D[YIt~NOVEMBER/DECEMBER 199$

David A. Balto is on attorney advisor to Chairman Robert Pitofsky of the Federal Trade Commission. This paper does not necessarily representthe views of the commission, of any individual commissioner, or of any other member of the staff.

PaymentSystems andAntitrust:Can theOpportunitiesFor NetworkCompetition BeRecognized?

David A. Balto

5 1iews of payment systems competitionhave evolved during the past genera-

SI’ tion. When automated tellermachine (ATM) netw-orks were firstcreated in the 1970s, policymakers consid-ered two models for these emergingnetworks: (1) a monopoly/public utilitynetwork model, with open accessobligations and (potentially) someform of regulation or (2) a competingnetwork model, with numerous networkscompeting in a lightly regulatedenvironment. This article describes how-these visions of network competition haveevolved. Even though the networkcompetition model was chosen in the1970s, because of a history of nonenforce-ment by antitrust agencies and regulators,it appears that by the close of this centurythe monopoly/public utility model may hevictorious. This article describes how thatchange occurred and considers whether itis appropriate.

The first section of the articledescribes general trends in antitrustenforcement affecting payment systexnsnetworks since the I 970s. The next

section examines the framework applied toantitrust analysis of ATM network mergersand is followed by a section applying thisanalysis to two recent ATM networkmergers. The article examines how themonopoly/public utility model appearsto have prevailed in the ATM networkmerger context.

THE SEARCH SOR. PAYMENTSYSTEMS CO.MPETITUON:

TRENDS ill EN.EORCEMENT

•The 1920-s — Providing theOpportunities hr Networkf/on-ipetition in New Marl,,ets

As the technology for automated pay-ment systems arose, Congress perceivedthe need to address the creation of thesesystems in a single forum, and created theNational Commission on Electronic FundsTransfer (NCEFT). The Antitrust Divisionof the Department ofjustice (the division)played an important role in informing theNCEFT whether and in what formcompetition could arise in the newlyformed networks.

One important question addressed bythe NCEFT was whether these networkswould be natural monopolies because ofthe substantial processing efficienciesinvolved. At the time, some commentatorsargued that, because a single networkcould serve all ATMs at lower cost, thesenetworks were natural monopolies. Basedon that conclusion, they argued that thenetworks should be open, that is,compelled to share their facilities with allfinancial institutions in a given area,

In proceedings before the T’ICEFT, theantitrust division opposed the concept ofmandatory sharing, in particular because itwould deter the incentives to create com-peting networks.’ The NCFFT adoptedthe division’s view. It observed thatmandatory sharing “would inevitably

‘See U.S. Deportment of Justice(1977).

FEDERAl RESERVE RANK OF ST. LOUIS

19

lltYI1~NOVEMBER/DECEMBER 1995

‘See EFT hr the United Stotes(1977).

At the time, b ecouse ATMnetworks were in their infancy,there were no significant barri-ers to entry.

‘See Rule, reprinted in Bakerand Brandel (1988).

‘See laderman (1990).

The division has a procedure,known as a business review,which ituses to indicatewhether it will bring onenforcement action.

‘See Baker (1977).

See Rule, (1988),

See infra notes 35-36 andaccompanyrng rest, he uoarddid appear to consider whethera proposed network “may rep-resent so large a proportion ofpossible ATM terminals in localmarkets that no other switclrescould successfully compete.”See Barclays larrk PLC(1985); CenterreBancarporotion (1983).

result in fewer competitors.... Maximumcompetition usually spells rapidtechnological improvement and lowerprices to consumers.” Thus the commis-sion expressly rejected any sharingrequirement, based on its assessment thatthere was potential for the creation of anumber of competing networks.’

The division continued to advocateits vision of network competition in anumber of forums. It actively opposedthe adoption of state sharing statutes.’The division argued that mandatorysharing would undermine the incentive tocreate networks in the first place bycreating a free rider problem. That is, ifthe creator of a network knew it wouldhave to share ownership with others andshare the fruits of its efforts after thenetwork succeeded, it might he deterredfrom creating the network in the firstplace. Moreover, the division suggestedthat mandatory sharing would lead to theformation of monopoly networks.

Despite the division’s intervention,many states adopted various forms ofmandatory sharing. Since these lawsrequire a netw-ork to admit any bank as amember, they dampened the opportunityfor intersystem competition. More recenteconomic analysis of these sharing lawssuggests that the division was correct insuggesting that mandatory sharing wouldnot, serve the interests of consumers. Inthose states with mandatory sharing laws,outptnt in terms of ATM deployment andcard usage is less than in those states thatdo not require sharing.’

In the I 970s, scores of ATM networkswere created. lwhen these networksappeared to interfere with the potential fornetwork competition, for example, bybeing too large or overinclusive, thedivision raised concerns and threatenedenforcement action. In 1977, the divisionissued a business review letter refusing toclear a proposed statewide electronic fundstransfer (PET) network in Nebraska,prianarily because of the proposedventure’s all-inclusive nature,” At the timeof the letter, the proposed networkcomprised 66 percent of the commercial

banks in the state, which collectivelyaccounted for 86 percent of deposits. Thenetwork attempted to justify its size basedon the amount of capital required, thedegree of risk, and the economies of scaleinvolved in operating an EET system. Thedivision concluded that these efficienciesdid not necessarily justify the all-inclusivenature of the proposed network.” Becauseof the division’s action, competingnetworks were created in Nebraska,and other networks avoided becomingover-inclusive.

The I 98’Os—&onamics of Ubiquitylake Center Staqe

In the 1980s, the division basicallydisappeared from the enforcement radarin payment systems. The lack of enforce-ment, especially in the merger area, wasbased on the recognition that there wereefficiencies from the consolidation of ATMnetworks. Charles Rule, former assistantattorney general of the Antitrust Division,discussed this factor in a 1985 speech.Rule stated that the division was focusingmore on the economies of ubiquity and theresulting consumer benefits achievableby widespread sharing of ATMs. Ruleobserved that the consolidation of ATMnetworks benefits consumers by amongother things, increasing the available ATMsin a single network; similarly increasingthe number of cardholders tends toincrease the deployment of ATMs. ThusRule indicated that the division would notchallenge the creation or merger of sharedATM networks based on size alone.’

Unsurprisingly the division did notchallenge, or even apparently investigate,any ATM mergers during the 1980s. TheFederal Reserve Board approved even’ATM merger before it because it viewedthe A’I’M network as primarily a systeun ofcomputers and consec~uent.lyfocusedalmost exclusively on the networks’“hack office” operations when approvingthese mergers.”

Perhaps the most notable merger wasthe 1988 acquisition of the Cashstreamnetw-ork by the MAC network in 1988—

FSDEBAI. RESERVE BANK OF ST. LOUIS

20

llF~IE~NOVEMBER/DECEMBER 1995

two mid-Atlantic networ-ks whichcompeted in Pennsylvania and NewjerseyThe division did not challenge the merger.Rather, the merger was the subject of a pri-vate antitrust challenge brought by TheTreasurer, a competing ATM network. Adistrict court rejected this challenge in TheTreasuret Inc. v. Philadelphia Tvational Bamh.The court adopted an approach similar tothe Board’s—that the relevant marketincluded anyone capable of providingcomputer processing and that market wasunconcentrated, MAC continued toacquire almost all of its neighboringnetworks, ultimately securing a dominantposition in Pennsylvania and manyadjoining states.

PULSE business review. The onematter that forced the division to confrontintersystem competition was a businessreview request submitted by the PULSEATM network in 1983. At the timethere was aggressive competition in Texasbetween two similar sized networks:PULSE and MPACT. MPACT, in particular,competed through an incentive priceprogram. First Texas Savings and Loan,a member of MPACT, sought to joinPULSE, and PULSE declined,

PULSE was faced with a peculiarquandary posed by the antitrust laws.If PULSE refused to admit the hank,First Texas could claim that its exclusionfrom PULSE constituted an illegalgroup boycott and it could seek trebledamages in a private antitrust suit.’° IfPULSE admitted First Texas, this wouldcreate a de facto merger with MPACT,and PULSE might face a governmentantitrust challenge because the networkhad become too large and the mergerelinnnated intersystem competition.’’

Faced with this dilemma. PULSEsought a business review from thedivision. PUT_SE posed three alternativesto the division: (I.) admitting FirstTexas; (2) generally admitting membersof competing networks; or (3) implementingan anti-duality rule, that would prohibitmembership to members of counpetingnetworks.

The division addressed only the firstalternative, saying that at the time,admitting First Texas would not pose anantitrust violation. The division notedthat the incremental consumer conveniencethat would result from admitting FirstTexas appeared to outweigh the loss ofrivalry that might occur between the twocompeting networks.” The other twoalternatives were not addressed becausethey were not considered ripe for reviewFaced with the lack of support from thedivision and the potential of a privateantitrust suit, PULSE admitted First Texas.The impact on intersystem competitionwas immediate; within six tnonths of thebusiness review letter, practically everyMPACT member joined PULSE. MPACTeliminated its incentive pricing. There wasa similar effect on consumers, as severalbanks increased their consumer fees.

The States intervene—The EntreeLose

Because of the division’s inaction,attention to intersystem competition issuesseemed dormant and ATM networkconsolidation seemed uncontroversial.This trend changed in the late 1980s withthe challenge by state attorneys general(the states) to the formation of the Entreenational point of sale (POS) joint venturebetween VISA and MasterCard.” At thetime, POS was in its infancy and was per-ceived as a competing (and perhapssuperior) technology’ to ATM networksand credit cards. VISA and MasterCardhad informed the division of the formationof Entree, hut no enforcement actionwas taken,

The states alleged that VISA andMasterCard violated the antitrust lawsthrough the formation of the Entree POSdebit program, their respective acquisitionsof interests in PLUS and CIRRUS (thenational ATM networks), and VISA’s acqui-sition of lnterlink, a California POSnetwork, The states alleged that byforming Entree and acquiring the ATMnetworks, VISA and MasterCard sought toretard the development of on-line P05

“Section 1 of the Sherman Act,15 U.S.C. § 1(1988), pro-hibits certain restmaints of trade.Courts have held that thedenial of membership in a jointventure mnoy violate Section 1 -

See Nocthwest WholesaleStationecs, Inc. mc PacificStall onecy & Pcin tiny Co., 472U.S. 284 (1985).

‘‘Up until that time both net-works were exclusive. If FirstTexas was a member of bothnetworks, it would serve as agateway and could eruoble anybank inane network to occessthe ATMs in the ather network.Dace the exclusiviti provisionswere bridged, arguably intersys-tem competition between thetwo netwarks would diminish.

