You are on page 1of 57

COMPETITION IN TWO-SIDED MARKETS: THE ANTITRUST ECONOMICS OF PAYMENT CARD

INTERCHANGE FEES
Author(s): Benjamin Klein, Andres V. Lerner, Kevin M. Murphy and Lacey L. Plache
Source: Antitrust Law Journal, Vol. 73, No. 3 (2006), pp. 571-626
Published by: American Bar Association
Stable URL: http://www.jstor.org/stable/40843688
Accessed: 23-07-2017 15:49 UTC

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted
digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about
JSTOR, please contact support@jstor.org.

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
http://about.jstor.org/terms

American Bar Association is collaborating with JSTOR to digitize, preserve and extend access to
Antitrust Law Journal

This content downloaded from 14.139.214.178 on Sun, 23 Jul 2017 15:49:06 UTC
All use subject to http://about.jstor.org/terms
COMPETITION IN TWO-SIDED MARKETS:
THE ANTITRUST ECONOMICS OF
PAYMENT CARD INTERCHANGE FEES

Benjamin Klein
Andres V. Lerner
Kevin M. Murphy
Lacey L. Plache*

I. INTRODUCTION

Standard economics provides a well-understood framew


competitive determinants of market prices that is now wid
for antitrust analysis. In "two-sided markets," where firms supp
demanded by two interrelated groups of consumers, these c
forces operate in a somewhat more complex way and unders
antitrust implications requires extending the standard fram
example, a newspaper publisher faces demand from both
advertisers. The publisher must balance demand on the two
market in determining two interrelated sets of prices, taki
of the fact that lowering subscription prices and thereb
readership will increase advertising prices. These "network
increased readership on advertising value are what make th
analysis unique and the antitrust implications somewhat un

Payment card systems also operate in a two-sided market, w


cards demanded by two distinct, interrelated groups of
cardholders (who use the cards to make purchases) and merc
accept the cards as payment for purchases) . The cardholder
include annual and other fees, interest charged on exten
and cardholder benefits (such as rewards in kind, cash-back

* Benjamin Klein is a Professor Emeritus of Economics, UCLA, and a Dir


Kevin M. Murphy is the George J. Stigler Distinguished Service Professor
University of Chicago. Andres V. Lerner and Lacey L. Plache are Princip
authors have served at various times as consultants to Visa U.S.A. Inc. The
in this article are solely those of the authors. We are indebted to Kevin Gr
at an early stage of the project. Research assistance was provided by Connie
Larrea, and Geralyn Villaflor.

571

This content downloaded from 14.139.214.178 on Sun, 23 Jul 2017 15:49:06 UTC
All use subject to http://about.jstor.org/terms
572 Antitrust Law Journal [Vol. 73

enhancements) that reduce the net cost of using


merchant price is the amount subtracted from th
payment is made to the merchant, known as the
The average merchant discount on American Expr
example, is currently about 2.6 percent, so that a
American Express 2.6 percent less than the amoun
chase.1

In contrast to "closed-loop" proprietary payment card systems, such


as American Express, Diners Club, and Discover, in which the single
company that owns the payment system issues cards to cardholders,
acquires transaction receipts from merchants, and directly sets card-
holder and merchant prices, Visa and MasterCard are "open-loop" pay-
ment card associations in which many independent members issue cards
to cardholders, acquire transaction receipts from merchants, and inde-
pendently set cardholder and merchant prices. The Visa and MasterCard
associations do not issue cards, acquire transaction receipts, or set card-
holder or merchant prices.2

Although Visa and MasterCard do not set cardholder and merchant


prices, they influence these prices by implementing interchange fees
that flow from the acquiring bank (where the merchant's account is
credited) to the card-issuing bank (where the consumer's account is
debited). For example, when a consumer uses a Visa card to make a
purchase at a particular merchant, the merchant's acquiring bank pays
an interchange fee on the transaction to the consumer's card-issuing
bank. The acquiring bank currently pays an average effective interchange
fee on Visa credit card transactions of approximately 1.80 percent to
the issuing bank.3
Because the interchange fee is a major element of the acquiring bank's
cost of processing the payment card transaction, it is a major element
in determining the amount the acquiring bank charges the merchant
to handle the transaction, that is, the merchant discount. For example,
the card-issuing bank compensates the acquiring bank by the amount
of the customer's transaction minus the interchange fee, say 1.80 percent.
The acquiring bank also has some other costs of handling the transaction,

1 American Express Company Annual Report 2005 at 30.

2 The traditional distinctions between payment card systems were blurred recently whe
American Express and Discover entered into agreements with some banks to issue t
cards. As a result, American Express and Discover no longer issue all cards or set pr
to all cardholders.

3 Data from Visa U.S.A. We focus primarily on Visa U.S.A. in our description of the
institutional facts. The MasterCard system historically has operated in an economically
similar way to Visa in most fundamental respects.

This content downloaded from 14.139.214.178 on Sun, 23 Jul 2017 15:49:06 UTC
All use subject to http://about.jstor.org/terms
2006] Competition in Two-Sided Markets 573

say 0.40 percent, which suggests that the merchant dis


be 2.20 percent. On the other side of the transaction, t
bank receives from the cardholder (when, and if, the c
his bill) the full amount of the transaction. The car
interchange fee revenue along with any fees and inter
holders must cover the cost of operating its card-issuin
ing the cost of cardholder defaults, fraud, marketing
provision of cardholder benefits.

Interchange fees have been criticized by various p


groups have complained that interchange fees impose a
on merchants and, therefore, indirectly on customers
higher retail prices.4 These complaints have led to a num
tions of interchange fees in other countries and, in som
ment regulation.5 Further, some economists hav
interchange fees lead to the inefficient imposition o
and check users.6 Interchange fees also have becom
recent antitrust litigation in the United States based on t
amount to the collective setting of a price above the co

In what follows we first present the fundamental eco


sided markets, demonstrating how competitive supplier
determine prices by balancing demand on the two side
In the case of payment card systems, this economic ba
leads to merchant and cardholder prices that involve m
a larger fraction of total system costs than do cardhol

4 See, e.g., Financial Services Issues: A Consumer's Perspective: Hearing B


on Financial Institutions and Consumer Credit, Comm. on Financial In
(Sept. 15, 2004) (statement of Food Marketing Institute).
5 For instance, the Reserve Bank of Australia (RBA) and the Austra
Consumer Commission conducted a study on interchange fees that r
of Visa and MasterCard credit interchange fees in 2003. Similarly
Commission settlement reduced Visa's interchange fees on cross-b
transactions in the European Economic Area. Visa International - Mu
Fee, 2002 OJ. (L 318) 17, 18-21 (Nov. 22, 2002). A parallel investiga
cross-border interchange fees is currently underway by the Euro
addition, an investigation of Visa's and MasterCard's interchange fees
in the United Kingdom by the British Office of Fair Trading (OF
investigations and regulation of interchange fees in 10 key countrie
E. Weiner & Julian Wright, Interchange Fees in Various Countries: Developm
4 Rev. Network Econ. 290 (2005).
6 See, e.g., Alan S. Frankel, Monopoly and Competition in the Supply and Exchange of Money,
66 Antitrust L.J. 313, 341-49 (1998); Dennis W. Carlton & Alan S. Frankel, Transaction
Costs, Externalities, and "Two Sided" Payment Markets, 2005 Colum. Bus. L. Rev. 617, 633-
35 (2005).
7 In re Payment Card Interchange Fee and Merchant Discount Antitrust Li tig., 398
F. Supp. 2d 1356 (E.D.N.Y. 2005).

This content downloaded from 14.139.214.178 on Sun, 23 Jul 2017 15:49:06 UTC
All use subject to http://about.jstor.org/terms
574 Antitrust Law Journal [Vol. 73

due to payment card system market power over m


demand sensitivity generally is much greater on t
the market than on the merchant side of the market. Cardholders decid-
ing which payment card to carry and use can and do play off competing
payment card systems; merchants, on the other hand, generally find it
profitable to accept cards from all major payment card systems so as not
to lose profitable incremental sales from consumers that use only one
payment card. Competitive balancing by payment card systems takes
account of network effects by pricing based on the relative demand
elasticities of cardholder and merchants. This results in an equilibrium
with a relatively higher merchant price than cardholder price. This
equilibrium occurs regardless of whether the payment card system is a
closed-loop system, such as American Express, which determines card-
holder and merchant prices directly, or an open-loop system, such as
Visa and MasterCard, which uses positive default interchange fees to
achieve such balancing. This equilibrium also occurs whether or not the
payment card system possesses market power.

In addition to using interchange fees to competitively balance the two


sides of the market, an open-loop payment card system must have default
interchange fees in order for the system to operate effectively in the
presence of an "honor-all- cards" rule. An honor-all- cards rule obligates
merchants who contract to accept a payment system's cards to accept
all cards of the payment system that are presented by consumers. This
"guaranteed acceptance" is one of the primary features of a payment
card system. Letting merchants who have agreed to accept a payment
card and display the payment system's symbol decide which of the sys-
tem's cards to accept and which to reject would impose a significant
cost on the payment system's brand name by undermining consumer
expectations of guaranteed acceptance.
Given an honor-all- cards rule, bilaterally negotiated interchange fees
in the absence of a default interchange fee would create an incentive
for individual card issuers in an open-loop payment card system to "hold
up" acquirers by demanding arbitrarily high interchange fees on transac-
tions made with their cards. In particular, individual Visa and MasterCard
issuers would have the incentive to take advantage of the fact that mer-
chants accepting Visa or MasterCard have agreed to honor-all- cards to
make unreasonably high interchange fee demands. Such demands would
impose an externality on the entire payment card system, and eventually
lead some merchants to drop acceptance of the payment system's cards.
A default interchange fee prevents such holdups by placing a cap on
interchange fees that can be bilaterally negotiated between individual

This content downloaded from 14.139.214.178 on Sun, 23 Jul 2017 15:49:06 UTC
All use subject to http://about.jstor.org/terms
2006] Competition in Two-Sided Markets 575

issuers and individual acquirers in the Visa and Mas


systems.

After demonstrating the necessity of interchange fees for an open-


loop payment system to function effectively, we show that interchange
fees are not a measure of payment card system market power. Inter-
change fees influence relative prices paid by cardholders and merchants,
not the total price of a payment card system, that is, the sum of the
prices paid by cardholders and merchants. The market power of a pay-
ment system determines the ability of the payment system to charge a
total price above costs, but has no predictable effect on relative prices.
The relative prices paid by cardholders and merchants are determined
by two-sided market balancing considerations. Accordingly, the level of
interchange fees has no particular relationship to the presence or
absence of market power. In fact, the economic effect of balancing,
either through interchange fees (for open-loop systems) or by directly
determining cardholder and merchant prices (for closed-loop systems),
is to maximize payment system output rather than to exercise market
power by restricting output.

Changes in the level of interchange fees over time in the United States
can be explained better by the competitive balancing of economic forces
on the two sides of the payment card market than by an exercise of market
power. In particular, the cause of the decline in Visa and MasterCard
interchange fees during the 1970s and 1980s was not a decline in Visa
and MasterCard market power. Interchange fees fell because of the
increased importance to Visa and MasterCard of broad merchant accep-
tance, which shifted the competitive balance more towards merchants.
The greater importance of this network effect for the Visa and Master-
Card payment systems compared to American Express explains why Visa
and MasterCard merchant discounts have been below the merchant
discounts of American Express, a payment card system with a substanti
smaller share of transactions.

In contrast, since 1998 increased competition among payment car


systems for issuance has shifted competitive balancing toward the card
holder side of the market, resulting in higher interchange fees an
merchant discounts. The development and promotion of issuer bran
names and the growth of co-branded and affinity cards with card-specif
reward programs has created increased cardholder loyalty to particular
issuers. This increased loyalty has increased the ability of issuers to pla
off Visa and MasterCard because issuers now have the ability to shi
most of their cardholders to one or another system by promising that

This content downloaded from 14.139.214.178 on Sun, 23 Jul 2017 15:49:06 UTC
All use subject to http://about.jstor.org/terms
576 Antitrust Law Journal [Vol. 73

system dedicated issuance. Such competition betwe


has led, as economics would predict, to an incr
level of interchange fees. This movement tow
arrangements and the resulting increase in in
resulted in increased rewards and lower fees for cardholders.

We use the economic implications of competitive balancing in two-


sided markets to assess whether there are inefficiencies associated with

interchange fees. In particular, we analyze the argument that interchan


fees are an exercise of market power which lead to the inefficient "cros
subsidization" of payment card users by cash and check users and show
that this is not the case. Any such cross-subsidization would not be limit
to customers using payment cards of open-loop systems with defau
interchange fees nor would it be related to the exercise of market powe
The same cross-subsidization would occur when customers use payment
cards of closed-loop systems, such as American Express and Discove
payment systems with no interchange fees and no market power. Balanc
ing in any two-sided market results in some consumers cross-subsidizin
other consumers. In fact, what is described as cross-subsidization is com-
mon in competitive one-sided markets, such as when merchants offer
without charge certain services, such as free parking, that only som
customers value and use. Furthermore, contrary to the usual antitrust
concerns regarding the restriction of output from the exercise of marke
power, it is argued that cross-subsidization of payment card users result
ing from interchange fees creates excess payment card output.

It has been proposed that the claimed excess payment card outpu
inefficiency could be controlled by regulating interchange fees, perha
at zero. However, this would amount to a rejection of the normal compet
tive process. In fact, interchange fees recently have increased due
increased competition for issuers among payment card systems. Anothe
criticism of interchange fees is that competition among merchants lead
merchants to pay too much for payment card acceptance because
merchant's decision to accept a payment card primarily has inter-retaile
effects, with little change in the aggregate demand facing all merchant
However, it makes no economic sense to conclude that competitio
among merchants must be controlled because of these inter-retaile
effects. These effects are the essence of the competitive process. If mer
chants could collude in their payment card acceptance decisions, inter-
change fees would no doubt be lower. Similarly, merchant spending on
advertising, parking, or other customer services also would be lower if
they could collude on the provision of these services. That does no
make such collusion desirable. A regulatory solution that mimics such
collusion on the part of merchants would be contrary to the goal o

This content downloaded from 14.139.214.178 on Sun, 23 Jul 2017 15:49:06 UTC
All use subject to http://about.jstor.org/terms
2006] Competition in Two-Sided Markets 577

antitrust and, by preventing competitive balancing, w


loop payment card systems at a competitive disa
marketplace.

II. COMPETITIVE PRICING IN TWO-SIDED MARKETS

A. Competitive Pricing in Two- Sided Markets Involves


Balancing Demand on Each Side of the Market

To illustrate the economic forces at work in the pricing o


supplied in two-sided markets, we initially consider the
newspaper that provides both a product for readers and a m
advertisers. Newspaper publishers must determine two sets
prices to readers (e.g., home delivery and newsstand prices)
to advertisers. The standard economic analysis of two-s
recognizes that a newspaper publisher in setting prices t
advertisers to maximize overall profit will take account of
of demand by readers to purchase the newspaper and the e
demand by advertisers to place advertisements in the n
addition, the newspaper publisher will take account of t
prices set on one side of the market may influence demand
side of the market. For example, if the newspaper publ
reader prices and thereby increases readership, this will incr
for advertising in the newspaper. These cross- effects are r
as "network effects," whereby the quantity on one side of
influences demand on the other side of the market.

To formalize these economic forces that operate in standard two-sided


markets, we denote the two sides of the newspaper market as R and A,
where R refers to readers and A to advertisers. PR and PA, QR and Qa,
and CR and CA then represent the respective prices, quantities, and costs
of supply on the two sides of the market. Suppliers of two-sided products
maximize total profit from the two sides of the market, defined in the
case of newspapers as

(PrQr-Cr) + (PaQa-Ca). (1)

That is, total profit is equal to total revenue minu


the newspaper to readers plus total revenue minu
advertisements to advertisers.

