Payments System Research Department, Federal Reserve Bank of Kansas City, 925 Grand Boulevard, Kansas City, Missouri, USA, 64198, e-mail: fumiko.hayashi@kc.frb.org , stuart.e.weiner@kc.frb.org . The authors wish to thank Nathan Halmrast, Jean-Charles Rochet, Julian Wright and seminar participants at the c Competition and Efficiency in Payment and Security Settlement Systems Conference, d co-sponsored by the Bank of Finland and the Center for Economic Policy Research, and the European Central Bank. The views expressed in this paper reflect those of the authors and do not necessarily reflect those of the Federal Reserve Bank of Kansas City or the Federal Reserve System.<br><br> 2 1. Introduction Interchange fees are an integral part of the pricing structure of credit and debit card industries. While in recent years the theoretical literature on interchange fees, and payment cards in general, has grown rapidly, the empirical literature has not.<br><br> There are several reasons for this. First, comprehensive data are hard to obtain. Second, the industries are very complicated, and empirical models need to incorporate many industry-specific features, such as payment-card network rules and government regulations.<br><br> And third, empirical studies may require a generalized empirical model since, typically, only a few payment card networks exist in a given country. However, because of the first and second reasons, generalizing empirical models may prove problematic. This paper seeks to provide a bridge between the theoretical and empirical literatures on interchange fees.<br><br> Specifically, the paper confronts theory with practice by asking, to what extent do existing models of interchange fees match up with actual interchange fee practices in various countries? For each of four countries 4Australia, the Netherlands, the UK, and the United States 4models that cbest d fit the competitive and institutional features of that country 9s payment card market are identified, and the implications of those model are compared to actual practices. Along what competitive dimensions is there alignment?<br><br> Along what competitive dimensions is there not alignment? What country-specific factors appear to be important in explaining deviations from theoretical predictions? The results suggest that a theory applicable in one country may not be applicable in another, and that similar interchange fee arrangements and regulations may well have different implications in different countries.<br><br> The paper is organized as follows. Section 2, which draws extensively from Weiner and Wright (2005), provides an overview of interchange fee developments and issues in ten key 3 countries and areas: Australia, Canada, Denmark, EU cross-border, Mexico, the Netherlands, Spain, Sweden, the UK, and the United States. Credit card, signature-based debit card, and PIN- based debit card markets are addressed separately.<br><br> Topics covered include interchange fee arrangements and regulations, network rules, recent and current controversies, and the role of public authorities. Section 3 of the paper surveys existing theories of interchange fees. The discussion is focused on assumptions regarding the degree of network competition, the degree of intra-network (issuing and acquiring) competition, and the behavior of consumers and merchants.<br><br> Section 4 of the paper then attempts to match the reality of Section 2 with the theory of Section 3 by examining in some detail interchange fee developments in four particularly interesting markets: Australia, the Netherlands, the UK, and the United States. These case studies provide useful insight into interchange fee competition issues. Finally, Section 5 offers a summary.<br><br> 2. Interchange Fee Arrangements Credit and debit card industries are examples of two-sided markets. The distinguishing feature of two-sided markets is they contain two sets of end users, each of whom needs the other in order for the market to operate.<br><br> In the case of credit and debit cards, the two end-user groups are cardholders and merchants. Payment card systems take one of two principal forms. They may be three-party systems: cardholders, merchants, and a single financial institution that offers proprietary network services, for example, American Express.<br><br> Alternatively, they may be four-party systems: cardholders, merchants, card-issuing banks, and merchant acquiring banks, using the services of a multi-party network such as MasterCard, Visa, or a domestic debit card network. In four-party systems, the interchange fee is an instrument that networks can use to achieve a desired balance of cardholder 4 usage versus merchant acceptance across the two sides of the market, in the same way that proprietary systems can do so directly. In other words, interchange fees are a mechanism that can be used to transfer revenues from one side of the market to the other in order to generate the desired level of card activity.<br><br> Interchange fee arrangements vary considerably across countries. This section of the paper documents these arrangements, as well as related developments and issues, in the following countries: Australia, Canada, Denmark, EU cross-border, Mexico, the Netherlands, Spain, Sweden, the UK, and the United States. These countries share some common features but also exhibit many differences in how interchange and related fees and rules operate in their respective markets.<br><br> In all of these countries, however, interchange fees currently are, to varying degrees, the focus of pointed industry and public-authority debate. 1 The section is organized as follows. The first three subsections discuss developments and issues in the credit, signature-based debit, and PIN-based debit card markets separately.<br><br> The discussion is guided by information presented in Tables 1, 2, and 3, respectively. In these tables, the ten individual countries and areas are listed as rows and, for each, key interchange-related information is provided in the columns: instrument market share (column 3); networks operating in that country (column 4); what bodies set interchange fees for those networks (column 5); recent movements in interchange fees (column 6); recent movements in merchant service charges (column 7); recent movements in cardholder fees (column 8); whether no-surcharge (column 9), honor-all-cards (column 10), net-issuer (column 11), and duality/exclusion rules (column 12) are in force in various networks; and finally, interchange-rated issues and controversies recently or currently under debate (column 13). The closing subsection delves more deeply into public 5 authority involvement in interchange discussions, documenting, with the aid of Table 4, actions and rulings taken by or pending before competition authorities and central banks in these ten countries.<br><br> 2.1 Credit cards Credit cards are an important payment instrument in many of the countries under review. 2 As noted in column 3 of Table 1, they are most prominent, in terms of share of overall noncash transactions volume, in Canada, the United States, Australia, the UK, and Spain. 3 Column 4 shows the credit card networks operating in the ten countries and areas.<br><br> Four-party networks are listed above the dotted line while three-party networks are listed below the dotted line. Networks with an asterisk behind them are networks available for international use only, that is, networks that issue cards to that country 9s residents for international use and are available for foreigner 9s use within the country. Another convention followed in Tables 1 through 3 relates to missing information.