“See letter fran William F.Baxter, Assistant AttorneyGeneral, Antitrust Division,to Donald I. Baker(Aug. 3,1983).

“See State of New York v. YISA,U.S.A. and /d,ostecfocd IntlNa. 89-Civ,-SD43 (S.D.N.y.filed July 26, 1989).

FEDRBAL RESERVS RANK OF ST. LOUES

21

DFYI[~NOVEMBER/DECEMBER I 995

“See State ofNew Yock mc VISAU.S.A., Inc., 1 99D-1 Trade (as.(CCH) 169,D16 (S.D.N.y.199D).

‘‘See Constantine (199D).

‘‘See “Bankers are Buryingthe Hatchet” (1994);Balto (1995).

“The switch fee is the feecharged by the network formoving a transaction overthe network’s switch. Theinterchonge fee is a fee paidbetween the nerchont bankand the cardholder’s (con-sumer’s) bank for processing acredit card or debit cord transoc-tion. Bath fees are set by thebankcard association.

See “Bank of America’(1994); “Debit Card War’(1994); ‘lcaaamics—MareIssuers” (1994) for a descrip-tion of campetiton ininterchange fees.

debit, a payment system they feared wouldcompete with and erode the profitability ofcredit cards. Entree, the states alleged, wasa combination of the five most likelyentrants into the POS market, The statesfurther alleged that as part of the joint ven-ture, MasterCard and VISA had agreed notto introduce their own separate systems tocompete with Entree. As parc of their alle-gations, the states challenged provisions inthe Entree agreement that limited its mem-bership to banks that were members ofboth associations, thereby excludingnonbanks such as Sears/Discover Card andAmerican Express.

The complaint sought divestiture ofCIRRUS (by MasterCard) and PLUS andInterlink (by VISA), as well as aninjunction against the implementation ofEntree. In T990, VISA and MasterCardagreed to abandon the Entree jointventure.” VISA kept its ownership ofTnterlink, and both card associations werepermitted to keep their interests in thenational ATM networks.”

Although arguments about theeconomics of ubiquity may have been per-suasive in other contexts, they did notpersuade the state attorneys generalinvolved in the Entree case, One couldargue that a single national POS networkwould have offered the opportunity forgreater customer convenience by puttingall of the POS terminals in a singlenetwork. Similarly aggregating all of thecardholders in a single network may havepersuaded merchants to use the new POSnetwork. A single network may have fos-tered development of the new technologyBut these arguments were unavailing. Thestates recognized that even if a single net-work might present some of theseefficiencies, they were outweighed by thepotential loss of competition betweencompeting P05 networks.

Five years after the settlement itappears that the states’ assessment wascorrect. After the settlement, VISA andMasterCard created their own independentPOS programs (Interlink and Maestro,respectively). In response to the concernsof the states, each of the national POS net-

works adopted anti-duality rules, whichprevent any bank member from belongingto a competing network. Competitionbetween the networks, in terms of productpromotion, product development, andpricing, has been aggressive and far moresignificant than that in the credit cardmarket, where duality is permitted.”

Each of the networks has competedvigorously to sign up both banks and mer-chants. Both networks have adopteddifferent switch and interchange fees, tomake more attractive packages forconsumers.” The fees charged by the net-works, including interchange fees, are farless than those charged by credit cardnetworks.” lnterlink charged additionalannual card service fees and merchantlocation fees. When Maestro entered, itdid not charge these fees, Of particularsignificance, Interlink initially charged atransaction servtce fee of $0.02 for everytransaction conducted by an Tnterlinkcardholder at an lnterlink terminal even ifthe transaction was actually processedthrough a regional network (in otherwords, if the bank attempted to bypass theInterlink network). Maestro enteredwithout such a bypass fee, and its entryforced Interlink to eliminate the fee.

In April T994, Maestro sought to elim-inate its anti-duality rule to permit issuerduality After considering the proposal forseveral months, the states rejected it inDecember T994, The states observed thatboth networks were competing aggressivelyand that the networks appeared to hethriving in terms of transaction volumesand merchant participation. Moreover,unlike other payment system markets,competition from nonbanking participants,such as Discover Card or AmericanExpress, was unlikely because debit cardservices are necessarily linked to afinancial institution’s demand depositaccount. Most important was the states’concern that eliminating Maestro’s anti-duality rule -‘would bring to an end theaggressive intersystem competitionbetween the two bankcard associations” inthe P05 market. Thus the states concludedthat they could not assure Maestro that

FEDERAl. RESERVE BANK OF ST. LOUIS

22

DFYI[~NOVEMBER/DECEMBER 1995

elitnination of their anti-duality rule wouldnot lead to an enforcement action.”

For the states, abstract argumentsabout efficiencies were simply a guise todeter the emergence of intersystem compe-tition. Their enforcement action led toincreased intersystem competition andconcomitant benefits for consumers.As important, the Entree case began toaffect how regulators and enforcementagencies assessed the opportunities fornetwork competition.

The I9’YLis-’-”kene’wed Atlerptk’n tofyfr’~,uc~rl.Carnpe4itian

Exclusive processing rule challenged—MAC ATM network settlement. Thereemergence of the division in the paymentsystem competition venue occurred inApril 1994, when the division challengedthe exclusivity rules of the MAC ATM net-work. In the six years since the divisiontook a pass on the Cashstream acquisition,MAC had acquired almost all of irs neigh-boring competing networks, had amonopoly in several mid-Atlantic statesand had become the largest ATM networkin the United States. At issue at this pointwas not a merger, hut rather certain exclu-sivity arrangements that VAC used toenforce its monopoly position. Thedivision challenged these restrictions asillegal tying and monopolization, undersections 1 and 2 of the Sherman Act.”

To understand the action, we setl’orward the different functions of an ATMnetwork. In its most basic sense, an ATMnetwork comprises a trademark, acomputer switch, and a set of rules. Somenetworks have their own computer systemthat drives the computer switch; other net-works contract for that service, Somenetworks engage in “processing,” that is,they drive (operate) their members’ ATMs;other necworks permit their members todrive their own ATMs or use third-partyprocessors, such as EDS Corp. Thismarket for “ATM processing” was thefocus of the division’s enforceunent action.At the time of the enfoi-cement action,Electronic Paymnent Services (EPS), which

operates the MAC network, was a jointventure of four bank holding counpanies:CoreStates Financial Corp., Banc OneCorp., PNC Bank Corp., and Society Corp.The MAC network has approximately a90 percent market share in Pennsylvaniaand a dominant position in adjacent mid-Atlantic states. The MAC network handles92 unillion transactions each month for27 million depositors at more than13,000 ATMs.

Most ATM networks are nonexclusive,that is, they permit their members tobelong to any of a number of networks.Until 1992, MAC generally did not permitits bank members to participate in rivalATM networks. These exclusivity rulescreated an almost impervious barrier tocompetitive entry because if a bankwanted to join a competing network itwould have to withdraw all of its ATMsfrom MAC. Faced with that all-or-nothingdecision, few banks chose to align withcompeting networks.” The rules helpedMAC acquire and maintain its dominantposition in the market. The rule againstmultiple affiliations was formally droppedin 1992 after being challenged in a privateantitrust suit.”

In this case, the division’s focus was onother rules which restricted the ability ofbanks to participate in other networks oruse competing third-party’ processors. Thedivision alleged that a rule that requiredbanks either to obtain ATM processingfrom MAC or to provide ATM processingin-house (which is prohibitively expensivefor many smaller banks, thrifts, and credit.unions) effectively made it impossible forthese smaller banks to belong to rival net-works while belonging to MAC. MACgenerally forbade its network meunhersfrom obtaining ATM driving from any ofthe several third-party’ processing firmsthat provided that service.

The MAC rules and practices, thecomplaint alleged, “prevent willing buyersand sellers from conducting business atcompetitively determined prices andterms.” By preventing banks frounobtaining ATM processing from others,MAC effectively prevented these banks

“See “State Antitrust Dificials’(1994).

“ United States v. ElectronicPayments Services, Inc.,No. 942D8 (D. Del. Apr. 21,1994), 59 Fed. Reg. 24,711(May 12, 1994), 59 Fed.keg. 44,757 (Dct. 14, 1994).

“As the division observed, ‘Thesmoll banks that wish to joinonother network (which mightoffer ATM network access atIawer prices) will nat be able todo so unless the other networkhas enough of o presence topravide small banks’ depositorswith sufficient ubiquity and con-venience. The entrant network,of course, cannot achieve thecriticot mass necessary toattroct banks.’ Elec. PaymentSen’s., 59 Fed. keg. 24,711,24,72D.

“See BuyPass ficp. mc New YorkSwitch Corp., No. 93-CV-32D1(ED. Pa. filed June 15, 1993).The rule lund survived a privateantitrust challenge, when PM(acquired Cashstream in 1988.See The Treasurer, Inc mcPhilodelphio NotionalBank,682 F. Supp. 269, 28D(D.N.J.) (upholding exclusivitypravisions which “were and areintended to structure [the own-er’si distcibuton of networkservices, ond to provide areturn.., far developing, main-taining and promoting the net-work and to prevent free ridingby competitors),’ off’d nem.,853 12d 921 (3d fir. 1988).Df course, in 1988, At~Chadfar less significant competitivepresence than it did in 1994.

FEDERAL RESERVE RANK OF ST. LOUIS

23

D1~Itt~NOVEMBER/DECEMBER 1995

“A royalty fee requires the ATMowner to pay a fee to the net-work for each tronsactan itchose to route through anolternatve network.

“See Public Comments anProposed Final Judgment,Electronic Payment Services,59 Fed. Reg. 44,757 (Aug.3D, 1994).

“See footnotes 6T69 andnccomponying text.

“First Data Corp., FTC File Na.951-DiD! (Sept. 21, 1995).The FTC brought on earlieraction against First Data inAugust 1994, when it intendedto bid on the assets of WesternUnian in o bankruptcy courtauction. First Financial was thesuccessful bidder and the FTC’ssettlement was never madefinal. First Data Carp., FTC FileNa. 931-DD9D (Aug. 18,1994).