Total profit, equation (1), can be rewritten as

Pr (Qr, Qa) • Qr - CR (QR) + PA (QR, Q0 ■ Qa - CA (Qa). (2)

This content downloaded from 14.139.214.178 on Sun, 23 Jul 2017 15:49:06 UTC
All use subject to http://about.jstor.org/terms
578 Antitrust Law Journal [Vol. 73

This assumes that PR and PA, the demand prices of


may be related to the quantities of both R and A.
market there exists the standard own elasticity of
demand price of R is related to the quantity of R
of A is related to the quantity of A. But there is
network effects, where the demand price of R is r
of A and the demand price of A is related to the

The profit-maximizing quantities of newspapers


and advertisements sold to advertisers, Qa, are giv
-'p pip

||QA + gQR + PA
These equations can be rewritten as

P^fJ-MC-Ho, (5)
PA(l+J-).MC.-gßR, (6)
where T|R and T|A are the own-price elasticities of demand by readers and
advertisers on the two sides of the market, that is, how much demand
by readers increases as the price of newspapers decreases and how much
demand for advertising increases as the price of advertising decreases;
MCR and MCA are the respective marginal costs of supplying a newspaper
to readers and of providing advertisements to advertisers; and 9Pa/5Qr
and 3Pr/3Qa are the cross (network) effects, or how much the value of
advertising to advertisers increases with increasing quantities of readers
and how much the value of the newspaper to readers increases with
increasing quantities of advertising.

Equations (5) and (6) state that, in general, profit-maximizing prices


will be determined on each side of a two-sided market where marginal
revenue (or P (1 + l/r|)) equals marginal cost minus marginal network
effects. If there are no network effects, prices on each side of the market
are determined entirely by marginal costs and elasticities of demand,

8 The equation assumes that CR and CA, the costs of supplying the newspaper to readers
and advertisements to advertisers, are related only to the respective quantities of R and A.

This content downloaded from 14.139.214.178 on Sun, 23 Jul 2017 15:49:06 UTC
All use subject to http://about.jstor.org/terms
2006] Competition in Two-Sided Markets 579

the traditional analysis of profit-maximizing prices. With ne


the relative prices paid by demanders on the two sides o
also will depend on the marginal network effects of the tw
customers. In general, firms selling goods or services in two
to two different groups of consumers will tend to charge
relative to marginal cost to the group that is more price-s
has a higher elasticity of demand) and generates greater m
work effects (i.e., where increased quantity has a larger ef
value of goods or services supplied on the other side of th
a supplier wishes to increase price, it will be more prof
on the side of the market where the demand response
effects are likely to be lower.

The importance of relative demand elasticities and relativ


effects in determining pricing will depend upon the parti
For newspapers, relative network effects explain why there
a higher price relative to marginal cost for advertisers
subscribers. Although there could be a greater own pr
response of readers to subscription price changes compared
price elasticity response of advertisers to advertising rate ch
important determinant of relative prices is the fact that t
effects of increased readers on the value of the product to
are generally much greater than the network effects of incr
ing on the value of the product to readers.9 The much larg
effects generated by increased readers on advertising v
to increased advertisers on reader value might even justify
a zero subscription price, that is, giving the newspaper awa

9 We say "generally" because there may be some newspaper ads that


valuable to readers, such as ads for groceries or movies, compared to, say
Because many individuals buy the local newspaper to check grocery or
few do so to check airline fare specials, additional grocery and movie
readers' demand price more than additional airline ads. Accordingly, at le
pers offer differential rates to grocery and movie advertisers. The Feder
sion has concluded that prohibiting such differential advertising rate str
the public interest. FTC v. Times-Mirror Co., 100 F.T.C. 252, 1982 FTC LEXIS 31 (1982).
10 In fact, in some circumstances the equilibrium prices given by equations (5) and (6)
may result in a negative subscription price. However, publishers must be careful not to
distribute copies that will not be read, which would increase their costs without increasing
advertising rates. In most cases, newspaper publishers cannot set a negative price because
consumers would "buy" unlimited quantities and merely throw them away. There are some
examples, however, where negative prices are sustainable. For example, the publisher of
Playbill, the program supplied to theater patrons, actually pays the theater (based on
attendance) for die right to distribute the program free of charge. Lisa Gubernick, Publish-
ing: Offstage, a Fight far Top Billing, Wall St. J., Apr. 26, 1999, at Bl.

This content downloaded from 14.139.214.178 on Sun, 23 Jul 2017 15:49:06 UTC
All use subject to http://about.jstor.org/terms
580 Antitrust Law Journal [Vol. 73

B. Competitive Balancing of Payment Card System Prices

Payment card systems also operate in two-sided markets, jointly


two classes of customers - consumers (who gain the ability to co
mate a transaction with the card rather than with some other medium
of exchange that may be less desirable) and merchants (who benefit
from the ability to increase their sales by accepting payment methods that
customers wish to use) . The demands of these two classes of customers are
interdependent, such that the value of a payment system to consumers
increases with the number of merchants that accept the card and the
value of a payment system to merchants increases with consumer use of
the card.

While the basic factors that influence pricing, demand elasticities and
network effects, are the same for a payment card system as for a newspa-
per, there are significant economic differences between the role of these
factors in determining pricing in these two types of two-sided markets.
In particular, while a newspaper publisher supplies two distinct but
interrelated products, newspapers for readers and a medium for advertis-
ers, a payment card system supplies only one product, payment card
transactions that are jointly consumed by a cardholder, who uses the pay-
ment card to make a transaction, and a merchant, who accepts the
payment card as a method of payment. Accordingly, the two sides of a
payment card system are not only interdependent, as are the two sides
of the newspaper market, but their consumption of payment card transac-
tions must be directly proportional. The two sides of the market can be
thought of as providing inputs into the supply by the payment system
of this single product. This fundamental distinction between pricing by
payment card systems and firms in other two-sided markets has important
implications for how network effects and demand elasticities influence
competitive pricing of a payment card system.
To examine the economic forces that determine the profit-maximizing
cardholder and merchant payment card system prices, we analyze a
closed-loop proprietary payment card system that directly operates on
both sides of the market, signing up and servicing both cardholders and
merchants. As we describe in Part III below, all payment systems must
balance the two sides of the market, regardless of whether they operate as
closed-loop or open-loop payment systems. The only difference between
closed-loop and open-loop systems is that while a closed-loop system will
balance the two sides of the market by setting prices directly to cardhold-
ers and merchants, an open-loop payment system will balance indirectly
through the use of system- determined default interchange fees. There-
fore, the analysis of the economic forces that determine pricing in closed-
loop systems also applies to open-loop payment systems.

This content downloaded from 14.139.214.178 on Sun, 23 Jul 2017 15:49:06 UTC
All use subject to http://about.jstor.org/terms
2006] Competition in Two- Sided Markets 581

Similar to newspaper publishers that set reader and ad


a closed-loop payment card system must establish prices
sides of the market - cardholder prices (such as card fee
interest rates), Pc, and merchant prices (merchant d
payment card system supplier earns money on the volum
that are jointly made by cardholders and merchants,
The profits of the payment system can be expressed as
Q and the total price per transaction, Pc + PM, minus th
transaction, C:12

QÌPm+Pc-Q. (7)
The volume of transactions can be
of merchants that accept the paym
per merchant, U.u We assume that
the payment card depends on the p
that cardholder usage depends on t
Thus, the profits of the payment c

N(PM) ■ U(Pc) -[Pm + Pc-Q- (8)

11 The following analysis of pricing in the two-sided payment


most closely to Richard Schmalensee, Payment Systems and Interch
Econ. 103 (2002). The original pathbreaking work on the economic m
card systems was William F. Baxter, Bank Interchange of Transactional P
Perspectives, 26 J.L. & Econ. 541 (1983). Other more recent analy
upon Baxter's framework include Jean- Charles Rochet & Jean Tiró
Competitors: Some Economics of Payment Card Associations, 33 RAND
Julian Wright, The Determinants of Optimal Interchange Fees in Payme
Econ. 1 (2004). For surveys of the literature, see Richard Schmale
A Review of the Literature, 1 Payment Cards Econ. Rev. 25 (Win
Review of the Literature] ; Sujit Chakravorti, Theory of Credit Card Ne
Literature, 2 Rev. Network Econ. 50 (2003); Jean- Charles Rochet, Th
Fees: A Synthesis of Recent Contributions, 2 Rev. Network Econ. 97
12 In contrast to a newspaper, where there are separate costs of p
to readers and a medium to advertisers, a payment card system in
providing a payment card transaction that is consumed jointly
merchant. Thus, we do not separate the costs of a payment card tran
and merchant-side costs. As we shall see, profit-maximizing relativ
holder prices do not depend on the costs of providing a payment
specification of the model also assumes that there is a constant
the payment card system and no fixed cost per cardholder that ca
or per merchant that accepts the card. If there are significant fix
system per cardholder or per merchant, such costs will also infl
cardholders and merchants.

13 We model the merchants' demand for the payment card as their decision as to whether
to accept the payment card. However, it is important to recognize that merchant demand
will not be determined entirely by merchant decisions regarding acceptance. As merchant
fees associated with a payment card increase, merchants may not only decide to refuse to
accept the payment card, but also may steer consumers to use alternative payment cards
and payment methods. See discussion infra pp. 586-87.

This content downloaded from 14.139.214.178 on Sun, 23 Jul 2017 15:49:06 UTC
All use subject to http://about.jstor.org/terms
582 Antitrust Law Journal [Vol. 73

The profit-maximizing prices are given by:


dN
^-. U[PM+PC-Q +NU=0 (9)

N-^-[PM + Pc-Q+NU=
Equations (9) and (10) imply that the profit-maxim
occurs where:

dN/N dU/U
dPM = dPc • (11)
Equation (11) states that profit-maximizing prices
such that the percentage change in merchant acceptan
increase (or decrease) in PM is equal to the percentag
holder use from an equal decrease (or increase) in
behind this equation is simple: If these terms are no
payment network can increase transaction volume and p
ing relative prices between cardholders and merchan
the total transaction price the same. For instance,
decrease in merchant acceptance from increasing PM b
is less than the percentage increase in cardholder use f
Pc by that same amount, then transaction volume can
raising PM and lowering Pc while keeping the total pr
same. By shifting the balance between cardholders and
manner, the payment system supplier, therefore, can in
volume and profits. Thus, the profit-maximizing relat
to merchants and cardholders, PM/Pc, will depend on th
ties of demand on the two sides of the market.

This result is equivalent to the result of our first model of newspape


readers and advertisers. A payment system supplier will set profit maxi-
mizing prices (PMand Pc) based on the relative demand sensitivities of th
r u i ^ u dN/N _ dU/ U
two sides oi r the u market, i which are repres
orM orc

These demand sensitivity terms a


in the newspaper model {r'R and
differences between competitive
such as newspapers, which suppli
products, and in a payment card
to two distinct groups of consum
trary to the newspaper model, in

This content downloaded from 14.139.214.178 on Sun, 23 Jul 2017 15:49:06 UTC
All use subject to http://about.jstor.org/terms
2006] Competition in Two-Sided Markets 583

by the terms (dPA/dQß) and (diVdQi) that were sep


from the price elasticity terms, the payment card sy
dN/N jdU/U UJUU. .... , lrr
- =77; - and - =r= - embody UJUU. both price sensit

Because cardholders and merchants jointly


payment card transactions, their consumpt
tions must be directly proportional. As a res
effects are directly related. If the paym
cardholder price, Pc, and, therefore, increa
proportionately increases the volume of tra
therefore, the value of the payment system
the payment system supplier lowers the pr
therefore, increases merchant acceptance
holder use of the payment card and, there
cardholders will also increase proportionate
situation like the newspaper example discus
effects operate primarily in one direction, t
to increased advertising value, payment car
effects in both directions. That is, incr
increases cardholder value and increased cardholder use increases mer-

chant value. Therefore, network effects will influence the payment sys
balancing of the two sides of the market.

Some economists have argued that network effects cannot explain


balanced pricing of the existing payment card systems, i.e., merchan
discounts and cardholder fees and benefits, because network effects are

14 Our specification of the model does not include all network effects. In particu
merchant acceptance of the payment card not only depends on the price to the mercha
but also on cardholder usage; and cardholder usage depends not only on the pric
cardholders, but also on merchant acceptance of the payment card. Accordingly,
profit equation for the payment system supplier, equation (8), can be more fully expre
as N{PM, U) • U(P& N) • [PM + Pc - C'- Such a formulation would allow the value of the
payment card system to merchants to increase more than proportionately with the number
of cardholders. This would occur, for example, if there were significant fixed costs to
merchants of participating in the payment system. Our formulation of payment system
profit is restricted to simplify the model and more clearly illustrate the fundamental
economic effects at work. The network effects that are omitted from the payment card
model are analogous to the network effects in the newspaper model, where the price to
advertisers depends on the number of readers and the price to readers depends on the
quantity of advertisements. The network effects in the payment card model, which derive
from the fact that both cardholders and merchants jointly consume a single product (a
payment card transaction) and, therefore, their consumption of payment card transactions
must be directly proportional, are not present in the typical two-sided market, such as
newspapers, in which two distinct but interrelated products are consumed by the two sides.

This content downloaded from 14.139.214.178 on Sun, 23 Jul 2017 15:49:06 UTC
All use subject to http://about.jstor.org/terms
584 Antitrust Law Journal [Vol. 73

no longer significant for these systems.15 Networ


may be important initially when a payment system
At that point in time a "chicken-and-egg" problem
chants value a payment system only if a sufficient
use it and cardholders value a payment card only
of merchants accept it.16 Accordingly, they argu
and-egg problem does not exist with the now wel
card systems. These economists argue that network
for large, well-established payment card systems, s
Card, because the large cardholder bases of the
there is little value to merchants of additional cardholders.17

However, network effects are not limited to those related to the


chicken-and-egg problem associated with new payment systems. Nor do
network effects in a payment card system diminish significantly when
the payment system has a large cardholder base. The value of the payment
system to merchants depends on the volume of transactions made by
cardholders. Greater cardholder use increases transaction volume and,
therefore, proportionately increases the value of the payment system to
merchants. Similarly, the value of the payment system to cardholders
depends on merchant acceptance of the payment card. These network
effects between the two sides of a payment system are important in
enhancing the value of the payment card system and influence current
equilibrium merchant discounts relative to cardholder fees and benefits
regardless of whether a payment system is nascent or established and
regardless of the size of the cardholder base.

15 See, e.g., Michael L. Katz, Reform of Credit Card Schemes in Australia II at 12-15
(RBA 2001), available at http://www.rba.gov.au/PaymentsSystem/Reforms/CCSchemes/
IICommissionedReport/2_commissioned_report.pdf; Carlton & Frankel, supra note 6, at
625, 639.
16 In these circumstances, no issuer will decide independently to issue any cards if there
is no merchant network and no acquirer will decide independently to sign up merchants
if there are no cardholders. Therefore, a payment card supplier initially may have to invest
by supplying the payment card system at a total price that is below total cost until the
payment card system gets established. For instance, to achieve a cardholder base large
enough to attract merchants, Bank of America did not charge annual card fees for its
BankAmericard and spent a considerable amount of money on mass mailings of credit
cards and on advertising to educate consumers about its cards, with the result that it
incurred substantial initial losses. &œ Joseph Nocera, A Piece of the Action: How the
Middle Class Joined the Money Class 25-33 (1994). If there is not a single supplier
operating on both sides of the market or another mechanism of internalizing the network
effects, there will be "network externalities" and less than the optimal incentive to invest.
See SJ. Liebowitz & Stephen E. Margolis, Network Externality: An Uncommon Tragedy, 8 J.
Econ. Persp., Spring 1994, at 133.
17 Supra note 15.

This content downloaded from 14.139.214.178 on Sun, 23 Jul 2017 15:49:06 UTC
All use subject to http://about.jstor.org/terms
2006] Competition in Two-Sided Markets 585

How network effects affect the balancing of the two side


i.e., the relative prices paid by cardholders and mer
largely on the relative sensitivities of cardholders a
changes in prices. A payment system will take into
effects in balancing the two sides of the market by low
side that is relatively more price-sensitive. Taking acco
effects by pricing according to relative demand sensit
payment system to increase total transaction volum
the payment system to both cardholders and merchan
level of the total payment system price, PM + Pc. Comp
payment system services will typically lead to merchant
share of the total costs of supplying the payment s
cardholders. This is because cardholders usually are mo
than merchants. In general, consumers consider alte
cards to be reasonably close substitutes and are willing
card usage between payment systems in response to
holder prices. In contrast, many merchants find it pro
all major payment cards because most consumers do
major payment card. Additionally, consumers that carry
are likely to prefer a particular card as a result, for exa
reward program. Therefore, merchants are likely to los
mental sales if they do not accept all cards.18

For example, consider a merchant's decision of wheth


tance of Visa or MasterCard. Since a non-trivial number of the merchant's
customers are likely to carry only one of the bankcards or have a prefer-
ence for using a particular payment card, not accepting either one of
the cards is likely to lead to an unprofitable loss of sales.19 In fact, a
merchant need not lose very many sales to make it uneconomic to drop
acceptance of either Visa or MasterCard. To illustrate, assume that it
costs merchants about 2 percent, on average, to complete a transaction
when a consumer pays with, say, a Visa credit card. We can further
reasonably assume that dropping Visa credit cards results in half of the
lost Visa credit transactions being made with other credit and charge

18 Because nearly all American Express cardholders also carry other cards (see infra note
19), many merchants do find it profitable to not accept American Express. However, the
merchant will still lose sales if some consumers have a preference for using American
Express.
19 Of U.S. households that have a major credit card, 34% have only one such card. David
S. Evans & Richard Schmalensee, Paying with Plastic : The Digital Revolution in
Buying and Borrowing 87, 232 (2005). Because relatively few consumers carry only an
American Express card (id. at 87), the one card carried by the overwhelming majority of
the households that carry only one card is likely to be Visa or MasterCard.