<br><br> When a statement or attribute is not attached to a specific network, it can be taken to hold for all networks in that country. When a statement or attribute is attached to a specific network, it applies only to that network, and comparable information not shown for other networks means that such information is not available. Finally, in the tables, cnap d denotes not applicable, and cneg d denotes negligible.<br><br> Interchange fees are set under a variety of arrangements (column 5). In some countries they are collectively set by members of the network, sometimes subject to regulatory limits; in others they are set by network management; in one country, Mexico, they are set by members of the Mexican Bankers Association; and in another, Sweden, they are set bilaterally. In Demark, 1 For institutional background on the payments card industry in various countries, see Bank for International Settlements (2003), European Central Bank (2001), Evans and Schmalensee (2005a), and Hayashi, Sullivan, and Weiner (2003).<br><br> 2 cCredit cards d includes charge cards as well as deferred debit transactions. 6 interchange fees are aligned with merchant service charges, which are subject to a regulatory cap. 4 In most countries interchange fees have declined or are declining (column 6).<br><br> In some countries, this is due to recent regulation or regulatory threat: Australia, Mexico, and EU cross- border. In Spain, interchange fees have been declining as a result of a 1999 agreement among banks, networks, merchants, and the Ministry of the Economy. 5 A notable outlier is the United States, where interchange fees on credit cards have been rising in recent years.<br><br> Although difficult to document, merchant service charge movements appear to have tracked interchange fee movements to some extent (column 7). Cardholder fee movements have also tended to move in the expected direction (column 8). In countries where interchange fees have declined, for example, Australia and Spain, annual fees have increased and, in the case of Australia, interest-free periods have shortened and rewards programs have become less generous.<br><br> In the United States, where interchange fees have risen, annual fees have declined, reward programs have become more generous, and zero introductory interest rates have become prevalent. Here the UK appears to be something of an outlier, with interchange fees falling but annual-fee and introductory-rate provisions remaining relatively generous. The ten countries/areas exhibit considerable variation across the principal categories of network rules: no-surcharge rules, honor-all-cards (HAC) rules, net issuer rules, and duality/exclusion rules.<br><br> No-surcharge rules prevent merchants from charging customers for the use of a particular payment mechanism, in this case, a credit card. Honor-all-card rules, as defined here, require merchants that accept a network 9s credit card to also accept that network 9s 3 In Spain, charge card transactions are predominant. 4 In addition, in many countries, MasterCard members are permitted to negotiate interchange fees bilaterally.<br><br> 5 A further lowering of fees may result from an April 2005 ruling by the Tribunal de Defensa de la Competencia. 7 signature-based debit card, if the latter exists in a given country. 6 Net issuer rules require acquiring banks to issue a minimum level of cards in order to participate on the acquiring side of the market.<br><br> Duality rules allow a bank that issues MasterCard credit cards to also issue Visa credit cards. Finally, exclusion rules prevent a bank that issues MasterCard or Visa credit cards from issuing other credit cards, for example, American Express and Discover. No-surcharge rules are presently in effect in Canada, Mexico, the United States, and Sweden (column 9).<br><br> They also are in effect 4for Visa 4for EU cross-border and Spanish transactions. On the other hand, surcharging is permitted in Australia, the Netherlands, and the UK and, for MasterCard, also in Spain and EU cross-border. 7 Honor-all-card rules have a particularly interesting history in the United States (column 10).<br><br> The so-called Wal-Mart case, brought by several merchants and trade associations against Visa and MasterCard in the mid-1990s and finally settled in 2003, eliminated Visa and MasterCard HAC rules. Net issuer rules are in place in roughly half of the countries under review (column 11). Duality is allowed in all countries but Canada; exclusion rules vary (column 12).<br><br> Turning to the last column of Table 1, industry participants and public authorities in virtually all of the countries have dealt with or are currently discussing issues and controversies surrounding credit card interchange fees and related matters. These range from regulations or agreements lowering interchange fees (Australia, Mexico, EU cross-border) or capping merchant service charges (Denmark), to regulations eliminating or permitting no-surcharge rules 6 In Europe, the term chonor-all-cards rule d is typically defined differently, namely, if a merchant accepts a MasterCard\Visa-branded credit card issued by Bank A, the merchant must also accept a MasterCard\Visa-branded credit card issued by Bank B and, similarly, if a merchant accepts a MasterCard\Visa-branded PIN debit card issued by Bank A, the merchant must also accept a MasterCard/Visa-branded PIN debit card issued by Bank B. Thus, in Europe, honor-all-cards rules typically do not tie debit cards to credit cards.<br><br> 7 Although MasterCard allows surcharging in Spain, the three domestic card networks do not. 8 (Australia, EU cross-border), to merchant dissatisfaction with interchange levels (United States, Spain, EU cross-border, UK), to complaints or concerns over three-party network schemes (Australia, United States), transparency (Spain, EU cross-border, Australia), and duality (Canada). Thus, there is no shortage of challenging issues confronting the industry.<br><br> 2.2 Signature-based debit cards Table 2 presents information on signature-based debit cards. Signature-based cards have an important presence in a few countries, for example, the United States and Spain, and in other countries, Canada, Denmark, and the Netherlands, they are essentially nonexistent (column 3). As the name suggests, they are debit cards that require a signature, not a PIN, for authorization.<br><br> MasterCard and Visa signature debit transactions run over MasterCard 9s and Visa 9s respective credit card networks. The three Spanish signature debit card networks, Euro 6000, ServiRed, and Sistema 4B, are stand-alone proprietary systems. 8 In this subsection and the next, to keep things a little simpler, the general terms cMasterCard d and cVisa d are used to denote the various MasterCard and Visa signature and PIN- based debit products.<br><br> Thus, Visa CheckCard (United States signature), Visa Debit (Australian signature), Visa Electron (European PIN), and Visa Interlink (United States PIN) are all referred to as Visa, while MasterCard MasterMoney (United States signature) and MasterCard Maestro (worldwide PIN) are all referred to as MasterCard. As is the case with credit cards, interchange fees on signature debit have tended to decline in recent years (column 6), often attributable to regulatory action (EU cross-border), regulatory threat (Mexico), or government-led industry agreement (Spain). Interchange fees on U.S.<br><br> signature debit, in contrast, have taken a different route. As part of the Wal-Mart settlement, 8 These networks also process MasterCard and Visa credit card and debit card transactions. 9 they were reduced by roughly a third over the period August to December 2003, but since that time they have risen to some extent.