“Consumer money transferservices involve the tronsferbetween twa parties af fundsthrough consumer money trans-fer agents, typically checkcasluirrg, private postal, or grocerystores. Customers wishing totronsfer money today begin theprocess by going to a consumermoney transfer agent, such aso check rasher or grocery store,and completing a transuctianform, which includes an expla-nation of how the recipient willidentify himself ar herself whenreceiving the rash. The senderthen gives the agent themoney to be tronsferred mud

from participating in other ATM networks.In turn, MAC’s rules made it substantiallymore difficult for other networks to enterinto MAC’s area of dominance, therebyexcluding competitors and maintainingMAC’s monopoly position.

The division alleged that regional ATMnetwork access and ATM processing wereseparate products and that MAC’s rulesand practices effectively’ forced itscustomers to purchase ATM processingfrom MAC. The monopolization claimalleged that MAC “w-illfully hasmaintained its monopoly power in themarket for regional ATM network access inthe affected states through exclusionarypractices,” including its processing rule.

The consent decree requires MAC toopen its network to independent ATMprocessors on a nondiscriminatory basis.MAC is prohibited from tying the use of itstrademark to the purchase of processingservices. Under it, MAC must permit itsparticipants to use third-party providers ofATM processing, to display multiplenetwork trademarks on all their ATMs,and to permit multiple branding of ATMcards issued by MAC members in areaswhere MAC has or could soon haveunarket power.

The objective of the decree is toprovide banks with the opportunity to useother networks or third-party processorsfor their processing services. MAC is alsorequired to sell its network services -‘atprices that will not vary with the processselected” and to provide a more open envi-ronment for third-party processors. Inaddition, MAC would he limited in theextent to which it can keep banks fromdisplaying symbols of other ATM networkson their ATMs and ATM cards.

The decree perunirs a wide range ofother activities that may’ raise exclusionary’concerns. First, MAC is permitted tocharge a royalty fee for transactionsprocessed outside the MAC switch.” Thisroyalty fee can hc as much as the fee for atransaction processed through the MACswitch. Second, MAC can prohibit itsmembers from hy’passing the switch, apractice known as subswicching. Third,

MAC is permitted to provide volumediscounts, but these must be provided on anondiscriminatory basis.

Whether the decree adequately solvedthe competitive problem is an open ques-tion. The consent decree received atremendous amount of adverse commentary;many competing networks stated that theproposed decree would permit MAC toachieve the same objective through avariety of other types of exclusionary con-duct.” In addition, as described later, theBoard staff raised concerns over the suffi-ciency of the relief when it examined theEPS-National City Bank merger.”

The division’s enforcement actiondemonstrated that the economics of ubiquityno longer rule the day The division wasable to do that by separating ATM servicesinto two separate product markets: ATMprocessing (or the back office operations)and branded regional ATM access (whichreflects the value of membership in thenetwork and the network mark). As thedivision observed, ATM processing can beprovided as a service distinct from brandedATM network access and can he perforunedin the facilities of the ATM switch, adepositor)’ institution’s own facilities, orin the facilities of a data processingservice organization.

Of course the irony here is that hadthe division not signed on to theeconomics of ubiquity’ bandwagon, andhad examined the nature of network coun-petition more carefully it may havechallenged the earlier acquisitions byMAC, and ultimately this enforcementaction ma)’ have been unnecessary

Payment systems merger challenge—consumer money transfer services. Theonly enforcement action brought against apayment systems merger was the challengeby the l’ederal Trade Commission (FTC)to the acquisition of the Western Unionconsumer unoney transfer system (ownedby First i’inancial Management Corp.) byFirst Data Corp., the owner of the Money-Gram svstem—in re First Deco Corp. (FirstData).” Consumer wire money transfersystems involve one-way’ money transfers,

FEDERAL RESERVE BAE,IK OF ST. LOUIS

24

DtYl1~NQVEMSRR/DECEMREB 1995

typically between two consumers.” Wiretransfer agents include a wide variety ofretail outlets including grocery stores andcheck cashing outlets.

Western Union has been the dominantfirm in the market and had been a regulatedmonopoly until the late I 970s. TheFederal Communications Commission(FCC) had deregulated Western Unionbased on the expectation that technologicaladvancement had reduced the baa-riers toentry” Those expectations were overlygenerous; entry was neither easynor timely

In the mid-1980s, Citibank attemptedto enter the market, hut their entrywas stifled by two factors: (1) developinga minimuun viable scale nationwidenetwork of money transfer agents; and(2) establishing name recognition and cus-tomer acceptance of its services throughlarge-scale advertising and promotion.Long-term agent contracts used byWestern Union made acquiring a sufficientagent network difficult. To build brandname recognition, substantial investmentwould be required over several years.Citibank’s attempted entry failed after sev-eral years of significant losses.”

MoneyGram, which was originallyowned by American Express, entered inthe late l980s. It was able to overcomethese barriers in part because it could relyon the trade name and the agent base ofAmerican Express. After several years oflosses, MoneyGram overcaune the harriersto entry and introduced competition intoan environment in which a monopolisthad dictated annual price increases.

Competition from MoneyGram led tolower prices, better services, and highercommissions for agents. Prior to Money-Gram’s entry Western Union imposedregular annual price increases of 5 percentto 8 percent. MoneyGram entered by’ com-peting aggressively on price; WesternUnion responded by’ iefraining froun priceincreases and offering special promotionsand discounts to customers.” In 1994,MoneyGram launched a frequent user dis-count program to increase sales andcustomer loyalty; Western Union

responded with a similar program. Non-price competition increased, includingincreased price advertising, the developmentof a more extensive will-call system, andfree long-distance telephone calls.

Competition between the twonetworks also led to almost a threefoldincrease in wire transfer agents, which pro-vided consumers with a dramatic increaseof convenience when using money transferservices. As both networks competed foragents, agent commissions increased, thenetworks provided greater amounts ofcash at more agent locations, and thenetworks increased their advertising.Competition created these consumerbenefits indirectly by pushing the compa-nies to pay their transfer agents highercommissions and significant bonuses forincreasing customer volume.

At the time of the FTC’s action,Western Union had approximately a90 percent market share. According to thecomplaint, MoneyGram and WesternUnion were the only two services in theU.S. consumer money transfer market andit would he difficult for new companies toenter the market, The complaint notedthat First Data’s acquisition of WesternUnion would create a monopoly in themarket. Further, the FTC contended thatentry was unlikely because of the difficultyof gaining brand name recognition andestablishing a nationwide network of retailoutlets. Thus absent the settlement, theFTC alleged that the acquisition wouldincrease the likelihood that, among otherthings, consumers would be forced to payhigher fees and receive less service andagents would be forced to acceptreduced coununissions.

The proposed consent agreement per-mits First Data to acquire Western Unioamas long as it divests either the MoneyGramor Western Union consumer money wiretransfer business. The divestiture packageincludes the MoneyGram or WesternUnion trade name, contracts \vith asufficient numher of retail sales agents tohave a minimum viable scale network, andenough other necessary’ assets to runthe husiness. The settlement also includes

pays the transaction fee. Thetransferring agent then inputsthe information into the database by computer (or by callingthe service supplier, who inputsthe information). This data-base allows the receiving cus-tomer to go to any receivingogent in that service’s agentnetwork and obtain the cashwith proof of identity.

A large portion of consumermoney transfer users do nothave banking relationships;they account for 2D percent to25 percent of U.S. households.For those consumers with limitedor nonexistent banking relation-ships, consumer moneytransfers offer the only meansto tronsfer money quickly fromone person ta another.

“See Graphnet Systems, Inc.,71 F.C.C. 2d 471,515(1979) (FCC observed that,‘We are confident that thepublic will be served by enablingmultiple entry into thismarket.’).

“See ‘Citicorp Express” (1987).

“ When First Data entered, itpriced domestic transfers ofS3DD or lesson $9; ot thetime western Union prrceithese tronsiers at between$13 and $29. Western Unionbrought an antitrust suit chargingthat First Data’s pricing waspredatory. The suit was unsuc-cessful. See Western UnionFinancial Services mc FirstData Carp., 2D Cal. App. 4th153D (1993).

FEDERAL RESERVE BANK OF ST. LOUIS

25

II [~IF~NOVEMRER/DECRMEER 1995

“15 U.S.C. § 18(1988).

Notional Boncard Carp.(‘Nailanco”) v. VISA U.S.A.,Inc., 596 F. Supp. 1231,1258 (S.D. Fla. 1984), aff’d,779 F.2d 592(11th Cir.), cect.denied, 479 U.S. 923(1986). Baxter (1977)argued that ‘no significantdegree of market power’ willexist in that strotum of commu-nities serviced by only one on-line system because ‘tIrepreexisting technology,” definedto include ‘currency, checks,off-line credit cards and checkguarantee cords subfect to floorlimits,” ‘will constrain pricingfreedom and service quality.’The Wailanca analysis is criti-cized in a recent article. SeeCarlton and Frankel (1995).It, however, was adapted by arecent district court decisioninvolving ATM network fee set-ting. See South trust Corp. mcPLUS System, bc, No. CV-93-P-2291-S (N.D. Alo.Aug. 1D, 1995).

The Treasurer, IncPhiladelphia Nailanal Bank,682 F. Supp. 269 (D.N.J.),off’dmem., 853 E2d 921 (3dCir. 1988)

“ 682 F. Supp. at 279 (empha-sis in original).

“For example, see, CB&TBancshares, Inc. (1984);Atlantic Bancorparation(1983); cf. Centerrelancarparotion (1983) TheBoard observed that the prod-uct market was ‘the provisionto unaffiliated financial institu-tions of data processing ser-vices, particularly the operationof an ATM network exchange”;Interstate Finonciol Carp.(1983) (same).

In other orders, it defined themarkets more narrawly. Forexample, see Citicorp (1986)(‘provision of ATM services’);Sovran Financial Carp. (1986)

various provisions designed to ensure thatthere would be an agent network sufficientto support the divested business. Finallythe settlement expressly permits First Datato provide data processing services to theacquirer of the MoneyGram or the WesternUnion assets, provided that First Data,among other things, shields any nonpublicinformation it receives from any First Dataemployees who are involved in First Data’sconsumer money wire transfer.

The importance of the First Dataaction was in differentiating between theimportance of the back office or systemsoperation, and the agent network andtrade name. Like che FCC, the FTC didnot contend that the back office operationposed an entry barrier. However, the yearsof experience gained since the FCCdecision had shown that ease of entry atthe back office level would not guarantee acompetitive market. Thus the proposedconsent order does not require the divesti-ture of the back office system and in factpermits First Data to provide back officeservices to the acquirer of the divestedassets. Rather, the FTC focused its reliefon the trade name and agent network,which it contended were the most signifi-cant harriers to entry

ESALUATING. MAP7METPOWER IN PAYMENTSYSTEMS CASES

Antitrust analysis examines the effectsof mergers on competition. The purposeof this analysis is to determine whether theeffect of an acquisition “may he stnhstantiallyto lessen competition or to tend to create amonopoly” Such analysis involves iden-tifying the relevant product sold by thefirms and the geographic scope of marketsin which they sell their products. Thissection discusses the nature of definingmarkets and assessing market power forpayment systems.

