, Plaintiffs
additionally are able to show harm to those same merchants’ customers on the other side of the
GPCC platform, as inflated merchant discount rates are passed on to all customers—Amex
cardholders and non-cardholders alike—in the form of higher retail prices. See infra Part V.C.
To the extent American Express argues that the NDPs foster rather than impede competition,
particularly with regard to competition in the separate but interrelated issuing market (including
that market’s co-brand, corporate card, and bank issuer aspects), the court views such arguments
as potential pro-competitive justifications for the challenged restraints, and assesses them in that
context. See infra Part V.E.
For the reasons set forth below, the court finds that Plaintiffs have discharged their initial
burden under the rule of reason by proving the challenged restraints have caused “actual,
sustained adverse effects on competition.” Ind. Fed’n of Dentists, 476 U.S. at 460-61 (noting
this determination is “legally sufficient to support a finding that the challenged restraint [is]
unreasonable even in the absence of elaborate market analysis”); see also Tops Mkts., 142 F.3d
at 96 (outlining the “two independent means by which [Plaintiffs may] satisfy the adverse-effect
requirement,” including “show[ing] an actual adverse effect on competition”); K.M.B.
Warehouse, 61 F.3d at 128-29; Todd, 275 F.3d at 206. Additionally, in conjunction with the
court’s finding that American Express possesses market power in the network services market,
the findings of fact contained in this section also establish the “other grounds to believe that the
defendant’s behavior will harm competition market-wide,” K.M.B. Warehouse, 61 F.3d at 129,
99
necessary to satisfy the indirect avenue of discharging Plaintiffs’ initial burden. See also Tops
Mkts., 142 F.3d at 97; Flash Elec., Inc. v. Universal Music & Video Distrib. Corp., 312 F. Supp.
2d 379, 394 (E.D.N.Y. 2004).
A.
The NDPs Impede Horizontal Interbrand Competition
The Sherman Act is premised on a congressional determination that “unrestrained
interaction of competitive forces will yield the best allocation of our economic resources, the
lowest prices, the highest quality and the greatest material progress.” N. Pac. Ry. Co. v. United
States, 356 U.S. 1, 4 (1958); Standard Oil Co. v. F.T.C., 340 U.S. 231, 248 (1951) (“The heart of
our national economic policy long has been faith in the value of competition.”). Among the
central facets of this system is competition on the basis of price—a recognition that suppliers
can, and often do, offer lower prices to induce buyers to purchase their goods or services rather
than those of a competitor. See United States v. Socony-Vacuum Oil Co., 310 U.S. 150, 224
n.29 (1940) (referring to price competition as the “central nervous system of the economy”);
Nat’l Soc’y of Prof’l Eng’rs, 435 U.S. at 692 (same); see also Anza v. Ideal Steel Supply Corp.,
547 U.S. 451, 482 (2006) (Breyer, J., concurring in part and dissenting in part) (“The basic
objective of antitrust law is to encourage the competitive process. In particular, that law
encourages businesses to compete by offering lower prices, better products, better methods of
production, and better systems of distribution.”). Price competition is a critical avenue of
horizontal interbrand competition, and yet it is frustrated to the point of near irrelevance in the
network services market as a result of American Express’s NDPs. By suppressing the incentives
of its network rivals to offer merchants, and by extension their customers, lower priced payment
options at the point of sale—short of triggering defection, merchant demand for network services
100
is largely unresponsive to changes in price unless merchants are able to steer customers among
the GPCC networks—American Express’s merchant restraints harm interbrand competition.
American Express’s merchant restraints impede competition by severing the typical link
between merchants’ demand for network services and the price charged for the same. In a
competitive market, for example, one would expect to see changes in price result in higher or
lower demand for or output of the product or service being sold, depending on the nature of the
price change. Here, by contrast, the NDPs disrupt the normal price-setting mechanism by
reinforcing an asymmetry of information between the two sides of the payment card platform.
Amex’s rules ensure that the set of customers responsible for driving demand for network
services (cardholders) cannot be influenced in their payment choice by the set of customers on
the other side of the platform, who are informed of and responsible for paying the swipe fees
associated with that decision (merchants). In other words, with the NDPs in place, customers do
not internalize the full cost of their payment choice or account for the costs of different forms of
payment when deciding which form to use, because without merchant steering, the cost to the
customer is the same regardless. Deprived of any meaningful ability to regulate their own
consumption of network services in response to differences in network pricing, merchants are
left with an all-or-nothing acceptance decision: either agree to be a passive consumer of
American Express’s network services, or refuse to accept Amex cards altogether. (Tr. at
2595:11-23 (Funda/Amex).) And once an affirmative decision has been made, merchants’
ability to leave the network in response to subsequent increases in the price of Amex’s network
services—the only means by which merchants can exercise price discipline on American
Express—is materially impeded by the readiness of Amex cardholders to shop elsewhere if
101
unable to use their card of choice. See supra Part IV.B (discussing the restraining effect of
cardholder insistence).
In disrupting the price-setting mechanism in this market, American Express’s NDPs
suppress its network competitors’ incentive to offer lower prices at the approximately 3.4 million
merchants where American Express is currently accepted, vitiating an important source of
downward pressure on Defendants’ merchant pricing, and resulting in higher profit-maximizing
prices across the network services market. (See Tr. at 3821:11-3822:4, 3846:3-15 (Katz).)
Steering is a lynchpin to inter-network competition on the basis of price. Without the ability to
induce merchants to shift share in response to pricing differentials, a credit card network like
Discover cannot increase sales or gain market share by offering merchants a more attractive price
than its competitors. (Tr. at 832:9-23 (Hochschild/Discover) (“Once you have acceptance at
th[e] merchant, lowering your price . . . does not drive incremental sales.”).) In effect, Amex’s
NDPs deny its competitors the ability to recognize a “competitive reward” for offering
merchants lower swipe fees, and thereby suppress an important avenue of horizontal interbrand
competition. (Id. at 3821:11-3822:4 (Katz).) In the absence of steering, therefore, each of the
credit card networks is largely insulated from the downward pricing pressure ordinarily present
in competitive markets. Indeed, the record demonstrates that the NDPs create a competitive
environment in which there is virtually no check on the networks’ incentive or ability to charge
higher prices to merchants, so long as the network’s pricing is below the level at which a rational
merchant would drop acceptance entirely.
American Express itself recognizes the absence of competition on the basis of merchant
pricing in the network services market. (See Tr. at 2667:22-2668:8 (Funda/Amex) (“I don’t
think anybody’s business strategy is to be cheaper than the next guy.”); PX0038 at ‘702 (“We
102
should not compete on costs with V[isa]/M[asterCard].”).) The conceptual value-based
methodology used by American Express in developing its pricing strategy, for example, does not
account for any downward pressure associated with its competitors’ swipe fees. (Tr. at 2595:112597:14 (Funda/Amex); PX1240 at ‘091.) Quite to the contrary, Amex uses Visa’s and
MasterCard’s rates as a floor when evaluating its own discount rate in various industries. The
three major networks similarly felt no pressure to lower their own prices or otherwise respond to
Discover’s efforts in the late 1990s to build its share in the network services market by offering
merchants prices well below those charged by its competitors. (See Tr. at 3821:11-3822:13
(Katz); 2665:4-2666:6 (Funda/Amex); PX0357 at ‘944-46 (Amex document listing pressures on
Visa and MasterCard pricing).) See also infra Part V.B.
Merchants also recognize the dysfunction in the network services market. Restrained by
Amex’s anti-steering rules, merchants cannot inject price competition into the network services
industry by encouraging their customers to use their lowest cost supplier, as they can in other
aspects of their businesses. (See, e.g., Tr. at 223:9-224:22 (Thiel/Alaska Airlines), 381:8-382:18
(Robinson/Ikea).) As described by one merchant witness, the “the market is broken” because the
GPCC networks do not compete on the basis of merchant pricing. (Tr. at 2440:4-15
(Priebe/Southwest); see also id. at 832:1-17 (Hochschild/Discover) (noting that once a GPCC
network has secured merchant acceptance, “lowering your price . . . does not drive incremental
sales”).) Indeed, the impetus for Defendants’ decision in the early 1990s to strengthen the NDPs
suggests an anticompetitive intent in doing so. See supra Part I.C.1 (discussing Amex’s efforts
to block rivals’ efforts to shift share by engaging in preference campaigns). See also Chi. Bd. of
Trade, 246 U.S. at 238 (encouraging courts to consider “[t]he history of the restraint, the evil
believed to exist, the reason for adopting the particular remedy, the purpose or end sought to be
103
attained . . . not because a good intention will save an otherwise objectionable regulation or the
reverse; but because knowledge of intent may help the court to interpret facts and to predict
consequences”); Dentsply, 399 F.3d at 196; Visa I, 163 F. Supp. 2d at 401, 404-05 (considering
Visa and MasterCard’s intent in enacting the exclusionary rules in rejecting proffered procompetitive justifications).
The history of Amex’s enforcement of its NDPs illustrates the manner in which
prohibitions on merchant steering adversely affect price competition. In response to the growth
of American Express’s network in the early 1990s, Visa recognized a need to “do a better job of
telling the Visa story to merchants” in order to more effectively compete with Amex’s new,
premium card offerings. (PX0132 at ‘882.) As part of these competitive efforts, Visa sought to
call merchants’ attention to what it viewed as a “key Amex vulnerability”: namely, American
Express’s higher merchant discount rates. (PX0132 at ‘879-80, ‘882; Tr. at 3317:15-3318:15
(Morgan/Visa).) For example, the “Profit Improvement Calculator” that Visa provided to its
merchants during this period framed the price difference between Visa and Amex as potential
profit for the merchant, and invited merchants to capture that incremental benefit by developing
“inoffensive, yet effective” means of steering their customers to Visa at the point of sale. (Tr.
at 3318:16-3321:20 (Morgan/Visa); PX0082 at ‘543.) Visa also encouraged merchants to shift
share to its network through the “We Prefer Visa” campaign, by which prominent merchants
stated an express preference for Visa cards, in addition to more traditional forms of point-of-sale
steering, such as posting signage that favors the merchant’s preferred form of payment (short of
making an explicit statement of preference) or asking customers, “Would you like to put this on
your Visa?” (See Tr. at 3321:21-3323:6, 3325:9-3326:3 (Morgan/Visa); PX0133 at ‘985; see
also PX0082 at ‘544.) For those merchants that chose to participate, these competitive efforts
104
were markedly successful at shifting spend to Visa’s network. (See Tr. at 3843:5-9 (Katz),
4487:7-16 (Chenault/Amex); DX7595 at ‘637; PX0133 at ‘986; see also Tr. at 3327:18-3328:2,
3330:3-8 (Morgan/Visa).) Yet notwithstanding the range of possible pro-competitive responses
to Visa’s steering campaigns available to Defendants, including reducing Amex’s discount rate
or improving its messaging to better communicate to merchants the value they received for the
premium price charged, 42 Amex’s primary response was to bolster its contractual restraints on
merchants in order to stifle any further steering or preference campaigns. (See Tr. at 4490:134491:18, 4492:16-4493:14, 4499:6-4504:20, 4531:17-4532:11 (Chenault/Amex); PX0163 at
‘030, ‘032-33, ‘035-36 (document reflecting Amex brainstorming on potential responses to
preference campaigns); DX7595 at ‘639-40.)
American Express observes these same facts and insists that preference campaigns, like
the “We Prefer Visa” initiative, represent a form of competition that is rightly suppressed by its
NDPs. Dr. Bernheim, for example, testified that such campaigns should not be viewed as
competition on the merits because they have the effect of undermining consumers’ perception of
the value of a competitor’s product rather than building the value of one’s own product. (See Tr.
at 6421:20-6422:19 (Bernheim); see also id. at 5059:24-5060:1 (Gilbert).) The court disagrees.
As an initial matter, it is not for the court to draw lines between “good” competition and “bad”
competition in the network services market; the federal antitrust laws reflect a steadfast
“legislative judgment that ultimately competition will produce not only lower prices, but also
better goods and services.” Nat’l Soc’y of Prof’l Eng’rs, 435 U.S. at 695; see also F.T.C. v.
Superior Court Trial Lawyers Ass’n, 493 U.S. 411, 423-24 (1990) (rejecting proffered
justifications for agreement among lawyers to fix fees, even though “the quality of
42
In Dr. Katz’s view, these responses, some of which Amex did pursue to a limited degree, represent competition
on the merits. (Tr. at 3843:10-19 (Katz).)