<br><br> 9 Merchant service charges appear, for the most part, to have moved in the same direction as interchange fees, although, like cardholder fee movements, complete data are hard to obtain (columns 7, 8). Typically, signature-based debit transactions are subject to the same no-surcharge rules as credit cards in the various countries, and by definition, all honor-all-cards rules are also the same (columns 9, 10). Net issuer and duality rules also coincide (columns 11, 12).<br><br> A number of signature debit issues are currently under discussion (column 13). In Australia, the Reserve Bank of Australia has proposed lowering interchange fees and eliminating the no-surcharge and HAC rules on the Visa signature card. In the United States, there is widespread merchant dissatisfaction with the level of interchange fees, and in Spain some merchants reportedly believe merchant service charges remain too high.<br><br> The EC, meanwhile, has MasterCard 9s fee policy under review, having reached an earlier agreement with Visa for Visa to lower its cross-border fees. 2.3 PIN-based debit cards Table 3 presents information on PIN-based debit cards. PIN-based cards have a large presence in many European countries as well as in Canada, Australia, and the United States (column 4).<br><br> MasterCard and Visa operate PIN-based systems in several of these countries; in addition, domestic systems are especially prominent in Australia, Canada, the United States, Denmark, and the Netherlands (column 4). 10 9 After rising in 2004, some signature-based debit interchange fees were lowered in April 2005. Most remain above post-settlement levels, however.<br><br> 10 In many cases, MasterCard and Visa PIN-based debit cards are co-branded with domestic schemes, and routing priority is given to the latter. 10 PIN-based networks display a variety of institutional features and practices across countries. In Australia, for example, EFTPOS PIN debit interchange fees are set bilaterally and, notably, are paid by the issuer to the acquirer.<br><br> In Canada and the Netherlands, interchange fees are set at zero by Interac and Interpay, respectively. Swedish PIN interchange fees are negotiated bilaterally. And in the United States, PIN debit interchange fees are often a competitive tool for attracting issuers 4the United States is characterized by a large number of good-sized domestic networks, the largest of which (Star, NYCE, and Pulse) compete vigorously with Visa (columns 4, 5).<br><br> Interchange fee movements have shown varying patterns in recent years (column 6). They have remained at zero in Canada and the Netherlands and have been stable in Australia. In Denmark, MasterCard and Visa fees have declined, while positive fees for Dankort transactions (the domestic network) were introduced at the beginning of this year but have since been eliminated again, all due to regulatory actions.<br><br> 11 In the United States, PIN debit interchange fees have been drifting up. As with credit cards and signature debit, a number of PIN debit issues are currently under discussion (column 13). The Reserve Bank of Australia has proposed lowering EFTPOS interchange fees.<br><br> The EC is reviewing MasterCard 9s cross-border interchange fees. In Canada, cards from participating issuers can now be used at U.S. merchants who belong to the NYCE network.<br><br> In the United States, merchants are dissatisfied with rising interchange fees; the United States has seen a striking consolidation of domestic networks in recent years, the outcome of which is still unclear. In the Netherlands, the Competition Authority recently fined Interpay for its pricing policies, although the objection process is still underway. And in Denmark, since March 1 of this year, surcharging has been eliminated on Dankort transactions.<br><br> 11 2.4 Public authority involvement Virtually all central banks have general oversight responsibility for the payments systems of their respective countries and areas. Explicitly or implicitly, most have a mandate to ensure that payments systems operate safely and efficiently. As retail payments systems around the globe migrate from paper to electronics 4and, in particular, as credit and especially debit card transactions become a dominant form of payment 4central banks are paying increasing attention to credit and debit card industries.<br><br> In most countries, however, specific interchange fee and other payment competition issues fall under the jurisdiction of competition (regulatory and antitrust) authorities. There are exceptions, of course: the Reserve Bank of Australia and Banco de Mexico have been very visible in interchange matters. And the Banco de Espana plays an important monitoring role.<br><br> But for the most part, it is the competition authorities that have taken the lead in evaluating and, at times, bringing about change, in credit and debit card markets. For the set of countries under review in this paper, this is true of Canada, the United States, EU cross-border, Denmark, the Netherlands, Spain, Sweden, and the UK. Turning to specifics, Table 4 documents various actions taken by and pending before public authorities on interchange and related issues.<br><br> Beginning with Australia, in early 2003, the Reserve Bank of Australia (RBA) eliminated credit card no-surcharge rules, and later that year mandated that credit card interchange fees be lowered. Early this year, it ruled that payments between American Express and Diners Club and their bank partners will not be regulated but that the two companies will reword certain restrictive clauses in their merchant agreements. American Express and Diners Club are now publishing their average merchant service fees; Bankcard, MasterCard, and Visa have been required to publish interchange fees since 2003.<br><br> 11 A further review of Dankort fees is possible by summer 2005. 12 Pending issues include RBA proposals to lower EFTPOS PIN debit interchange fees and Visa signature debit interchange fees. The RBA also has proposed eliminating the honor-all- cards rule (HAC) linking Visa 9s credit and signature debit cards.<br><br> The Bank will review existing regulations for credit card schemes in 2007. The Banco de Mexico is the other central bank that has taken a prominent role in addressing credit and debit card issues. It has been working with the Mexican Bankers Association to lower interchange fees and make the HAC rule more flexible.<br><br> Among competition authorities, the European Commission in 2001 came to an agreement with Visa whereby Visa agreed to gradually reduce its cross-border interchange fees on credit cards and signature and PIN debit cards by December 2007. At the same time, Visa was permitted to keep its no-surcharge, HAC, and net issuer rules. Currently pending are EC investigations of MasterCard cross-border interchange fees and exclusion provisions in Visa membership rules.<br><br> In the UK, the Office of Fair Trading (OFT) ruled in September 2005 that the 2000-2004 MasterCard multilateral credit card interchange fee agreements were anti-competitive; MasterCard is currently appealing this decision. In October 2005, the OFT issued a statement of objections against Visa 9s domestic multilateral interchange fee agreements for its credit, deferred debit, and charge card transactions. A period of discussion is currently underway.<br><br> Competition authorities have been active elsewhere as well. Last year, the Netherlands Competition Authority (NMa) fined Interpay and its member banks for charging cexcessive d merchant service charges, although the objection process is still underway. Also last year, the United States Supreme Court upheld the U.S.