This content downloaded from 14.139.214.178 on Sun, 23 Jul 2017 15:49:06 UTC
All use subject to http://about.jstor.org/terms
586 Antitrust Law Journal [Vol. 73

cards and the other half being made with othe


The transaction fee savings on transactions that
with a Visa credit card will depend upon the avera
and charge cards and non-credit alternative pa
assume that the transaction cost of using other c
is also about 2 percent and that the cost of non-cr
is about half the cost of credit, or 1 percent,21 th
transaction fee savings from dropping the Vis
about 50 basis points (the difference between the
credit and the average of non-Visa payment m
Assuming this potential cost savings and an av
margin of approximately 25 percent,22 a merchan
a 2 percent decrease in the sales that would hav
credit cards (or 1 in 50 sales) for it not to be prof
to drop acceptance of Visa credit.23 A similar cal
any other payment card used by the merchant's

Consequently, the demand for a particular sy


is likely to be significantly less price-sensitive fo
cardholders. However, it is important to reco
demand elasticity will not be determined entirely
regarding acceptance. As merchant fees associated

20 Consumers that would have paid with a Visa credit car


alternate credit or charge card rather than some other pay
another card, which approximately two-thirds of cardholder
have assumed for purposes of this illustrative calculation tha
will be made with other credit and charge cards.
21 For the purposes of this illustrative calculation, we base
2000 FMI estimates, which indicate merchant costs of approx
pays with cash and 0.75% if a customer pays with a check. Se
It All Adds Up: An Activity- Based Cost Study of Retail Payments (2000) at 14-15.
But see infra note 96 regarding limitations of the FMI data. Also, based on 2004 effective
interchange fees, signature debit transactions cost approximately 35 basis points less than
credit transactions, or 1.65%, and PIN debit transactions cost roughly 65 basis points less
than signature debit transactions, or 1%. Signature debit and credit interchange fees: Visa
U.S.A. PIN debit interchange fees: EFT Data Book, 2006 ed., ATM & Debit News (Sept.
15, 2005).
22 The average retail gross margin defined as gross sales less cost of goods sold as a
percent of gross sales was 28% in 2002. U.S. Census Bureau, Annual Benchmark Report
for Retail Trade and Food Services: January 1992 through February 2004 at ix,
46 (Mar. 2004).
23 For a merchant to maintain or exceed the same total profit after dropping acceptance
of Visa credit cards, the merchant's lost profit on lost Visa credit sales, 2% x 25%, must
be equal to or less than the transaction cost savings of 50 basis points on the Visa credit
sales that move to other payment methods. The percent decrease in sales necessary not
to drop acceptance of Visa credit cards would be even lower for higher margin retailers,
such as jewelers and furniture stores, which have a higher gross profit margin than 25%.

This content downloaded from 14.139.214.178 on Sun, 23 Jul 2017 15:49:06 UTC
All use subject to http://about.jstor.org/terms
2006] Competition in Two- Sided Markets 587

increase, merchants may not only decide to refuse to accept


card, but also may steer consumers to use alternative payme
Specifically, many merchants now encourage consumers to a
use merchant co-branded cards that offer significant reward
also may offer cash discounts at the time of transaction as a
for cardholders to use another means of payment, provided
is non-discriminatory between a Visa card and a "compar
Thus, cash discounts at the time of transaction are permitte
tions made with a number of payment methods, including
PIN debit cards, and proprietary store cards. Moreover, me
offer non-cash rewards at the time of transaction to custom
type of payment card, such as a hotel offering a room upgr
use of, say, an American Express card.26 In addition, merchan
transaction volume to a competing payment card by agreein
that payment card. In fact, as Visa's and MasterCard's merch
have increased relative to American Express's, more mer
decided to accept American Express.27
However, because demanders on the cardholder side of the market
are much more willing to substitute across payment systems than demand-
ers on the merchant side of the market, merchants generally will bear
a disproportionately greater share of the total costs of payment system
supply compared to cardholders. In fact, in contrast to the newspaper
example, the equilibrium price charged to consumers on the cardholder
side of the market may be negative. Many cardholders who pay their
bill in full each month ("transactors") currently pay a negative price for
use of a payment card, receiving free float as well as rewards without
paying any annual fees.28 Negative prices are sustainable in this case
because cardholders must make transactions to collect the benefits.
Therefore, there is no risk of a cardholder collecting the benefits with

On the other hand, for lower margin retailers, such as discount stores, the decrea
sales required to make dropping the cards unprofitable would be higher.
24 Visa U.S.A. Inc. Operating Regulations, Nov. 15, 2004, Rule 5.2.B.3, provides that
"The Merchant may request or encourage a Cardholder to use a means of payment other
than a Visa Card."

25 A "comparable card" is defined as "any other branded, general purpose payment card
that uses the cardholder's signature as the primary means of cardholder authorization."
See id. Rule 5.2.D.2.
26 See id. Rule 5.2.B.3.

27 See infra discussion of Figure 3.


28 According to data from the Visa Payment Panel Survey, 17% of bankcards in 2003
were rewards cards with no annual fees. In addition, even cardholders who do not pay
their bills in full each month ("revolvers") may pay a negative price if they hold small
enough balances relative to the value of the rewards and other benefits that they receive.

This content downloaded from 14.139.214.178 on Sun, 23 Jul 2017 15:49:06 UTC
All use subject to http://about.jstor.org/terms
588 Antitrust Law Journal [Vol. 73

actually "buying" the product, as could occur i


prices for newspapers.29

This competitive balancing equilibrium holds not


loop proprietary payment card systems analyzed h
loop payment card systems. However, in an ope
system, balancing the two sides of the market can
directly setting cardholder and merchant prices b
default interchange fee.

III. OPEN-LOOP PAYMENT CARD SYSTEMS REQUIRE DEFAULT


INTERCHANGE FEES

A. The Efficiencies of an Open-Loop Payment Card System

Payment card systems historically have fallen into two groups


first group consists of closed-loop systems, such as American
and Discover, that issue cards, acquire merchant transaction r
and set cardholder and merchant prices. These systems are propr
systems in that each system may retain any profits earned by the
The second group of payment card systems includes open-loop as
tions of financial institutions, such as Visa and MasterCard. Unlik
loop payment card systems, the open-loop systems do not issue c
acquire merchant transaction receipts, or set prices to cardholde
merchants. These functions are performed by the multiple issue
acquirers that are members of these open-loop systems. Visa is o
as a pass-through joint venture in that the association performs
aspects of payment card services for its members on a nonprofit

Many of the earliest payment cards in the United States were p


tary cards that could be used at only one merchant (e.g., Sears, M
or Texaco cards). Beginning in the 1950s, proprietary travel and
tainment cards, such as Diners Club and American Express, ex
the convenience benefits of payment cards by offering consume

29 See discussion supra note 10.


30 As described supra note 2, recent changes have somewhat blurred the dis
between closed-loop and open-loop payment systems, in that American Expres
cover no longer issue all cards or set all cardholder prices. In addition, MasterCa
was an open-loop, pass-through joint venture until 2002, was converted at that t
private share corporation, formally separating ownership from membership, an
its structure again in May 2006 when an initial public offering led to the cre
publicly traded company. Therefore, MasterCard is no longer a pass-through join
payment card system, but rather is operated as a proprietary payment card system.
MasterCard remains an open-loop payment card system and still does not dire
cards, acquire transaction receipts from merchants, or set cardholder or mercha

This content downloaded from 14.139.214.178 on Sun, 23 Jul 2017 15:49:06 UTC
All use subject to http://about.jstor.org/terms
2006] Competition in Two- Sided Markets 589

ability to use a single "general purpose" payment car


different merchants.31 These early travel and entertain
of little use to the majority of consumers who made m
chases at local merchants, a limitation that was solved by
of the general purpose bankcard. Bank of America
BankAmericard, the predecessor to the Visa credit ca
BankAmericard payment card program, originally a close
tary payment card program, became an open-loo
expanded to include other financial institutions, first a
system through a series of licensing agreements in
quently through the creation of a non-proprietary, pas
venture among financial institutions in 1970.32

The Visa and MasterCard open-loop payment card sy


efficient way for financial institutions to cooperate in
mon platform from which they then could compete ind
supply of payment card services. Large economies of sc
individual financial institutions from being able to ente
card business without such cooperation. Economies o
part, because of significant network effects, which imply
payment card system must operate at a fairly large min
scale to develop large merchant and cardholder bases. In
are significant supply-side economies of scale in deve
brand with worldwide consumer recognition and goodw
ing and maintaining the computer infrastructure neces
payment transactions efficiently. Given these large eco
in developing and sustaining a payment card system, it w
individual financial institutions to allocate the supply o
to the Visa and MasterCard associations. The joining tog
individual financial institutions in open-loop payment c
take advantage of these efficiencies by providing a paym
platform from which each financial institution can comp

Ol ine JJiners i^iuo cara was introduced in lyou. uinmgon tne l,ujj, rNEWSWEEK,jan. zy,
1951, at 73. The American Express card made its first appearance in 1958. Peter Z.
Grossman, American Express: The Unofficial History of the People Who Built
the Great Financial Empire 280-81 (1987). For a detailed history and discussion of
the payment card industry, see Evans & Schmalensee, supra note 19.
32 Visa U.S.A. Inc., About Visa USA: History, http://usa.visa.com/about_visa/about_
visa_usa/ history.html. The joint venture originally was called National BankAmericard
Inc. (NBI) . The name of the joint venture was changed to Visa in 1976. Id. The MasterCard
joint venture, originally called the Interbank Card Association, was formed in 1966
MasterCard International, The MasterCard Story, http://www.mastercardinternational.com/
corporate/the_mastercard_story.html.

This content downloaded from 14.139.214.178 on Sun, 23 Jul 2017 15:49:06 UTC
All use subject to http://about.jstor.org/terms
590 Antitrust Law Journal [Vol. 73

payment card services has been recognized by th


tive.33 An open-loop payment system takes advan
of a broad-based system with the innovation and v
consumers that results from having many issuers

While the structure of the major card association


tion among their members on most of the service
ers and merchants, the associations themselves m
aspects of the payment card system for an open-lo
to function efficiently. These key central functi
system for the authorization, clearing, and settle
between members and establishing payment card
interactions between members, such as the allocation of certain costs
and risks (e.g., fraud and non-payment). The associations also perform
brand-level promotion and advertising, invest in designing new variants
of payment cards, and make other system-wide investments.34 The large
economies of scale in providing these functions make it efficient for
the associations to perform these functions rather than the individual
members. Moreover, determining system-level default interchange fees
is also a central function of an open-loop payment card system because,
as we show in the next section, it is necessary if the system is to func-
tion effectively.

B. Default Interchange Fees Are Necessary for the Effective


Operation of an Open- Loop Payment Card System

The underlying competitive economic forces that determine pr


in closed-loop payment card systems also apply to open-loop p

33 See, e.g., United States v. Visa U.S.A., Inc., 163 F. Supp. 2d 322, 332 (S.D.N
(noting "The members of MasterCard and Visa work together through each of th
tions to achieve benefits for themselves they could not provide independently, i
globally recognized brands and sophisticated computer networks for processing t
tions.") According to data from Visa U.S.A., as of November 2004, there w
"licensed" Visa issuers, i.e., issuers that control the terms and pricing of card
approval of applications, and 6,770 financial institutions issued Visa cards as "spo
issuers, in which case another party (often a larger bank) determines card fea
addition, there were 2,856 financial institutions that acquire Visa transactions, a
to November 2004 data from Visa U.S.A. Due to economies of scale, many ban
outsource many elements of their acquiring business to a small number of thi
processors, such as First Data Corporation and Total System Services. Evans &
lensee, supra note 19, at 17-19.
34 To pay for these key central functions, the Visa and MasterCard association
various system fees on their members. These member fees are not designed to m
the association's profits, but only to cover its operating costs and investments. After
these costs, any remaining surplus is normally used to maintain a capital fun
merchants in the event of a member bank failure and to pay for other unan
expenses. U.S. v. Visa U.S.A., 163 F. Supp. 2d at 332.

This content downloaded from 14.139.214.178 on Sun, 23 Jul 2017 15:49:06 UTC
All use subject to http://about.jstor.org/terms
2006] Competition in Two-Sided Markets 591

card systems, such as Visa and MasterCard. All paym


must optimally balance the two sides of the market
cardholder and merchant prices. The difference bet
and open-loop payment card systems is that a closed-lo
system can balance market forces on the two sides o
directly setting the merchant discount as well as car
benefits and the level of promotion. In an open-loo
system, on the other hand, issuers and acquirers, not t
system, set cardholder and merchant prices. Therefore
be accomplished through interchange fees. By specifyin
payments that flow from acquiring banks to issuing ban
tion, the payment system can indirectly influence mer
and cardholder fees, benefits, and promotion.
For example, when an open-loop payment card syst
change fees, the costs to acquiring banks increase, l
merchant discounts that make the card less attractive to merchants. On
the other side of the market, the higher interchange fees make card
transactions correspondingly more attractive to issuing banks, which now
have a greater incentive to issue the payment system's cards and compete
for cardholders and transaction volume by increasing card promotion,
decreasing cardholder costs (such as annual fees, per transaction fees,
and finance charges), and/or increasing credit limits and cardholder
rewards, all of which make the card more attractive to cardholders.

It has been argued that system- determined default interchange fees


amount to a collective price fixing agreement that is not necessary for
an open-loop payment card system to exist or to operate efficiently.35
One proposed alternative is to eliminate system- determined default inter-
change fees and substitute bilaterally negotiated interchange fees
between each issuer and each acquirer.36 Proponents of this alternative
assert that the extra transaction costs involved in individual negotiations
would be much less than in the past due to the high concentration of

35 Supra note 6. Such claims were raised and rejected in National Bancard Corp. (NaBanco)
v. Visa U.S.A., Inc., 596 F. Supp. 1231, 1241 (S.D. Fla. 1984), affd, 779 F.2d 592 (11th
Cir. 1986).
36 This alternative initially was proposed by the plaintiff in Nabanco, 596 F. Supp. at 1241,
1261, and rejected by the court due to high transaction costs and potential opportunistic
behavior by issuer banks that would try to hold up merchants by charging higher fees.
Bilateral agreements between issuers and acquirers to replace system- determined default
interchange fees have been advocated again more recently. See, e.g., David Bal to, The
Problem of Interchange Fees: Costs Without Benefits?, 4 E.C.L.R. 215, 219-20 (2000). However,
even economists who have criticized interchange fees have described forcing all U.S.
financial institutions to enter into a multitude of bilateral interchange fee agreements as
"not sensible." Dennis W. Carl ton & Alan S. Frankel, The Antitrust Economics of Credit Card
Networks: Reply to Evans and Schmalensee Comment, 63 Antitrust LJ. 903, 914 (1995).

This content downloaded from 14.139.214.178 on Sun, 23 Jul 2017 15:49:06 UTC
All use subject to http://about.jstor.org/terms
592 Antitrust Law Journal [Vol.73

payment card issuing and acquiring today.37 H


interchange fees, individual issuers would have
bilateral interchange fees at much greater than
imizing levels. This is because payment card sys
rules that require merchants that agree to accep
to accept all such cards that are presented by

An honor-all- cards rule is the essence of a pay


it assures each cardholder that his card will be
that display the mark of the card's payment ca
could decide which issuer's cards of a particula
would accept, the benefits of guaranteed acc
mined and the value of the payment card syst
Individual merchants cannot independently
accept a particular issuer's version of the paym
tially damaging the reputation of the entire pa
quently, all payment card systems have an hon
not leave it up to the discretion of individu
refuse payment cards of individual issuers once
to accept the payment system's cards.