Market Deflru’tIon Issues

Antitrust analysis of payment systemmergers or other competitive activity

depends critically on whether the systemhas market power. This is typically a diffi-cult question to answer in part because thedelineation of relevant markets is itself acomplex and uncertain undertaking. Thedefinition of the relevant market has bothproduct and geographic marketcomponents. Tn both respects, the marketsdefined have become more precise andnarrow over time.

Product market definition. One ofthe uncertainties in counseling paymentsystems is traceable to the difficultiesin defining the relevant product marketfor purposes of measuring marketpower. Many different approacheshave been used. Product market definitionhas become more precise as regulatorshave become more sensitive to thecompetitive problems raised by networkcompetition. Tn particular, both thedivision and the Board have begun to dif-ferentiate between the back office andtrademark aspects of a network in definingthe market. Typically fact finders definethe product market from the perspective ofthe cardholder (the retail market) and thecard issuing bank (the wholesale market).

A payment systems market. One ofthe earliest cases, NaBanca, involveda challenge to a credit card interchangefee. The district court defined a broadretail market consisting of all paymentsystems, which it defined further as:

a market consisting of VISAand all payment services usedin retail sales. This marketincludes VISA, MasterCard,T & E cards, merchants’proprietary cards, merchants’open book credit, cash,travelers cheques, ATM cards,personal checks and checkguarantee cards.”

The court acknowledged that none ofthese was a perfect substitute hut relied onan exannination of cross-elasticities ofsupply and demand to determine that they

FEDREAL RESERVE BANK OF ST. LOUIS

26

HEVIEWN*VIMSII/SIflafla S05

were sufficiently close substitutes for theVISA card.

A data processing market. In terms of awholesale market, in early cases factfinders emphasized the data processingfunctions of ATM networks. For example,in The Treasurer, the district court adopteda broad definition of the relevant productmarket.” That case involved a challengeby The Treasurer ATM network in NewJersey to the acquisition of the Cashstreamnetwork by Philadelphia National Bank,the owner and operator of the MACnetwork. Although he ultimatelydismissed the case for lack of antitrustinjury Judge Politan also examined thecase on the merits. In so doing, he definedthe relevant product market as “electronicdata processing to all AIMs plus all ofthose institutions that have unaffiliatedATM systems and those institutionsthatdo not currently have AIMs but have thecapacity to install them and use markettechnology to its fullest.”M In other words,the market included all finns capable ofperforming the electronic communicationfunction performed by an ATM network.

Similarly, in the 1980s, in ordersapproving bank holding companies’ acqui-sitionsof voting stock in shared Efinetworks, the Federal Reserve Board typi-cally defined the relevant market as “theprovision of data processing services tounaffihiated financial institutions.”” Inaddition, the Board noted that the marketfor data processing and related ATMservices is “unconcentrated, with manycompetitorsand few barriers to entry”M

An ATM services and network switchingmarket. In more recent decisions andenforcement actions, fact finders havedefined more narrow markets, focusingprimarily on demandside factors. Forexample, in the Financial Interchange arbi-tration, which involved ATM networkinterchange fees, the arbitrator rejectedproposed markets of all payment systemsand all means ofobtaining cash, similar tothe approach taken by the Board and thecourts in NaBanco or The Treasurer.

Instead, it identified a narrow retail marketof “ATM services” on the grounds “thatthere is a significant group of ATM userswho value the characteristics of ATMs andfor whom other means of obtaining cashare not reasonable substitutes.”7

In addition, in Financial Interchange,the arbitrator identified a wholesalemarket for network switching, andconcluded that PULSE had market powerbecause “existing subnetworks, regionalnetworks and national networks do notpresently provide a reasonable substitutefor the [switchingi service PULSE providesto its members.””

In the EPS consent decree, theantitrust division tooka similar approach,albeit focusing on the wholesale side of themarket.’~First, itdefined a market forregional ATM service, based on the needsof banks to provide depositors “ubiquitousaccess to their accounts.” Itobserved that“while a bank can deploy its ownATMs,the advantage to a shared ATM network isthat a banks depositors will be able to useATMs at many more locations thanonebank alone could practicably support. Theareas a bank seeks to serve through ashared ATM network include the areas inwhich its depositors live, work and shop,and the broader areas in which they moveregularly. A banks ability to offer itsdepositors access to other banks’ ATMs,and thereby to offer its depositorsconvenient access to their accounts, is inmost bankers’ view necessary to attractand retain deposits.... Because no otherservice constitutes a reasonably close sub-stitute for regional ATM network access,regional ATM networks constitutes aproduct market.”’°

Similarly it defined a second marketfor ATM processing. This marketinvolvesproviding the data processingservices and telecommunicationsfacilitiesand services used in providing regionalATM access.4’

Network access, network services, and ATMprocessing. In its analysis of the EPS-National City Bank merger (hereinafterBane One Corp.), the Federal Reserve

(sowe); lideys Bwk PLC(1985) rcmi,effiiflthe provWon dATM orMseatesl.

MFAJ exm* Sonmi FkmcidCmp. (1986).

“hi ,eASoIko beMceea F1~tliesLw Ass’n & FkoS—~it, 55tSReg.Rep.(ONfl4O,356 ~. 25,1988)(FWKS biSnige) (aholimi deSie by ftoiessor

mtdwd&~mi

91.ot3S5. 0IIwre~oSnePwetwere fotmd bbs mdy— dwm*s’es forIexinAIM owners mid stishmildhiSs fnSr.j the noboSMwet’— mS~*S,dwrsSerkebyhm*slorhcdoeiwot,S

weSSthed)were S to iiçe4e cooçe*SfrflnoboSnefwet, PWSSORIUS. Idot353~54.Forosiróx

—. ~k acm note 4.fl144 (osseSgAlMnet’works ii Sn ~SesSemidSAiMsenkes).

~S~Kft~tSm*ss&,No.94208 ffl.0etAp.2~1994),5PFed.Reg.24711~oy12,1994).

•S9Fed.Iteg.24713(Moyl2,1994).

M9Fe&Reg.24712(Moyl2,1994).

filiAl tilliVi PANE •l ST. LOVES

r

OtYIt~NOVIMRiR/DECIMOER 2995

92 Banc One Carp. (1995), pp.

491, 494.

See NoBanco, 596 F Supp. at1259 (where the panicsogreed thotthe market wasnationwide); see alsaComplaint, ‘]fl] 77-80 in NewYork v. VISA USA., Inc., Na.89{iw5043 (S.D.N.Y. July26, 1939) (the states allegeda nabanwide morket for creditcards and point of sole debitcards marketed by nananal]ointventure).

See Financial Interchange, 55Trade keg. Rep. (BNA), No,1380, pp. 356 (although thegeogrophic boundaries of theretail market were not directlyaddressed in this proceeding,retail markets are presumablylocal because consumers willuse only ATMs close to wherethey are).

See fPScarrsent decree, 59Federal Register 24,711; BoorOne Corp. (1995), pp. 494.See Mchndrews and Kauffman(1993).

Banc One Corp. (1995),p.494.

Board further refined the divisionapproach by defining three markets:(1) network access (access to an ATMnetwork identified by a commontrademark or logo displayed on AIMs andATM cards); (2) network services (theswitching functions for the network); and(3) ATM processing (the data processingand telecommunications facilities usedto operate, monitor, and support ahank~ATMs).42

According to the Board, network accessincludes: (1) the right to brand ATMs andATM cards with the trademark or logo ofthe ATM network; (2) the ability of theATM cardholder with an account at onemember depositor)’ institution to initiatewithdrawal arid other account transactionsat an ATM owned by another depositoryinstitution that is a member of the samenetwork; and (3) minimum standards fornetwork performance and products offeredthrough the network.

Similarly the Board defined networkservices as including the switchingfunctions performed by the ATM switchand gateway services with other networks.Finally the Board defined ATM processingas including the provision of terminal dri-ving, transaction routing and authorization,and account reconciliation services.

An obsen’cuion. How a fact finder analyzesthe relevant product market in casesinvolving hank networks depends in parton how much weight is accorded to thevalue of the network trademark. If onelooks only to the data processing functionof shared ATM networks, it may beplausible to conclude, as did the frcastircrcourt, that the data processing industry isunconcentrated, that there are numerousalternatives available to financialinstitutions to perfortn their dataprocessing, and that a network—even adominant regional network—does nothave market power. On the other hand, ifthe network is viewed not so much as avendor of undifferentiated data processingservices, hut rather as the purveyor of aunique branded product marketed underthe network logo, the fact finder may reach

a very different conclusion, as in Bane OneCorp. or Financial Interchange.

In this respect, the Board’s decision inBane One Corp. is a significant analyticaladvancement. By identifying a “networkaccess” market which focuses on the“branded product” aspect of the network,the Board’s decision provides a mechanismfor more careful and precise analysis ofmarket po\ver.

Geographic market definition. The geo-graphic market can be defined only withreference to a specific product or servicemarket, and there are uncertainties here aswell. Markets have been defined asnational, regional. or local depending onthe product market selected.

For example, early court opinions thataddressed the geographic market applicableto a payment systems market suggestedthat it is national:” If the focus of a factfinder is a product market defined in termsof data processing for unaffiliaced institu-tions or network switching services, thegeographic market should be nationalbecause those services are generallyprovided on a national basis. On the otherhand, in cases such as Financial Interchange,whoch focused on a retail market. the geo-graphic market was assumed to be localin scope.~

The most recent decisions havedefined ATM networks as participating inregional markets~’In Bane One Corp., theBoard observed that most networks wereregional in scope (that is, consisting ofseveral states), and a Federal Reservestudy found that the markets for ATMnetwork access were at least regional.’~The Board decided that the appropriategeographic market in which to analyzethe competitive effects of the mergerwas MAC~Mideast Region (westernPennsylvania, Ohio, Indiana, Kentuckyand West Virginia), where National Cityhad a competitive presence.