<br><br> Department of Justice 9s successful 2001 court case eliminating MasterCard and Visa exclusion rules that prohibited member banks from issuing 13 American Express and Discover credit cards. Earlier actions or agreements in other countries include: (i) the Danish Competition Authority 9s administration of the 2003 Amendment to the Danish Act on Certain Payments, whereby MasterCard and Visa PIN debit merchant service charges were reduced and a positive (non-zero) interchange fee was introduced on Dankort chip PIN debit card transactions; (ii) the Canadian Competition Authority 9s mid-90 9s Consent Agreement with the principal member of Interac, which resulted in expanded representation on the Board, liberalization of network access rules, and removal of constraints on product innovation and price competition, including the removal of the prohibition against surcharging; and (iii) the Spanish Ministry of the Economy 9s 1999 agreement with banks, networks, and merchants to lower credit card and signature debit card interchange fees, and the Tribunal 9s 2005 request to the networks for new interchange fee setting procedures. Interaction among competition authorities and central banks on credit and debit card issues varies across countries (column 9).<br><br> In some countries, there is some degree of consultation and occasional representation on joint task forces 4this is, or has been, true in Spain, the EC, the UK, and Australia, for example. In other countries, the degree of interaction between competition authority and central bank appears limited, at least formally. Whether closer ties will be forged in light of the issues facing the card industry bears watching.<br><br> 3. Interchange Fee Theories Interchange fees and related payment card issues have been the subject of a growing body of theoretical work in recent years. 12 This section of the paper surveys a portion of this work, focusing on models that examine various factors potentially affecting interchange fees.<br><br> In order 12 Other surveys are provided by Schmalensee (2003), Evans and Schmalensee (2005b), Roson (2005), and Weiner and Wright (2005). 14 to review this literature somewhat systematically, we group possible factors into four categories: assumptions regarding networks, assumptions regarding issuers and acquirers, assumptions regarding end-users (consumers and merchants), and assumptions regarding other possible factors. A single factor, by itself, is highly unlikely to determine the level of interchange fees.<br><br> Rather, interaction among factors, in some or all of these four categories, typically proves critical. 3.1 Assumptions Table 5 presents a summary of many of the key theoretical articles on interchange fees written over the last several years. The papers, organized by the assumed level of network and intra-network (issuer and acquirer) competition, are listed in the third column of the table.<br><br> As will be discussed in Section 4 below, use of these two categories proves to be a useful cfirst- step d sorting mechanism when comparing model assumptions and predictions with actual interchange fee arrangements. The first organizational division, reflected in the first column of Table 5, is the assumption regarding network competition. Many models assume there is no competition among card networks, either explicitly, by assuming a monopolistic network, or implicitly, by not considering network competition in the setup.<br><br> Other models assume there is competition among networks. In some cases, these networks are defined as identical, competing within the same payment instrument (for example, credit vs. credit or debit vs.<br><br> debit). In other cases, these networks are defined as asymmetric, competing across different payment instruments (for example, credit vs. debit or PIN debit vs.<br><br> signature debit), across different network schemes (three-party vs. four-party), or within the same payment instrument but facing different cost structures. 15 The second organizational division, shown in the second column of Table 5, is the assumption regarding intra-network competition.<br><br> A key feature of most models is the assumed degree of competition among card-issuing banks and among merchant acquiring banks. This degree of competition is typically modeled with reference to the price-cost margins of issuers and acquirers. A zero margin is taken to imply perfect competition.<br><br> A positive margin is taken to imply some market power. As seen in Table 5, some models assume both issuers and acquirers operate in perfectly competitive markets, some assume both issuers and acquirers have some market power, and still others assume only issuers have market power. The remaining cAssumption d columns in Table 5 list other important factors assumed or incorporated in the respective models.<br><br> Column 4 focuses on network attributes. In addition to the assumed degree of network competition, three additional network attributes are considered: whether the model in question assumes a three-party or four-party scheme; what the network(s) 9 objectives are; and whether there is a single or multiple interchange fee structure. Possible network objectives include maximizing the number of transactions or market share; maximizing network profits (in a three-party scheme); and maximizing members 9 joint profits (in a four-party scheme), perhaps weighted more on the issuer or acquirer side.<br><br> In addition, networks may seek to address any imbalances between the costs and revenues of issuers and acquirers, and between the demand of consumers and merchants. Finally, models may incorporate either a single interchange fee or, alternatively, multiple interchange fees that vary according to type of industry or transaction volume. Column 5 in Table 5 focuses on intra-network attributes.<br><br> In addition to the assumed degree of competition on the issuing and acquiring sides of the market, three additional attributes are considered: the degree of pass-through of interchange fees from issuers and acquirers to 16 cardholders and merchants, respectively; the relative cost structures facing issues and acquirers; and whether issuers and acquirers are the same or different entities. The next two columns turn to assumptions regarding the end users in payment card models. Consumer characteristics (column 6) include the demand for products (elastic or inelastic); the demand for cards (exogenous or endogenous; singlehoming or multihoming); and the demand for specific card transactions (homogeneous or heterogeneous transactional benefits).<br><br> The types of fees and rewards that consumers face also vary by model. Merchant characteristics are listed in column 7. Some models assume that merchants are strategic in their card-acceptance behavior, that is, they are competitive.<br><br> Others assume that merchants are monopolistic. Models also differ according to whether merchants are assumed to derive homogeneous or heterogeneous transactional benefits, and whether they pay per- transaction fees and/or fixed fees. Finally, column 8 shows other factors that are built into various models.<br><br> Chief among them are the presence or absence of assorted network rules and bylaws. These include no-surcharge and non-discrimination rules, honor all cards rules, and net issuer rules. 3.2 Results It is probably fair to say that the results of the papers summarized in Table 5 vary as much or more as the underlying assumptions in these papers.<br><br> Key results are listed in column 9. Perhaps the most important general result involves network competition. The effect of network competition on interchange fees is not uniform but varies widely depending on other factors.<br><br> Some key factors include consumer and merchant demand characteristics, and the nature of intra-system competition. 17 To the extent consumers are singlehoming, that is, use only one payment card, networks can act as monopolies, so interchange fees are not reduced by network competition (Rochet and Tirole, 2002). However, as consumers become multihoming, merchant resistance to interchange fees increases, and network competition lowers interchange fees (Rochet and Tirole, 2002, 2003; Guthrie and Wright, 2003).