105
representation” might have been improved as a result). As explained by the Supreme Court,
“[t]he statutory policy underlying the Sherman Act ‘precludes inquiry into the question whether
competition is good or bad.’” Trial Lawyers, 493 U.S. at 424 (quoting Nat’l Soc’y of Prof’l
Eng’rs, 435 U.S. at 695); see also N. Pac. Ry. Co., 356 U.S. at 4 (stating “the policy
unequivocally laid down by the [Sherman] Act is competition”). The court further recognizes
that merchants will be amenable to a network’s invitation to steer only if they perceive an
economic justification for doing so—for instance, if one network offers lower prices, or is
willing to offer a lower price in exchange for merchants’ efforts to shift share, or if the network’s
pricing structure incentivizes the merchant to focus its charge volume on that network in order to
qualify for a rate reduction tied to charge volume. During the “We Prefer Visa” campaign, for
example, the evidence suggests that merchants responded positively to Visa’s messaging because
American Express was, in fact, markedly more expensive to accept than Visa. (See Tr. at
3317:15-3318:15 (Morgan/Visa), 3842:19-3843:4, 3844:1-14 (Katz); PX0132 at ‘879-80, ‘882.)
Nonetheless, Amex’s anti-steering rules continue to block pro-competitive efforts by its
network rivals to gain share by incentivizing merchants to engage in preference relationships.
For example, when American Express enforced its NDPs in 2003 to prevent Travelocity from
stating on its website that “Travelocity Prefers MasterCard” pursuant to a preference agreement
with that network, it undermined the online travel agency’s ability to direct additional share to
MasterCard and diminished the “value of MasterCard as Travelocity’s preferred brand,”
resulting in the merchant receiving less financial remuneration from MasterCard. (Tr. at
3244:1-3252:2 (Biornstad/MasterCard); PX0385; PX1324.) See also supra Part I.C.2. After
Travelocity was compelled to change its promotion to refer to MasterCard as an “Official Card,”
rather than its preferred form of payment, the partners’ joint television, print, and radio
106
advertising touting the partnership “basically stopped” and MasterCard saw a reduction in the
share being shifted to its network. (Tr. at 3248:17-3249:4 (Biornstad/MasterCard).) As
described by Dr. Katz, this change represented “a lessening of competitive pressure” in the
relevant market, and exemplifies the manner in which rules against steering reduce competition
among the credit card networks. (Id. at 3844:15-3845:1 (Katz).)
In sum, by preventing merchants from influencing their customers’ payment choices,
Defendants’ anti-steering rules render merchant demand for network services less responsive to
changes in the price charged for those services. In so doing, the NDPs effectively remove the
incentive for American Express or its network competitors to compete with one another by
offering merchants a lower price, as without merchant participation in the point-of-sale payment
decision, a lower price will not translate into increased volume for the network. In undermining
the competitive process and price-setting mechanism in the market for GPCC card network
services, the challenged restraints impede a critical form of horizontal, interbrand competition.
B.
The NDPs Block Low-Cost Business Models
American Express’s merchant restraints also render it nearly impossible for a firm to
enter the relevant market by offering merchants a low-cost alternative to the existing networks.
Indeed, the failure of Discover’s low-cost provider strategy in the 1990s provides direct evidence
of how anti-steering rules like Defendants’ NDPs impede modes of competition that likely would
benefit consumers on both sides of the GPCC platform. Amex’s rules effectively deny another
GPCC network—whether Discover or a potential new entrant—the opportunity to pursue a
business model that differentiates itself by offering merchants a low price in return for greater
volume.
107
Discover launched in 1985 by offering a combination of breakthrough value propositions:
Cardholders could receive the first GPCC card with a rewards feature at no annual fee, and
merchants were offered a low-price alternative to the existing GPCC networks. (Tr. at 821:8-16
(Hochschild/Discover).) Discover pursued its low-price strategy by pricing its network services
“very aggressively for merchants,” setting all-in discount rates significantly below those of its
competitors. (Id.)
Sensing an increase in merchant dissatisfaction in the late 1990s amidst a series of price
increases by its competitors, Discover saw an opportunity to leverage its position as the lowestpriced network to gain share. (See id. at 832:24-835:17 (Hochschild/Discover); PX1277 at
‘094.) See also Visa I, 163 F. Supp. 2d at 333 (noting Discover’s status as the lowest priced
network). In 1999, the network launched a “major campaign” aimed at highlighting the pricing
disparity between it and its competitors in order to persuade merchants to “shift their business to
[Discover’s] lower-priced network.” (Tr. at 833:4-11 (Hochschild/Discover).) In a speech
before an industry group in April 1999, David Nelms, then-President of Discover, outlined the
network’s plan: Discover intended to partner with merchants in helping them control payment
costs and proposed that they steer customers to the lower-cost Discover cards. (Id. at 834:13-20
(Hochschild/Discover); PX1277 at ‘090, ‘094-95 (noting Discover wanted “to help [merchants]
save money by encouraging their customers to pay with Discover Card”).) To that end, Discover
sent a letter to every merchant on its network, alerting them to their competitors’ recent price
increases and inviting the merchant to save money by shifting volume to Discover. (Tr. at 836:6837:18 (Hochschild/Discover).) Discover representatives also met with a number of larger
merchants to offer discounts from the network’s already lower prices if they would steer
customers to Discover. (Id. at 837:2-25 (Hochschild/Discover).) The network suggested a
108
number of means by which merchants could achieve this share shift, including point-of-sale
signage (id. at 839:22-842:3 (Hochschild/Discover); PX1292 at ‘991-94), and also suggested that
merchants use the savings to lower their own prices and thereby invest in generating customer
loyalty for themselves (Tr. at 847:8-848:14 (Hochschild/Discover)). The additional volume
recognized as a result of these efforts would be Discover’s reward for offering lower prices to its
merchant base. (See id. at 3836:9-3838:18 (Katz); see also id. at 837:19-25
(Hochschild/Discover) (Discover believed offering further discounts to large merchants would be
profitable for the network by virtue of greater transaction volume, and resulting increases in
discount and interest revenue).)
Discover’s efforts, however, failed to produce “any significant movement in share” due
to the anti-steering rules maintained at the time by Visa, MasterCard, and American Express.
(Tr. at 848:15-849:15 (Hochschild/Discover).) In its conversations with a number of merchants,
Discover learned that the merchant restrictions imposed by the other payment networks denied
merchants the ability to express a preference for Discover or to employ any other tool by which
they might steer share to Discover’s lower-priced network. (Id. at 848:15-849:15, 852:24-853:15
(Hochschild/Discover) (“[T]he limitations placed by the other networks didn’t give [merchants]
any effective strategies to shift share.”); see also PX0075 at ‘028-29, ‘032 (“Merchants have
largely not responded to simple, low prices and our challenge to drive Discover share.”).)
Notably, Defendants do not strenuously dispute the evidence regarding the effect of anti-steering
rules on Discover’s low-price model, or that such restrictions effectively raise a barrier to entry
in the relevant market for firms pursuing a low-price strategy.
Recognizing that its lower prices would not drive incremental volume to its network in a
market subject to limitations on merchant steering, Discover abandoned its low-price business
109
model in 2000 and began raising discount rates in order to more closely align its merchant
pricing with that of Visa and MasterCard. (Tr. at 853:19-854:15 (Hochschild/Discover); see also
id. at 832:1-23 (Hochschild/Discover) (testifying that “lowering your price . . . does not drive
incremental sales”).) In the company’s view, “[t]o the extent that offering a lower price was not
going to give [Discover] any business benefits, it was leaving money on the table.” (Id. at 854:715 (Hochschild/Discover) (noting that giving merchants a “discount without getting anything in
return didn’t make business sense”).) Discover described this transition in its internal documents
as a move from a “Low Cost Provider Strategy” to a strategy titled “Close Competitive Gap,”
pursuant to which Discover raised its average effective discount rate nearly 24% from 2000 to
2007. (PX1285 at ‘474; Tr. at 862:1-24 (Hochschild/Discover).) Today, Discover’s prices are
similar to those offered by Visa and MasterCard, and the network has also adopted the more
complicated “unbundled” pricing model used by those networks. (See id. at 863:25-864:8
(Hochschild/Discover).)
In the court’s view, the failure of Discover’s low-price value proposition is emblematic of
the harm done to the competitive process by Amex’s rules against merchant steering. 43 Since
customers can neither independently access nor account for the costs of different forms of
payment when deciding which to use, a lowest-cost provider strategy cannot succeed in the
network services market if merchants are unable to shift share among the various networks. (See
Tr. at 853:19-22 (Hochschild/Discover), 3821:13-3822:13, 3840:24-3841:10 (Katz).) Absent
merchant participation in the point-of-sale payment decision, a supplier in the network services
43
The court, of course, recognizes that prior to the consent decrees entered into by Visa and MasterCard in this
case, both networks also maintained anti-steering restrictions akin to Amex’s NDPs, and that all three networks’
restraints likely contributed to Discover’s decision to abandon its low-price model. Yet, based on the testimony
adduced at trial, the court finds that a similar outcome is likely in a market subject to American Express’s NDPs
alone, even if these restraints do not cover every merchant in the United States. (See Tr. at 3841:3-9 (Katz).)
110
market cannot realistically expect to receive any competitive benefit for offering a price below
that of its competitors, even if such a move would benefit merchants and their customers alike.
C.
The NDPs Have Resulted in Higher Prices to Merchants and Consumers
American Express’s merchant restraints have allowed all four networks to raise their
swipe fees more easily and more profitably than would have been possible were merchants
permitted to influence their customers’ payment decisions. Beyond Plaintiffs’ evidence
concerning the manner in which the NDPs undercut the competitive process, the record
demonstrates that these restraints have resulted in higher all-in merchant prices across the
network services market, providing additional proof of their actual anticompetitive effect.
Plaintiffs have established, for instance, that American Express’s prohibitions on
merchant steering aided the network’s efforts to profitably raise its discount rates on merchants
accounting for 65% of the network’s annual U.S. charge volume as part of its Value Recapture
initiatives in the late 2000s. (See PX0121 at ‘459.) By precluding merchants from directing
transactions to other networks, Amex’s merchant restraints blocked an important safety valve
that would have moderated its efforts to increase discount rates. (See Tr. at 3846:1-15, 3850:817 (Katz).) Among large merchants, for example, American Express did not even account for
the possibility that merchants would respond to its price increases by attempting to shift share to
a competitor’s network when assessing the likely profitability of Value Recapture, and instead
considered only whether merchants would cease acceptance altogether as a result of the
initiative. (See id. at 3849:10-3850:17 (Katz); PX1099 at ‘555.) By contrast, at smaller
merchants where the network had greater difficulty monitoring steering or suppression of its
cards, Amex did consider the effect steering would have on its ability to profitably increase
price, and concluded that such efforts would not defeat the repricing initiative. (See Tr.
111
at 3846:1-15, 3847:8-3849:9 (Katz); PX1753A at ‘033.) Merchant testimony presented at trial
confirmed that were large merchants able to do so, they would have attempted to steer customers
away from American Express to blunt the effect of Amex’s price hikes. (See, e.g., Tr. at 2418:317 (Priebe/Southwest); see also id. at 3851:1-12 (Katz).) In preventing such mitigation, the
NDPs were integral to American Express’s Value Recapture increases and thereby caused
merchants to pay higher prices.
American Express disputes that its premium discount rates are supracompetitive, faulting
Plaintiffs for evaluating only the merchant side of the GPCC platform and not proffering
empirical evidence that the NDPs have resulted in a higher two-sided price—i.e., that the price
charged across Amex’s entire platform, accounting for both discount revenue and the expense of
providing cardholder rewards, increased as a result of the network’s anti-steering rules. (Defs.
Post-Trial Br. at 9-10.) Yet, as the court has previously noted, neither party has presented a
reliable measure of American Express’s two-sided price that appropriately accounts for the value
or cost of the rewards paid to cardholders. See supra Part IV.D. Even without such data,
however, Plaintiffs have provided sufficient circumstantial evidence and expert testimony for the
court to conclude that Amex’s Value Recapture price increases were not wholly offset by
additional rewards expenditures or otherwise passed through to cardholders, and resulted in a
higher net price. (See Tr. at 3853:3-3854:17, 4039:16-4040:20 (Katz) (economic theory supports
a finding that where, as here, prices are inflated due to buyers being less responsive to price,
“there will be less than a hundred percent passthrough” and that “networks are going to keep
some of [the higher discount rate] for themselves”).) Indeed, Amex’s Chief Financial Officer
told investors in June 2013 that Amex “drop[s]” part of its premium to the bottom line even as it
invests part in creating value for cardholders. (PX1475 at 2; see also Tr. at 3853:3-24 (Katz).)
112
Dr. Katz further concluded that American Express spends less than half of the discount fees it
collects from merchants on cardholder rewards. (Tr. at 3853:3-24 (Katz).)