Given that the Visa and MasterCard payment


all-cards rules, each issuer negotiating a bilate
a default interchange fee would be able to
demanding an arbitrarily high interchange fe
have artificial bargaining power in bilateral ne
because merchants accepting Visa or Master
ability to reject the individual issuer's cards. Th
especially small issuers, would have the ince
high interchange fees for Visa or MasterCard t
cardholders, knowing that merchants would not
these transactions without deciding to not acc
cards. To the extent that such opportunistic is

37 See, e.g., Bal to, supra note 36.


38 Even the merchant plaintiffs in the Wal-Mart litigatio
benefits to the honor-all- cards rule and did not challeng
MasterMoney Antitrust Litig., No. 96-CV-5238 (JG), 20
2003). Specifically, the Wal-Mart plaintiffs accepted the
of the honor-all- cards rule, but challenged the claim tha
cards to debit cards was necessary to achieve these bene
also has recognized the value of the honor-all- cards rul
honor-all- cards rule in 2001 and concluding that abolish
competition substantially. Commission Decision 2001/78
(L 293) 24, 37-38.

This content downloaded from 14.139.214.178 on Sun, 23 Jul 2017 15:49:06 UTC
All use subject to http://about.jstor.org/terms
2006] Competition in Two- Sided Markets 593

widespread among issuers, merchants would drop out of


or attempt to steer customers to other forms of pay
individual issuers would make their interchange fee d
considering such externalities that their actions woul
payment card system as a whole. As a result, an open-lo
system could not function effectively without default in

The alternative of freely negotiated bilateral interchan


absence of default interchange fees, therefore, really
elimination of the honor-all- cards rule. Only if mercha
strained to accept every issuer's card will such bilateral
lead to severe holdup problems. However, the eliminatio
acceptance would fundamentally undermine open-loo
systems. System- determination of default interchange f
a critical procompetitive function of open-loop paymen

Under current Visa and MasterCard rules, individual issuers and


acquirers may agree on a separately determined bilateral interchange
fee. In fact, individual issuers do negotiate side payments in return for
a merchant's decision to promote a co-branded card. However, these
individually determined arrangements are negotiated within the frame-
work of a system- determined default interchange fee and thus can be
negotiated without the coercion of a one-sided holdup threat.

Besides eliminating system- determined default interchange fees in


favor of bilaterally negotiated interchange fees, another proposed alter-
native to system- determined default interchange fees is a "par collection
system," that is, a payment system with a mandated zero interchange

39 The role of default interchange fees in preventing holdup problems in open-loop


payment card systems was first recognized by Baxter, supra note 11, at 576-77, and has
been restated in the economics literature by, e.g., Howard H. Chang & David Evans, The
Competitive Effects of the Collective Setting of Interchange Fees by Payment Card Systems, 45 Anti-
trust Bull. 641 (2000); and John Small & Julian Wright, The Bilateral Negotiation of
Interchange Fees in Payment Schemes (Jan. 2002) (unpublished manuscript on file
with authors).
w Accordingly, courts that have looked at this question have concluded that interchange
fees should not be considered per se illegal. For example, the court in NaBanco concluded
that the interchange fee "on balance is procompetitive because it was necessary to achieve
stability and thus ensure the one element vital to the survival of the VISA system -
universality of acceptance." NaBanco, 779 F.2d at 605. The court further stated that "VISA
is a joint venture-type enterprise in which the [interchange fee] acts as an internal control
mechanism that yields procompetitive efficiencies that its members could not create acting
alone, and helps create a product that its members could not produce singly." Id. Therefore,
based on Broadcast Music, Inc. v. CBS, 441 U.S. 1 (1979), interchange fees were analyzed
under the rule of reason.

This content downloaded from 14.139.214.178 on Sun, 23 Jul 2017 15:49:06 UTC
All use subject to http://about.jstor.org/terms
594 Antitrust Law Journal [Vol. 73

fee.41 However, this alternative would not elimina


centrally determined default interchange fees but
such interchange fees at zero. Moreover, such regu
an open-loop payment card system's ability to bala
the payment card market, thereby hindering the p
ability to compete effectively against closed-loop p
such as American Express and Discover, that ba
the market directly. Additionally, as we now will show
basis for regulating interchange fees because int
an indication or a consequence of payment card sy

IV. THE LEVEL OF INTERCHANGE FEES DOES NOT REFLECT


THE EXERCISE OF MARKET POWER

A. Interchange Fees Are Independent of the Degree of


Payment System Market Power

Some economists have argued that interchange fees amoun


collective setting of prices above the competitive level which re
supracompetitive merchant discounts.42 However, the economic
of pricing in two-sided markets indicates that interchange fees
a measure of payment card system market power. Market powe
in principle affect the total price collected by the payment syste
cardholders and merchants. However, the role of interchange f
influence relative merchant and cardholders prices and not t
price collected by the payment system.
To see this, we initially consider a closed-loop proprietary p
card system that directly sets prices to cardholders (Pc) and me
(PM). The total price collected by the payment system (PM +
depend on the market power of the closed-loop proprietary
system, as well as short-run demand and supply conditions in the
system market. All else equal, as the payment card market beco
competitive, the total price paid by merchants and cardhold
increase. This effect of competition on the total price collected
payment system is analogous to pricing in one-sided markets.43

41 This alternative initially was proposed by the plaintiff in Nabanco, 596


1241. A par collection system has been advocated again more recently. See,
supra note 36, at 222-23. However, not all proposed alternatives involving the r
of default interchange fees feature a zero interchange fee. For instance, M
proposed that interchange fees be regulated based on certain costs. Katz, sup
at 33-35.

42 Frankel, supra note 6, at 341-46; Carlton & Frankel, supra note 6, at 633-35, 637-38.
43 As the product a payment system sells becomes more differentiated, the total price to
the cardholder and merchant may become greater than the marginal cost of a transaction.

This content downloaded from 14.139.214.178 on Sun, 23 Jul 2017 15:49:06 UTC
All use subject to http://about.jstor.org/terms
2006] Competition in Two-Sided Markets 595

However, the relative prices of the two sides of the m


are independent of the total price collected by the s
for example, a hypothetical payment card system with a
share that competes against numerous rivals which con
to collect a total price that equals total costs. Despite bei
constrained to a price that only covers its costs, such a
would likely find it optimal to balance the two sides of
collecting a larger share of its total costs from the mer
market because of the differential elasticity considerat
Part II.B. In fact, balancing by a payment system in a h
market may even lead to a negative price on the cardh
market and a price greater than total costs on the mer
closed-loop proprietary payment system is not only ab
two sides of the market in this way despite facing signif
but competition is likely to compel the payment syste
this manner. Moreover, there is no reason to expect
system in a very competitive market will balance the t
market any differently than a payment system with co
power. The fact that a payment system collects almost
supplying the payment system from the merchant sid
any information about the total price that the syste
relative prices to merchants and cardholders cannot
market power.

However, this does not mean that the payment system necessarily p
market power. All that is implied from the fact that a firm can p
marginal cost is that the firm is facing a less than perfectly elastic d
a condition generally present even for firms that operate in highly c
fact, almost every firm in the economy (except the wheat farmers in
produces goods for which there are not perfect substitutes and, the
firm faces a negatively sloped demand. In measuring antitrust m
circumstances, one must distinguish between the firm's own-price
and the firm's ability to influence market prices. Almost every firm
some power to control its own prices, i.e., its demand will not go to
a small amount. However, this does not mean that all such firms h
market prices, i.e., possess any antitrust market power in the sense o
market output and thereby significantly raise market prices above t
See Benjamin Klein, Market Power in Antitrust: Economic Analysis A
Econ. Rev. 43, 63-71 (1993) . Recent antitrust case law is fully consiste
of antitrust market power as the ability to significantly influence mark
than the ability of a firm to control its own prices and to price abo
Benjamin Klein & John Shepard Wiley Jr., Competitive Price Discrim
Justification for Intellectual Property Refusals to Deal, 70 Antitrust L.J
44 Competition from other rival systems would not prevent the p
charging a price on the merchant side that is greater than the total
payment system (and a negative price on the cardholder side) bec
system that offers a lower merchant discount will also have high
and a smaller cardholder base. Therefore, the rival system will p
to merchants.

This content downloaded from 14.139.214.178 on Sun, 23 Jul 2017 15:49:06 UTC
All use subject to http://about.jstor.org/terms
596 Antitrust Law Journal [Vol. 73

These economic principles are the same for a


card system in which interchange fees are used t
relative cardholder and merchant prices. When
assessed on acquirers are fully passed through by
in the form of lowered fees, cash-back, rewards i
tions, interchange fees determine relative prices
total prices. Interchange fees influence both mer
prices. As interchange fees increase, one price (th
increases, but the other price (the cardholder
same amount, leaving the total price collected by
unchanged. In a payment system in which inte
passed through to cardholders, interchange fees
price collected by the payment system because the
from merchants to cardholders. A higher interch
cate that the payment system is collecting a high
indicates that the payment system has decided
payment system is enhanced by collecting a high
price from merchants rather than cardholders. I
interchange fees is to balance the two sides of
influencing relative cardholder and merchant
increase the total price collected by the payment

It has been argued that interchange fees may


through in an open-loop payment system because
power.45 However, the evidence indicates that ther
tion among issuers for cardholders. Issuers com
cardholder fees, offering cash-back and rewards
airline tickets) to cardholders, and promoting th
grams. Higher interchange fees give issuers th
for cardholders by offering greater rewards or lo
fact, since interchange fees have increased, issuer
fees and greater benefits in the form of cash-back
card enhancements to cardholders.46

As described in the payment system model presented in Part II.B, the


profit-maximizing relative prices charged to merchants and cardholders
in all payment systems will depend on the relative demand elasticities of

45 See, e.g., Michael L. Katz, What We Know About Interchange Fees and What Does It
Mean for Public Policy? 14 ( Tune 2005) (unpublished manuscript on file with authors).
46 See infra note 82. Even if system- determined interchange fees are not fully passed
through to cardholders, it does not mean that these interchange fees are supracompetitive
in the sense that they restrict market output. In fact, even when not fully passed through,
system- determined interchange fees are likely to increase, not restrict, payment system
output. See discussion p. 610.

This content downloaded from 14.139.214.178 on Sun, 23 Jul 2017 15:49:06 UTC
All use subject to http://about.jstor.org/terms
2006] Competition in Two- Sided Markets 597

merchants and cardholders for a payment card system.


are invariant to the absolute elasticities of demand on the cardholder
and merchant sides. For instance, even if demand on both the merchant
and cardholder sides becomes very elastic, relative prices will not chang
as long as the relative elasticities on the two sides of the market stay th
same.47 Because the role of interchange fees is to indirectly determine
relative prices and not the total price collected by the payment system
interchange fees also will depend on the relative demand elasticitie
of cardholders and merchants and not on the absolute elasticities of
demand.48 Accordingly, only if there is a change in the relative elastici
on the two sides of the market will there be a change in relative pric
to cardholders and merchants or in the interchange fees of an open-lo
payment system. However, there is no reason to expect that increase

47 The invariance of relative prices to the absolute elasticities of demand facing a paymen
system shows that two-sided market balancing of demand is distinct from "price discrimin
tion" (i.e., differential pricing relative to marginal cost across customers). In the stand
economic model of price discrimination, as demand becomes more elastic, prices in b
segments of the market decrease towards marginal costs. Thus, a firm's ability to p
discriminate is diminished as demand from each of the two customer segments becom
more elastic. In balancing the two sides of a payment system, on the other hand,
elasticities of demand on the two sides of the market increase, optimal relative pri
remain unchanged as long as the relative elasticities on the two sides of the market rem
the same. Thus, payment system balancing is invariant to the absolute elasticities of dem
on the two sides of the payment system market. Neither balancing in two-sided mark
nor price discrimination, however, imply the existence of market power. Because m
firms in real-world markets sell a somewhat differentiated product, differential pric
relative to marginal cost across customers can be expected to be a normal and perva
part of the competitive process, even in markets that include highly competitive fi
possessing very small market shares. Antitrust law has begun to recognize the ubiquit
nature of such price discrimination, including the likelihood of its existence in competi
markets. See, e.g., In re Brand Name Prescription Drugs Antitrust Litig., 186 F.3d
786-87 (7th Cir. 1999); see also Illinois Tool Works Inc. v. Independent Ink, Inc., 12
Ct. 1281, 1296 (2006).
48 As equation (11) in Part II. B shows, profit-maximizing relative prices also do not
depend on per transaction payment system costs ( Q . Nor are relative prices affected by
the division of these costs into merchant- and cardholder-side costs because such costs
are incurred by the payment system for providing a transaction that is jointly consumed
by a cardholder and a merchant. See supra note 12. However, interchange fees are influ-
enced by the division of per transaction costs between issuers and acquirers, which is
determined by the payment card system based on efficiency considerations. For instance,
the payment card association assigns liability to issuers for many types of fraud and non-
payment because issuers, who sign up cardholders, have a greater ability than acquirers
to control the costs associated with such liabilities. The association also allocates between
issuers and acquirers the costs of promoting the payment system and operating the payment
network. These allocations of per transaction costs between issuers and acquirers influence
optimal interchange fees because they affect prices to cardholders and merchants. In
particular, merchant discounts are likely to equal the interchange fee plus the cost of
acquiring. Similarly, cardholder prices are likely to equal the cost of issuance minus the
interchange fee. Thus, the payment system will determine the interchange fees to achieve
the optimal relative prices to merchants and cardholders by taking into account issuer
and acquirer costs.

This content downloaded from 14.139.214.178 on Sun, 23 Jul 2017 15:49:06 UTC
All use subject to http://about.jstor.org/terms
598 Antitrust Law Journal [Vol. 73

payment system competition will have a greater ef


of demand on the merchant side than on the cardholder side of the
market and therefore have the effect of decreasing interchange fees. I
theory, increased payment system competition could lead to higher or
lower interchange fees depending on how the relative elasticities i
the two sides of the market change. As we discuss in Part IV.C below,
competition between payment systems for issuance recently has led to
increased interchange fees. Consequently, a reduction in payment card
system competition, say if Visa and MasterCard merged, may in fa
lead to lower, not higher, interchange fees. This clearly indicates that
interchange fees cannot be a measure of payment system market power

B. Differences in Merchant Discounts Across Systems and


Over Time Reflect Differences in Competitive Balancing,
Not Market Power

The economic theory of two-sided markets indicates that relat


on the two sides of the market are independent of the degree o
tion faced by a supplier in such a market. While total pric
influenced by competition, relative prices are determined b
balancing of demand on the two sides of the market. The
evidence with regard to payment system pricing is consistent
economic theory. In particular, changes in Visa's and MasterCa
change fees over time reflect changes in competitive balancin
two sides of the payment card market, and not changes in th
power of the Visa and MasterCard associations.
To illustrate, consider Figure 1, which presents Visa's average
interchange fees over time. After Bank of America moved in
open-loop payment card structure (first known as NBI and ul
Visa), its initial interchange fee was 1.95 percent. Visa's interc
remained constant at 1.95 percent until 1977, and then
decrease. Visa's interchange fees were generally falling throug
1980s, then remained relatively constant from 1991 to 19
increasing from 1998 to the present.49

49 Visa average effective interchange fees for 1971-1977: Visa U.S.A.; 1982, 1
Schmalensee (2005), supra note 19, at 154-56; 1991-2005: Visa U.S.A. We
consistent estimates of Visa's average effective interchange fees between 197
After 1977, Visa moved from a single interchange rate of 1.95% to addin
component to its standard interchange fee and began to apply other interch
qualifying transactions, such as the "electronic" and "terminal" incentive inte
We are not able to estimate the average effective interchange fees based on t
data. Hence, except for estimates for 1982 and 1983 from Evans & Schma
the lines drawn between the intervening years of our early observations do

This content downloaded from 14.139.214.178 on Sun, 23 Jul 2017 15:49:06 UTC
All use subject to http://about.jstor.org/terms
2006] Competition in Two-Sided Markets 599

2.2%

2.0% a

1.2%

J .0%

0.8%

0.6%

0.4%

0.2%

It
th
en
M
t
ac
t
te
b
a
p
w
w

ac
ar
in
50
nomic Analysis, National Income and Product Accounts, All NIPA Tables: Table
2.3.5 Personal Consumption Expenditures by Major Type of Product (A) (Q), available at http://
www.bea.doc.gov/bea/dn/nipaweb/SelectTable.asp?Select=N. Visa and MasterCard pur-
chases for 1971: The Nilson Rep. 38 (Feb. 1972), at 3; 2003: The Nilson Rep. 805 (Feb.
2004), at 6.