<br><br> To the extent that merchants are homogeneous, with an inelastic demand for transactions, network competition leads to a lower (or equal) interchange fee than non-competition. However, if merchants are heterogeneous (elastic demand), the competitive interchange fee can be higher than the monopolistic interchange fee (Guthrie and Wright, 2005). Network competition lowers interchange fees for both strategic (competitive) and monopolistic merchants.<br><br> However, interchange fees for monopolistic merchants are lower than those for strategic merchants whether the network is competitive or not (Guthrie and Wright, 2003). Intra-system competition is similarly influential. Several models show that differences in issuers 9 and acquirers 9 margins affect equilibrium interchange fees (Maneti and Somma, 2002; Rochet and Tirole, 2002; and Guthrie and Wright, 2005).<br><br> Differences in these margins also affect competing networks 9 profits (Maneti and Somma, 2002). A number of other interesting results involving network competition fall out of these models as well. These include: (1) network competition lowers the total fees charged across the issuing and acquring sides of the market (Rochet and Tirole, 2003; and Chakravorti and Roson, 2004); (2) network competition may raise interchange fees if consumers hold a single card and merchant demand for transactions is elastic (Guthrie and Wright, 2003, 2005); and (3) if the network is a monopoly, interchange fees can vary depending on the interaction of network 18 objectives and issuer and acquirer margins (Gans and King, 2002; Schmalensee, 2002; Wright, 2003, 2004; and Schwartz and Vincent, 2004).<br><br> Clearly, the nature of network competition is central to many of the results of the models in Table 5. Another important role is played by the various network rules and bylaws. Most of the models, for example, explicitly assume a no-surcharge rule and implicitly assume an honor all cards rule.<br><br> Relaxing these assumptions can lead to differing results. If merchants are allowed to surcharge, for example, interchange fee levels may change depending on any number of additional factors, including the effective cost of surcharging to merchants, merchant competitiveness, and the price elasticity of consumer demand for goods (Gans and King, 2002; Katz, 2001; Wright, 2003; and Schwartz and Vincent, 2004). What one comes away with after surveying this rich theoretical literature is an appreciation for the many factors that may affect interchange fees.<br><br> Even a single factor may impact interchange fees differently depending on other factors. Determining the actual impact of such variables is, in the end, an empirical question. We attempt to take a step in this direction in the next section.<br><br> 4. Country Case Studies The previous section surveyed some of the important contributions in the theoretical literature on payment card interchange fees. This section details actual market conditions in four countries 4the United States, Australia, the UK, and the Netherlands 4and compares these conditions with theory.<br><br> The key question asked is, to what extent do actual interchange fee practices cline up d with model assumptions and predictions? For each country, we first characterize the credit and debit card industries by the level of network and intra-network 19 competition. We then try to match up a country 9s experience with existing theory, suggesting additional assumptions and institutional features that may help explain that country 9s situation.<br><br> 4.1 United States 4.1.1 Network competition Network competition exists in the United States, both within and across payment card instruments. The United States has six credit card networks. The three largest 4Visa, MasterCard, and American Express 4compete aggressively with one another.<br><br> Visa has the largest market share, followed by MasterCard and American Express. Visa 9s market share has declined somewhat in recent years, as measured by purchase value, number of transactions, and number of cards (Chart 1). The remaining three credit card networks 4Discover, Diners Club, and JCB, have relatively small market shares (Chart 1).<br><br> The United States has two signature-based debit card networks (Visa Check Card and MasterCard MasterMoney) and thirteen PIN-based debit card networks. Competition has been especially pronounced in the PIN debit market, and especially among the four largest networks. The market share of Visa 9s Interlink network has trended up in recent years while those for Star, NYCE, and Pulse have fluctuated (Chart 2).<br><br> These large PIN-based networks also compete vigorously with the two signature-based networks. 13 It is unclear to what extent credit cards and debit cards compete. Overall debit card market share (signature plus PIN) has been rising in recent years, and in 2003, the number of debit card transactions exceeded the number of credit card transactions for the first time (Chart 3).<br><br> However, in terms of purchase value, the difference between credit cards and debit cards has 13 For discussion of PIN vs signature debit competition, see Hayashi et.al (2003). 20 been stable over the last five years, suggesting perhaps that debit card transactions are largely substituting for paper-based (check and cash) transactions and not for credit card transactions (Chart 3). One can safely say, however, that there is competition within the credit card industry, within the PIN debit card industry, and across the PIN and signature debit card industries.<br><br> 4.1.2 Intra-network competition Both the acquiring and issuing sides of the card market appear to be competitive in the United States. With regard to the acquiring market, although the largest acquirers 9 market share has increased slightly in the last ten years, acquirers 9 margins per transaction reportedly have been declining (Chart 4). 14 On the issuing side, top credit card issuers 9 market shares have increased significantly in the last ten years (Chart 5).<br><br> Nevertheless, this market appears to be quite competitive. No annual fees, generous reward programs, and free or low introductory interest rates are typical in the industry, as issuers compete aggressively for customers. The story is somewhat different with respect to debit cards.<br><br> Here, market shares of top issuers are much smaller than in the credit card market (Chart 6, 7), but the degree of competition is hard to gauge. Because debit cards are tied to demand deposit accounts, it is costly for consumers to switch issuers. At the same time, however, many banks use their debit products as a strategic tool, providing rewards for signature card transactions and charging so-called PIN fees for PIN card transactions.<br><br> On net, it is probably fair to view card issuing 4both credit and debit 4as a competitive environment. 14 According to the Star 9s fee structure, a processing fee is around 3 cents per transaction and according to the FMI, the acquirer 9s processing charge is between 2.5 cents to 6.5 cents per transactions, these fees have declined slightly in the last several years. 21 4.1.3 Matching theory and practice As noted above, both network and intra-network payment card competition exists in the United States.<br><br> In terms of network competition, competition between Visa and MasterCard in the credit card market, and among the top networks in the PIN debit market, fit well with the identical four-party network schemes assumed in Guthrie and Wright(2003, 2005) and Rochet and Tirole (2002, 2003). Competition between Visa/MasterCard and American Express, on the other hand, fits well with Maneti and Somma (2002). In terms of intra-network competition, although both the acquiring and issuing sides of market are competitive, it is difficult to judge which side is more competitive or which side experiences lower margins per transaction.