Even if Amex did fully pass through the higher discount fees attributable to its NDPs to
cardholders, the court finds that prohibitions on merchant steering—including Visa and
MasterCard’s anti-steering rules, which were abandoned as a result of this case—have also
enabled American Express’s competitors to charge higher all-in fees. Visa and MasterCard, for
instance, were able to increase their average all-in merchant rates through a variety of means by
more than 20% from 1997 to 2009, without fear of other networks undercutting their prices in
order to gain share. (PX0357 at ‘959; Tr. at 2663:24-2665:3 (Funda/Amex).) Similarly, after
Discover was forced to abandon its low-price strategy as a result of its competitors’ merchant
regulations, that network was able to radically increase its merchant pricing over a relatively
short period of time, in order to match the rates set by its competitors. See supra Part V.B.
Discover was able to raise its rates with virtual impunity, relying on the restraining effect of antisteering rules to ensure that it would not be undercut by a competitor offering a lower price to
merchants. These examples provide further support for the court’s finding that without affording
merchants the ability to influence their customers’ credit and charge card decisions, there is little,
if any, downward pressure on the price charged to merchants.
The NDPs have also resulted in increased prices for consumers. Merchants facing
increased credit card acceptance costs will pass most, if not all, of their additional costs along to
their customers in the form of higher retail prices. (See Tr. at 3840:10-23, 3854:18-3855:25
(Katz) (testifying that “an economically rational merchant is going to pass [the higher costs of
accepting payments] on to its customers,” and “prices are going to go up with the merchant for
everybody”); see also id. at 1405:22-1407:11 (Rein/Walgreen); DX2214 at ‘983.) See also
113
Freedom Holdings, Inc. v. Cuomo, 624 F.3d at 38, 56 (2d Cir. 2010) (noting a “tax increase, like
any cost, will likely be passed on to consumers in the form of higher prices”); F.T.C. v. H.J.
Heinz Co., 246 F.3d 708, 719 (D.C. Cir. 2001) (“[T]he antitrust laws assume that a retailer faced
with an increase in the cost of one of its inventory items ‘will try so far as competition allows to
pass that cost on to its customers in the form of a higher price for its product.’” (quoting In re
Brand Name Prescription Drugs Antitrust Litig., 123 F.3d 599, 605 (7th Cir. 1997))). Higher
retail prices affect not only those customers who use American Express cards, but also shoppers
who instead prefer to pay using a lower-rewards GPCC card, debit card, check, or cash. (Tr. at
3854:18-3855:25 (Katz).) Even if American Express passed through every cent of its premium
or the incremental revenue realized from its Value Recapture price increases to cardholders—
which it does not—customers who do not carry or qualify for an Amex card are nonetheless
subject to higher retail prices at the merchant, but do not receive any of the premium rewards or
other benefits conferred by American Express on the cardholder side of its platform. (See Tr. at
3852:3-3853:17 (Katz), 5252:10-5253:13 (Gilbert); PX1475 at 2.) Thus, in the most extreme
case, a lower-income shopper who pays for his or her groceries with cash or through Electronic
Benefit Transfer—and the same is true of any consumer who does not use an Amex card or
comparable high-rewards product from Visa, MasterCard, or Discover—is subsidizing, for
example, the cost of the premium rewards conferred by American Express on its relatively small,
affluent cardholder base in the form of higher retail prices. See generally Adam J. Levitin,
Priceless? The Economic Costs of Credit Card Merchant Restraints, 55 UCLA L. Rev. 1321,
1356 (2008) (noting this cross-subsidy is also highly regressive). The court views this
externality as another anticompetitive effect of Defendants’ NDPs.
114
D.
The NDPs Stifle Innovation
Plaintiffs additionally point to evidence from a number of merchant witnesses in support
of their theory that Amex’s NDPs stifle innovation in the network services market. See Visa II,
344 F.3d at 241 (affirming district court’s finding that defendants had harmed competition
because “product innovation . . . ha[d] been stunted by the challenged policies”); Aventis Envt.
Sci. USA LP v. Scotts Co., 383 F. Supp. 2d 488, 504 (S.D.N.Y. 2005) (noting “retardation of
innovation and subsequent decrease in the quality of [defendant’s product] . . . could qualify as a
harm to competition and consumers”). This argument is less convincing than the theories
discussed above. Many of the examples cited by Plaintiffs as evidence of its innovation theory—
including efforts by at least two merchants to deploy new technologies that would allow them to
steer customers to lower-cost cards by updating in real time the price displayed to the customer
based on the form of payment presented at the point of sale—do not relate to stifled
improvements in the network services provided to merchants, but instead illustrate novel forms
of steering that have been unable to thrive under the NDPs. (See Tr. at 3158:4-3171:9
(Gibson/Sinclair) (discussing Sinclair’s proposal to “roll back” prices on gasoline pumps and on
their mobile commerce application depending on the type of card used by the customer),
6144:14-6151:7 (Mitchell/Official Payments).)
But other examples of stunted innovation in the relevant market are more pertinent to the
court’s antitrust analysis. For example, Defendants’ NDPs and similar anti-steering rules
formerly maintained by Visa and MasterCard are responsible for inhibiting the development of
several proposed merchant-owned payment solutions. Project Monet, for example, was a
network venture proposed by Discover in the early 2000s, whereby merchants would receive
equity in the network and be able to directly control their payment costs by influencing future
115
pricing decisions. (Tr. at 838:20-839:11 (Hochschild/Discover).) Similarly, a group of forty
large retailers have recently created a joint venture under the name Merchant Customer
Exchange (“MCX”) to develop a new payment platform that would operate on customers’
mobile devices and significantly reduce the participating merchants’ payment processing costs.
(Id. at 2433:6-2435:4 (Priebe/Southwest).) Both ventures have been impeded by restrictions on
merchant steering. Project Monet was ultimately abandoned by Discover when it became clear
that merchant-investors would be unable to encourage customers to use the preferred cards by
traditional forms of steering (id. at 956:11-17 (Hochschild/Discover)), and under the NDPs,
MCX’s capacity to develop a viable brand as the low-cost alternative to traditional GPCC cards
is endangered by merchants’ inability to “compar[e] and contrast[]” MCX’s payment services
with those offered by American Express (id. at 2436:1-20 (Priebe/Southwest).
Accordingly, while the court finds limited evidentiary value in Plaintiffs’ innovation
theories that relate to new and novel forms of steering, American Express’s anti-steering rules—
and those previously maintained by Visa and MasterCard—are responsible for impeding
development of novel payment solutions that would have injected or potentially may inject
greater diversification into the network services industry, and thus improve the quality of
offerings therein.
E.
Removal of the NDPs Would Benefit Merchants and Consumers
Elimination of American Express’s anti-steering rules would restore merchants’
responsiveness to changes in network pricing, and, in turn, unlock an important avenue of
competition among the credit card networks. Plaintiffs have proven through direct and
circumstantial evidence that not only are merchants and networks likely to engage in point-ofsale steering if the NDPs are lifted, but also that such activities will inure to the benefit of both
116
merchants and customers alike. American Express’s efforts to discredit Plaintiffs’ proposed “but
for” world are unavailing.
American Express itself recognizes the pro-competitive benefits of steering. Defendants
regularly use such tactics in American Express’s travel agency business, which is one of the
largest in the United States, to reward certain of its airline, hotel, and car rental vendors with
increased travel volume in return for offering a lower rate or entering into a “preferred supplier”
relationship with Amex. (Tr. at 3460:7-3461:23, 3467:2-3468:15, 3472:4-11, 3473:2-5
(Corbett/Amex); see also PX1685 at ‘686 (noting “we try to sell and promote only preferred
suppliers, and we actively sell away from nonpreferred suppliers” and that “what keeps our
preferreds coming back to us is their fear of how aggressive we actually are against
nonpreferreds”); PX1007 at ‘930-31, ‘941-42, ‘947 (discussing successful Amex travel agency
campaign to shift share away from a British airline, which included an agreement with a
competitor airline to lower rates in return for increased share).) Merchant steering in the network
services market has also been to Amex’s own benefit. For instance, in the early 1990s Amex
entered into a preference relationship with Ticketmaster whereby phone operators for the
merchant would inform callers: “Our card of choice is American Express. Would you like to
use your American Express card today?” (See PX2766; PX0355 at ‘411; see also, e.g., PX0150
at ‘036 (Amex entered into a sponsorship relationship with Radio City Music Hall whereby the
venue informed patrons that “Radio City Welcomes the American Express Card” and designated
one box office window exclusively for the use of Amex cardholders); PX2602 at ‘938, ‘944-45,
‘950 & Tr. at 4557:4-4572:4 (Chenault/Amex) (Amex contracted to be the “Official Payment
Services Product Provider” for Universal Studios Theme Park Properties, which involved
117
multiple forms of steering, including offering discounts on certain food, beverages, and
merchandise when using an Amex card).)
Providing merchants the freedom to participate in their customers’ payment decisions
will foster greater interbrand competition among the GPCC networks and restore downward
pressure on their merchant prices. Eager to control the costs associated with running their
businesses, merchants routinely seek lower prices for necessary goods and services by promoting
competition among multiple suppliers, often by rewarding competitive bidders with increased
purchase volume. (See, e.g., Tr. at 219:6-223:22 (Thiel/Alaska Airlines), 381:2-382:18
(Robinson/Ikea), 1343:22-1347:13 (Rein/Walgreen).) Similar competition cannot be generated
in the network services market under the NDPs, however, despite the fact that credit card fees
represent a significant cost for many merchants. (See id. at 192:9-21, 224:11-22 (Thiel/Alaska
Airlines), 381:11-18, 387:1-7 (Robinson/Ikea), 1222:5-17 (Kimmet/Home Depot), 1347:14-16
(Rein/Walgreen), 1608:14-18 (Brennan/Hilton), 2318:22-2319:17 (Bruno/Crate & Barrel),
2440:5-15, 2440:23-2441:3 (Priebe/Southwest).) Were merchants able to direct a greater share
of their charge volume to lower-cost credit or charge card networks, whether by offering
discounts to customers for using such cards, posting the relative costs of different modes of
payment, or engaging in another form of point-of-sale steering, they would be better able to
control and lower their costs of credit card acceptance. (See id. at 1276:4-8 (Kimmet/Home
Depot).)
Depending on the magnitude of the charge volume available to be shifted at any given
merchant, removal of the NDPs would also restore the networks’ incentive to offer merchants
lower rates in the hope of capturing additional share. (See id. at 1614:22-1617:7
(Brennan/Hilton) (discussing Hilton’s potential partnership with Visa by which Visa “would
118
offer a rebate back on the interchange fees collected” at the end of the year “if [Hilton] could
increase Visa’s relative share of Hilton charge volume”); see also Tr. at 3258:25-3259:13,
3298:7-13 (Biornstad/MasterCard) (noting MasterCard likely would pursue preference
relationships with select vendors if permitted), 3845:6-14 (Katz).) For larger merchants, in fact,
the networks may be induced to competitively bid for the additional share the merchant is able to
funnel to a preferred (and likely cheapest) payment network. (See id. at 408:24-409:17
(Robinson/Ikea), 2328:10-2329:7 (Bruno/Crate & Barrel); see also id. at 3670:10-25
(Silverman/Amex) (discussing competitive bidding process by which merchants invite networks
and partner banks to compete for co-brand partnerships).) Even if a merchant were not inclined
to engage in steering, its freedom to do so in the future would enhance its bargaining position
relative to American Express and its competitor networks, placing additional downward
competitive pressure on rates. (Id. at 581:4-8 (Bouchard/Sears), 2353:14-2354:11
(Bruno/Crate & Barrel), 3219:17-22 (Gibson/Sinclair).)
The restoration of downward competitive pressure on merchant prices would, in the
court’s view, result in lower swipe fees charged to merchants by American Express and its
competitors. Indeed, American Express itself recognizes that several of its larger merchants
desire the ability to steer volume to less expensive networks, and that removal of its NDPs may
require Amex to reduce its premium discount rates. (See Tr. at 2738:4-20 (Funda/Amex); see
also PX0701 at ‘584 (U.S. Airways asked Amex to remove the NDPs during negotiations for a
new acceptance agreement); PX1846 at ‘018-19, ‘087 (Southwest tried to modify the NDPs in its
acceptance agreement); PX0842 at ‘285 (internal Amex email stating United Airlines “insists on
right to preference Amex competitors that have lower discount rates and this of course is
unacceptable to us”).) For example, when Congress signaled in 2010 that it might permit
119
merchants to engage in differential discounting among GPCC card brands as part of the DoddFrank Act, 44 the network recognized that one “strategic option[]” included “[r]emov[ing the]
[e]conomic incentive for merchants to discount/cancel” American Express by “selectively (by
industry) reduc[ing]” American Express’s discount rate “to the point where discounting or
cancelling is not economically advantageous for merchants.” (PX0090 at ‘473; see also PX1176
at ‘383-84 (considering potential responses to “[m]itigate the likelihood of merchants . . .
offering incentive to use another form of payment,” including lowering discount rates in “highrisk” industries); PX1239 at ‘943 (noting differential discounting would put pressure on Amex’s
rates).) Amex witnesses also testified that the network would face increased pressure to reduce
its rates if merchants could shift share to a less expensive network. (See Tr. at 702:3-10
(Quagliata/Amex), 2693:25-2694:4, 2694:20-23 (Funda/Amex).) In addition, Discover’s
President and Chief Operating Officer, Roger Hochschild, testified that his network would
“aggressively pursue a strategy of lowering [its] prices” were merchants permitted to steer
transactions to Discover. (Tr. at 872:3-17 (Hochschild/Discover); see also id. at 3841:13-3842:6
(Katz) (stating it would be economically rational for Discover to lower prices in the absence of
the NDPs because steering “would increase [Discover’s] incentives to lower prices” by
unlocking the ordinary competitive reward for doing so).)