In Bane One Corp., the Board alsoseems to suggest that in some cases thegeographic market may he national inscopei The Board observed that compa-nies are able to provide A[M processing

HOiRAL flSlRVi BANK OF ST. LOUIS

28

IIFyIu~NOVEMSEE/DScSMBER 1995

and network services through dataprocessing facilities regardless ofgeographic proximity and that somefirms provide these services on anationwide basis.

One issue that arises in ATM cases is~5~hethernational ATM networks (for

example, PLUS and CIRRUS) competewith regional networks, In the FinancialInterchange arbitration, the arbitrator heldthat national ATM networks did notprovide an adequate alternative to PULSEbecause neither could duplicate thecoverage of the PULSE network. Theantitrust division in the EPS consentdecree has taken a skeptical position aboutthe level of competition offered by nationalnetworks. In its Competitive Impact State-ment it observed the following:

National ATM networks exist, butthese are by design networhs of lastresort, used only where the twobanks involved in a transaction donot both belong to any one regionalATM network. National ATM net-work transactions are typically moreexpensive, and those networks pro-vide only a subset of the transactionsavailable through regional ATMnetworks 48

Meosun’na Market Power

There is relatively little guidance asto what statistical base should be usedas a surrogate for measuring the powerof a particular network. In Financial Inter-change, the arbitrator variotosly examinedthe share of all ATM transactions (which“understate]d the venture~]position in themarket”), the share of interprocessorswitching transactions, the share of avail-able AIMs, and the cardholder base.~’

In The Treasureo; Inc. rc PhiladelphiaNat’l Bank, the court suggested that marketpower should be measured by the nunuherof AIMs.” it wrote that “the principal

competitive advantage of any ATMnetwork is the number of AIMs utilized bythe systean.” The court also examined

financial institution deposits in holdingthat measurement of the market cannotbe confined to network ATMs, but musttake account of the large number ofunaffihated ATMs that are open territoryfor competition.”

Other possible measures for assessingmarket power include the number of ATMlocations (as distinct from number of ATMmachines), the value of ATM transactions(as distinct from number of transactions),the number of member institutions, andthe value of retail deposits accessible byATM. The interpretation of any statisticalmeasure must be tempered by therecognition that ATMs, cardholders, andinstitutions may have simultaneous accessto multiple networks. Ultimately, in thebank network context, statistical anarketshare evidence—at least in terms of a shareof ATM transactions—may be an imperfectmeasure of market power. Because of theavailability of alternative networks, histor-ical market share may overstate the marketpower of a network. Yet because of thedifficulty competing networks may haveacquiring the necessary critical mass,market shares may tend to understatemarket power. Similarly because of thesignificance of entry barriers in the ATMaccess market, market shares will alsounderstate market power. Thus a factfinder must exercise caution before relyingon any individual statistical measure.’’

Aria!ysis of Entry BarriersEssential to the analysis of market

power in payment system cases is consid-eration of the existence of entry harriers.Where entry is “easy,” ii. is difficult for anetwork to raise prices or reduce outputsince that exercise will lead new firms toenter the market and cease the competitiveopportunity. According to the antitrustdivision and the FTC entry is “easy” onlyif it would he timely, likely and sufficientin magnitude to counteract the competitiveeffects of concern.

In the network environment analysisof entry becomes more complex because ofthe critical mass nature of networks. .A

“59 Fed. Reg. 24719 (May 12,19941. The Federai ReserveBoard has taken a similorpasiflan. Banc One Carp.,p.494, n.21.

55 Trade Rca. Rep. (BNA), No.1380, pp. 353, 356.

“682 F Sapp. 269 (O.NJ.),off’dmom.,853F2d 921 (3dCir. 1988).

‘Id. p. 279.

‘lid.

“See Blamenthai (1989).

FIDIRAL RBSSRVE SANK 01 ST. LOUIS

29

ll[YIF~NOVSMBSR/DECFMSSR I 995

(PS, 59 Fed. Reg., p. 24,720.

network may not be able to effectivelyenter unless it acquires a sufficient numberof participants to offer a viable product.This poses a “chicken and egg” problem;potential members are reluctant to joinunless they are assured that a sufficientnutuher of other firms will join to makethe network viable.

Moreover, network externalities mayalso impose significant entry barriers.ATM networks provide an example thatillustrates the difficulty a challenger facesin duplicating the network externality ofan incumbent firm. ATM networks exhibita positive externality: large networks yieldincreased convenience to consumers, thusincreasing the network’s value to the con-sumer. Thus a new network is unlikely tosucceed unless it can demonstrate that asubstantial number of transactions andcardholders within the market will beavailable on a long-term basis. Effectiveentry requires that a new ATM networkoffer the same (or better) convenience andubiquity offered by the incumbentnetwork. As the division observed in theEPS competitive impact statement, inorder to be competitive, a network mustprovide “enough of a presence to provide[their] depositors with sufficient ubiquityand convenience.””

As in the analysis of relevant productmarket, the analysis of entry barriers in thenetwork context has varied significantlyOne approach, which focuses on competi-tion at the “back office” level, has been tosuggest that entry can be accomplishedrelatively easily For example, in The Trea-surer, the court focused on competition inproviding automated data processing ser-vices to banks. In this market there were anumber of potential entrants includingthird-party processors, regional andnational ATM networks. Of course, TheTreasurer was decided in 1988, in a contextin which there were large nuanbers ofbanks that were unaffiliated with any net-work and in which no network wasdominant. Thus, the potential for a newnetwork to arise and compete with MACwas far more significant than it is today

A more sophisticated approach to

analysis of entry was provided by the arbi-trator in the Financial Interchange matter.The PULSE network argued that barriersto entry might not be significant. Facedwith the exercise of market power, PULSEsuggested, individual banks could useother networks or form their own quasi-network by bypassing the PULSE networkswitch. Although these opportunities forbypass existed, the arbitrator suggestedthat entry barriers were significant becauseof both network externality and criticalmass factors. Although there was theopportunity for the formation of smallernetworks through individual bypass betweenmember banks, this was insufficient toalleviate the concern over market power.Expert testimony established that a newATM network could not succeed withoutproviding consumers a level of conveniencecomparable with that of the PULSEnetwork. The arbitrator found that a newnetwork could not support the number ofAIMs required to furnish such conveniencewithout achieving “major defections” fromPULSE, and that such defections wereunlikely. These findings ultituately led thearbitrator to conclude that the PULSE net-work did have market power, even thoughthe complainants could have bypassedPULSE and created their own local network.

Analysis of entry harriers is essentialto determining whether networks have theability to exercise market power. Thisanalysis should focus on whether potentialentrants have the ability to attract asufficient number of firms to join a newnetwork and whether that network has theability to deter the exercise of marketpower. This analysis should focus oncompetition at the brand or ATM accesslevel, where network externalities and crit-ical mass play an important role.

/TEM NETWORK MEROEfl—.~~THE LEGAL FRAMEWORK

Since the mid-1980s tremendous con-solidation among ATM networks hasoccurred. The number of regional ATMnetworks has been reduced substantiallyand in relatively few areas is there head-to-

lIbERAL RES~RVB BANK OF ST. LOUIS

30

II FYI [~NOVIMBU/DECEMREB 1995

head competition between networks.Some commentators have predicted theremay be as few as 10 regional networks bythe end of the century” In this section,we discuss the legal framework foranalyzing ATM network mergers.

il-re reqw l~rnni,a,I,c.~neMergers and acquisitions of ATM net-

works may be challenged under either theSherman Act or the Clayton Act by thedivision, state attorneys general, or privateparties. To prevail, the plaintiff mustdemonstrate that the merger or acquisitionmay have a significant adverse effect oncompetition, and this, in turn, requires theplaintiff to prove a relevant product andgeographic market. In addition, a privateparty unlike the government, must alsoprove that the challenged merger or acqui-sition will cause it to suffer antitrustinjury Where the private-party plaintiff isa competitor of the merging parties, thiswill be a difficult burden to satisfy becausethe plaintiff must demonstrate that it willbe injured by higher prices charged by themerging parties.”

Mergers and acquisitions betweenATM networks nnay also require regulatoryapproval. Thus, for example, where thenetwork’s shareholders are bank holdingcompanies, the shareholders typicallynnust receive the approval of the iederalReserve Board (or the relevant FederalReserve Bank) before acquiring anothernetwork.” The parties to a network acqui-sition may also he required to filenotification with the FTC and the antitrustdivision under the Hart-Scott-Rodino Act,although the size of anost network acquisi-tions and the parties making them willusually he below the size thresholds.Other exemptions may also apply undercertain circumstances.’”

To date, there have not been any chal-lenges to ATM network mergers by thedivision, and the Board has declined tostop any mergers. The only decided caseinvolved a private challenge to a regionalATM network merger. In 1988. The Trea-surer network sought a preliminary

injunction to stop the acquisition of theCashstream network by PhiladelphiaNational Bank, which operated the MACNetwork.’” The court dismissed the suiton the grounds that the plaintiff hadsuffered no antitrust injury as requiredunder sections 4 and 16 of the Clayton Actand hence lacked standing to sue. Afterthe acquisition, MAC also acquiredThe Treasurer

—~ ~ ““ ~

Both the antitrust division and theFederal Reserve Board have given renewedattention to ATM network mergers.Reportedly both agencies have investigatedthe NYCE-Yankee 24 and the EPS-NationalCity Bank mergers (discussed later), hutneither has taken any enforcement action.’”

Enforcement officials at the divisionhave provided some guidance about theirnew interest in ATM network mergers.The division no longer adheres to the cate-chistn of economics of ubiquity and is nowsubjecting ATM anergers to much greaterscrutiny Robert Litan, the former antitrustdivision deputy assistant attorney general,said that the division is revisiting theassumption of economics of ubiquity”F1e suggested that the division is notconvinced that ATM networks arenatural monopolies. Rather than taking adoctrinal view in favor of ATM mergers,titan suggested that these mergers willreceive greater scrutiny and that thenetworks would carry a significant burdenof proof. He also observed that theprocompetitive benefits of mergers mighthe acquired through less restrictivealternatives: “Ii] t is very possible that theycan achieve the same economies of scalewithout going to full-scale mergers.”

The EPS consent decree suggests howthe division is likely to consider someissues that arise in ATM network mergers.First, defining the product market is thefirst step in merger analysis. In EPS, thedivision identified separate markets forATM processing and Regional ATM access.Thus the division will look at the competi-

“See The Bankers Raundtable(1994).

“See Brunswick Corp. v. PuebloBawl~O-Mat Inc., 429 U.S.477, 489 (1977); Car/IL Inc

Manfartof Colorado. Inc.,479 U.S. 104 (1986).