<br><br> Revenues (not margins) are much higher for issuers than acquirers, but their costs per transaction are unknown. It does appear that pass-through of interchange fees is 100 percent on the acquiring side, while on the issuing side it is less than 100 percent. 15, 16 On balance, network objectives are likely to be weighted more heavily toward issuers than acquirers in the U.S.<br><br> One reason is that even the largest nonbank acquirers do not have voting power in association networks and market share of nonbank acquirers is fairly large (Chart 8). A second reason is that large bank acquirers are typically large issuers as well (Table 6). Therefore, maximization of issuer profits, number of transactions, or the weighted sum of end-user surplus (with a high weight on consumers) appears to be plausible assumptions in the U.S.<br><br> case. These assumptions are made by Guthrie and Wright (2003, 2005) and Rochet and Tirole (2002, 2003). 15 A typical merchant fee consists of three components, an interchange fee (to the issuer), a processing fee (to the acquirer), and a switch fee (to the network).<br><br> 22 Other important factors in the United States are merchant demand for card transactions and consumer cardholding behavior. Most industries in the United States are quite competitive. As a result, merchants clearly have a strategic motive to accept cards.<br><br> In addition, unlike in most other countries, interchange fees in the United States are set in a very detailed manner according to industry category (Table 7). Thus, a single interchange fee applies to a relatively homogeneous set of merchants, and this industry-specific fee less likely impacts consumer cardholding behavior, which is consistent with Rochet and Tirole (2002). U.S.<br><br> households typically hold multiple credit and debit cards, that is, they are multihoming. 17 However, also consistent with Rochet and Tirole (2002), these multihoming cardholders often appear to prefer a particular card over the others. 18 Taken in sum, the assumptions in Rochet and Tirole (2002) fit the U.S.<br><br> payment card market well. However, the model does not predict that network competition raises interchange fees, which, arguably, is what is being observed in the United States. The model also predicts that competition among issuers lowers interchange fees, which also seems to contradict the U.S.<br><br> case (Chart 9). The only model that predicts that network competition may raise interchange fees is Guthrie and Wright (2003, 2005). However, to generate this result, the model assumes that heterogeneous merchants face a single interchange fee, which is not the case in the United States.<br><br> Can theory and fact be reconciled? Additional considerations may help explain the U.S. situation.<br><br> For example, modeling issuers 9 behavior may prove critical. Oligopolistic issuers may alter their card portfolio, if not change networks, according to profitability. Network competition, 16 According to a large credit card issuer 9s annual report, the average growth rate of interchange fee income (after deducted the costs of reward program) exceeds the average growth rate of transaction value.<br><br> This suggests that interchange fee does not pass-through 100 percent on issuing side. 17 According to the BIS, the number of debit cards and credit cards issued in the U.S. in 2002 were 260.4 million and 709 million, respectively.<br><br> The U.S. population in the same year was 288.2 million. 18 Some studies pointed out this.<br><br> For example, the 2004 Preferred Card Study by Edgar, Dunn, and Company concluded that crewards dominate the reasons to use a specific credit card for 6 in 10 Americans. d 23 therefore, gives networks a strong incentive to try to attract issuers as much as possible. One of the strategies for doing so is to provide issuers with higher interchange fees. Per-transaction costs for issuers may not be fixed, as many papers assume, but rather be endogenous with respect to interchange fees.<br><br> Higher interchange fees allow issuers to provide more generous reward programs to cardholders, but of course reward programs are costly. As noted, U.S. interchange fees are set by industry.<br><br> As a result, modeling consumer cardholding and merchant card acceptance under a single interchange fee does not fit the U.S. case. In a given industry, perhaps the merchant 9s card acceptance does not influence their customers 9 cardholding behavior.<br><br> 4.2 Australia 4.2.1 Network competition Network competition likely exists in Australia. There are six credit card networks in the country. The three largest 4Visa, MasterCard, and Bankcard 4have a combined market share in excess of 80 percent.<br><br> The remaining market is divided among American Express, Diners Club, and JCB. Individual network share data are not available for recent years, but in 2001-2002, shares in terms of number of credit cards were 53.4 percent for Visa, 22.7 percent for MasterCard, 15.4 percent for Bankcard, 6.5 percent for American Express, 1.9 percent for Diners, and essentially negligible for JCB. Over the 2002 to 2005 period, the combined American Express/Diners share has increased slightly (Chart 10).<br><br> There are two debit card networks in Australia, EFTPOS and Visa Debit. EFTPOS is PIN-based, while Visa Debit is signature-based. Based on statistics furnished by the Building Society to the Reserve Bank of Australia, EFTPOS 9s share of the overall debit network is 24 roughly 90 percent while Visa Debit 9s is roughly 10 percent.<br><br> 19 Visa Debit cards are primarily issued by credit unions and building societies that were precluded from issuing credit cards. EFTPOS cards, in contrast, are issued by all types of financial institutions. Credit card and EFTPOS debit card transactions have exhibited an interesting growth pattern in recent ten years (Chart 11).<br><br> In 1995, credit and EFTPOS debit transaction volume were about the same. From 1996 to 1998, debit volume exceeded credit volume, but from 1999 to 2004, credit volume exceeded debit volume. In 2005, volume for the two instruments has essentially been the same again.<br><br> This may imply that, in Australia, credit and EFTPOS debit are relatively close substitutes and, hence, credit card networks and the EFTPOS network see each other as competitors. 4.2.2 Intra-network competition Both the acquiring and issuing sides of the card market appear to be competitive in Australia. While the acquiring market is highly concentrated, a large portion of recent interchange fee reductions has been passed through to lower merchant service charges (MSC).<br><br> The four largest banks in Australia acquire about 95 percent of transaction volume and 85 percent of transaction value. 20 However, according to the Reserve Bank of Australia, the average MSC for four-party networks in Australia has declined from 1.46 percent prior to regulation to 0.97 percent since regulation. This roughly 50 basis point decline is in line with the decline in interchange fees pre- and post-regulation.<br><br> 19 Building Society Comments on RBA Draft Standards for Visa Debit and EFTPOS (April 29, 2005). 20 Reserve Bank of Australia and Australian Competition and Consumer Commission (2000) 25 Regarding credit card issuing, the four largest banks have a 55 percent market share in terms of number of cards and a 70 percent share in terms of transaction volume. 21 Although many banks reportedly have cut reward-program benefits as a response to lower regulated interchange fees, they still provide rewards to their cardholders.