Removal of the NDPs would also benefit consumers. In the short term, consumers would
benefit by taking advantage of the inducements offered by merchants in order to sway their card
choice. For instance, by agreeing to pay with a merchant’s preferred card, customers might
receive a 5% discount off of the retail price when using MasterCard (id. at 526:5-13
(Satkowski/Enterprise)), a free night at a hotel or day of rental car use for using Visa or Discover
(id. at 1616:11-20 (Brennan/Hilton), 497:12-498:18 (Satkowski/Enterprise)), free shipping when
44
Such a provision ultimately was not included in the Durbin Amendment.
120
they use American Express, designated checkout lanes when paying with Discover (id. at 845:310 (Hochschild/Discover)), or any number of other offers or inducements used by merchants to
direct transactions to a preferred network. Ultimately, and essentially, it is the customer’s
decision whether to accept the merchant’s offer or to pay with his or her card of choice. Thus,
even if a merchant is inclined to steer away from American Express, the cardholder would still
have the freedom to use an Amex card if the cardholder decides the rewards offered by American
Express are of greater value than the discount, in-kind perk, or other benefit offered by the
merchant. (See Tr. at 6678:1-6680:10 (Katz) (consumers would benefit from having options
because “it is giving the cardholder more choice, more chance to decide . . . whether he or she
wants to get the rewards for that given purchase . . . from the card issuer or whether he or she
would rather get some sort of reward from the merchant, for example”).) 45 Allowing merchants
to actively participate in their customers’ point-of-sale decisions would remove the artificial
barrier that now segregates merchant demand from the price of network services, and allow
merchants and cardholders alike to jointly determine how the prices charged on each side of the
GPCC platform weigh against one another. In the longer term, the court expects that merchants
will pass along some amount of the savings associated with declining swipe fees to their
customers in the form of lower retail prices. (See Tr. at 382:19-383:7 (Robinson/Ikea), 1278:1-
45
See also id. at 688:23-689:12 (Quagliata/Amex) (testifying that if steering were allowed, “cardmembers will still
have a choice”), 2736:14-2737:18 (Funda/Amex) (agreeing that, if merchants were permitted to offer discounts by
card brand, customers would have a choice of benefits and could “make a decision one way or another”), 6188:1425 (Mitchell/Official Payments) (Official Payments offers customers discounts for using cheaper cards with the
understanding that customers “can make their own choice as to whether they want to use their rewards card or not or
if it is more important for them at that moment in time to save some money”); PX1176 at ‘379 (“With selective
discounting, the risk of loyal Amex Cardmembers walking away (not returning) is much lower; they are being
offered a discount but still allowed to use Amex if they prefer.”).
121
14 (Kimmet/Home Depot), 1346:19-1347:13 (Rein/Walgreen), 3150:15-25 (Gibson/Sinclair
Oil); cf. id. at 3840:10-23, 3854:18-3855:25 (Katz).) 46
Underpinning this “but for” world lies a determination that, without Amex’s contractual
restraints, merchants actually are likely to steer customers between various forms of payment or
GPCC networks. The costs associated with accepting credit and charge cards are among many
merchants’ highest, and accordingly they have a strong economic incentive to take steps to
reduce these expenses as much as possible without alienating either the networks or customers.
(Tr. at 192:14-21 (Thiel/Alaska Airlines) (Alaska Airlines’s credit card costs are approximately
twice as much as its U.S. labor costs, and more than what it spends on food and beverages),
387:1-7 (Robinson/Ikea) (payment acceptance costs are Ikea’s fourth highest cost, after labor,
advertising, and rent), 2318:22-2319:23 (Bruno/Crate & Barrel).) To that end, multiple merchant
witnesses testified that they have asked American Express in the past to relax its NDPs and grant
them greater freedom to participate in their customers’ card choices. (See, e.g., id. at 264:4-15
(Thiel/Alaska Airlines), 1257:25-1258:19 (Kimmet/Home Depot), 1541:11-15 (O’Malley/Best
Buy), 1697:13-1698:13 (Dale/Sprint).) Other merchants testified that they would, in fact, steer if
given the opportunity. (See id. at 497:12-499:25 (Satkowski/Enterprise), 580:9-581:18
46
To the extent merchants realize cost savings as a result of steering, Defendants dispute that any degree of
merchant savings will be passed on to customers in the form of lower retail prices. Primarily, Amex relies on a
report by the Reserve Bank of Australia (“RBA”) from April 2008 that noted there was no “concrete evidence” that
cost savings attributable to the RBA’s decision to regulate interchange rates and expressly permit surcharging in
Australia had been passed on to consumers. (See DX4026 at 23; Tr. at 5817:18-5820:22 (Gilligan/Amex).) Yet
when read in its entirety, the RBA’s report actually supports the court’s determination that lower discount rates
resulting from removal of the NDPs will benefit consumers as merchants translate some amount of their lower credit
card costs into lower prices. In full, the RBA noted that “[n]o concrete evidence has been presented to the [RBA]
regarding the pass-through of these savings, although this is not surprising, as the effect is difficult to isolate. The
Bank had previously estimated that the cost savings would be likely to lead to the CPI being around 0.1 to 0.2
percentage points lower than would otherwise be the case over the longer term (all else constant). It is very difficult
to detect this against a background where other costs are changing by much larger amounts and the CPI is increasing
by around 2 1/2 per cent per year on average. Despite the difficulties of measurement, the Board’s judgment
remains that the bulk of these savings have been, or will eventually be, passed through into savings to consumers.
This judgement is consistent with standard economic analysis which suggests that, ultimately, changes in business
costs are reflected in the prices that businesses charge.” (DX4026 at 23 (emphasis added).)
122
(Bouchard/Sears), 1613:23-1614:7 (Brennan/Hilton).) Moreover, the trial record further
establishes that many merchants, whether subject to Amex’s standard NDPs or to a customized
version, attempt to steer customers to the extent permitted under their respective acceptance
agreements. (See id. at 1537:8-1540:4, 1541:11-23 (O’Malley/Best Buy) (Best Buy steers
toward its private label card at the point of sale, to the extent permitted under its customized
NDPs), 1699:5-1702:6 (Dale/Sprint), 2324:21-2326:25 (Bruno/Crate & Barrel) (noting his
company has reduced its overall cost of acceptance “considerably” by steering to a private label
card).)
American Express sharply disputes Plaintiffs’ theory that merchant discount rates and
retail prices would decrease in the absence of the NDPs. Specifically, Defendants point to the
roughly 3 million merchant locations that comprise Amex’s merchant coverage gap, where
merchants today are free to engage in steering as a result of the consent decrees entered into by
Visa and MasterCard in this case. 47 (Defs. Post-Trial Br. at 14-15; see also Tr. at 6456:106457:17, 6459:13-20 (Bernheim).) Amex correctly notes there is little evidence of widespread
steering at these merchant locations, or that price competition among the networks has increased
in the four years since Visa and MasterCard agreed to abandon their anti-steering rules for the
merchants operating these locations. (See Tr. at 4237:3-11, 4240:10-13 (Katz), 6456:10-6457:17
(Bernheim). But see PX1337A at ‘054-58, ‘073-82; Tr. at 6144:14-6146:21 (Mitchell/Official
Payments) (discussing Official Payments’s efforts to exploit this opportunity).) Indeed, it
appears likely that discount rates at these locations have continued to rise during this period. (Tr.
at 950:7-16 (Hochschild/Discover), 6457:5-17 (Bernheim).)
47
Although Discover retains its anti-steering rules at nearly all of these locations, the court does not view those
provisions as a significant explanation for the lack of merchant steering, as discussed later in this section.
123
Yet the court cannot agree with American Express that the “natural experiment”
occurring at these three million merchant locations is an accurate predictor of the consequences
of eliminating American Express’s NDPs. First, the testimony received at trial indicates that
Discover did, in fact, study the feasibility of lowering its prices at these merchants in the wake of
Visa’s and MasterCard’s consent decrees in order to induce steering. (Tr. at 985:23-987:4
(Hochschild/Discover).) But upon learning that its 100 largest merchants remained bound by
Amex’s NDPs, the network concluded that it did not make business sense to continue pursuing
these opportunities. (Id.) The court declines to second-guess Discover’s business judgment in
this regard, which appears to be a reasonable conclusion given the present state of the network
services market. Moreover, as Discover indirectly found, the vast majority of the locations
identified by American Express are very small merchants, often with potential annual Amex
charge volume well below $50,000. 48 The court finds little predictive value in the fact that these
merchants have not yet begun widespread steering given that: (1) there is no indication that the
Government or any GPCC network took meaningful steps to alert these small business owners of
their new freedom to participate in their customers’ card choices; (2) the potential savings for
merchants of this size would be small and may well be outweighed by the costs of steering; and
(3) small merchants ordinarily have very little, if any, direct contact with the GPCC networks.
(See Tr. at 4231:14-4235:22 (Katz).) Rather, it is more likely that large merchants will be the
vanguard of widespread steering in the United States, as occurred in Australia after the Reserve
Bank of Australia issued regulations capping interchange rates and permitting merchants to
differentially surcharge among the various credit card networks—a particularly strong form of
48
See Tr. at 2859:19-22, 3002:13-3004:2 (Pojero/Amex) (“[M]ost of the[] merchants in the gap are very, very small
merchants. . . . The vast majority sit well below [$]20,000 in AMEX annual charge volume potential.”); PX0021 at
‘127 (showing 91% of the gap is made up of merchants with under $50,000 in annual Amex volume); PX0890 at
‘353-54 (noting 75% of merchants in the gap are “probably half the size” of “your local florist”).
124
steering not sought by Plaintiffs in this case. (See Tr. at 5811:3-19, 5813:5-5814:14, 5817:11-16
(Gilligan/Amex); PX1126 at ‘629.) Today, however, the vast majority of these large merchants
remain bound by Amex’s NDPs. Finally, the court notes that American Express’s insistence that
steering would be unlikely to occur in the absence of its NDPs is wholly inconsistent with the
dire consequences it suggests could result from their removal. See infra Part VI.A.
The court is similarly unconvinced by American Express’s position that practical hurdles
to steering would defeat merchants’ efforts to promote price competition among the credit card
networks. For example, Amex argues that even if merchants wished to post their costs of
accepting GPCC cards on the various networks at the point of sale, they likely would be unable
to do so accurately, since few merchants are able to understand the complicated product-based
pricing structures used by Visa, MasterCard, and Discover. (See, e.g., Tr. at 651:15-652:21,
675:2-16 (Quagliata/Amex), 5549:17-5550:10 (Landau/DryBar).) Yet this concern appears to
rest on an overgeneralization (see, e.g., id. at 385:10-386:19 (Robinson/Ikea), 2380:11-20,
2381:20-2382:4 (Priebe/Southwest)), and the court believes that Amex’s competitors and/or the
merchant’s acquiring bank or processor would be in a position to provide merchants with an
accurate measure of their true costs of acceptance were there demand for such information (i.e.,
in a world where steering was allowed) (see id. at Tr. 651:1-10, 662:24-664:9 (Amex calculated
Riggins Oil’s “all in net effective Visa/MasterCard discount rate” using the merchant’s
statements from its acquiring bank), 664:16-665:16 (Quagliata/Amex), 916:14-19, 922:18-923:8
(Hochschild/Discover)). 49 Further, to the extent American Express is concerned that
unsophisticated merchants might steer customers away from the network based on “a prevailing
49
Similarly, the fact that some larger merchants with custom acceptance agreements with American Express would
ultimately choose not to disclose their negotiated effective discount rates at the point of sale due to confidentiality
concerns (see, e.g., Tr. at 1306:9-1307:3 (Kimmet/Home Depot), 2211:22-2212:3 (Haslam/OfficeMax)), does not
establish that this particular form of steering would not be pursued by the millions of merchants subject to Amex’s
standard rate tables.