“See 12 U.S.C. 1843 (1988);12 C.FR. Parts 225.21-25(1994).

“See 15 U.S.C. 1 Ba (1988);16 C,F.R. Parts 801-803(1994).

Ore Treasuret Inc.Philadelphia National Bank,682 F Sapp. 269 (D.N.J.),aff’drnern., 853 F2d 921(3d Cir. 1988).

“See Bank ol New York Ca.(1994) appraving merger ofYankee 24 and NYCF; BaacOne Carp. (1995) appravingmerger of FPS and NationalCity Bank.

“See “Bank, AIM Mergers’(1994), p.2.

Id.

FEDERAL RESERVE SANK OF ST. LOUIS

31

DFYIF~NOVEMEER/DECEMRER 199S

“59 Fed. Req. 24712 (May 12,

1994).

“Bank af New Yark Ca. (1994).

Id.

tive effects of mergers in both markets.The ATM network access market is likelyto raise more competitive issues becausesome competitors, including third-partyprocessors, do not provide effective alter-natives in that market.

Second, apparently the only competi-tive alternatives in the ATM networkaccess market are regional networks. Thusarguments that other types of networks orprocessors offer competitive alternativesmay not succeed. In particular, nationalnetworks, although they offer a degree ofcoverage comparable to regional networks,are unlikely to be seen as competitivealternatives. The Competitive ImpactStatement in the EPS case noted thatnational ATM networks are “by design net-works of the last resort.”””

Finally exclusivity rules, such as thosechallenged in the EPS case will he an impor-tant part of the analysis; these rules mayprevent the entry of alternative networksinto the market. If many of the availablebanks are committed to long-term exclusivedealing arrangements with a dominantnetwork, an alternative ATM network maybe unable to acquire the critical mass ofbanks necessary to achieve a minimumviable scale. Where these rules are present,antitrust enforcers should be especiallyviligant to ensure that the merger will notprevent the entry of competing networks.

RECENT ATM MERGEROEC1SfONS—REPAVlNGTHE ROAD E.G REGiONALMONOPOLY

The remainder of this article addressesthe decisions of the Federal Reserve Boardin two recent mergers—Yankee 24-NYCLand EPS-National City Bank and the impli-cations of those decisions for futurenetwork competition.

Ya.nkee 24-NUT

A recent network merger that receiveda great deal of scrutiny by both the Board

and the division was the merger of theNYCE and Yankee 24.’” NYCE was thethird largest network in the United Stateswith 95 million transactions monthlymore than 13,000 ATMs and a dominantposition in New York. Yankee 24 was theninth largest network, with 23 milliontransactions and more than 4,000 ATMs,and competed throughout New England.Both networks competed in parts of NewEngland, primarily in Massachusettsand Connecticut.

Even though there was direct competi-tion between the two networks, it did notreceive a great deal of attention in theBoarchv decision. The Board did notaddress the nature of the head-to-headcompetition between the networks or itssignificance. In approving the merger, theBoard did not appear to believe that theloss of competition between the twonetworks would be significant. It observedthat “a number of factors should mitigatethe loss of Yankee 24.. .as an independentcompetitor.”” ln particular, the Boardobserved that other providers of FF1 ser-vices would remain in the market, includingthird-party processors and other regionaland national ATM and POS networks.

Further analysis of the nature of coa’n-petition would have been useful. Forexample, the Board did not discuss oridentify the nature of competition betweenthe two networks. Its observations oncompetitive alternatives also deserved elab-oration. Although third-papty processorsoffer competition in the ATM processingmarket, they do not compete in either thenetwork access or network services mnarket.s.The only other regional network in themarket, MAC, had a coanpetitive presenceonly in New Hampshire. The competitivesignificance of national networks is limited,as noted in the El’s decree. Thus the Board’sreasons for finding there was no significant.loss of cotnpetition seem open to question.

The most interesting aspects of theorder were not the observations about thelevel of current competition, but ratherwhat the Board had to say about themerged networks commitment to an opennetwork structure, the existence of poten-

FEDERAL RESERVE SANK OF ST. LOUIS

32

OFYIEWNOVEMR~R7DEtEMRER 1995

tial efficiencies and how these factors justi-fied the loss of competition.

Operating rules—the importance of anopen network structure. The criticalfactor from the Board’s perspective was thenew operating rules offered by thenetwork, which permitted all non-equitymembers to bypass the network and enterinto arrangements with alternativenetworks or third-party processors. Thenetwork’s operating rules permit: (1)third-party processors to participate in thenetwork; (2) members to participate inother networks; (3) card issuers todetermine routing; and (4) institutions toparticipate on a nondiscriminatory basis.

The first and second of these rulesprovide member banks with possible alter-natives, including processing from thirdparties and ATM switching services fromother networks. The third and fourthrules provide mechanisms by which smallinstitutions can enhance their ability toobtain competitively priced services fromthe network. Of particular importancemay be the card-issuer routing rule, whichwould permit banks to choose lower costnetworks if the merged network attemptedto raise prices.’”

Efficiencies. The Board also found thatthe merger would result in public benefitsthat outweighed any loss of competition.These were primarily in economies ofscope and reduced costs, including:(1) increased transaction volume, whichwould increase economies of scale andreduce costs (primarily in transaction pro-cessing); (2) increased ability to offer POSservices to retailers; and (3) increased con-sumer convenience.

Benc One Ccn-e,----The EPS-NB,tk,nai’Oty Bank Merger

Sometimes networks expand by admit-ting new financial institutions in adjacentareas as owners. One such merger thatreceived a lot of scrutiny by the FederalReserve Board was the application to admitNational City Bank of Ohio as an owner of

EPS; the Board approved the application ina 5-1 vote in March 1995.”

Compared with a merger of aneighboring networks, adding new ownersmay he a preferable (and less expensive)method of expanding geographicallyAntitrust enforcers, however, should treatthese transactions as mergers because inmany cases they may result in the diminu-tion of competition between the twonetworks. For example, if the expandingnetwork has some sort of exclusivityarrangement (either dejtrre or dejacto), thetransfer of one institution’s ATNs coulddrive the neighboring network below theminimum efficient scale needed to operate.Once the neighboring network is drivenbelow aninimum efficient scale, its compet-itive significance will cease. In otherwords, the net result could be the same asa merger.

National City Bank ftCB) sought tojoin EPS as a 20 percent equity memberand in turn, FPS would acquire NationalCity’s branded ATM network (MoneyCenter), which operates in Ohio, Indiana,and Kentucky (it has just under 900ATMs). NCB was a member of Money Sta-tion, a neighboring joint venture ATM inOhio. Money Station filed a protest. TheBoard staff considered the application forseveral months, received several pleadingsfrom the parties, and conducted aninformal meeting.

The loss of competition. Money Stationclaimed that the acquisition wouldeliminate actual and potential competitionand would increase the barriers to entry orexpansion by existing or potential ATMnetworks. NCB was one of MoneyStation’s largest members. By acquiringMoney Station, EPS would have a substan-tial share of AIMs in several Ohiomarkets, including Cleveland andColumbus. In Money Station’s view bypermitting the acquisition, NCB would heeliminated as an actual or potentialcompetitor because as an equity ownerof EPS, it would have no incentive toparticipate in alternative networks. Inaddition, the merger would increase the

“Fart discassian of the inpar-tanre of rar&issuer roufingrules, see Grimm and lolta(19931.

Banc One Carp. (Marth 6,1995) (Vice Chairman AlanBlinder dissentingl.

FEDERAL RESERVE DANK OF ST. LOIFIE

33

V FYI F~NOVEMRER/DECEMRER 1995

“Letter from Stephen A.Rhaodes, Assistant Dirertar,Division of Research andStatistics, to Allen Raiken, et al.(Feb. 15, 1995).

“The Board’s understanding ofthe purpose of consent decreesappears mistaken. The purposeof the decree is to remedy theconpeflflve problem at tImetime the decree is entered,nat daring the pendency ofthe decree.

difficulty for existing or potentialcompeting ATM networks to retain orassemble the necessary critical mass ofterminals and cardholders required by eco-nomic considerations, such as economiesof scale and ubiquity to be effectivecompetitors of MAC.

The Board rejected the argumentbecause the facts of record did notsupport the view that NCB would be par-ticularly likely to enter the marketindependently or through another jointventure in competition with MAC if thisproposal were denied. Of particularimportance was that NCB abandonedits attempts to form a new regional ATMnetwork with other large banking organi-zations in 1992 and instead became aparticipating member of the MACnetwork. NCB also ceased offeringATM processing services to unaffiliatedthird parties thus the loss of actual compe-tition in network services was minimal.Thus in the Board’s view, NCB did notcompete in either the ATM access orATM processing markets. In addition,MAC would remain subject to actualand potential competition from otherproviders of FF1 services. Thus the Boardconcluded there was no significant lossof competition.

Operating rules. The Board relied heavilyon the role the division consent decreewould play in ensuring that the marketremained competitive. In particular, theBoard appeared to believe that by openingthe MAC network to third-party processors,banks could easily find a competitive alter-native to MAC. Moreover, the Board heldthat these third-party processors couldprovide a channel for entry by competingregional ATM networks.

Money Station contended that variousMAC rules permitted the network tothwart any procompetitive effects achievedunder the division consent decree. TheBoard staff investigated the effects offour rules: (1) MAC’s prohibition of sub-switching between members; (2) MAC’srights under the consent decree to charge aroyalty fee if subswitching were to he per-

mitted; (3) MAC~requirement thatnational network transactions be routedthrough the MAC network; and (4) MAC’sholding company rule that generallyrequires membership of all affiliatedbanks. The Board staff specifically askedthe parties what would he the coanpetitiveeffect of changing these rules.” Withoutsecuring any evidence, the Boardconcluded that modification of theserules was not necessary (although ViceChairman Alan Blinder would haverequired the changes). The Board did sobecause “the consent decree recentlybecame effective, and that its terms aredesigned to achieve procompetitive effectsover time during the 10-year duration ofthe decree.”

Efficiencies/public benefits. The Boardconcluded that there were potential publicbenefits because NCB would make cashinfusions that would enable FP5 “tocontinue and expand its research anddevelopment efforts,” improving itsability to offer innovative electronicbanking products.

Dissent. Vice Chairman Blinderdissented. He noted that although theloss of competition was modest, thepublic benefits did not outweigh thisloss of competition. He observed thatthe application “demonstrates no suchbenefits to the public, in my view,”as required by Sec. 4(c)(8) of theBank Holding Company Act. Thevice chairman would have requiredmodification of MAC’s operating rules,apparently as suggested by the staff, tomeet the public benefits test.