This content downloaded from 14.139.214.178 on Sun, 23 Jul 2017 15:49:06 UTC
All use subject to http://about.jstor.org/terms
600 Antitrust Law Journal [Vol. 73

50% j

40% "~" """" ^ _^'^__

35%

*~ *. a. / '^ American Express


V ""V
15%

10%

Discover

5% /*"""*" * *^«~-~«
""""" N*N-- -- -""""'^ >«

0% «

r*^ t^^ t^*- oo oo oo oo oo o^ ov Çf^ o^ o^ ^5 <*^ ^> ¿*s


^jv ¿^ ^' ^' ¿^ Ç' <^' c^ &' $' ^' ^^ o^ ^5 <^^ ^5 ^5

Fiffure 2. Credit and Charge Card Shares

were falling.51 The fact that interchange fees were falling from 1977 to
1991 while Visa's and MasterCard's importance in the economy was
growing, and that both Visa's and MasterCard's interchange fees remain
lower now than they were in 1971, seems inconsistent with the level of
interchange fees being an appropriate measure of market power.
Furthermore, Visa's and MasterCard's interchange fees always have
been at levels such that the merchant discounts on Visa and MasterCard
transactions were below the American Express merchant discount.52 This
is because Visa and MasterCard historically desired greater merchant

51 The Nilson Rep., various issues, except American Express 1974-1983 and Diners
Club 1975-1983, 1985: Visa U.S.A. Shares are measured by gross dollar volume. Diners
Club includes Carte Blanche through 1994. MasterCard includes Diners Club in 2005.
52 Acquirer costs must be added to interchange fees to calculate the acquirer-determined
merchant discounts for Visa and MasterCard transactions. These added costs have
decreased over time, from approximately 0.7% in 1983 to 0.4% in 2003. 1983: Ev
Schmalensee, supra note 19, at 156. 2003: Visa U.S. A. The resulting average mer
discount for Visa was 2.3% in 1983 and 2.1% in 2003. In contrast, American E
average merchant discount was 3.5% in 1983 and 2.6% in 2003. Visa merchant discou
1983: Visa U.S.A.; 2003: Estimated from 2003 average effective interchange fee. Am
Express merchant discount for 1983: Visa U.S.A.; 2003: American Express Company
Report 2005.

This content downloaded from 14.139.214.178 on Sun, 23 Jul 2017 15:49:06 UTC
All use subject to http://about.jstor.org/terms
2006] Competition in Two-Sided Markets 601

acceptance than American Express. American Expre


chant discounts because it adopted a marketing strate
aimed to deliver a smaller, more valuable (higher
customers to a smaller group of merchants. Diners
of credit and charge card transaction volume has been
since the early 1970s and was less than 1 percent in 2
absorbed into MasterCard by Citigroup, adopted a sim
always charged higher merchant discounts than the
merchants on Visa and MasterCard transactions.53 Dis
major proprietary payment card system, in contrast, s
stantial cardholder base from its association with Sears
itself differently in the marketplace with lower merc
Declining Visa and MasterCard interchange fees duri
and 1980s can be explained in large part by the increa
to Visa and MasterCard of greater merchant accept
although Visa always had aimed for higher merchan
American Express, Visa began to expand its merchant
mid-1980s - at first slowly, and then more rapidly in
part of this expansion effort, Visa initiated its well-k
campaign, "It's everywhere you want to be," in 1985.55
"universal acceptance" of Visa cards in all sectors of th
a critical aspect of Visa's brand name and marketing f
Card competed in a similar way.56 The importance
tance, which has continued to the present, has placed d
on interchange fees.57

53 Early 1970s: The Nilson Rep. No. 119, July 1975, at 3. 2004:
828, Feb. 2005, at 8. When Citigroup added the MasterCard sym
cards, it meant that all merchants that accepted MasterCard als
Club and that Diners Club transactions would be processed as M
and presumably subject to MasterCard interchange fees. Daniel Rom
Pact Props up Diners; MasterCard Tie Might Mean Big Boost for a Faded
Chicago Bus., June 20, 2005, at 16.
54 As part of its drive to expand merchant acceptance, Visa int
supermarket interchange fees in the early 1990s. MasterCard follow
Denise Gray, The New Super Market for Cards, Credit Card Mgmt.
55 Evans & Schmalensee, supra note 19, at 193.
56 It is the universal acceptance feature of the Visa and MasterCa
permits a significant number of Visa and MasterCard cardholder
payment card. See supra note 19.
57 When optimally balancing the two sides of the payment card m
Card must take into account the importance of broad merchant acc
the value of their payment card systems. Once enough key mer
of Visa or MasterCard, the consumers who wish to carry only one p
towards the one that retains the greater level of merchant acceptan
also will make it easier for some of the remaining merchants to
card as well. Moreover, merchants not only may decide to refu

This content downloaded from 14.139.214.178 on Sun, 23 Jul 2017 15:49:06 UTC
All use subject to http://about.jstor.org/terms
602 Antitrust Law Journal [Vol. 73

1.4% | y' °-70

S 1.2% | s***^^ ' / V>^/ 0.60

g 1.0% ! / Xv / "^X *** 0.50 1

5
I I ; v vx' !
0.8% I >. 0.40 ^

E | >.
~ 0.6% | >, 0.30

I _- _ (AmEx -Visa) Merchant Discounts >^ q20


' ° !

^ '
0.0% i

S£3SooMoooeq^SSSosSosgs^gs§oo8So
CTV ^p- »^ CT^ ^^ C7^ ^P- ^^ ^P- CT4 C7^ w^ CT^ Cy> w^ CT^ Vs w1^ ^7* C5 C5 C3 C? f^j

Figure 3. Comparison of Ame


Discounts and Merchant

American Express also change


cardholder sides over time in r
landscape. Figure 2 shows that af
acceptance in 1985, American
that this decline accelerated afte
Express attempted to stop its sh
between its merchant discount and the merchant discounts on Visa and
MasterCard transactions. This change in American Express policy is
illustrated in Figure 3, which presents time series of the difference
between the American Express and Visa average merchant discounts and
of the relative merchant acceptance of the two payment systems.59

card, but also may steer consumers to use alternative payment cards and payment methods.
The commitment Visa and MasterCard each has to universal acceptance as well each
payment system's concern about merchants steering consumers to use alternative payment
methods will influence each payment system's market-balancing decision.
58 The decrease in American Express's share of credit and charge transaction volume
resulted primarily from Visa and MasterCard increasing their transaction volume at the
expense of alternative payment methods, such as cash and checks.
59 Visa merchant discounts for 1982-1999: Visa U.S.A.; 2001: David S. Evans, Bank Inter-
change Fees Balance Dual Demand, Am. Banker, Jan. 26, 2001, at 17; 2002-2005: Estimated
from 2002-2005 average effective interchange fees. American Express merchant discounts
for 1982-1993: Visa U.S.A.; 1994-2005: American Express Company Annual Reports,
various years. Visa and American Express merchant outlets: The Nilson Rep., various issues.

This content downloaded from 14.139.214.178 on Sun, 23 Jul 2017 15:49:06 UTC
All use subject to http://about.jstor.org/terms
2006] Competition in Two- Sided Markets 603

Figure 3 indicates that, although Visa has always h


tance than American Express, the fraction of Visa
accepted American Express started to increase in respon
ing of the gap between the American Express and V
counts.60 Between 1990 and 2005, as the gap betwe
Express and Visa merchant discounts narrowed fro
0.37 percent, the fraction of outlets that accepted Am
addition to Visa increased from 44 percent in 1990
2005. 61 This increase in the number of Visa merchant outlets that also
accepted American Express was accomplished in spite of the fact that
Visa merchant acceptance was growing rapidly during the period. For
instance, between 1990 and 2005, the number of U.S. merchant outlets
that accepted Visa more than doubled, from 2.7 million merchant outlets
to 6.1 million merchant outlets.62 During that same period, however,
merchant acceptance of American Express more than tripled, from 1.2
million merchant outlets to 3.9 million merchant outlets.63 Figure 2
shows that, as a result of the narrowing gap between the American
Express and Visa merchant discounts and the increase in the American
Express merchant base, American Express stabilized its share of overall
credit and charge card transaction volume by 1996 and recently began
to increase its share. These changes in American Express' merchant
discount over time, and the changes in Visa's and MasterCard's inter-
change fees described above, are consistent with changes in competitive
balancing of the two sides of the payment card market, and not changes
in the market power of the payment card system.

C. Interchange Fees Increased in the Late 1990s as Payment


Card System Competition for Issuers Increased

Figure 1 shows that the Visa average effective interchange fee


to increase in 1998, reversing a general long-term trend of first

60 This increase in merchant acceptance of American Express relative to Visa w


during 1993-1995. It was during this period that Visa was engaged in an
advertising and marketing campaign for its signature debit card. Evans & Sch
supra note 19, at 206-07. As a consequence of its effective brand extension, V
a more valuable card for merchants to accept, resulting in a substantial expansi
number of Visa outlets in the mid-1990s.

61 This statement assumes that all outlets that accept American Express also accept Visa.
Although this is a useful way to summarize the ratio of the number of American Express
outlets relative to the number of Visa outlets, a very few outlets (most notably Costco)
accept American Express but not Visa.
62 1990: The Nilson Rep. No. 500, May 1991, at 7; 2005: The Nilson Rep. No. 851,
Feb. 2006, at 10-11.
63 1990: The Nilson Rep. No. 520, Mar. 1992, at 5; 2005: The Nilson Rep. No. 851,
Feb. 2006, at 10-11.

This content downloaded from 14.139.214.178 on Sun, 23 Jul 2017 15:49:06 UTC
All use subject to http://about.jstor.org/terms
604 Antitrust Law Journal [Vol. 73

and then relatively constant interchange fees.


change fees can be explained by the increased com
and MasterCard for issuers that occurred during

Payment card system competition for issuers inte


developed increased cardholder loyalty that per
fully determine the choice of payment card system
Issuers created such cardholder loyalty by deve
their own brand names independent of the Visa a
names. For example, Capital One and Citigroup
independent brand names in large part through i
expenditures.64 Issuers established independent re
ing valuable rewards, good service, and reasona
issuers also developed loyal cardholders through
programs that cardholders value, such as co-bran
programs that target cardholders with loyalty to
organizations and offer card-specific rewards, such
chant-specific discounts, and donations to char
affinity programs were first developed in the early 1
cards that featured airlines, such as Citigroup
Aadvantage program, and subsequently expanded
ety of companies and organizations.65 By 2003, co
cards accounted for 40 percent of transactions on
credit cards.66

When cardholders have greater loyalty to a particular issuer, such as


Citigroup, than to a particular payment system, such as Visa or Master-
Card, issuers can switch their cardholders among payment systems with-
out losing many cardholders. Loyal cardholders are unlikely to change
issuers if the issuer does not offer the cardholder's preferred payment
card system. In these circumstances, a payment card system cannot have
an interchange fee that is significantly below the interchange fees of
competing payment card systems. If it did, issuers would offer their
cardholders to a competing payment card system with a higher inter-
change fee. Increased cardholder loyalty to individual issuers, therefore,
makes issuers more sensitive to interchange fee differences, and this

64 For example, Capital One advertising grew nearly 550% between 1996 and 2004.
Capital One Annual Reports, 1998-2004. Capital One and Citigroup ads promote the
Capital One and Citi brand names without mentioning Visa or MasterCard.
65 For example, J.P. Morgan Chase, one of the largest issuers of Visa payment cards,
was offering 57 different co-branded cards and 122 different affinity cards on its website
in December 2005. See http://chase.com.
66 Visa U.S.A.

This content downloaded from 14.139.214.178 on Sun, 23 Jul 2017 15:49:06 UTC
All use subject to http://about.jstor.org/terms
2006] Competition in Two- Sided Markets 605

increase in the elasticity of demand faced by each paym


will lead to higher interchange fees.67

Economically, it is useful to think of an issuer as acting


ing agent for its cardholders as a group and restricti
individual cardholders to exercise their specific paym
preferences, if any. The issuer is thereby internalizing
independent buying decision and increasing the elast
that each payment card system faces in indirectly biddin
through issuers. Each payment card system now knows th
a large fraction of an issuer's cardholders if it offers th
superior interchange fees and/or other terms compared
system rivals. Visa and MasterCard are engaging in "ex an
for cardholders as a group by trying to convince indivi
choose their particular payment system for a group of
holders, rather than the "ex post competition" each paym
would face if it directly competed for cardholders, for ex
to convince individual customers to check off the paym
desired when applying for an issuer's credit card. Ex an
substantially increases the elasticity of demand each paym
faces compared to ex post competition. Thus, the growt
loyalty to individual issuers and the increased competiti
ment systems for issuance can be expected to lead to hi
fees in an attempt to attract issuance.68

Cardholder loyalty to an issuer increases the issuer's ab


more effectively with a payment system because the is

67 See, e.g., William Sheedy, Credit and Commercial Interchange Reimbursem


Visa Bus. Rev.-Gen. (Visa U.S.A., Foster City, CA) (Feb. 2004) at 1,
competition for issuance, among other factors, as one of the reas
interchange fees in 2004. Increased competition with American Exp
a consequence of the United States v. Visa decision (which allow
payment systems to compete for agreements with banks to issue the
upward pressure on interchange fees. United States v. Visa U.S.A.,
322, 408-09 (S.D.N.Y. 2001). While competing for issuance, paymen
avoid raising interchange fees substantially above those of its compet
possibility of merchant steering and the importance of "universal me
See supra note 57.
68 This is similar to the use of a drug formulary by an insurance com
benefit manager, where pharmaceutical manufacturers of alternative d
category offer better prices in return for the promise of being chos
drug in an insurer's formulary. The potential for share shifting with
"elasticizes" the demand facing each drug manufacturer. See Kennet
E. Mills, The Distribution and Pricing of Prescription Drugs, 4 Int'lJ. E
and the more general application of this analysis to retail stocking
Benjamin Klein, Kevin M. Murphy & Joshua D. Wright, Exclusive D
Management in Retail Distribution (2006) (unpublished manuscript o

This content downloaded from 14.139.214.178 on Sun, 23 Jul 2017 15:49:06 UTC
All use subject to http://about.jstor.org/terms
606 Antitrust Law Journal [Vol.73

to deliver its cardholders to the featured pay


the issuer possesses any monopsony power.
too small a share of the total market for this
relevant.69 This situation is similar to the low
hospitals, even a single relatively small hosp
has significant control over its patients' deman
share between pharmaceutical manufacturers
products.70 Likewise, the price advantage gain
between payment systems for a group of th
nothing to do with monopsony power of issue
but is based on the fact that issuers (even relative
cardholders) can commit to shift cardholders

In addition to higher interchange fees, incre


also has enabled issuers to negotiate more f
processing fees in return for a long-term prom
Citigroup appears to have been the first iss
agreement, demanding from both Visa and M
financial terms (that included discounted m
fees) in return for a promise of dedicated issu
the competitive bidding in February 199973
"Member Business Agreement" dedicating 85 p
to MasterCard.74 Citigroup's credit card portf

69 Specifically, it is not that issuers are so large that


payment system's product raises the market price on all
this, have an incentive to reduce their demand for the
exercise of monopsony power.
70 See In re Brand Name Prescription Drugs Antitrust L
Cir. 1999).
71 These fees which are paid by issuers to the payment
than one-tenth the level of interchange fees received
membership and processing fees, see Visa U.S.A. Inc.
General Rules, Nov. 15, 2004, ch. 9; and Visa U.S.A. Inc
tion 3.02.