125
misperception . . . among merchants as to what they’re paying for Visa/MasterCard versus
American Express”—in other words, that merchants incorrectly believe they are paying more to
accept American Express when compared to its competitor networks—the fault for allowing
such a misconception to persist and the burden for remedying it lies with Defendants. (Id. at
675:17-676:8 (Quagliata/Amex).) Even if the practical difficulties of steering were as significant
as Defendants suggest, the record provides several examples of how new technologies under
development by several merchants, including real-time pricing tools being developed by Official
Payments and Sinclair Oil (Tr. at 3158:4-3171:9 (Gibson/Sinclair) (discussing Sinclair’s “roll
back” pricing technology at gasoline pumps and its new mobile commerce application), 6144:146151:6 (Mitchell/Official Payments)), would help overcome any remaining logistical hurdles to
interbrand competition associated with merchant steering.
Finally, the court does not find that Discover’s anti-steering policies represent a
meaningful impediment to the type of inter-network competition likely to result from removal of
the challenged restraints. Unlike American Express, Discover has never sought to enforce its
anti-steering rules or threatened to terminate a merchant for attempting to influence a customer’s
card choice. (Tr. at 984:16-985:3 (Hochschild/Discover).) Discover has also suggested that it
may not enforce its anti-steering rules were Plaintiffs to prevail in this suit, and may abandon
them altogether. (Id. at 866:12-867:1, 985:13-22 (Hochschild/Discover).) In any case, even if
Discover were inclined to prevent merchants from steering, it likely would be unsuccessful given
the network’s relatively low market share. (See id. at 369:22-370:12 (Thiel/Alaska Airlines),
504:1-23 (Satkowski/Enterprise), 1544:14-23 (O’Malley/Best Buy); see also id. at 1608:7-13
(Brennan/Hilton) (testifying “[i]t would be easier to drop Discover than American Express . . .
[because t]he volume of charges that come through on Discover are far less than through
126
American Express”), 4226:18-4227:14 (Katz) (noting it would be difficult for Discover to
enforce its anti-steering rules if merchants were particularly motivated to steer due to a
significant pricing disparity given Discover’s low market share).)
*
*
*
In sum, Plaintiffs have proven by a preponderance of the evidence that American
Express’s Non-Discrimination Provisions have imposed actual, concrete harms on competition in
the credit and charge card network services market. The challenged restraints interrupt the
ordinary price-setting mechanism in this market by taking away a network’s reward for
competing on the basis of price, and thereby removing any network’s incentive to do so. In
rendering merchants less responsive to changes in price, the NDPs ensure that no competitor will
attempt to differentiate itself by being the lowest cost supplier, and consequently result in higher
prices for merchants and their customers. This conduct is prohibited by Section 1 of the
Sherman Act.
VI.
PRO-COMPETITIVE JUSTIFICATIONS
Upon Plaintiffs’ discharging their burden to establish that American Express possesses
market power in the network services market and that the challenged restraints have caused
actual anticompetitive harm therein, “the burden shifts to the defendants to offer evidence of the
pro-competitive effects of their agreement.” Geneva Pharm., 386 F.3d at 507. However, as
Judge Jones aptly noted in Visa, “the broad sweep of the rule of reason ‘does not open the field
of antitrust inquiry to any argument in favor of a challenged restraint that may fall within the
realm of reason,’” but rather, “‘focuses directly on the challenged restraint’s impact on
competitive conditions.’” Visa I, 163 F. Supp. 2d at 344 (quoting Nat’l Soc’y of Prof’l Eng’rs,
435 U.S. at 688). In the event Defendants provide a valid justification for the challenged
127
restraints, Plaintiffs must prove either that the challenged restraints are not reasonably necessary
to accomplish Defendants’ legitimate objective(s), or that the same objective(s) may be
“achieved by less restrictive alternatives, that is, those that would be less prejudicial to
competition as a whole.” Capital Imaging, 996 F.2d at 543; Visa II, 344 F.3d at 238. The rule of
reason then requires the court to “engage in a careful weighing of the competitive effects of the
agreement—both pro and con—to determine if the effects of the challenged restraint tend to
promote or destroy competition.” Geneva Pharm., 386 F.3d at 507; see also Capital Imaging,
996 F.2d at 542-44.
Here, Defendants propose two pro-competitive justifications for its NDPs, arguing that
the restraints are reasonably necessary (1) to preserve American Express’s differentiated
business model and thus the company’s ability to drive competition in the network services
market, and (2) to prevent merchants from “free-riding” on the network’s investments in its
merchant and cardholder value propositions. For the reasons set forth below, these purported
justifications do not offset, much less overcome, the more widespread and injurious effects of the
NDPs on interbrand competition in the relevant market.
A.
Defendants’ Ability To Drive Competition
American Express first proposes that its anti-steering rules are necessary to ensure its
cardholders enjoy a frictionless and consistent point-of-sale experience when using their
American Express cards—what the network terms “welcome acceptance”—which it asserts is
critical to the survival of Amex’s differentiated business model. (Defs. Post-Trial Br. at 24-25;
Tr. at 4477:12-20 (Chenault/Amex), 6356:10-6357:24 (Bernhiem); see also id. at 4943:94945:20, 4961:11-17 (Hayes/Amex) (noting “welcome acceptance” is central to Amex’s brand
promise).) By delivering a unique set of products and services to its merchants and cardholders,
Amex argues that its premium, spend-centric model has served as a key driver of innovation and
128
competition in the marketplace as its competitors have attempted to mimic Amex’s successes.
(See Tr. at 5115:1-21 (Gilbert), 6397:12-6398:4 (Bernheim).) Therefore, according to
Defendants, the NDPs are defensible under Section 1 of the Sherman Act because they safeguard
American Express’s ability to compete effectively in the credit card industry, and thereby
enhance interbrand competition. For the reasons that follow, the court finds that Defendants’
proffered justification is neither legally cognizable nor supported by the record.
Amex’s theory of pro-competitive effect largely rests on the proposition that, were the
network unable to rely on the NDPs to control merchants’ conduct toward its cardholders at the
point of sale, the company’s ability to pursue its differentiated business model would be
invariably and irreparably harmed. (Tr. at 4477:12-20 (Chenault/Amex).) Specifically, Amex
suggests that if merchants are permitted to “discriminate” at the point of sale by encouraging its
cardholders to use another form of payment—which, in Defendants’ view, is a “negative”
payment experience for the customer, regardless of how that steering is accomplished—its
cardholders will be less likely to use their Amex cards, not only at the steering merchant, but also
on subsequent transactions due to the effects of spillover. (See Tr. at 3067:14-3068:12
(Pojero/Amex), 4787:12-4788:25 (Glenn/Amex)); see also id. at 928:16-24
(Hochschild/Discover) (agreeing that, from a network’s perspective, attempts at point-of-sale
steering can damage a network’s relationship with its cardholder).) As merchants shift share
away from American Express to less expensive alternatives, 50 the network theorizes, it could
50
Defendants view a freely competitive network services market—that is, one without the NDPs—as inherently
tilted toward Visa and MasterCard, given those networks’ ubiquity and scale advantages. (See Tr. at 4434:214435:10 (Chenault/Amex) (“The reality is it’s a total[ly] unlevel playing field.”).) For instance, Amex contends that
because Visa and MasterCard have more cards in force when compared to American Express, merchant steering will
be a one-way street away from Amex, since merchants will not be willing to steer customers toward a card their
customers may not have. (See id. at 4475:22-4478:12 (Chenault/Amex); 6097:2-6098:20 (McNeal/Amex).) As a
factual matter, Amex is correct that Visa and MasterCard have more cards in circulation. Moreover, an Amex
cardholder almost always has a Visa or MasterCard product in his wallet, but the reverse is not true to the same
extent. (See Jt. Stmt. ¶ 18; Tr. at 4193:2-8 (Katz).) Yet Defendants’ inferential leap that this disparity will render
129
enter a “downward spiral” that would endanger the viability of its differentiated business model,
and by extension, its ability to promote competition in the credit card industry. Specifically, the
argument proceeds as follows: If merchants are permitted to steer, all merchants will see fewer
customers presenting Amex cards at the register due to spillover effects; as the importance of
accepting Amex cards declines, fewer merchants will do so; declining merchant coverage will
negatively affect cardholders’ perceptions of coverage and card utility; and, as a result, Amex’s
ability to acquire new cardholders and drive spending by existing cardholders will erode, leading
to further degradation of Amex’s acceptance network, and so on. (See, e.g., Tr. at 4477:124478:12 (Chenault/Amex), 5752:19-24, 5833:7-17 (Gilligan/Amex).) Therefore, in American
Express’s view, removing the NDPs and opening the door for unrestricted competition on
merchant pricing invites the demise of its differentiated model.
According to American Express, the prospect of a negative feedback loop resulting from
removal of the NDPs is exacerbated by the spend-centric nature of its current business model.
Defendants posit that any decline in the network’s discount revenue—whether attributable to
declining charge volume or a decision to reduce discount rates—will impair the company’s
ability to invest in its premium value propositions and continue to differentiate itself in the
marketplace. (See Tr. at 3771:24-3772:25 (Silverman/Amex).) Defendants aver that the
company’s ability to deliver a differentiated product to both sides of the GPCC platform,
including its premium cardholder rewards, is not only an important means of limiting merchants’
them unable to compete in a market where merchants can steer customers among the various card brands lacks
support in the record. First, Amex’s predictions about merchant willingness to steer customers to its network is
belied by the steering toward Amex that already occurs in the co-brand and “Official Card” contexts. Additionally,
there is no evidence in the record that merchants would be unwilling to attempt to steer customers if some customers
do not carry their preferred form of payment. In other words, if merchants were economically incentivized to steer
toward American Express—for example, due to lower discount rates or other remuneration—merchants may well be
willing to undertake such efforts even if fewer than 100% of their customers would be able to take advantage of the
offer.
130
willingness to steer and cardholders’ responsiveness to such efforts (see Tr. at 3717:19-3718:20,
3772:13-3773:12 (Silverman/Amex) (“[W]e would be in a vicious cycle where we would be
cutting benefits that we believe are absolutely critical and then we’d be in an even worse place
where merchants are steering against us and we don’t even have benefits and services that are
unique and competitive in the marketplace.”)), but also supposedly ensures that American
Express is able to continue serving as a competitive check on what Defendants view as the
effective duopoly of Visa and MasterCard (Tr. at 3541:1-19 (Silverman/Amex) (noting that to
compete “[i]n a network effect business like this, you’re either the biggest or you’re the best,”
and discussing the necessity of Amex’s ability to differentiate itself from its competitors),
4719:18-4720:18 (Glenn/Amex) (testifying that Amex has focused on providing differentiated
value to overcome Visa’s and MasterCard’s scale advantage)). American Express accordingly
views the challenged restraints as reasonably necessary to preserve its existing business model,
and theorizes that removing these protections from its merchant contracts would threaten its
ability to drive innovation and compete effectively with the dominant firms in the credit card
industry, thereby harming overall interbrand competition.
To the extent Defendants maintain that the NDPs drive interbrand competition in the
credit card industry, they focus primarily on the interrelated card issuance market. (See Tr. at
3544:1-3545:6 (Silverman/Amex) (describing issuing business as “fiercely competitive”),
6415:11-25 (Bernheim).) For example, Defendants repeatedly stressed at trial that competition
among American Express and various network/issuer tandems (e.g., Citibank and Visa) for cobrand relationships is intense. These ventures are often subject to a competitive bidding process,
and a large merchant looking for a new co-brand partner may expect to receive between six and
eight competitive bids in addition to Amex’s proposal. (See Tr. at 3670:10-25, 3672:21-3673:14
131
(Silverman/Amex) (describing “competition for . . . co-brand relationships” as “brutal”).) Even
in this limited context, Amex’s competitors enjoy a number of structural advantages. Visa and
MasterCard possess broader acceptance networks, and the issuing banks are able to leverage
their existing banking relationships with both merchants and prospective cardholders in order to
design more attractive bids. (See Tr. at 3673:15-3675:6 (Silverman/Amex).) Assuming, as
Amex does, that clearing the way for price competition on the merchant side of the credit card
platform—by removing the NDPs—would cause Amex’s coverage gap to grow and place
pressure on its ability to include generous remuneration packages in its co-brand proposals,
Defendants would be further disadvantaged in the race to secure these lucrative contracts, and
interbrand competition would suffer. (Tr. at 3676:2-3677:12 (Silverman/Amex).) Defendants
assert that a similar dynamic would play out in the third-party issuer and corporate card segments
should the NDPs be stricken, resulting in less overall interbrand competition, not more.