A,ssesscnentThe Board’s approach in these

cases is very much a mixed bag. Someaspects of their decision making appear togive credence to the opportunities for net-work competition, yet ultimately theyseem to assume that a regional monopolyis foreordained.

FEDERAL RESERVE RANK OF ST. LOUIS

34

II FYI F~NOVEMRER/DECEMSER ‘995

Defining the relevant market. Critical tounderstanding the analysis of networkmergers is disaggregating the differentdimensions of the network and analyzingthe effect of mergers on competition foreach dimension. A network has severalcomponents, including a trademark, acomputer switch, and operating rules. Asnoted earlier, too often enforcers and regu-lators have focused on the unconcentratednature of back office operations, and havegiven too little attention to competition atthe brand level.” Differentiating betweenthe two is important because there may berelatively few firms capable of competingat the brand level. Moreover, the barriersto entry may he dramatically different inthe hack office or brand level. Similarlyeven though there may be efficiencies fromconsolidation at the systems level, theseefficiencies may not outweigh the loss ofbrand competition.

The most encouraging aspect of theBoarth decision in Bane One Corp. wastheir effort to disaggregate the dimensionsof competition in their analysis of the rele-vant product market. As noted earlier, theBoard had previously viewed the relevantmarket as basically the networks backoffice operations—an unconcentratedmarket in which entry harriers would berelatively trivial.

In Bane One Corp., the Boardrecognized the distinction between thehack office and brand aspects of competi-tion. As noted earlier, it defined threerelevant markets: network access, ATMprocessing and network services. The lasttwo markets reflect the value of the backoffice operations and the network switch,respectively The first market reflects thevalue of the brand name, reputation, andagreements between the network andits members.”

Competitive effect analysis. Critical inthe analysis of any merger is a determina-tion of the competitive effects of themerger, that is, what will be the ability ofthe merged firm to exercise market powerafter the merger. In both Yankee 24 andBane One Corp., the Board appeared to rely

on the general structure of the market andthe operating rules (discussed later) inconcluding that anticompetitive effectswere unlikely In both cases the competi-tive analysis of the Board was ratherlimited. Particularly in Yankee 24, wherethe two networks had competed directlyin Connecticut and Massachusetts,an analysis of the impact of that competi-tion on both banks and consumerswould have been useful. Some relevantissues, similar to those in First Data,would have included the impact of networkcompetition on network fees, fees toconsumers, output (in terms of ATMs andtransactions), advertising, and revenue tohank members.

Another important issue in Bane OneCorp. was whether NCB’s incentives inparticipating in alternative networkswould he altered because of hecoaning anequity owner of EPS. If NCB’s incentiveswere altered and it dedicated its AT Msexclusively to MAC, Money Station mightfall below minimum viable scale and itscompetitive viability might he in doubt.The Board concluded that this concernwas “too speculative at this time to repre-sent a significant potential adverse effect,”because MAC no longer requiredexclusively for its members.

The Board’s analysis of the likelihoodof defaeto exclusivity may be deficientby failing to recognize how NCB’s owner-ship interests in EPS would affect itsincentives. NCB has no ownership inMoney Station. As an owner of EPS, itis in NCB’s interest to direct as manytransactions as possible through MAC.Thus it seems simple to predict that thelikely outcome is that NCB will dedicateits transactions to the network that willenhance its revenue. That a financialinterest can create defaeto exclusivityhas been recognized by the divisionand the FTC in several recent casesin nonbanking markets and inthe recently issued Health CarePolicy Statements.”

The importance of network operatingrules. The Board’s approach to the

“See Bokar (1993).

“The bard eaploined that net-work access includes: (1) theright to brand ATMs and ATMcards with the trademark orlogo of the ATM network; (2)the obility of the ATM cardholderwith an account at one memberdepository institution to initiatewithdrawal and other accounttransactions at on ATM ownedby another depositary institu-tion that isa member of thesome network; and (3) mini-mum standards for networkperformance and productsoffered through the network.

“Home Oxygen, FTC File Na.9Ol~OlO9(Nov. 2,1993) andHomecore thqgen, FTC File Na.901-0020 (Nov. 2,1993)(heolthcare joint ventures werede lana exclusive; physicianownership interests deferredincentive to participate incompeting ventures); U.S.Deportment of Justice andFederal Trade Commission,Statements of EnforcementPolicy and Anolytical PrinciplesRelating to Heolth Care andAntitrust (Sept. 27, 1994), at69-JO (describing factors thatlead to de [onto eaclusivity).

FEDERAL RESERVE SANK OF ST. LOUIS

35

V FYI 1WNOVEMRER/DECEMSEB 1925

A suMetwork could be an alter-native AIM network,

“55 Trade teg. Rep. (BNA), No.13BO,at353,

“Id.

importance of operating rules seemsconfusing. In Yankee 24-NYCE, thecommitment to an open network structurethat permitted anembers to bypass the net-work and enter into arrangements withalternative networks or third-party proces-sors appeared critical to the Board’sconclusion that there was little potentialfor exercise of market power.

Yet in Bane One Corp., the Boardseemed unwilling to fofiow that precedent.The Board staff appeared concerned thatMAC rules that imposed restrictions onsubswitching between members wouldmake it difficult for members to bypass thenetwork. Vice Chairman Blinder wouldhave preferred that the Board require thatMAC amend these rules. If the Board wascorrect in Yankee 24, that would seem thepreferable approach.

Amending network rules may be nec-essary to resolve concerns over theexercise of market power, but is itsufficient? Should network rules thatcreate an open architecture in and of them-selves immunize a merger where themerged firm will have market power? Isthe opportunity to form suhnetworksbetween individual network memberssufficient to alleviate concerns aboutmarket power?

The Board is basically sailing onuncharted waters in this area. TheonE’ case to address the issue, theFinancial Interchange arbitration, didnot provide clear guidance on whetheropen architecture would alleviate theconcerns o( market power. (In this case,the network (PULSE) peranitted itsmembers to route transactions throughsubnetworks) “In determining whetheralternative routing would diminishthe threat of market power, thearbitrator wrote:

Becatese ATM owners control routingof ATM transactions, they couldchoose in some in stnunces to elect torotet e transactions within a stthnet—work. If. for example, the inter-change fee within the subnetwork ishigher than that of PULSE, the ATM

owner has the incentive to use sub-network routings if available. Thesame could be trtre in reverse ifissuers could control routing. Thiscompetition within the existingstructure could decrease PULSE~revenue Interprocessor stibnet-works functioning within the PULSEsystem can provide some limit onPULSE~freedomto establishinterchange fees.”

Nonetheless, the arbitrator discounted thesignificance of this open architecture inpart because of the universal access offeredby PULSE:

The very fact that all Texas subnet-works are PULSE members at leastsuggests that they perceive the needfor sharing on a broader basis, Thenumber of cards and ATMs in each ofthese networks is far smaller than inPULSE. Moreovet single processor‘capability is limited. Even withinlocal markets such as Dallas orHouston, the access provided by sub-networks falls far short of that ofPULSE. Unless cardholders areindifferent to the added accessPULSE participation provides,intraprocessor switching is not anadequate substitute; reliance solelyon such switching woceld place [juan,—cial instittotions at a significant dis-nunlvantag The combination ofexisting subnetworks might c4 e~urseprovide nan alternative to PULSE...but single subnetworlas nas they nowexist are no real snhstiture.”

Ultimately, individual suhnetworks (orthird-party processors) were not a viablecomupetitive alternative because they didnot offer the level of universal access pro-vided by PULSE. Similarly in Bane One,although individual third-party processorsmight he capable of entering into the areadominated by MAC, it seems unlikely anyof thetn could provide the level ofuniversal access provided by MAC. Asimportant, third-party processors can

FEDERAL RESERVE BANK OF ST. LOUIS

36

VIYIIWNOVEMBER/DECEMBER 1995

offer competition only at the back officelevel; they do not provide competition atthe network access or network servicesmarkets.

Of course, at this stage there is littleevidence that the division consent decreewith MAC has resulted in significant entryby third-party processors or competingnetworks. Even if the consent created anopen network structure, there are severalreasons why that structure might notensure that a network—especially adominant network—cannot exercisemuarket power.

Eirst, even with an open architecture,a network might attempt to impose defacto exclusivity through other types ofrules or fees that raised the costs ofentering into alternative arrangements.For example, a network could set a “royaltyor bypass fee” that would make usingalternative networks financially unfeasihie.In addition, other incentives such as o\vn-ership in the network, may discourage theuse of alternative arrangements.

Ultimately open architecture may bean illusory solution. If members start tobypass the network to any significantextent, free-rider problems will arise; inturn, members may become increasinglyreluctant to invest in the network. Thenet\vork may respond by closing thenet\vork, hanning subswitching orimposing a fee for hypassed transactions.For example, a network could impose a feeon transactions routed outside thenetwork. These free-riding/routingdisputes are some of the most contentiotmsin the ATM area.”

The Board’s failure to address the oper-ating rules in Bane One Corp. sends aconfusing message to ATM networks. Ifthese rules are important to reducing thelikelihood of the exercise of market power,they should he imposed where that threatis present. But even if the Board believesthat operating rules can remedy the threatof anarket power, relying on this factor is atbest a second-rate solution. If operatingrules are important, a preferable positionmight be that taken by the states inEntree—to prevent the merger and permit

the networks to compete in terms of oper-ating rules. Moreover, approving mergershased on operating rules will place the Boardin the position of increasingly regulatingthese networks and eventually arbitratingthe intranetwork disputes over these rules.

The importance of efficiencies/networkexternalities. In merger cases, theenforcement agencies evaluate whether theefficiencies that may arise from a mergermay outweigh the potential for competitiveharm. Prominent in network merger casesare argumnents that efficiencies in terms ofnetwork externalities will outweigh anycompetitive harm. Network externalitiesreflect the view that the value of a networkto a consumer depends on the number ofusers and the identities of specific users.The larger the network, the greater thenumber of consumers who will join it,and, conversely the smaller the network,the fewer the number of consumers whowill join it. Network externalities areespecially common in electronic networkssuch as payment systenms.”