72 U.S. v. Visa U.S.A., 163 F. Supp. 2d at 369. The court believed that the associations
had legitimate, competitive business rationales for dedicated contracts, the procompetitive
nature of which we outline below. Although the court granted member bank issuers a
two-year period in which to rescind their current dedication agreements, this was not
because it found such agreements anticompetitive ("the [dedication] agreements them-
selves are not inherently anticompetitive"). Rather, the court gave issuers this period
because it concluded the associations' rules historically had prevented American Express
and Discover from competing for the contracts and wished to create an opportunity for
such competition. Id. at 408-09.
73 Id. at 369-71; Paul Beckett, Credit Cards: Spurning Visa, Citigroup Turns to MasterCard,
Wall St. J., Feb. 10, 1999, at Bl.
74 U.S. v. Visa U.S.A., 163 F. Supp. 2d at 370. In addition, Citigroup successfully negotiated
the ability to advertise the Citicard without the MasterCard logo (a contract term that

This content downloaded from 14.139.214.178 on Sun, 23 Jul 2017 15:49:06 UTC
All use subject to http://about.jstor.org/terms
2006] Competition in Two- Sided Markets 607

from 61 percent Visa and 39 percent MasterCard in 199


24.1 million Visa cards and 15.5 million MasterCard card
Visa and 72 percent MasterCard in 2004 (consisting of 2
cards and 75.1 million MasterCard cards).75 To a larg
MasterCard's Member Business Agreement with Citigro
similar agreement that MasterCard entered into with C
Bank in 1999,76 that caused the large increase in Master
credit and charge card volume during 1999-2002 illustrat

In response to losing a significant share of issuance fr


then its second and fourth largest issuers,78 Visa forma
"Issuer Partnership Program" in 1999. Similar to Master
Business Agreements, this program offered discounted
to issuers that agreed to dedicate their portfolios to th
For example, Bank of America made an early commitm
Partnership Program. As a result, Bank of America, wh
a predominantly Visa issuer in 1997 with 66 percent Vi
MasterCard (consisting of 9.5 million Visa cards and 5
Cards), by 2004 had moved to 95 percent Visa and o
MasterCard (consisting of 58.1 million Visa cards and 3.
Cards).80

Issuers have obtained favorable financial terms from the payment


systems by offering these dedicated (near exclusive) issuance agreements

now has been generally instituted by most Visa and MasterCard issuers) as well as the
right to place the MasterCard logo on the back of the Citicard (which Citi and other
issuers have not done). See supra note 73.
75 1997: The Nilson Rep. No. 660, Jan. 1998, at 8-9; 2004: The Nilson Rep. No. 826,
Tan. 2005, at 8-9.
76 U.S. v. Visa U.S.A., 163 F. Supp. 2d at 369-71; Paul Beckett, More Big Banks Choose
Sides in Card Wars, Wall St. J., Jan. 26, 2000, at Bl.
77 In addition to the agreements with Citigroup and Chase, MasterCard entered into
Member Business Agreements with other members, including board members Household,
Metris, and Peoples Bank, who committed to dedicate 80% or more of their portfolios
to MasterCard. U.S. v. Visa U.S.A., 163 F. Supp. 2d at 369-71. MasterCard lost the Chase
account in 2005 when Chase shifted back to Visa after Chase acquired Bank One, an
almost exclusively Visa issuer, in 2004. Lavonne Kuykendall & Isabelle Lindenmayer, JPM
Chase Going with Visa, Mostly, Am. Banker, Feb. 23, 2005. at 1.
78 The Nilson Rep. No. 708, Jan. 2000, at 8-9.
79 U.S. v. Visa U.S.A., 163 F. Supp. 2d at 368-69. See also Evans & Schmalensee, supra
note 19, at 204. The program requires issuers to generate 90% of their total credit card
volume through the Visa system within a certain time period. Id.
80 1997: The Nilson Rep. No. 660, Jan. 1998, at 8-9; 2004: The Nilson Rep. No. 826,
Jan. 2005, at 8-9. Similarly, Fleet decided in 2000 to dedicate card issuance to Visa. W.A.
Lee, Fleet's New Card Chief Wants Deeper Relationships Bankwide, Am. Banker, Dec. 17, 2001 ,
at 1. In fact, by the end of 2000, some 439 Visa members, accounting for more than 60%
of Visa's transaction volume, had committed to Visa under the Issuer Partnership Program.
U.S. v. Visa U.S.A., 163 F. Supp. 2d at 368.

This content downloaded from 14.139.214.178 on Sun, 23 Jul 2017 15:49:06 UTC
All use subject to http://about.jstor.org/terms
608 Antitrust Law Journal [Vol. 73

because such agreements increase the elasticity


payment system faces in bidding for a large f
cardholders.81 The increase in cardholder loyalty
therefore, has had significant procompetitive eff
of the market. Moreover, given inter-issuer comp
the ability of issuers to obtain more favorable ter
systems has created incentives for issuers to attr
greater variety of improved benefits and terms. C
ers have benefited from the ability of issuers to b
with payment systems.82

It has been suggested that, although issuers may


change fees as incentives to issue Visa and Mast
in the early years of the associations, it is no l
interchange fees to create incentives for banks to
It is important to recognize, however, that the rol
fees resulting from the increased inter-payment s
issuers is not to provide incentives for issuers to
generally. Rather, increased competition for issue

81 If some cardholders value a particular payment system bran


the issuer may not find it profit-maximizing to offer a full exc
exclusive, e.g., 80-90%, of new cardholders. This permits c
preference for a particular payment system brand to obtai
that the issuer does not lose these cardholders to competin
form taken by the "dedicated issuance" payment system cont
and 79 and accompanying text. Many "exclusive" retail arra
efficient limited exclusivity form so that customers with a s
obtain their desired brands. See Klein et al., supra note 68.
O£ issuers nave ottered greater oenents to caranoiaers since mtercnange tees nave
increased as a result of the increased payment system competition for issuers. Credit cards
with rewards programs have proliferated, with the percent of total cardholder solicitations
that include cards with rewards increasing from 22% in 2001 to 51% in 2004. Moreover,
the number of rewards cards with annual fees has decreased from 25% in 2001 to 12%
in 2004. Visa U.S.A. More credit cards have featured enhancements, such as auto rental
insurance, travel and emergency assistance, and warranty management service. For exam-
ple, 50% of Visa credit cards offered such enhancements in 2004, compared to 24% in
1995. In addition, the average annual fee on Visa and MasterCard credit card offers has
fallen from $14.48 in 2000 to $8.52 in 2004. William Sheedy, Interchange Reimbursement
Fees Delivering Value and Driving Innovation, Remarks to Federal Reserve of Kansas City
Payment Sys. Conf. 6-7 (May 2005), available at http://www.kc.frb.org/FRFS/PSR/PDF/
SheedyPanelRemarks.pdf. The increase in cardholder benefits has been reported in the
trade press. See, e.g., Dan Moreau, Credit Card Perks Get Sexier to Seduce Customers, Investor's
Bus. Daily, July 14, 2000, at Bl; Kate Fitzgerald, Giving More and Charging Less: Issuers
Cutting Fees and Interest Rates for Rewards Cards, Card Mktg., May 2002/June 2002, at 1;
Jeremy Quittner, Loyalty Programs' Use - and Expectations - Rising, Am. Banker, May 6, 2003,
at 6A; Ron Lieber, Credit-Card Companies Escalate Battle of the Perks, Wall St. J., Sept. 14,
2004, at Dl.
83 See, e.g., Balto, supra note 36, at 219.

This content downloaded from 14.139.214.178 on Sun, 23 Jul 2017 15:49:06 UTC
All use subject to http://about.jstor.org/terms
2006] Competition in Two-Sided Markets 609

for payment card systems to compensate issuers to pro


lar system over rival payment systems.

Further, inter-payment system competition for issue


terized as "distorted" because it leads to higher prices
change fees.84 However, as we discussed in Part IV.A, h
fees do not result in a higher total price being charge
system. Interchange fees influence relative prices. As
increase, one price (the merchant discount) likely
other price (the cardholder price) likely decreases. Th
to associate higher interchange fees with a higher tot
by the payment system. In this case, a higher interch
indicates that the payment card system has chosen to
fraction of the total price from merchants relative to c
ily as a consequence of increased competition for
increased cardholder loyalty to individual issuers.

The fact that the increased competition between Vis


for issuers has led to higher interchange fees since 1
with interchange fees being a measure of payment ca
power. Visa and MasterCard did not achieve any signi
their combined share of credit and charge card trans
1998 that permitted them to increase interchange fees
power, if any, Visa or MasterCard possessed over m
change after 1998. Instead, a significant change in de
the other side of the market, as the intensity of comp
between Visa and MasterCard increased substantial
increased interchange fees.

D. Interchange Fees Do not Restrict Output

The essence of market power is the ability to raise marke


restricting market output. In a one-sided market a higher
lower output. Likewise, a firm with market power can raise
by restricting market output. In contrast, there is no such inver
ship between the level of interchange fees and the output
volume) of a payment card system. The output of a payment
as in a one-sided market, depend on the total price ch
payment system. However, as we discussed in Part IV.A
loop payment system in which issuers fully pass through in
revenues to cardholders, interchange fees do not influe
price of the payment system but, rather, only influence the

84 See, e.g., Frankel, supra note 6, at 345; Balto, supra note 36, at 221.

This content downloaded from 14.139.214.178 on Sun, 23 Jul 2017 15:49:06 UTC
All use subject to http://about.jstor.org/terms
610 Antitrust Law Journal [Vol. 73

between merchants and cardholders. For any g


payment system can only increase profits by balancin
market in order to increase output. Therefore, th
system will maximize profits by choosing interch
imize the total output of the system.

Even if interchange fees are not fully passed th


using interchange fees to balance the two sides
likely to expand payment system output comp
payment system that is prevented from balancin
market, for example, because of a mandated z
Because interchange fees increase one price an
price, interchange fees will have a greater impact
on the total payment system price. Moreover, th
that an open-loop payment system achieves throu
change fees is greatest when demand elasticities
market are significantly different. For example,
cardholder demand is higher than the elasticity of
payment system can increase output by indirectl
prices relative to merchant prices through the use
fees. Because cardholder demand is significantly
chant demand, the gains in output as a result of
use considerably outweigh the loss in output from
acceptance. Thus, in such circumstances, balancin
fees achieves significantly greater output than if
not allowed to balance as a result of mandated zer

The empirical evidence is inconsistent with Visa


change fees being at output-restricting levels. Sp
indicates that if an open-loop payment card system
fees are reduced by regulation, these systems are
disadvantage relative to closed-loop systems in te
balance the two sides of the market, resulting in

85 Another proposed alternative to system- determined defa


change fees set by bilateral agreements between issuers and
a discussion of the problems with this alternative.
86 Instead of addressing the question of whether a mandated
lead to higher or lower output, the theoretical economics liter
system- determined profit-maximizing interchange fees lead
This literature shows that it is ambiguous whether payment s
determined interchange fees is larger or smaller than the soc
note 98. However, this question is different from whether sys
fees restrict or expand the output of a payment system r
interchange fee.

This content downloaded from 14.139.214.178 on Sun, 23 Jul 2017 15:49:06 UTC
All use subject to http://about.jstor.org/terms
2006] Competition in Two-Sided Markets 611

output. For example, consider the Australian regula


MasterCard credit card interchange fees. Under this re
change fees are based on issuer costs, but eligible is
include all issuer expenses. Specifically, marketing expe
of providing cardholder rewards and other benefits ar
the calculation of reimbursable expenses.87 As a result
reduction in interchange fees, the average Visa and
chant discount was reduced 42 basis points.88 Due to
regulation, American Express and Diners Club now poss
competitive advantage over Visa and MasterCard. Amer
Diners Club still can optimally balance the two sides of
their payment cards while Visa and MasterCard are stu
optimal interchange fee levels. Therefore, American Ex
Club are able to compete more effectively for cardhold
Visa and MasterCard. In particular, when competing
MasterCard issuers for cardholders, American Express
now can use their relatively higher, unregulated mercha
nue to provide superior cardholder benefits and lower
and to engage in more promotion relative to Visa and M
Because cardholders have more elastic demand than merchants, the
competitive advantage that interchange fee regulation gives American
Express and Diners Club from being able to provide greater cardholder
rewards and lower cardholder fees outweighs the benefits that Visa and
MasterCard receive in terms of increased merchant acceptance resulting
from a lower merchant discount.

The result of the American Express and Diners Club competitive


advantage over Visa and MasterCard has been an increase in the Ameri-
can Express and Diners Club share of Australian credit and charge card
purchase volume. Specifically, the American Express and Diners Club

87 The Australian regulation requires that the weighted average of the credit interchange
fees set by each association be no higher than a cost-based benchmark based on specific
issuer costs, including the costs incurred in processing, dealing with and preventing fraud,
providing authorization, and funding the interest-free period. RBA, Payments System
Board Annual Report 2004: Competition and Efficiency 8 (Nov. 26, 2004), available at
http://www.rba.gov.au/PublicationsAndResearch/AnnualReports/PaymentsSystem2004/
competition_and_efficiency.pdf [hereinafter Competition and Efficiency]', RBA, Reform of
Credit Card Schemes in Australia IV 26 (Aug. 2002), available at http://www.rba.gov.
au/PaymentsSystemPayments Policy/Reforms/CreditCardSchemes/Final Reforms [here-
inafter Reform of Credit Card Schemes] . Similarly, the European Commission settlement
required Visa's interchange fees on cross-border transactions to be based on costs "benefit-
ing" merchants, including the costs of providing processing, the payment guarantee, and
the interest-free period. Commission Decision 2002/914 of July 24, 2002, Case No. COMP/
29.373 Visa International- Multilateral Interchange Fee, 2002 OJ. (L 318) 17, 18-21.
88 RBA, Competition and Efficiency, supra note 87, at 9-10.

This content downloaded from 14.139.214.178 on Sun, 23 Jul 2017 15:49:06 UTC
All use subject to http://about.jstor.org/terms
612 Antitrust Law Journal [Vol. 73

combined share of purchase volume increased fro


financial year before the interchange fee regulat
2002-June 2003) to 16.5 percent during the most
(July 2004-June 2005). 89 The increase in the
Diners Club share of purchases has been accom
American Express and Diners Club taking advanta
fee regulation to reach agreements with three of
banks to issue American Express or Diners Club
share of the merchant discount.90 This is a clear p
share away from Visa and MasterCard in favor o
Diners Club.91 This loss in share by Visa and Mas

89 Merchant Service Fees and Market Shares for Credit and Cha
(Aug. 2005). The Payments System Board has noted that it
the full extent of cardholder switching from Visa and Mas
and Diners Club that may eventually occur. RBA, Competition
at 13.

90 See RBA, Competition and Efficiency, supra note 87, at 13; Andrew Cornell, American
Express Aims for Expansion, Austl. Fin. Rev., Apr. 18, 2005, at 54. These agreements may
explain why there was substantial growth in the number of cards after the regulation was
imposed. In particular, the total number of cards, which grew at an average annual rate
of 1.02% during 7/00-7/03, grew at an average annual rate of 7.04% during 7/03-6/05.
Payments Data available at http://www.rba.gov.au/PaymentsSystem/PaymentsStatistics/
payments_data.html. This increase in growth presumably occurred because of a large
number of additional American Express and Diners Club cards now held by individuals
who had held only Visa and MasterCard cards before the increase in fees. The Australian
Payments System Board recently examined the payments made by American Express
and Diners Club to their partner banks in return for the banks' promotional activities,
recognizing that these payments are in a fundamental economic sense analytically similar
to the interchange fees received by issuers in open-loop payment card systems, but decided
that it was not currently in the public interest to regulate these payments. See RBA,
Competition and Efficiency, supra note 87, at 13; Payments System Reform, RBA Media Release,
Feb. 24, 2005, available at http://www.rba.gov.au/MediaReleases/ 2005/mn_05_02.html.
91 The reduction in the Visa and MasterCard shares of credit and charge card purchases
is a better measure of the negative impact of the Australian regulation than the reduction
in total real credit and charge card purchase volume by all payment card systems. There
are many other factors that must be taken into account to explain changes in the overall
total credit and charge card market, which was decelerating both before and after the
regulation. See Howard Chang, David S. Evans & Daniel D. Garcia Swartz, The Effect of
Regulatory Intervention in Two-Sided Markets: An Assessment of Interchange-Fee Capping in Austra-
lia, 4 Rev. Network Econ. 1 (2005) . Specifically, total real credit and charge card purchase
volume grew at an average rate of 8.5% in the two years following the onset of regulation
compared to an average rate of 14.0% during the previous three years. However, the
previous three years showed a declining rate of growth, from 16.8% to 13.3% to 11.7%
(for the 7/00-6/01, 7/01-6/02, and 7/02-6/03 periods, respectively). This growth rate
decline continued in the post-regulation period, with the rate falling to 9.9% and 7.1%
(for the 7/03-6/04 and 7/04-6/05 periods, respectively). Payments Data supra note 90.
Australian CPI: Bloomberg. It is not clear how much of the reduction in total credit and
charge card purchase volume is due to the regulation because of the difficulties of measur-
ing in this dynamic environment what would have occurred in the overall market absent

This content downloaded from 14.139.214.178 on Sun, 23 Jul 2017 15:49:06 UTC
All use subject to http://about.jstor.org/terms
2006] Competition in Two- Sided Markets 613

the interchange fee regulation is convincing econom


pre-regulation Visa and MasterCard interchange fees w
competitive levels in the sense that the interchange fe
and MasterCard output.
The effect of the Australian regulation on Visa's a
shares highlights the economically significant differen
use of system- determined default interchange fees as a
tively balance a two-sided market and an anticompetit
cartel. Because a price-fixing cartel restricts output
prices, the breakup of the cartel (or the regulation of t
necessarily leads to increased output by the cartel mem
theories of cartels also predict that because "fringe" f
part of a cartel typically gain share as a result of the c
of the cartel results in the decreased market share of
regulation of interchange fees in Australia has had t
effect. It has led to a decrease in Visa's and MasterCard's shares of credit
and charge card purchase volume, and to an increase in the shares of
the unregulated payment systems, American Express and Diners Club.