Though perhaps intuitively appealing, Defendants’ putative justification is inconsistent
with both the law and the factual record. It is axiomatic that the federal antitrust laws “were
enacted for ‘the protection of competition, not competitors.’” United States v. Apple Inc., 952 F.
Supp. 2d 638, 709 (S.D.N.Y. 2013) (emphasis in original) (quoting Brown Shoe, 370 U.S. at
320). 51 American Express recognizes as much, and insists that it prohibits merchant steering and
thereby preserves its differentiated business model not for its own sake, but for competition’s
sake. Yet the underlying premise of Defendants’ position highlights the flaw in this theory—i.e.,
that Amex’s current business model could not survive if exposed to the full spectrum of
interbrand competition. To find the NDPs to be reasonable restraints on trade because they
51
See also Tops Mkts., 142 F.3d at 97 (noting “[t]he Sherman Act protects competition as a whole in the relevant
market, not the individual competitors within that market”); Stamatakis Indus., Inc. v. King, 965 F.2d 469, 471 (7th
Cir. 1992) (stating the antitrust laws exist to “protect consumers from suppliers rather than suppliers from each
other”).
132
shield American Express’s preferred business strategy from a legitimate form of interbrand
competition, especially competition on the basis of price, would amount to “nothing less than a
frontal assault on the basic policy of the Sherman Act.” Nat’l Soc. of Prof’l Eng’rs, 435 U.S. at
695. Indeed, it is telling that American Express cites no legal authority—and the court similarly
finds none—to support the remarkable proposition that a restraint that effectively blocks
interbrand competition on price across an entire market may be justified under Section 1 because
the defendant firm would be less able to compete effectively in its absence. 52 See NCAA v. Bd.
of Regents of Univ. of Okla., 468 U.S. 85, 116-17 (1984) (rejecting pro-competitive justification
in the horizontal restraint context premised on an “assumption that the product itself is
insufficiently attractive to consumers” and thus needs to be insulated from “the full spectrum of
competition” as “inconsistent with the basic policy of the Sherman Act”); Nat’l Soc’y of Prof’l
Eng’rs, 435 U.S. at 696 (“In sum, the Rule of Reason does not support a defense based on the
assumption that competition itself is unreasonable.”).
Here, the point-of-sale “discrimination” from which American Express has insulated both
itself and, effectively, its competitors, represents a valid expression of interbrand competition in
the network services market. The court does not suggest that American Express does not have a
52
In its post-trial briefing, American Express cites only two cases for the proposition that “it is widely recognized
that vertical restraints can increase interbrand competition by facilitating product differentiation and competition on
quality.” (Defs. Post-Trial Br. at 24.) Neither case is directly applicable to the court’s analysis of the NDPs. In
Leegin, for example, the Supreme Court found the defendant’s vertical agreements setting a floor below which its
retailers could not price the goods in question were justified because “[m]inimum resale price maintenance can
stimulate interbrand competition—the competition among manufacturers selling different brands of the same type of
product—by reducing intrabrand competition—the competition among retailers selling the same brand.” 551 U.S.
at 890. Similarly, in New York v. Anheuser-Busch, Inc., a district court held that non-price agreements restricting
intrabrand competition among defendant’s wholesalers to be justified because the vertical restraints promoted
interbrand competition by ensuring the quality, consistency, and vigorous promotion of its product. 811 F. Supp.
848, 874-78 (E.D.N.Y. 1993). Defendants’ argument concerning its NDPs is quite different from the justifications
accepted by the courts in Leegin and Anheuser-Busch. Rather than restricting intrabrand competition in the name of
fostering greater interbrand competition, as was the case in the two decisions cited by American Express, the NDPs
impede one form of interbrand competition (competition on price) in the purported name of another (increasing the
number of rival firms with “welcome acceptance”). Therefore, the decisions cited by Defendants have no direct
bearing on the issue before the court in this case. And as described below, American Express may not decide on
behalf of an entire market which avenues of interbrand competition are open and which are closed.
133
legitimate business interest in ensuring its cardholders have a positive experience when using
their Amex cards. It undoubtedly does. (See, e.g., Tr. at 928:16-24 (Hochschild/Discover),
4943:9-4944:21, 4959:15-4961:17 (Hayes/Amex) (discussing impact of user experience on the
Amex brand, perceptions of coverage, and customer satisfaction); DX3974 at ‘279, ‘288; see
also Tr. at 6356:23-6357:24 (Bernheim).) But the concept of “welcome acceptance” developed
at trial describes marketplace conditions that have resulted in the absence of inter-network price
competition at the point of sale. This point is made plain by the fact that Amex’s restraints
preclude merchants from disclosing the relative costs of acceptance or otherwise encouraging
their customers to use a particular card brand, even when American Express is not mentioned. 53
See supra Parts I.C, V.A (discussing scope of Amex’s NDPs and their ability to suppress price
competition across the network services market).
In asserting this defense, American Express would also require the court to balance the
restraints’ pro-competitive effect in a separate, though intertwined, antitrust market against their
anticompetitive effect on the merchant side of the GPCC platform, a proposition for which
Defendants cite no legal authority. By effectively suppressing competition on merchant pricing,
Defendants’ anti-steering rules shift the bulk of interbrand competition in the credit and charge
card industry to the cardholder side of the platform. As noted earlier, it is in the card issuance
market that American Express and its rivals fiercely compete to acquire new cardholders and
53
Testimony by a number of American Express witnesses further illustrates that merchant steering and a positive
payment experience are not necessarily mutually exclusive. (See Tr. at 4575:7-10, 4576:12-18 (Chenault/Amex)
(suggesting that if MasterCard offered a discount, free shipping, or another “benefit” to a customer at Whole Foods
for using her MasterCard, it would not interfere with “welcome acceptance” so long as it occurred pursuant to a
marketing agreement), 4550:6-4551:23 (Chenault/Amex) (noting that when a merchant offers a discount for using a
co-brand card it “does not interfere with welcome acceptance”), 4557:22-4563:6, 4564:2-4565:16 (Chenault/Amex)
(testifying that steering does not run afoul of “welcome acceptance” if done pursuant to a “sponsorship”
relationship); see also id. at 790:7-24 (Quaglita/Amex); DX7525 at ‘382 (Amex’s consumer research concerning the
“We Prefer Visa” campaign concluded that “being exposed to a payment preference message does not make CMs
[cardmembers] and Non-CMs feel unwelcome,” and results in negative opinions of the merchant rather than the
network).)
134
capture share of wallet by, among other things, securing lucrative co-brand deals, signing
corporate card clients, and offering cardholders ever more robust suites of rewards and other
ancillary benefits intended to induce them to spend on a particular card. Thus, in essence, while
American Express professes that its NDPs enhance overall competition in the credit and charge
card industry, that pro-competitive end is accomplished by inhibiting competition in the network
services market for merchants—thereby ensuring that Amex’s spend-centric model continues to
be fueled by high merchant discount fees—in favor of greater competition in the interrelated but
distinct issuing market. As a general matter, however, a restraint that causes anticompetitive
harm in one market may not be justified by greater competition in a different market. 54 Whether
this rule precludes jointly weighing the relative gains and losses to interbrand competition in two
separate, yet interrelated, markets that together comprise a single two-sided platform has yet to
be explicitly considered by the Second Circuit. However, even if such cross-market balancing is
appropriate under the rule of reason in a two-sided context, 55 here Defendants have failed to
establish that the NDPs are reasonably necessary to robust competition on the cardholder side of
the GPCC platform, or that any such gains offset the harm done in the network services market.
54
See United States v. Topco Assocs., Inc., 405 U.S. 596, 610 (1972) (explaining that “the freedom to compete . . .
cannot be foreclosed with respect to one sector of the economy because certain private citizens or groups believe
that such foreclosure might promote greater competition in a more important sector of the economy”); United States
v. Brown Univ., 5 F.3d 658, 675 (3d Cir. 1993) (dismissing argument that a restraint led to increased competition in
other forms of competition as a “mere consequence of limiting price competition”). But see Sullivan v. Nat’l
Football League, 34 F.3d 1091, 1112 (1st Cir. 1994) (suggesting in dicta that courts are capable of balancing harms
inflicted to intrabrand competition in one market with gains to interbrand competition in another, distinct market).
A similar concept has been rejected in the merger context as well. See United States v. Phila. Nat’l Bank, 374 U.S.
321, 370 (1963) (rejecting the possibility that under the Clayton Act “anticompetitive effects in one market could be
justified by procompetitive consequences in another”).
55
American Express has previously rejected the propriety of such balancing, arguing before the Second Circuit in
Visa II that “no amount of issuer competition can eliminate the effects of increased prices, or reduced output, choice,
or innovation at the network level. . . . Decreased competition in the sale of an input or intermediate good, such as
network services, is harmful to consumers no matter how competitive the downstream market may be.” Brief of
American Express Co. as Amicus Curiae in Support of Affirmance at 6-7, Visa II, 344 F.3d 229 (2d Cir. 2003),
2002 WL 32828497, at *6-7.
135
Furthermore, the law does not permit American Express to decide on behalf of the entire
market which legitimate forms of interbrand competition should be available and which should
not. See Nat’l Soc’y of Prof’l Eng’rs, 435 U.S. at 695 (asserting that a defendant firm cannot
“impose[ its] views of the costs and benefits of competition on the entire marketplace”); see also
Trial Lawyers, 493 U.S. at 423-24. In force and effect, the NDPs represent a decision made by
Defendants on behalf of all participants in the network services market that networks will not
compete for additional share of GPCC spending by lowering their merchant pricing; rather, by
suppressing competition on the merchant side of the GPCC platform, Amex has effectively
compelled its rival networks and their issuing partners to focus their competitive efforts on
cardholders. (See, e.g., Tr. at 6482:19-6483:24 (Bernheim) (suggesting the NDPs do not impede
Discover’s ability to compete for share because, rather than lower its discount rates, Discover
could simply provide more robust rewards to its cardholders—in other words, it could be more
like American Express).) This effect is evident in Visa’s and MasterCard’s decisions to
introduce premium interchange categories in order to equip issuing banks with the resources
necessary to more directly compete with the rewards and other benefits provided by American
Express by virtue of its premium discount rate. Yet not all networks necessarily prefer to
compete primarily on the cardholder side of the platform. Without the NDPs in place, for
example, “Discover would aggressively pursue a strategy of lowering [its] prices and providing
incentives to merchants that would steer incremental volume to Discover.” (See Tr. at 872:3-10
(Hochschild/Discover).) Defendants’ restraints therefore deny other firms the ability to
differentiate themselves on the basis of their merchant pricing. Indeed, Defendants’ argument
that net competition in the credit card industry will decline if its NDPs are eliminated fails to
136
consider the likely increase in interbrand competition on the merchant side of the GPCC platform
that would result from unlocking price competition in the network services market.
Even if Defendants’ proffered justification were legally cognizable, the court finds that
their dire prediction of how business will be impacted by removal of the NDPs—namely, that
American Express will cease to exist or be relegated to a niche competitor in the GPCC
market—is not supported by the evidentiary record. It is not uncommon for antitrust defendants
to assert that their businesses will suffer irreparable injury if the challenged restraints are deemed
unlawful, and such predictions frequently prove inaccurate. (See Tr. at 5257:5-5260:14 (Gilbert)
(acknowledging that a Visa executive testified in the course of the Visa litigation that removal of
the exclusionary rules would be a “fatal blow” and that an adverse finding could result in the
“destruction” of the Visa venture in the United States), 6599:10-6601:11 (Bernheim) (noting that
a MasterCard witness similarly testified that a Government victory “could be a shattering blow to
MasterCard”); see also PX2248 at 25-27 (similar testimony in United States v. Microsoft).)
Here, Defendants have presented no expert testimony, financial analysis, or other direct evidence
establishing that without its NDPs it will, in fact, be unable to adapt its business to a more
competitive market and will instead cease to be an effective competitor in the GPCC industry.
Furthermore, the testimony adduced from various American Express executives and Defendants’
experts, discussing the viability of Amex’s current business model in a market in which
merchant steering is permitted, was notably inconsistent. Mr. Chenault, for example, took the
extreme position that “if the NDPs are eliminated [Amex] will not survive as a company,” and
that “if the NDPs go away, [Amex] will go away.” (Tr. at 4633:5-4634:7 (Chenault/Amex).)
But other witnesses, including Mr. Gilligan and Dr. Bernheim, took a more measured stance,
noting that removal of the NDPs would “have a very negative effect on [Amex’s] ability to
137
compete” (id. at 5833:7-17 (Gilligan/Amex)), and would require American Express to adapt its
current business model to compete in both the issuing and network services markets (see id. at
6486:10-6487:16 (Bernheim)).