<br><br> This may imply that a portion of interchange fee revenue remains passed through to cardholders, and that credit card issuing is competitive. Also indicative of competition is the fact that two of the four largest banks now issue and promote American Express cards as well as Visa and MasterCard cards. Regarding EFTPOS debit card issuing, the combined market share of the four largest banks is large.<br><br> Issuers typically charge per-transaction fees to their cardholders after a certain number of free transactions. Issuers seem to compete by using the per-transaction fees and/or free-transactions as their strategic tools. 4.2.3 Matching theory and practice As suggested above, the Australian payment card market probably can be characterized as exhibiting both network and intra-network competition.<br><br> In light of this, which theoretical model(s) best cline up d with Australian interchange fee practices? None of the models appear to closely fit the Australian market over a large number of parameters. For example, competition between Visa and MasterCard, between Visa/MasterCard and Bankcard, and between credit cards and EFTPOS can all be characterized as four-party scheme network competition.<br><br> Although the competition between Visa and MasterCard can be regarded as identical, the other two competitive relationships cannot. A number of important papers adopt four-party schemes, but all of them assume identical networks (Guthrie and Wright 2003, 2005; Rochet and Tirole 2002, 2003). Chakravorti and Roson (2004) assume asymmetric network competition, but adopt either a three-party scheme or an issuer-controlled four-party 21 Author 9s calculation from Nilson Report.<br><br> 26 scheme. EFTPOS cannot be regarded as issuer-controlled because interchange fees flow from issuers to acquirers in this market. (Note: On the other hand, competition between four-party scheme and three-party scheme may fit well with Maneti and Somma 2002.) And most importantly, of course, interchange fees are now regulated in Australia, which likely has fundamentally changed pricing dynamics.<br><br> This cnew regime d must be taken into account in analyzing current Australian conditions. Other factors to consider in addressing the Australian situation include differences in acquirer and issuer margins, merchant strategy, consumer cardholding, and surcharging. Acquirers appear to maintain a constant margin regardless of interchange fee levels, while issuers 9 margins appear to be influenced by the level of interchange fees.<br><br> Most models assume constant margins on both sides of the market; only Wright (2004) considers interchange fee pass- through. Regarding merchant strategy, it is generally believed that the Australian retail industry is more concentrated than that in the United States. It is unclear, however, how competitive Australian merchants are in practice.<br><br> Merchants likely have a strategic motive to accept cards. Unlike in the United States, each network sets a single interchange fee for a typical point-of-sale transaction, that is, interchange fees do not vary by industry. This implies that heterogeneous merchants face a single interchange fee, consistent with Guthrie and Wright (2003, 2005); Rochet and Tirole (2003); Chakravorti and Roson (2004), Schmalensee (2002); and Wright (2004).<br><br> Regarding consumers, consumers typically pay an annual fee for credit cards with an interest-free period. To join a reward program, an additional annual fee is charged. Such endogenous cardholding with a fixed cost is assumed by Chakravorti and Roson (2004), Katz (2001), and Wright (2003).<br><br> 27 Merchants were not allowed to surcharge prior to credit card reform. Since reform, surcharging has been permitted, but few merchants reportedly have elected to do so. According to a recent survey, however, nearly half of Australia 9s merchants plan to apply surcharges to credit card transactions in 2006.<br><br> 22 To sufficiently capture developments in the Australian payment card market, future models will probably need to explicitly assume the option of merchant surcharging as well as interchange caps for four-party schemes. 4.3 United Kingdom 4.3.1 Network competition It is unclear to what extent network competition exists in the UK. Whether the two dominant networks, Visa and MasterCard, compete against each other in both credit and debit markets, in just one market (likely debit), or whether Visa focuses on credit and MasterCard focuses on debit, is an open question.<br><br> There are five credit card networks in the country. The two largest, Visa and MasterCard, together have a more than 90 percent market share. 23 In addition, the number of Visa and MasterCard credit/charge cards has been increasing in recent years, while the sum of those of other networks (American Express, Diners, and JCB) has not.<br><br> Purchase values show the same trend (Chart 12). In the debit card market, the two networks, Visa and MasterCard, have essentially equal (50-50) market shares (Chart 13). A typical UK bank is a member of both the Visa and MasterCard networks, but in issuing debit cards banks choose one brand or the other.<br><br> According to Cruickshank (2000), Switch 9s (now MasterCard) interchange fee was considerably lower than Visa 9s in 2000, suggesting that, on revenue grounds, Visa would be more attractive. However, 22 See, for example, American Banker vol. 170, No.<br><br> 148. 28 potentially offsetting this is the fact that MasterCard 9s debit card, Maestro, is more popular throughout Europe. Two other facts make the UK card market interesting.<br><br> First, unlike in most other European countries, debit cards have not markedly outstripped credit cards in terms of usage (Chart 14). Second, unlike in Australia, credit card-debit card network competition is subtle, if it exists at all, because there is no third network equivalent to EFTPOS. 4.3.2 Intra-network competition Both the acquiring and issuing sides of the card market are competitive in the UK.<br><br> The acquiring market is relatively concentrated: in 2002, the top two acquirers had 40 percent and 30 percent market shares, respectively. However, it is likely that the market share of the top three acquirers has declined recently, and the difference between merchant service charges (MSCs) and interchange fees in the UK is comparable to the average difference in the United States. 24 The issuing market is clearly competitive.<br><br> With respect to credit cards, no annual fees, free or low introductory interest periods, and cash rebates are broadly used. In addition, several U.S. issuers, including Capital One, Citibank, and MBNA, have entered the UK market in recent years, and their combined market share now accounts for 20 percent of credit cards issued.<br><br> Smaller UK banks have also entered the market (Table 8). With respect to debit cards, banks 9 debit card market shares correspond closely with the current account market shares. 23 European Payment Cards Yearbook (2004-5).<br><br> 24 According to the 2002 MSC Survey by Payment Systems Europe Ltd, average credit card MSC in UK has been stable around 1.5-1.6% from 1995 to 2002. According to Cruickshank (2000), average credit card interchange fee was 1-1.1%. In the United States, reportedly average MasterCard and Visa credit card MSC is around 2% and average interchange fees is around 1.5%.<br><br> 29 4.3.3 Matching theory and practice As noted above, the degree of network competition in the UK is difficult to gauge. Intra- network competition, on the other hand, exists. To the extent Visa and MasterCard compete in the credit card market, it can be characterized as an identical four-party scheme (Guthrie and Wright, 2003, 2005; Rochet and Tirole, 2002, 2003).