In Bane One Corp.. the Boardrecognized the importance of networkexternalities. It observed that.:

as an ATM network expands thenumber ofitsfinancial institutionmembers and available ATMs, itsvalue to network cardholnlersincreases due to the greateraccessi-bility o~their deposit accounts.Similarly as the number of cnerdholnl—ers increases, so will the number oftransactions and hence the economicreturn on ATM terminals nleployed inthe network. ‘This increased econom-ic return provides incentives forbanks to establish additional ATMs,thereby’ further enhancing the net-works value to cardholders.Accordingly, banks tend to place agreater value on membership in anetwork as its membershipexpands.”

Some commentators have suggested

“See Grimm and bIte (1993).

“See Stevens (1993); fart andShapiro (1985).

Banc One Carp. 11995),p. 494 n. 20.

FEDERAL RESERVE BANK OF ST. LOUIS

37

VFYI[~NOVEMBER/DECEMBER 2995

“See Guerin-Colvert (1994).

“See Baxter, Coetner end Scott11977).

/992 Horizontal MergerGuidelines of the Department oflastine and the Federal TradeCam,nissian, 4 Trade leg. Rep.(CCH) ¶ 13,104, Section 4.0(April 2, 1992). This bittstatement by the two agenciesstates that ‘the Agency willreject claims of efficiencies ifeqeivalent ar comparablesavings can reasonably beochieved by the parties thraughother means... . load that the]expected net efficiencies roustbe greater the name significantore the competitive risks identi-fied .,., See FTC a.University Health, Inn., 938E2d 1206(11th Cir. 1991) (a‘defendorrt wha seeks to over-come a presumption that a pro-posed ocquisitiot wouldsubstantially lessen competitionmust demonstrate that theintended ocquisition wouldresult in significant economiesand that these economiesultimately would benefitcompetition and,hence, consumers.’).

“See booc One Carp. (1995).p. 497.

that the existence of network externalitiesmay counsel for a more laissez-faireapproach in analyzing payment systemsmergers.” Although the existence of net-work externalities may suggest greaterpotential for the existence of efficiencies,that does not mean that those potentialefficiencies should lead to less antitrustenforcement. Eirst, many of those efficien-cies could be achieved by less restrictivealternatives. In the ATM context, forexample, a subswitching arrangement(between the two networks) may permitthe networks to achieve a level of ubiquity(and consumer convenience) withoutelimninating competition at the brand level.

Moreover, network externalities arenot without limit. William Baxter, theformer assistant attorney general in chargeof the antitrust division, has observed thatalthough ATM joint ventures can achieveefficiency benefits related to economies ofscale, these efficiencies will cease to he sig-nificant once a joint venture reaches acertain size. Beyond the point where theseefficiencies are significant, Baxter suggeststhat it is preferable to limit the size of thenetwork to encourage the creation of com-peting networks rather than one largenetwork.”

The Board’s overall analysis of efficien-cies in these cases seems lighthanded andsuperficial. The approach taken by theFTC and division and the courts requirethe parties to demonstrate that the thereare no less anticompetitive means forachieving the efficiencies and that thesebenefits will be passed on to consumers.”The Board did not consider these factors ineither Yankee 24-NYCE or Bane One Corp.In Bane One Corp., the argument—accepted by the Board—that NCB wouldmake cash infusions that would enableEPS to continue and expand its researchand development efforts would not passthis test because there are a number ofalternative sources of revenue to fund suchresearch. Similarly the economies of scalerecognized in Yankee 24-NYCE could havebeen achieved through a more limitedmerger of the two networks’ back officeoperations, while preserving competition

between the networks at the networkaccess level—similar to the ETC approachin Eirst Data.

The vision of the regional networkmonopoly Although the Board’s analysisin these areas seems conventional, oneaspect of the decision in Bane One Corp.poses an “ominous cloud on the horizon.”In response to the concerns about the lossof competition, the Board articulated avision of regional network monopoliesapparently fated by economics.

JT]he signmficant position oJ aregional ATM network is not, stand-ing alone, contrary to the publicinterest. Network externnmlities, suchas the economies of ubiquity, tend topromote consolidation oJ regionalATM networks. As a result, in vari-ous geographic areas, like theMideast region, dominant ATM net-works have been emerging through-out the EET industry. One recentstudy indicates that the ten largestregional networks now account for80 percent of all regional ATM net-work transactions in the UnitedStates. In this light, the Boardbelieves that, as a result of economicand market structure conditions,regions are likely to have one dotni-nant ATM network.”

The Board appears to view the road toregional mnonopoly as foreordained anddictated by the economics of networks. Ischat vision correct? The enforcementactions taken by the states in Entree andthe FTC in First Data suggest thatmonopoly is not a foregone conclusion,even in settings where there may appear tohe significant network externalities. Inboth cases, the antitrust enforcers wereable to spur network competition byfocusing on the impediments to entry atthe brand level and carefully assessing effi-ciencies at the systems level.

Ultimnately the Board’s view seemus toharken back to the day when economics ofubiquity placed ATM network mergers into

FEDERAL RESERVE BANK OF ST. LOUIS

38

NOVEMBER/DECEMBER 995

the per se legal category. Although its Baxter, William F, Paul H. Ceotner end Kenneth F. Scott. Retail Borrkingdecision in Bank One Corp. appears to in the Flentrami Age: TIre fow and Enonervrins of flentrenin Fundsadvance the analytical model, the Board’s liaesfer(1977), p.117.conclusion appears to he that competition blumenthal, William. ‘Three Veming Issues Under the Essentiol Facilitiesis not worth the candle. If the Board’s view Doctrine: ATM Networks as Illustration,” Antitrust tawlaurnal, vol.prevails, the road to regional monopoly 58 (1989), pp. 855, 864.may turn into a superhighway Corlton, Dennis W., and Alan S. Fronkel. ‘The Antiftust Economics ol

Credit Cord Networks,” Antitrust Law Jaurna( vol. 6311995),p.228.

Network mergers are particularly com- Cb&T Bancshores, Inc., Federal Reserve Balleffn, vol.70(1984), p. 589.plex because the)’ require careful Centerre Boncorporation, Federal Reserve Bulletin, vol. 69(1983),differentiation of the elements of competi- p. 643.tion and thoughtful assessment of thepotential for efficiencies. Too often, Citicorp, Federal Reserve Bulletin, vol.72(1986), p. 583.antitrust enforcers have quickly grasped ‘Citicorp Express and Western Union Escalate War of the Wires,”the potential for theoretical efficiencies AmetinanBankeF (Nov. 18, 1987).without giving sufficient attention to the Constantine, Lloyd E. Presentation before the Charles River Associates,opportunities for network competition. Antitrust Issues in Regulated lndastties (Dec. 6, 1990).Payment systems networks play anincreasingly itnportant role in the today’s ‘Debit Cord War Faces Tough Choices,’ Amerinan Bankeç vol. 17economy A monopoly/regulatory (Feb. 7,1994).model—which may he the result of the ‘Economics—More Issuers Get Debit lnterchanae,’ PUS NewsBoard~recent ATM decisions—may lead to (Jon. 1,1994).less competition and higher prices.

EFTin the United States Finol Report of the Notional Commission anElectronic Fund Tronsfers, 57, 11977).

Financial Interchonge, 55 Trade leg. Rep. (BNAI, Ne. 1380,

Atlantic Bancarparotion, Federal Reserve Bulletin, vol. 69(1983), p. 356.p. 639. Grimm, Karen L, and David A. bolto. “Consumer Pricing farATM

Baker, Donald I. ‘Compulsory Access to Network Joint Ventures Under Services: Antitrust Constroints and Legislative Alternatives,” Geergiathe Sherman Act: Rules or Roulette?” Utah Law Review (19931, State Law Review, vol. 4, (Spring 1993), p. 839.p. 999. Guerin-Colvert, Margaret F. ‘Key Economic Issues in Network Merger

__________ letter to William B. Brandt, in the Nebraska Bankers Analysis,” fnanamists Ink (FoIl 1994).Assaniatian (March 7,1977). Interstate Finorrcial Carp., Federal Reserve Bulletin, vol. 69(1983),

Balta, David A. ‘Dualitl and Payment Systems: Antitrust Issues,” p. 560.Review ef Bankmg and Fmancial Servines, vol. 11 (May11, 19951, Katz, Michoel I., and Carl Shapiro. ‘Network Externalities, CompetitionP~105. and Campatabilit~’Amerinan franemir Revie~vol. 75(19851,

Banc Ore Carp., Federal Reserve Bulletin, vol. 81(19951. p. 424.

‘Bank, ATM Mergers Facing Tougher Antitrust Scrutiny,” American laderman, Elizabeth S. “The Public Policy Implications of State lawsBanker (April 7, 1994), p. 2. Pertaining to Automated Teller Machines,” Federal Reserve Bank of

San Frannina Eranamtn Review (Winter 1990).“Bank of America is Going to Bat far Maestro,’ Ameniran Bankec vol.

10 (April 5, 1994). McAndrews, James]. and Robert J. Kauffman. “Network Emternolitiesand Shared Electronic Netwark Adoption,” Working Paper Na. 93-18,

Bank of New York Ca., Federal Reserve Bulletin, vol. 80(1994), Fcenamic lesearch Divisiorr, Federal Reserve bank of Philadelphiap.1107. (November 1993).

‘Bankers are Burying the Hatchet to Join Forces far Debit Push,”Rule, Charles F “Antitrust Anolysis of Joint Ventures in the Banking

American Banket vol. 20, (Feb. 8, 1994).lndustrl: Evaluoting Shared ATMs,” remarks reprinted in Danald I.

Barclays Bank PLC, Federal Reserve BulletTh, vol. 71(1985), Baker and Roland F. Bmndel, Tire law efFlentronin Fund Transferpp. 113-14. Systems, appendim F (1988), n. 5, p. A-1 41.

FEDERAL RESERVE BANK OF ST. LOUIS

39

V F\’I 1~NOVEMBER/DECEMBER 1995

Sovrnn Financial Corp., Federal Reserve Bulletin, vol. 72(1986),pp. 347~B.

‘State Antitrust Officials Criticize Mastercard Debit Rule,” AmetinanBanker(Dec.17, l994),pp. 1,17.

Stevens, John M. “Antitrust Law and Open Access to the NREN,’ 38VillanavalawReview57l (1993), pp. 59T8.

The Bankers Roundtable, Banking’s Role in Tomorrow’s PaymentsSystem, vol. 47. Furosh II Ca., 1994.

U.S. Department of Justice, Antitrust Division. Paliny Statement enSharing ta the National Commission an Elerironin Fund liansfers (Jan.13, 1977).

FEDERAL RESERVE BANK OF ST. LOUIS

40