Defendants are not, as they might have the court believe, powerless in a world in which
merchants and customers are able to jointly determine which payment product is used at the
point of sale. American Express is the single largest issuer of credit and charge cards in the
United States by purchase volume (see PX1560 at 8), maintains a worldwide acceptance network
with millions of accepting merchants, possesses one of the most valuable brands in the world
(DX6763 at ‘842; PX2143 (Amex has the most valuable brand of any financial services
company, including its issuer and network competitors)), is a highly profitable enterprise that
earned over $5 billion in post-tax profit in 2013 (PX1412 at ‘135 tbl.1), and is operated by a
tremendously qualified and resourceful set of employees and executives, many of whom testified
at trial. (See Tr. at 4005:12-4008:12 (Katz).) Additionally, American Express has a proven
track record of transforming itself and adapting its business model to suit changing competitive
landscapes and market conditions. 56 (See id. at 3132:1-3133:20 (Pojero/Amex), 4630:4-4632:18
(Chenault/Amex); see also id. at 5747:4-20 (Gilligan/Amex); PX1438 at ’22-24 (touting to
investors the company’s “flexible business model” and its ability to adapt to changing regulatory
conditions); PX0131 at 7 (same); PX1473 at 11 (same); PX1442 at 18 (same).) Thus, the court
56
Although the court declines to rely on examples of how American Express has adapted its business in those
foreign countries where merchant steering is permitted as evidence of how the company might react in the United
States, such evidence does illustrate Amex’s adaptability as an institution. In Canada, for example, merchants since
2010 have had the ability to differentially discount by credit card brand and American Express continues to operate
a profitable business in that country, maintaining its commitments to superior customer service and cardholder
rewards. (See Tr. at 2694:24-2695:10 (Funda/Amex), 5737:10-5738:6 (Gilligan/Amex).) Similarly, in Australia,
after the RBA allowed merchants to impose differential surcharges by card brand, American Express was able to
adapt to the new competitive landscape and today runs a highly profitable business in that country. (See PX1442
at 18 (Chenault telling investors “we’ve experienced surcharging in Australia. And while we had a momentary
adjustment period obviously, to that, I think we were able to surmount the issues and we’re running a very profitable
business in Australia”); Tr. at 5821:6-5823:15 (Gilligan/Amex) (noting Amex’s business in Australia “started
growing at a very healthy rate” after surcharging began and “we gained market share in Australia”).)
138
finds that American Express possesses the flexibility and expertise necessary to adapt its
business model to suit a market in which it is required to compete on both the cardholder and
merchant sides of the GPCC platform. (See Tr. at 4005:12-4008:12 (Katz) (testifying that Amex
will likely adapt its model “in order to prevent the downward spiral from happening” and that
“there is no basis for concluding that [the NDPs] are essential to protecting [Amex’s]
differentiat[ed] model”).)
American Express itself has identified a range of potential, permissible steps that the
company could take in order to protect its ability to deliver a differentiated product if steering is
permitted. In 2010, in response to the possibility that the Durbin Amendment might include a
provision permitting merchants to engage in differential discounting among credit card brands,
Amex convened a “Durbin Task Force” to analyze the competitive options available to the
network. Recognizing that the network “would look to protect the business . . . [by] looking to
incent the right behavior” among merchants, the Task Force identified a number of potential
competitive responses the network might take to mitigate the likelihood that its merchants would
be willing to engage in steering, including: engaging with merchants to reinforce Amex’s value
proposition and better explain the benefits of accepting American Express; targeting highvisibility “Anchor” merchants with its own steering programs to preserve Amex’s relevance in
key industries; shifting share away from non-friendly merchants; and reducing the network’s
discount rate in industries where merchants might be particularly inclined to steer. (Tr. at
2742:17-2743:18, 2752:4-9 (Funda/Amex); PX1176 at ‘383-84; PX0090 at ‘473; PX0091 at
‘906-07.) Relatedly, American Express may also combat merchant steering by improving its
value proposition to cardholders through, for example, increasing rewards and/or better
communicating how the benefits of using an Amex card are more valuable than the enticements
139
offered by merchants at the point of sale, so as to increase cardholder insistence and thus render
its cardholder base more resistant to merchants’ efforts to persuade customers to use another
form of payment. (PX1176 at ‘383, 385; PX0090 at ‘473; PX0091 at ‘906-07; see also Tr. at
2747:3-2748:7 (Funda/Amex).) These same competitive options are available to American
Express in the wake of this litigation. (See Tr. at 2746:16-2747:2, 2748:8-17, 2753:17-23,
2754:14-20 (Funda/Amex).)
The court recognizes, of course, that these potential responses to merchant steering are
not without their attendant costs. (Tr. at 2811:5-23 (Funda/Amex), 6425:8-24 (Bernheim)
(noting the Durbin Task Force looked at the range of possible responses, without necessarily
considering cost).) Yet the outcome of the court’s analysis in this case cannot be dictated by a
concern that Amex may be a less profitable enterprise if it is required to compete on the basis of
price with its fellow networks. 57 Moreover, in the event American Express is forced to decrease
its merchant discount rates in order to dissuade merchants from steering, the court finds it
unlikely that the downward spiral described by Defendants’ counsel will occur. 58 (See Tr. at
57
See Cargill, Inc. v. Monfort of Colo., Inc., 479 U.S. 104, 117 (1986) (describing the “perverse result” that would
accompany a “hold[ing] that the antitrust laws protect competitors from the loss of profits due to . . . price
competition”); Stamatakis Indus., 965 F.2d at 471 (“[A] producer’s loss is no concern of the antitrust laws, which
protect consumers from suppliers rather than suppliers from each other.”); Drug Mart Pharmacy Corp. v. Am. Home
Prods. Corp., 472 F. Supp. 2d 385, 402 (E.D.N.Y. 2007) (“[T]he antitrust laws are not intended to protect profit
margins but consumer welfare.” (internal quotation marks omitted)).
58
At the extreme, Defendants caution that the NDPs provide important protections against the network services
market “tipping” to the dominant networks, Visa and MasterCard, and that their removal could result in the market
being dominated by an effective duopoly. (See Tr. at 5051:11-22 (Gilbert).) Again, this argument overstates the
value of anti-steering rules in enhancing interbrand competition. Tipping describes “the tendency of one system to
pull away from its rivals in popularity once it has gained an initial edge,” which is particularly relevant in markets
subject to network effects, where the positive feedback loops are self-reinforcing. (DX0229 at ‘105-06; Tr. at
4201:17-4202:8 (Katz).) Here, however, the likelihood of the network services market tipping to Visa and
MasterCard is limited both by Amex’s provision of a differentiated product and by the fact that merchants and
cardholders typically are multi-homing—i.e., merchants accept multiple card brands, which “greatly diminishes the
[likelihood of] tipping and makes it feasible – or more likely that multiple networks will survive.” (Tr. at 4250:254251:18 (Katz); see also id. at 4208:1-17 (Katz), 5052:2-17 (Gilbert).) Defendants present no empirical evidence to
suggest that this market is particularly likely to tip if steering is permitted, and the quantitative evidence that was
presented to the court suggests that the market is less susceptible to this dynamic than Amex contends. For example,
the evidence shows that although Amex’s share fell 20% between 1990 and 1995, the market did not tip to Visa and
MasterCard. (Id. at 5154:11-17.)
140
4006:1-4008:12 (Katz) (concluding that a downward spiral is unlikely to result from removal of
the NDPs, based on market evidence and Defendants’ capacity to take steps to prevent such an
outcome).) Defendants’ depiction of the “vicious cycle” they contend would result from removal
of the NDPs not only disregards the various responses the firm has available to mitigate the
degree and efficacy of merchant steering, but also ignores the countervailing incentives and
market forces that suggest that Amex’s current model will be more resilient than Defendants
imply. (Tr. at 4011:22-4014:2 (Katz).) If, as Defendants have strenuously insisted, American
Express truly offers merchants a differentiated and premium set of services as compared to its
competitors—i.e., access to higher-spending cardholders, closed-loop analytics, and the like—
merchants will take that additional value into account when deciding whether and to what extent
to steer customers to other forms of payment. (Id.) Merchants also can be expected to consider
the reaction of insistent Amex cardholders when deciding whether and how much to steer. Put
plainly, assuming American Express actually offers premium value to its merchants, the market
will tolerate Amex charging a premium price for its network services, even in the absence of the
NDPs—albeit, in all likelihood, at a smaller premium than it charges today. In such an
environment, removal of the NDPs will benefit interbrand competition, as the net price
established across the two-sided GPCC platform will be jointly determined by the two sets of
consumers served by American Express, and not artificially inflated by contractual restraints that
isolate supply from demand. 59
The court nonetheless shares American Express’s concerns about disrupting the
competitive landscape in such a concentrated, complex market. (See, e.g., Tr. at 5126:19-5127:5
(Gilbert).) As the court noted at the beginning of this Decision, it would have strongly preferred
59
In striking this equilibrium, it is possible that American Express will be forced to raise its fees to cardholders in
order to offset declining discount revenue—in effect, requiring the recipients of its premium rewards to bear a
greater share of their cost.
141
the parties to have resolved this dispute among themselves. Absent such an agreement, the court
is compelled to enforce Section 1 based on the prevailing law in this circuit and the facts
developed at trial. American Express plainly faces a number of structural challenges in this
market: It is a discretionary card for the vast majority of cardholders, it has significantly fewer
cardholders than the other networks, and it is accepted at roughly three million fewer merchant
locations. But American Express is not permitted to resort to unreasonable vertical restraints in
order to ensure what it subjectively views to be a “level playing field.” (See, e.g., Tr. at 2749:4-9
(Funda/Amex).)
Notwithstanding the outcome of this litigation, American Express’s card products and
cardholders will remain protected from unfair denigration and discrimination at the point of sale.
Plaintiffs do not challenge, and the court does not find unlawful, those aspects of the network’s
Merchant Regulations and acceptance agreements that prohibit merchants from
“mischaracteriz[ing]” Amex’s products, “engag[ing] in activities that harm [Amex’s] business or
the American Express brand (or both),” or requiring customers to pay a fee when using their
American Express card that is not also charged when using another card brand. (See Am.
Compl. ¶ 28; see also Tr. at 5828:8-5829:14 (Gilligan/Amex); PX2754 (demonstrative exhibit
highlighting NDP language not challenged in this case).) Thus, even after a remedial order is
issued pursuant to this Decision, under the surviving provisions of Amex’s merchant rules, a
merchant likely will still be prohibited from, among other things, posting a sign purporting to
show the networks’ relative costs that is inaccurate or that mischaracterizes the relative cost of
American Express, refusing to accept an Amex card when presented at the point of sale, or
disparaging American Express to the merchants’ customers. Second, to the extent American
Express has expressed a concern that Visa and/or MasterCard may seek to undermine Amex’s
142
success cycle at the point of sale in order to create a “contagion effect” that spreads through
Amex’s business at non-steering merchants (see Tr. at 6418:3-6420:4 (Bernheim); see also
DX5378 at ‘405 (discussing Visa strategy in Australia where prohibitions on steering are more
relaxed)), Defendants retain the right and resources to seek enforcement of the federal and state
antitrust laws if they feel their competitors have strayed beyond the confines of legitimate
competition (Tr. at 4009:3-4010:15 (Katz)). See, e.g., Apple, 952 F. Supp. 2d at 708 (“Another
company’s alleged violation of antitrust laws is not an excuse for engaging in your own
violations of law.”).
B.
Free-Riding
American Express also asserts that its NDPs enable competition by reducing merchants’
ability to “free-ride” on the network’s various investments in its merchant and cardholder value
propositions. Present even in purely competitive markets, the so-called “free-rider effect” occurs
when a competitor’s incentive to make a particular pro-competitive investment is undercut by the
diversion of its expected return to another firm without compensation. 60 (See Tr. at 3994:113995:15 (Katz), 6426:14-6427:8 (Bernheim).) See also Leegin, 551 U.S. at 913-14; Major
60
As explained by Judge Easterbrook, the classic example of free-riding occurs in the context of retail distribution:
Manufacturer produces a product or improvement that requires explanation or demonstration—
perhaps a television set with an improved degaussing coil. Retailer #1 demonstrates the effects to
consumers in a showroom filled with TV sets, some with and some without the feature. Such a
demonstration is costly in merchandise, in staff time, in floor space. Neither Manufacturer nor
Retailer #1 can charge the consumer for this information. Its value is too uncertain to expect the
consumer to pay for access to the sales floor, and a retailer is not apt to gain customers by
threatening to charge them if they leave without buying. So a consumer may leave the store with
valuable information; Retailer #1 recovers the cost of supplying this information in the purchase
price of the product, not with a separate charge. Yet a consumer armed with the information may
order the product from Retailer #2, which offers no information. Retailer #2 can make a profit at a
lower price than does Retailer #1, for Retailer #2 has lower costs. To compete, Retailer #1 must
lower its own price, and that means lowering its costs too—cutting cost by cutting services that
consumers value.