<br><br> To the extent Visa and MasterCard compete in the debit card market, competition is again four-party, but it may or may not be identical network competition. According to Cruickshank (2000), Visa debit and Switch (MasterCard) interchange fees were quite different in 1998: the Visa debit interchange fee was at least twice as much as the Switch interchange fee. However, the more recent European Payment Cards Yearbook (2004-5) reports that the average interchange fee on Visa debit is thought to have fallen sharply from the figure reported in Cruickshank (2000).<br><br> Depending on how close the two networks 9 interchange fees now are, they may be regarded as almost identical. If they are not identical, Chakravorti and Roson 9s (2004) asymmetric network competition model may fit well. Although their model assumes a three-party scheme, it can also accommodate an issuer-controlled four-party network.<br><br> Since Visa debit issuers typically are not Switch issuers, an issuer-controlled four-party assumption may be valid. In terms of intra-network competition, it is hard to judge which side is more competitive or which side receives higher margins. However, in the UK, all networks are still subject to so- called cnet issuer rules d 4only issuers can be acquirers.<br><br> In addition, many aspects of merchant acquiring, such as transaction processing and recruitment of retailers, are outsourced to third- party service providers who do not have voting power. Therefore, a network 9s objective is likely to be weighted more on the issuer side. As a result, maximizing issuers 9 profits, maximizing the 30 number of transactions, or maximizing a weighted sum of end-user surplus (with a higher weight on the consumer side) are plausible assumptions.<br><br> These assumptions are made in Guthrie and Wright (2003, 2005) and Rochet and Tirole (2002, 2003), with network competition; and in Wright (2003), Schwartz and Vincent (2004), and Schmalensee (2002), without network competition. There are additional factors to consider as well. One, although the degree of competition among merchants is unknown, merchants likely have a strategic motive to accept cards.<br><br> If they did not, they may not have complained about credit card interchange fees to the Office of Fair Trading in the early 1990s. Two, credit and debit card interchange fee schedules are not publicly available. However, according to Cruickshank (2000), credit card interchange fees vary according to a number of factors, including whether a transaction is domestic or cross-border, whether it is a face-to-face or a mail order transaction, and on the level of information about the transaction that is provided to the issuer.<br><br> Visa 9s pricing in the UK may, therefore, be somewhat similar to Visa 9s pricing for EU cross-border transactions and, unlike in the US, a single rate may typically apply to retail POS transactions in the UK. This implies that heterogeneous merchants largely face a single interchange fee, as assumed by Guthrie and Wright (2003, 2005), Rochet and Tirole (2003), Chakravorti and Roson (2004), Schmalensee (2002), and Wright (2004). Consumer factors to consider: Consumers can hold credit cards with no annual fees, so endogenous cardholding with no fixed fees might be an apt description in the UK credit card market (Guthrie and Wright, 2003, 2005; Gans and King, 2002).<br><br> On the other hand, since the debit card is a demand deposit account product, debit card holding might be exogenous. Other factors to consider: UK merchants are prohibited from surcharging debit card transactions, but they are permitted to surcharge credit card transactions. However, most 31 merchants choose not to surcharge for credit card transactions; surcharging may require some costs to merchants.<br><br> Thus, interchange fees are not neutral, unlike the Gans and King (2002) prediction. On balance, the assumptions in Guthrie and Wright (2003, 2005) appear to fit UK payment markets well if Visa and MasterCard compete against one another. The model predicts that under network competition and merchant competition, merchant heterogeneity may raise interchange fees if networks place more weight on consumer surplus than on merchant surplus.<br><br> Since credit card reward programs are very popular in the UK, consumer surplus is likely weighted more heavily than merchant surplus. However, unlike the model 9s prediction, UK credit card interchange fees have been declining. The decline in interchange fees may not be a result of market equilibrium but may be due instead to regulatory pressure from the Office of Fair Trading.<br><br> If, on the other hand, Visa and MasterCard do not compete in the UK, none of the models closely match the UK market. However, one of Schmalensee 9s (2002) key results 4that under bilateral monopoly, the more weight the issuer has in the network 9s objective function, the higher the interchange fee 4may describe differences between credit and debit card interchange fees. In the UK, credit card interchange fees are higher than debit card interchange fees, and they are considered to be anticompetitive.<br><br> The Schmalensee result may fit if credit card network members are close to a bilateral monopoly and debit card network members are more competitive. 4.4 Netherlands 4.4.1 Network competition Network competition appears to be limited in the Netherlands. The country has two credit card networks (Visa and MasterCard) and one debit card network (Interpay).<br><br> 32 With respect to credit cards, MasterCard has a larger market share than Visa. This is true for both number of cards and overall (domestic plus foreign) transaction volume (Chart 15). Credit card usage in the Netherlands is quite low compared with debit card usage, however.<br><br> Interpay is the sole debit card network. It is PIN-based. Its share in the total Dutch payment card network (credit plus debit) is 74 percent in terms of number of cards, 90 percent in transaction value, and 95 percent in transaction volume.<br><br> In contrast with credit cards, Interpay 9s transaction volume and value have been increasing in recent years (Chart 16). 4.4.2 Intra-network competition It is difficult to gauge the degree of intra-network competition in the Netherlands. On the acquiring side, the Interpay network itself used to be the sole acquirer of PIN debit.<br><br> However, acquiring contracts have recently been transferred to individual banks. Data are not available to determine whether merchant service charges have since declined. On the issuing side, until a few years ago, only one Dutch bank (Fortis) issued Visa credit cards, while the other major Dutch banks (as well as Interpay) issued only MasterCard credit cards.<br><br> Today, several banks issue both brands, but credit card volume and value continue to be dwarfed by debit. Retail banking is highly concentrated in the Netherlands, with the top four banks accounting for 85 percent of Dutch current account market. As a result, concentration of debit card issuing is probably similarly high.<br><br> 4.4.3 Matching theory and practice Overall, the Schmalansee (2002), Wright (2004), and Gans and King (2002) models appear to best fit the Dutch card environment. All three are consistent with the lack of network competition. All three are also consistent with acquirers and issuers having some degree of market power.<br><br> Although surcharging is allowed in the Netherlands, only a small fraction of 33 merchants actually surcharge. In addition, although the Dutch debit card market does not have explicit interchange fees, it had implicit ones at the time Interpay was the sole acquirer 4 Interpay paid dividends to its owner financial institutions who were PIN card issuers. Today, acq<br><br>

less