The competitive reaction of American Express and Diners Club to


the lower regulated Visa and MasterCard interchange fees also highlights
the considerable differences between a price-fixing cartel and competi-
tive balancing by an open-loop payment card system through system-
determined default interchange fees. In particular, American Express
and Diners Club decreased their average Australian merchant discount
by only a fraction (13 basis points) of the decrease in the average Visa
and MasterCard merchant discount (42 basis points).92 Classic economic
theories of cartels predict that fringe firms operating under a cartel
price "umbrella" will benefit by increasing their prices by almost as much
as the cartel firms.93 This suggests that a breakup of the cartel should
lead to fringe firms also lowering prices by almost as much as firms
participating in the cartel. These predictions regarding the breakup or
regulation of a price-fixing cartel are inconsistent with the competitive

regulation. However, there are few important factors other than the regulation that can
explain the clear shift between different credit and charge card products in the market
after interchange fee regulation.
92 REA, Competition and Efficiency, supra note 87, at 9-10, 12-13. This may be somewhat
of an overestimate of how much the regulated decrease in Visa and MasterCard interchange
fees caused the American Express and Diners Club average merchant discount to decrease
if, as in the United States, American Express already was decreasing its Australian merchant
discount in an attempt to increase merchant acceptance and its share of purchases.
93 Fringe firms will often raise prices by slightly less than the cartel firms in order to
gain market share.

This content downloaded from 14.139.214.178 on Sun, 23 Jul 2017 15:49:06 UTC
All use subject to http://about.jstor.org/terms
614 Antitrust Law Journal [Vol.73

reaction of American Express and Diners Clu


Visa and MasterCard interchange fees. Ame
Club took advantage of the regulation of Vi
change fees by lowering their average merc
fraction of the mandated decrease in Visa and
fees. The fact that American Express and
increase their share by increasing the different
discounts and Visa and MasterCard, merchant d
and MasterCard interchange fees were not a
before their regulation.
These Australian developments highlight t
able to determine interchange fees for open-loo
such as Visa and MasterCard, to compete o
proprietary payment card systems.94 Without
interchange fee, an open-loop payment car
merchant and cardholder demand so as to opti
petitiveness and output. The Australian exp
fee regulation is consistent with the view tha
MasterCard interchange fees were not set ab
they had been, lower Visa and MasterCard inte
led to higher Visa and MasterCard shares of
fact that regulation of interchange fees led to
shares clearly suggests that the higher intercha
tion was imposed were not an output restricti
output- expanding factor in the competitive pa
fact, the concern regarding interchange fees h
change fees have led to an anticompetitive
rather to an inefficient expansion of payment

V. ARE THERE INEFFICIENCIES ASSOCIATED WITH


INTERCHANGE FEES?

A. Do Interchange Fees Cause Excess Payment Card Usage?

Some economists argue that interchange fees lead to a so


ciency, namely to "excessive" payment card use relative t

94 Some economists have contended that regulating Visa and MasterCar


fees also may constrain American Express merchant discounts. See, e.g., Carl
supra note 6, at 641. However, the Australian experience clearly shows that
constraining the merchant discounts of closed-loop payment systems, th
interchange fees has benefited these systems, which have increased their s
and charge card transaction volume by maintaining significantly higher mer
than Visa and MasterCard.

This content downloaded from 14.139.214.178 on Sun, 23 Jul 2017 15:49:06 UTC
All use subject to http://about.jstor.org/terms
2006] Competition in Two-Sided Markets 615

cash and checks.95 These economists proceed by first a


current interchange fee levels, the cost to a merchant
uses a payment card (as measured by the discount m
the transaction) is greater than the merchant cost
customer's use of cash or check.96 Because merchan
uniform price for the products they sell regardless
method used by a customer, the extra merchant cost w
use payment cards is assumed to be reflected in higher
merchant prices paid by both payment card custome
who use other means of payment. Consequently, it i
customers who use payment cards impose a "usage exte
tomers who use other, cheaper forms of payment,
checks, and that customers who use cash and checks therefore "cross-
subsidize" payment card users.97 Because customers using payment cards
do not bear the full cost of their use, they are said to overuse payment
cards relative to other payment methods from a social perspective. This

95 See, e.g., Dennis W. Carl ton & Alan S. Frankel, The Antitrust Economics of Payment Card
Networks, 63 Antitrust LJ. 643, 661 (1995); Frankel, supra note 6; Katz, supra note 15,
at 25-29. The RBA followed Katz's 2001 report in adopting this theory as a stated motivation
for its regulation of interchange fees. RBA, Reform of Credit Card Schemes, supra note 87, at 34.
96 For example, a merchant's average cost of accepting a Visa or MasterCard credit card
is approximately 2.00% while, according to FMI estimates, a merchant's cost of accepting
a check is approximately 0.75% for an average check transaction. Cost of Visa credit cards:
Visa U.S.A. Cost of checks: Food Marketing Institute (FMI), It All Adds Up: An
Activity- Based Cost Study of Retail Payments (2000). These estimates, however,
understate check costs compared to credit card costs if merchants were to attempt to
replace all credit card transactions with checks. Merchants that currently accept a high
proportion of checks relative to other payment methods (e.g., supermarkets) are typically
merchants with a high number of repeat customers who are able to minimize fraud losses
from bad checks through such methods as issuing check cashing cards, maintaining lists
of customers who have written bad checks in the past, and developing personal relation-
ships with customers. Merchants in all segments are unable to take such precautions.
Therefore, if all merchants accepted higher proportions of checks, fraud losses and the
cost of checks would increase substantially from what is estimated by the FMI. The simple
fact that many merchants refuse to accept checks or have stringent requirements to do
so, while freely accepting Visa and MasterCard credit cards, illustrates that the FMI figures
do not capture all relevant costs and benefits of different methods of payment and cannot
be generalized to non-supermarket merchants. Garcia Swartz et al. discuss other limitations
of the FMI's payment cost estimates, such as the failure to include the cost of float, and
attempt to measure the net social costs and benefits of different payment methods to
affected parties other than merchants, including consumers, banks, and the government.
Daniel D. Garcia Swartz, Robert W. Hahn & Anne Layne-Farrar, The Economics of a Cashless
Society: An Analysis of the Costs and Benefits of Payment Instruments 43 (Oct. 2004) (mimeo, AEI-
Brookings Joint Center for Regulatory Studies) . See also Richard A. Epstein, The Regulation of
Interchange Fees: Australian Fine-Tuning Gone Awry, 2005 Colum. Bus. L. Rev. 551, 571-75.
97 See supra note 6. We are assuming for the purpose of this discussion that cash and
checks are cheaper than payment cards. However, as discussed in supra note 96, this may
not be the case.

This content downloaded from 14.139.214.178 on Sun, 23 Jul 2017 15:49:06 UTC
All use subject to http://about.jstor.org/terms
616 Antitrust Law Journal [Vol. 73

excess usage of payment cards forms the basis fo


change fees.98

It has been argued that interchange fees should


(a "par collection" system) in order to avoid any c
payment card users by cash and check customers.
been proposed that any cross-subsidization should
ing issuers to cover their costs by collecting inte
from cardholders rather than from acquirers (a
mechanism supposedly would maintain the revenu
promote cards and provide benefits to cardholders
the inefficiencies that are said to arise from inte
important to recognize that these two proposed a
determined default interchange fees are the same.
collect all revenues from cardholders rather than
a portion of their revenues from acquirers in the
fee is equivalent to a mandating a zero interc
imposed on cardholders do not amount to a tra
side of the market to the other and therefore do not affect the relative

98 Some economists have argued there may be an offsetting inefficiency from the
presence of issuer market power. In particular, these economists assert that the increased
quantity of payment card use that is induced with an interchange fee offsets the decreased
quantity of payment card use that otherwise would result from issuer market power. See,
e.g., Schmalensee, Review of the Literature, supra note 11; Rochet, supra note 11; Wright,
supranote 1 1. In theoretical economic models, the primary reasons why system- determined
interchange fees may deviate from socially optimal interchange fees are the excess usage
of payment cards resulting from the "usage externality" and the insufficient use of payment
cards resulting from issuer market power. See, e.g., Schmalensee, Review of the Literature,
supra note 11; Rochet, supra note 11. Because these factors have opposite effects on
payment system output, this theoretical economics literature concludes that one cannot
know if system- determined interchange fees are above or below socially optimal levels.
For example, Evans and Schmalensee conclude that "[o]ne cannot presume on the basis
of theory alone that the interchange fee set collectively by an association is greater than,
less than, or equal to the socially optimal interchange fee." David S. Evans & Richard
Schmalensee, The Economics of Interchange Fees and Their Regulation: An Over-
view 40 (MIT Sloan Working Paper 4548-05, May 2005) , available at http://www.ssrn.com/
abstract=744705.

99 See supra note 41.


100 See, e.g., Carl ton & Frankel, supra note 6, at 640. Placing the costs of providing payment
cards on cardholders is said to be efficient because, as a result, prices reflect the cost of
serving different customers and cross-subsidization across customers is avoided. However,
competition often does not result in customers being charged for all of the specific costs
associated with their purchases. For example, providing promotional services to customers
free of charge is the essence of promotion. Promotion is effectively a targeted discount
to marginal consumers, who are more sensitive to promotional services than inframarginal
consumers. Charging marginal consumers for the cost of promotion would defeat the
purpose of using promotion to attract marginal consumers. See Benjamin Klein & Kevin
M. Murphy, Vertical Restraints as Contract Enforcement Mechanisms, 2 J.L. & Econ. 265,
283-85 (1988).

This content downloaded from 14.139.214.178 on Sun, 23 Jul 2017 15:49:06 UTC
All use subject to http://about.jstor.org/terms
2006] Competition in Two-Sided Markets 617

prices to cardholders and merchants, which is the essence


the two sides of the payment system through the use
fees. As we have discussed, a mandated zero interchan
eliminate an open-loop payment card system's ability to b
sides of the payment card market and would hinder its abi
effectively against closed-loop payment card systems such
Express and Discover.
Moreover, any cross-subsidization of payment card user
check customers would not be limited to the use of defau
fees in open-loop payment card systems. The same cros
would occur with closed-loop systems, such as America
Discover, which do not use interchange fees. In fact, beca
Express transactions have a higher average merchant disc
externality problem would be even greater for these tran
for Visa and MasterCard transactions. Furthermore, the criticisms of
interchange fees are really criticisms of balancing in all two-sided mar-
kets. However, competitive balancing often leads to some consumers
paying a higher share of the cost of supplying the product than other
consumers.

The cross-subsidization across consumers that is said to occur in two-


sided markets also can occur outside the context of two-sided markets.
It is common for merchants to incur costs for various services that do
not benefit all customers to the same extent without passing on these
differential costs to the particular consumers using the services. For
example, merchants frequently offer without charge a number of amenit-
ies that only some customers use, including parking, gift wrapping,
extended store hours, and delivery. Another obvious example is the
practice of restaurants to offer free coffee refills. This presumably create
a usage externality by consumers of multiple cups of coffee and an excess
coffee consumption inefficiency can be said to exist. However, such so-
called cross-subsidization between consumers is common throughout
the economy and is imposed by firms without any market power.
Some economists have asserted that the situation for the Visa and
MasterCard payment card systems is fundamentally different from t
many clearly competitive examples in the marketplace in which differ
customers are cross-subsidized and impose externalities on one anothe
These economists contend that open-loop payment card systems u
their market power to impose cross-subsidies on all card-accepting me
chants and their customers via the use of interchange fees.101 Therefo

101 See, e.g., Frankel, supra note 6; Katz, supra note 15, at 46-47; Carl ton & Frankel, supr
note 6.

This content downloaded from 14.139.214.178 on Sun, 23 Jul 2017 15:49:06 UTC
All use subject to http://about.jstor.org/terms
618 Antitrust Law Journal [Vol. 73

it is said the decision of how much to subsidize dif


ers is not left up to individual merchant choic
However, as we have seen, interchange fees ar
payment system market power but of competitiv
card systems, including those with relatively sma
as American Express and Discover, must balanc
market in order to compete effectively. Balancin
payment system results from competitive forces
exercise of market power. Balancing by payment
possess no market power also could be said to resu
by consumers using cash and checks. However,
would not arise from the exercise of market pow
not provide a valid basis for antitrust intervention

Merchants also have the ability to eliminate any


payment card users by cash and check users by ch
a higher price than cash and check customers. It
merchants do not charge differential prices to cu
differences in the cost of payment method us
payment card system rules. In particular, paym
surcharge" rules, which prohibit merchants from
on payment card transactions, are said to prevent
from overriding any cross-subsidization of paym
check users.102 However, if merchants wished to
subsidization of payment card users by cash and
accomplish this by charging a lower retail price t
cash or checks.103 Payment card system rules per
United States to offer such discounts.104 Merchant

102 See, e.g., Frankel, supra note 6, at 343, 348; Carlton 8c Fran
103 For example, assume the price of a product is $10.00 a
when a customer uses a payment card rather than cash or c
we assume for the purpose of this illustration that cash an
then this would result in net prices received by the merch
check customers and $9.80 from payment card customers. Th
used payment cards and 50% used cash or checks, merchan
average, $9.90 net per sale and there would be a cross-subsidy
by cash and check customers and excessive payment card us
increased the product price to $10.10 and simultaneously ga
and check sales, they would be receiving $9.90 on card sale
sales and customers would have the correct incentive to us
payment cards would be used by consumers only when the ad
more to them than the added $0.20 cost borne by the mercha
fees would be "neutral" in the sense of having no effect on t
by consumers using different forms of payment and would h
of payment card transactions.
104 See supra note 25.

This content downloaded from 14.139.214.178 on Sun, 23 Jul 2017 15:49:06 UTC
All use subject to http://about.jstor.org/terms
2006] Competition in Two-Sided Markets 619

any interchange fee that subsidized a payment card sy


at the expense of cash and check payers with lowe
prices.105 A discount for cash and checks is analyticall
surcharge for credit.106

Although merchant discounts for cash and checks a


payment card system rules, U.S. merchants almost u
offer such discounts. A few U.S. merchants have charg
for cash in the past (for example, some U.S. gas stat
discounts in the 1980s), but even these relatively isolat
largely been abandoned.107 This absence of discounts fo
has been attributed to transaction costs associated with
ing different prices for their products based on what
the customer uses.108 But it is highly unlikely that th
incidence of discounts for customers using cash an
explained entirely by the extra transaction costs of institut
ential pricing. Charging different prices does not se
given, for example, the prevalence of coupons and oth
programs (unrelated to the method of payment used b

105 This neutrality result is formally derived in several economic a


nomic conditions for neutrality summarized in Joshua S. Gans &
Neutrality of the Interchange Fees in the Payment Systems, 3 Topics in
(2003) Article 1 . The ability of merchants to charge different prices to
on whether they use a payment card is not a necessary condition
neutrality. Alternatively, some economic models derive neutrality by
ing that merchants are operating in a perfectly competitive retail en
customers always have a comparable merchant available that does
sells the identical product at a lower price.
106 Given this equivalency of discounts and surcharges as a means
prices to different customers, an obvious question is why payment c
cash and check discounts but have rules that prohibit merchants from
on payment card transactions. One reason is that by surcharging,
potential to produce a negative externality on the payment card
Merchants that place a significant surcharge on transactions made w
ment card would undermine the honor-all- cards rule and, therefore
ers of the payment card brand. Specifically, a merchant that puts a
in its window to attract consumers but then institutes a large surchar
is misleading consumers who expect, when they see the payment
without any payment surcharge. The merchant that imposes a large
does not really accept the card in a way that is consistent with card
about the brand. Meeting such customer expectations is particularl
brands as Visa and MasterCard that have built their brand images on
merchant acceptance at par. Consequently, a merchant that surch
externality on the entire payment card system by reducing the over
for the system.
107 John M. Barron, Michael E. Staten &John Umbeck, Discounts for C
Marketing, 10 Contemp. Pol'y Issues 89, 89 (1992).
1U8 See, e.g., frankel, supra note b, at 343, 348; Carlton & Frankel, sup

This content downloaded from 14.139.214.178 on Sun, 23 Jul 2017 15:49:06 UTC
All use subject to http://about.jstor.org/terms
620 Antitrust Law Journal [Vol.73

that likely are more costly to administer than a


and check payments.