Chi. Prof’l Sports Ltd. P’ship v. Nat’l Basketball Ass’n, 961 F.2d 667, 675 (7th Cir. 1992) (citing Lester G. Telser,
Why Should Manufacturers Want Fair Trade?, 3 J.L. & Econ. 86 (1960)). (See also Tr. at 6437:23-6438:6
(Bernheim) (describing the same free-rider problem by referencing the “expression that Best Buy is Amazon’s
showroom”).)
143
League, 542 F.3d at 305 (referring to the “‘free-rider’ problem” as “one entity’s cashing in on the
efforts of another”); Chi. Prof’l Sports Ltd. P’ship v. Nat’l Basketball Ass’n, 961 F.2d 667, 675
(7th Cir. 1992) (“Free-riding is the diversion of value from a business rival’s efforts without
payment.”). The Supreme Court has recognized that prevention of free-riding is a legitimate,
pro-competitive justification for vertical restraints on trade. See Leegin, 551 U.S. at 590-91
(finding resale price maintenance to be justified in part because in its absence, services “that
enhance interbrand competition might be underprovided because discounting retailers can freeride” on firms “who furnish services and then capture some of the increased demand those
services generate”); GTE Sylvania, 433 U.S. at 55 (noting vertical restraints may promote
interbrand competition by remedying “market imperfections such as the so-called ‘free rider’
effect”). Here, however, to the extent Defendants have identified potential avenues of free-riding
foreclosed by its NDPs, the court finds that the competitive benefits of preventing these forms of
merchant behavior do not offset the significantly more pervasive harms done to interbrand
competition by the same restraints.
American Express first proposes that the NDPs are justified in order to prevent merchants
from free-riding on the analytics-based services the network provides directly to its merchants.
(Defs. Post-Trial Br. at 26.) In the absence of these restraints, Defendants suggest, for example,
that merchants could draw Amex cardholders to their establishments through a targeted
marketing campaign facilitated by Amex’s investments in its closed-loop model—efforts which,
according to the network, are both effective and have persistent benefits for participating
merchants—only to steer those cardholders to a less expensive card network at the register. (Tr.
at 6429:13-6430:8, 6431:11-6433:17, 6440:1-9, 6442:13-6443:1 (Bernheim).) Were steering
permitted, American Express asserts its ability to “deliver[] useful advertising products to
144
merchants” would be hampered, reducing beneficial competition among the networks and
merchants. (See id. at 6443:20-6444:20 (Bernheim).) Similarly, American Express is concerned
about merchants’ ability to exploit its other market intelligence products, as merchants
conceivably could use the insights contained therein to more effectively attract Amex
cardholders and then steer them to another card product. This free-riding justification is flawed
in a number of respects.
For example, American Express’s efforts to justify its anti-steering rules as a necessary
measure to protect its pro-competitive investments in delivering data-analytics services is
significantly undercut by the fact that the network can—and, indeed, often does—price and sell
these ancillary benefits separately from its core network services. Where, as here, “payment is
possible, free-riding is not a problem because the ‘ride’ is not free.” Chi. Prof’l Sports, 961 F.2d
at 675 (“What gives this the name free-riding is the lack of charge. Retailer #1 does not charge
the customer for a valuable service; Retailer #2 does not pay Retailer #1 for delivering this
service.” (emphasis in original)); United States v. Microsoft Corp., No. 98-CV-1232 (TPJ), 1998
WL 614485, at *21 (D.D.C. Sept. 14, 1998); see also 13 Areeda & Hovenkamp ¶ 2223b3.
American Express acknowledges that it sells its merchant analytics to some merchants for a fixed
fee, separate and apart from what the merchant pays for Amex’s network services, and often
pursuant to a separate contractual agreement that ensures Amex is paid for the benefit conferred.
(See Tr. at 1072:5-15 (Quagliata/Amex) (“At times . . . we charge for [data-analytics products]
and at times we don’t, depending upon the nature of the analysis.”); see also id. at 396:25-397:16
(Robinson/Ikea) (Ikea received a “one-time freebie” of merchant analytics, but understood it
would have to pay for any additional market analyses it wanted in the future), 2438:13-2439:12
(Priebe/Southwest); DX5391; DX4852; Tr. at 4889:18-20 (Glenn/Amex).)
145
In other words, unlike the service-providing retailer in the paradigmatic free-rider
scenario, American Express is not reliant on the sale of its core network services to recover the
cost of providing these additional benefits; to recoup its investment and obviate the “free” benefit
captured by the merchant, American Express can simply charge all merchants a fee for its
analytics, rather than charging only some. (See Tr. at 3995:19-3997:3 (Katz) (“[T]o the extent
there’s freeriding [in the provision of direct benefits to merchants] it’s America[n] Express’s
choice whether to allow it. . . . [I]t’s their decision whether or not to charge for [merchant
analytics].”).) Amex’s marketing services are similarly divisible from its provision of network
services; consequently, a less restrictive fee-for-service alternative to the NDPs is available for
use by American Express should it choose to do so. 61 (Id. at 3996:14-3997:3 (Katz); see also id.
at 5120:13-5121:9 (Gilbert) (noting “[Amex] could still . . . ask for a direct payment instead of
using the merchant discount rate,” while noting such a structure may reduce the value of the
program).) Thus, American Express’s ability to separately price and sell the data-analytics
services it claims are susceptible to free-riding, as well as its concomitant ability to exclude a
merchant from receiving these benefits if Amex believes it is free-riding or otherwise paying an
insufficient amount for these services, 62 leads the court to conclude that the network possesses
61
Drs. Bernheim and Gilbert dispute the feasibility of a fee-for-service model for American Express’s marketing
services. (Tr. at 5120:13-5122:11 (Gilbert), 6447:17-6451:8 (Bernheim).) Under the current model, they argue,
American Express recovers the cost of supporting its merchants’ marketing efforts through the merchant discount
rate. They note this compensation structure effectively aligns the interests of both the network and merchant over
the course of the promotion—both parties aim to increase spend—and that a fixed-fee structure would be less
effective in doing so. (Id. at 6448:1-6449:11 (Bernheim).) Yet Plaintiffs do not suggest that the only feasible
pricing structure for Amex’s marketing services is a fixed fee, and Defendants have presented insufficient
justification as to why joint marketing agreements could not have incentive-based compensation structures that are
equally effective at eliminating moral hazard and reducing potential free-riding.
62
American Express does not dispute that it retains the capacity to end free-riding by a specific merchant should it
so choose by, for example, ending the specific marketing campaign or even terminating that merchant’s acceptance
agreement. For example, when Defendants learned Marquis Jet was steering customers away from its network
during a joint marketing promotion with American Express—i.e., precisely the kind of free-riding envisioned by
Defendants—the network terminated its promotion contract with Marquis Jet. (See id. at 6440:14-6441:11
(Bernheim); DX2016 at ‘814.) Defendants have provided no reason to why a similar approach, including active
monitoring of the merchant’s charge volume during and after the promotion, would be neither effective nor feasible
146
equally effective and significantly less restrictive means of preventing this form of free-riding. 63
Therefore, as compared to the alternatives, the marginal pro-competitive benefit of the NDPs in
this regard is minimal.
American Express next proposes that the NDPs are necessary to prevent merchants from
indirectly free-riding on its cardholder investments, including rewards. (Defs. Post-Trial Br.
at 26; Tr. at 6428:15-6429:22 (Bernheim).) Plainly, however, investments tied to card use (such
as Membership Rewards points, purchase protection, and the like) are not subject to free-riding,
since the network does not incur any cost if the cardholder is successfully steered away from
using his or her American Express card. (See Tr. at 3997:4-3998:5 (Katz), 5120:8-12 (Gilbert),
6428:4-6429:12 (Bernheim).) American Express instead suggests that certain “fixed expenses”
undertaken by the network to build a positive and trusting relationship with its cardholders are
undermined when merchants direct its cardholders to less expensive competitors. (Id. at 6912:13 (Closing Argument); see also id. at 6429:13-22 (Bernheim).) Though Defendants make no
effort to identify the fixed expenses to which its experts referred or to explain how they are
subject to free-riding, the court surmises the free-rider problem potentially at issue. Suppose, for
example, an American Express cardholder who self-selects a particular merchant specifically
because that merchant accepts American Express cards, but, at the point of sale, the cardholder is
induced by the merchant to use another form of payment in order to receive a 5% discount off of
the retail price. The merchant would then appear to have benefitted from American Express’s
among the limited number of merchants with whom American Express conducts such programs. (See Tr. at
3996:14-3997:3 (Katz).)
63
American Express also overstates the potential harm of free-riding on these types of investments. The network’s
ability to separately price and sell its merchant analytics, for example, ensures an independent incentive for the
network to continue investing in its ability to provide these analytics-based services. Put simply, the network has a
means to ensure a return on its investment separate and apart from the discount revenue associated with those
cardholders who are successfully steered to a different card product.
147
efforts to direct its cardholders to Amex-accepting merchants without having to incur the cost of
actually accepting Amex on that purchase.
Avoidance of this type of potential free-riding does not suffice to overcome the marketwide harms effected by the NDPs, however. First, the court understands that the free-riding
problem identified above is limited; if the customer in question is not directed to the merchant by
virtue of its acceptance of American Express, but instead—in what the court finds is a more
credible or likely scenario—walks into the merchant’s establishment intending to make a
particular purchase regardless of the card options available, no free-riding occurs when that
customer is steered to a non-Amex card. (Tr. at 6434:6-6435:12 (Bernheim).) More
importantly, the anticompetitive effect associated with free-riding of this type is matched by a
countervailing benefit for consumers on both sides of the platform. An economically rational
consumer will not accept a merchant’s invitation to use another card product unless he believes
that what the merchant is offering is of greater value than the rewards or other benefits he
receives for using his Amex card. Thus, where a merchant is able to steer the self-selecting
customer away from American Express, it appears that both the merchant-consumer and
cardholder-consumer derive a net benefit. Moreover, the court remains unconvinced that this
type of steering will reduce the network’s incentive to invest in creating value for cardholders.
To the contrary, as one Amex executive testified, the network may choose to increase its
investments in rewards in order to make its cardholders more resistant to merchants’ efforts to
steer them to other card brands. (Tr. at 2748:3-17, 2754:7-20 (Funda/Amex); see also PX0090 at
‘473 (outlining potential Amex responses to steering toward debit cards).) 64 As a result, while
64
To the extent Amex argues that merchants could free-ride by benefitting from its cardholders’ tendency to spend
more on average without having to incur the associated costs (see Tr. at 5119:9-17 (Gilbert)), the network has
presented no evidence to suggest that customers steered to another card product would still spend as much as they
would have if they used their Amex cards.
148
Defendants have identified a potentially viable avenue of free-riding, the court does not agree
that the potential competitive harm posed by this type of merchant behavior outweighs the
negative effect its NDPs have on competition in the relevant market.
Lastly, Defendants propose that the challenged restraints promote competition by
ensuring merchants cannot free-ride on American Express’s investments to enhance its brand
value. (Defs. Post-Trial Br. at 26.) Specifically, Defendants argue that merchants derive a
benefit from linking their brands with that of American Express—a phenomenon known as
“brand association” or “credentialing”—when they represent to consumers that they accept the
networks’ credit and charge cards; Defendants maintain that when merchants steer away from
American Express, they effectively free-ride on Amex’s investments in strengthening its own
brand. (Tr. at 6444:21-6445:10 (Bernheim).) The court finds this purported justification to be
without merit. As Dr. Katz correctly noted, any alleged credentialing is “not the result of a
specific investment” by Defendants, and is instead “an ancillary benefit [of] their business
model,” such that any free-riding would not endanger the network’s underlying incentive to
promote its brand. (Tr. at 3998:21-4000:1 (Katz).) Additionally, Amex’s own survey data
concerning cardholder perceptions of Amex-accepting merchants indicate that the network’s
ability to confer a credentialing benefit trails that of its competitors, casting doubt on whether
there is in fact any particular benefit associated with accepting Amex that is subject to freeriding. (See PX0815 at ‘294-95; Tr. at 4002:1-4004:12 (Katz).) Thus, the court cannot conclude
the NDPs are justified to prevent free-riding on Amex’s brand.
149
s/ Nicholas G. Garaufis
Disclaimer: Justia Dockets & Filings provides public litigation records from the federal appellate and district courts. These filings and docket sheets should not be considered findings of fact or liability, nor do they necessarily reflect the view of Justia.
Why Is My Information Online?