Contrary to the assumption that merchants wo


relatively higher prices to payment card users if
most merchants would be reluctant to offer such a discount for cash
and check transactions, even if there were no associated transaction
costs, because of rational merchant pricing. Payment card users are likely
to be a merchant's relatively more price-elastic customers for a number
of reasons. First, payment cards are accepted at a much larger number
of merchants than personal checks. Even when a merchant accepts
checks, it may do so only for regular customers who have an ongoing
relationship with the store. Therefore, there are a significantly larger
number of alternative merchants potentially available for payment card
consumers, increasing the payment card consumer's elasticity of demand
for shopping at an individual merchant. In addition, consumers who use
payment cards have an increased flexibility to take advantage of a mer-
chant's low prices. Payment card consumers have a greater ability to
make large durable good purchases when they see favorable prices or
to stock up on staple items that are on sale because, in contrast to cash
or checks, a credit card provides the consumer with access to all funds
up to the card's credit limit. This relatively higher demand elasticity for
consumers who use payment cards explains why most merchants do not
offer discounts for cash and checks.109 In fact, inter-merchant competition
for payment card customers will prevent most merchants from providing
discounts for cash and checks.

Merchants can also avoid any cross-subsidization of payment card


customers by individually deciding not to accept a payment system's
cards. An individual merchant's decision to accept a payment card is

109 Thjs js consistent with the fact that surcharging is fairly limited in countries where
such pricing prohibitions have been eliminated. For example, the OFT's 2005 decision
on MasterCard's interchange fees contained several studies on the incidence of surcharging
in the United Kingdom. These studies found (and the OFT concluded) that, even though
merchants have had the right to surcharge in the United Kingdom since 1991, relatively
few actually have done so. Moreover, most of the merchants that surcharge do so only
on transactions meeting certain criteria (e.g., small transactions) and were concentrated
in selected merchant sectors (e.g., travel agencies, taxis, and ticket agencies). Decision of
the Office of Fair Trading, No. CA98/05/05, at 87-98 (Sept. 6, 2005).) See also IMA
Market Development AB: Study Regarding the Effects of the Abolition of the
Non-Discrimination Rule in Sweden, Prepared for European Commission Competi-
tion Directorate General (2000), available at http://europa.eu.int/comm/competition/
antitrust/cases/29373/studies/sweden/report.pdf; E. Vis & J. Toth, The Abolition of
the No-Discrimination Rule, ITM Research: Amsterdam (Mar. 2000), available at
http://europa.eu.int/comm/competition/antitrust/cases/29373/studies/netherlands/
report.pdf. (regarding the prevalence of surcharging in Sweden and the Netherlands,
respectively) .

This content downloaded from 14.139.214.178 on Sun, 23 Jul 2017 15:49:06 UTC
All use subject to http://about.jstor.org/terms
2006] Competition in Two- Sided Markets 621

based on the fundamental economic calculations described in Part II.B,


which indicate that a merchant need not lose many sales from its failure
to accept a particular payment system card to make it worthwhile for
the merchant to accept the card. It has been argued, however, that this
individual merchant decision to accept a payment card does not reflect
an accurate measure of the social costs and benefits of card acceptance.110
In particular, merchants are said to be willing to accept a means of
payment with higher costs because of the incremental sales that the
merchant gains at the expense of other merchants.111 It is asserted that
because these incremental sales net out to zero in the aggregate and
thus largely represent a redistribution of sales across merchants and not
an increase in aggregate sales, an individual merchant's private gain
from payment card acceptance is greater than the benefits to merchants
as a group from acceptance.112 To merchants as a group, this is then a
largely zero-sum game with little or no net benefit from payment card
acceptance. Because the aggregate social value of payment cards to all
merchants as a group in terms of incremental sales is maintained to be
close to zero, the costs that merchants incur to accept payment cards
are argued to be socially excessive.113 The inter-merchant externality
resulting from the incremental sales of payment card acceptance is
the fundamental economic reason why inter-merchant competition in
economic models of interchange fees may lead to interchange fees that
are too high from a social perspective and to an overprovision of credit
card services.114

However, this same logic applies to many other competitive activities


used by merchants to attract customers from their rivals. We do not
believe, for example, that competition by supermarkets in terms of main-
taining clean and attractive store facilities or other forms of nonprice
competition are excessive because in the aggregate they are unlikely to
have much effect on the aggregate consumption of food. Competition
that primarily has inter-retailer effects with little change in aggregate
demand facing retailers as a group cannot be said to be excessive or
inefficient. These forms of competition are the essence of the competitive
process. Consumers value the nonprice competition, as evidenced by

110 See Katz, supra note 15, at 26-27.


111 Id.

112 Id.

113 Payment cards create a social benefit because they offer cardholders relatively cheap
access to a line of credit. See Tom Brown & Lacey Piache, Paying with Plastic: Maybe Not So
Crazy, 73 U. Chi. L. Rev. 63, 71-74 (2006).
114 See Rochet & Tiróle, supra note 11; Rochet, supra note 11; Wright, supra note 11.
Some economists call this a "strategic motive" for merchant acceptance of payment cards.
See, e.g., Wright, supra note 11, at 4.

This content downloaded from 14.139.214.178 on Sun, 23 Jul 2017 15:49:06 UTC
All use subject to http://about.jstor.org/terms
622 Antitrust Law Journal [Vol. 73

their switching retailers in response to its supply,


to a merchant's decision to accept a payment card
as a group would prefer not to compete along th
gously, merchants as a group would prefer not
another by accepting payment cards. However, r
in interchange fees or otherwise controlling the e
from this competition would be equivalent to col
all merchants to prevent the competitive process

B. Do Interchange Fees Result in Wasteful Issuer Promotion?

It has been argued that competition among merchants to a


tomers by accepting payment cards is different from other
merchant competition because the latter results in social
investments by merchants (e.g., clean stores) , while merchan
of payment cards at claimed socially excessive interchange f
"inefficient" and "wasteful" rent dissipation by issuers.115 Ho
petition among issuers results in significant cardholders ben
of this issuer competition occurs in the form of reduced
fees and increased cardholder rewards, including cash a
rewards.116 Such forms of competition are not inefficient an
Reduced cardholder fees and cash-back rewards are effec
prices to cardholders. Moreover, because there are no con
issuers offering cash-back to cardholders, as many do, the fact th
tition among issuers results in rewards in kind, such as
tickets, indicates that some cardholders prefer receiving suc
over cash.117

It also has been asserted that promotional and marketing e


to obtain new cardholders are socially inefficient because re
are wasted in the competitive process, for example, in the m

115 Frankel, supra note 6.


116 As discussed above, cardholder benefits increased after interchange f
in 1998. See supra note 82. Similarly, decreases in interchange fees resu
cardholder benefits. For instance, after regulation lowered Australian in
Australian issuers responded by increasing annual and other cardholder f
MasterCard credit cards while reducing cardholder rewards. In addition
have increased the requirements to earn rewards while other issuers ha
amount of the rewards that can be earned. RBA, Competition and Efficiency
at 1 1 . The effects of the Australian regulation of interchange fees are ana
et al., supra note 91.
117 According to one survey, 26% of credit cards in 2003 offered cash-bac
38% offered rewards in kind. Synovate Inside Track, http://www.synovate.c

This content downloaded from 14.139.214.178 on Sun, 23 Jul 2017 15:49:06 UTC
All use subject to http://about.jstor.org/terms
2006] Competition in Two-Sided Markets 623

mailings of payment card solicitations that have a low


For example, the number of credit card offers has in
billion per year in 1996 to 5.23 billion per year in 200
46.7 annual offers per household, with the response r
1.4 percent to 0.4 percent.119 However, much of issuer
the provision of information about superior program
grams with reduced finance charges and/or better
that this is wasteful activity ignores the fact that the
essence of competitive promotion. Indeed, there are n
which competitive firms advertise their products a
lower response rates than credit card solicitations. For
redemption rates in the United States are estimate
1 percent.120 Similarly, the "response rate" of viewers
tisements (i.e., the number of viewers that decide to
as a fraction of the number that receive the message)
low. Mailings are likely a relatively more important e
tional campaigns compared to television or radio ads f
than for other products because mailings for paymen
application with which consumers readily can "purcha
whereas television and radio ads for payment cards
effort on the part of the consumer who must make a
application before purchase is possible.

In any event, whatever one thinks about the soci


advertising, engaging in such promotional activity ce
to do with market power and is part of the normal c
Advertising is undertaken by many small firms in th
all firms in the payment card industry. Firms operating
marketplace will use the form of competition that they fi
Antitrust should not be about regulating or controllin
forms of competition chosen by business firms but ab
anticompetitive creation or exploitation of market pow

118 See, e.g., Frankel, supra note 6, at 345-46. This criticism m


regulatory authorities have reduced interchange fees and, in p
narrower set of issuer costs that does not include promotional exp
when determining the "appropriate" level of interchange fees. See
119 Number of credit card offers and response rates: Synovate Mail M
synovate.com, accessed 1/10/2006. Number of U.S. households
Statistical Abstracts of the United States: 1996-2004, available at http://
www.census.gov/compendia/statab.
^"According to CMS Inc., a promotion management services firm, the U.S. national
coupon redemption rate in 2005 was 0.93%. More Coupons Issued in 2005, but Redemption
Fell, Feb. 13, 2006, http://www.thewisemarketer.com/news/read.asp?lc=f98458mxl633zr.

This content downloaded from 14.139.214.178 on Sun, 23 Jul 2017 15:49:06 UTC
All use subject to http://about.jstor.org/terms
624 Antitrust Law Journal [Vol. 73

The paradox of the criticisms of interchange f


inefficiencies resulting not from restrictions of
the traditional antitrust concerns regarding the
but rather from socially excessive competition a
These criticisms emphasize that competition
merchants to accept payment cards despite the
cash and checks because merchants do not take into account that the
incremental sales each merchant gains as a result of accepting a paym
card are a zero -sum game from a social perspective and therefore hav
little social benefit. As a result of this merchant competition through
payment card acceptance, it is argued that merchants are willing to p
merchant discounts that are too high from a social perspective, which
leads to socially excessive competition by issuers, including the ov
provision of cash-back and rewards in kind and the over-promotion o
payment cards. This socially excessive competition among merchan
and among issuers is said to result in the overuse of payment car
relative to cash and checks. These inefficiencies attributed to interchan
fees are contrary to the usual inefficiencies resulting from the exercise
market power, namely the restriction of competition and market outp

Antitrust policy should not attempt to micro -regulate the manner


which firms compete or the outcome of the competitive process base
on a theoretical standard of social optimality. For instance, wheth
issuer competition for cardholders takes the form of cash-back, rewar
in kind, or promotion should be left up to the competitive process rat
than regulated based on theoretical arguments about "inefficient" and
"wasteful" competition. Similarly, merchant competition through pay
ment card acceptance should not be regulated because it results in
zero-sum game with little effect on aggregate demand. Merchant comp
tition that primarily has inter-retailer effects is the essence of the compet
tive process. Regulating such competition because of considerations of
an externality among merchants amounts to a rejection of the compe
tive process and mimics a collusive agreement among merchants. Rath
than micro -regulating the specific form and outcome of the competit
market process on the basis of theoretical considerations of social opt
mality, the goal of antitrust is to assure that the competitive market
process is not restrained by the exercise of market power.

VI. CONCLUSION

The use of system- determined default interchange fees is ne


for the effective operation of open-loop payment card systems.
system- determined interchange fees, incentives would be crea
individual issuers in an open-loop payment system to take adva

This content downloaded from 14.139.214.178 on Sun, 23 Jul 2017 15:49:06 UTC
All use subject to http://about.jstor.org/terms
2006] Competition in Two- Sided Markets 625

the honor-all- cards rule to hold up merchants by dema


high interchange fees. Given honor-all- cards, merc
the ability to reject an individual issuer's cards, but mu
all the cards of a payment system. Therefore, individu
have the incentive to free ride and impose external
payment system participants by demanding a high int
open-loop payment card system could not effectively co
payment systems without setting a default interchange
individual issuers from engaging in such hold-up behav

The particular level of the default interchange fee ch


loop payment card system is determined by the compe
of demand on the cardholder and merchant sides of the market. Such
competitive balancing occurs in all two-sided markets, and specifically
by all payment card systems. Although closed-loop payment card system
do not use interchange fees, they similarly balance demand on the two
sides of the market through the direct setting of merchant discounts an
cardholder fees and rewards. Interchange fees are the way competitive
balancing and the resulting relative prices paid by cardholders and mer
chants are determined in an open-loop payment card system where ther
are multiple issuers and acquirers. Competition compels all paymen
card systems to optimally balance by taking account of demand on the
two sides of the market and in this way to increase the output and valu
of payment card systems.

We have seen no evidence that any actual balance between merchants


and cardholders that has been chosen over time by the Visa and Master
Card associations in their setting of default interchange fees involves a
exercise of market power and the restriction of output. Balancin
between the two sides of a payment card market generally leads to
relatively higher merchant prices than cardholder prices because card-
holders usually are willing to switch their card usage between payment
systems and, therefore, have greater demand elasticities across paymen
systems than merchants, who usually find it economic to accept cards
from all major payment card systems. This fundamental economic deter
minant of relative cardholder and merchant prices operates in all pay-
ment card systems, including those payment systems, such as America
Express, with relatively small market shares. Moreover, changes in the
level of Visa and MasterCard interchange fees over time in the United
States - the generally declining level of interchange fees through th
early 1990s and the increasing level of interchange fees since 1998 - ca
be explained not by changes in payment system market power, but by
changes in the competitive balancing of economic forces on the tw
sides of the payment card market. In particular, the growing cardhold

This content downloaded from 14.139.214.178 on Sun, 23 Jul 2017 15:49:06 UTC
All use subject to http://about.jstor.org/terms
626 Antitrust Law Journal [Vol. 73

loyalty to individual issuers recently has increase


Visa and MasterCard for issuers and shifted th
towards the cardholder side of the market and aw
side of the market.

To determine if market power has been exercised, one must look not
at relative prices on the two sides of the market, or the level of inter-
change fees in an open-payment card system, but rather at the total price
charged by the payment card system, that is, the sum of the prices paid
by cardholders and merchants. We have found no evidence to suggest
that the total price charged by open-loop payment card systems is abnor-
mally high. In fact, the price paid by cardholders has been decreasing
recently, as the competitive process on the issuing side of the market has
led to an increase in rewards along with the increase in interchange fees.

The criticisms of interchange fees have not been based on empirical


examinations of the total price charged by payment card systems but
rather on the theory of an inefficiency caused by interchange fees,
namely that interchange fees lead to the cross-subsidization of credit
card users by cash and check customers. However, such theoretical cross-
subsidization would not be limited to the use of default interchange fees
by open-loop payment systems. An identical cross-subsidization would
occur with all payment card systems, including closed-loop systems, such
as American Express and Discover, that do not use interchange fees and
clearly do not possess market power. Therefore, we conclude that any
cross-subsidization resulting from interchange fees is a consequence of
the normal competitive process and does not provide a valid basis for
antitrust intervention.

This content downloaded from 14.139.214.178 on Sun, 23 Jul 2017 15:49:06 UTC
All use subject to http://about.jstor.org/terms

You might also like