The Card and Marks One important distinction between the early Interbank and the BankAmericard sys- tems was their approach to the physical design of the card and usage of the adver- tising marks. Because the Interbank system was comprised of many pre-existing regional systems with different established names and marks, Interbank required only a small “i” logo to be placed in one corner of the card’s face, while the rest of the card used the local name and marks. This made it difficult for cardholders to realize that their cards were accepted at merchants outside their local areas, and for merchants to recognize acceptable cards from different regions, as the primary marks on the cards and merchant windows would be completely different. The com- mon “i” logo was simply too small and insignificant to be recognized intuitively by merchants and cardholders as the common mark. The early BankAmericard system, on the other hand, was built by licensing the BankAmericard name and its blue-white-and-gold (BWG) bands design to other banks, so the domestic card designs and merchant signage were almost entirely uni- form. A BankAmericard issued by BofA in California looked almost identical to a BankAmericard issued by banks in New York or any other state, the only differ- ence being that the local bank was allowed to include their own name in addition to “BankAmericard,” but only in smaller, less-prominent type. Many of the inter- national licensees were allowed to use more culturally-appropriate names, but they were still required to use the BWG design across the entire face of the card, making it fairly obvious that they were at least related if not the same. The NBI operating regulations continued this approach, and as I will describe in Chap. 6, the international and domestic cards were eventually brought under a common name and graphic design philosophy. This made it easier for cardholders traveling abroad to quickly and reliably identify merchants that accepted the card, even if the cardholder and merchant did not speak a common language. It was not until the late 1980s that the BWG band design was reduced to a smaller logo on the card face; by then, the recognition of, and expectations about, the marks were established firmly enough that there was less risk in giving the member banks more of the card face for graphical differentiation.7 The graphical consistency of the cards and merchant signage also allowed NBI to conduct national advertising on behalf of the member banks, the details of which I 7Chutkow (2001), p. 215. 58 3 Crafting the Social Dynamics: Staffing, Operating Regulations, and Advertising will describe later in this chapter. Although this could be seen as simply an efficient centralization of a common task, we shall see that NBI’s intentions for these early advertising campaigns were actually far deeper, and the effects were much more powerful. Inter-Organizational Work and Fees The second relevant area of the operating regulations covered how the inter- organizational work was to be accomplished, what fees would be charged for the various interchange services, and which party would be responsible for fraudu- lent or disputed transactions. Although the operating regulations did stipulate some of the rules by which acquirers interacted with merchants (e.g., merchant screen- ing, inspection and auditing) and issuers with cardholders (e.g., common require- ments for cardholder agreements), most of these regulations pertained to the inter- organizational work conducted between member banks and NBI; how the banks interacted with their merchants and cardholders was largely left up to the individual banks, subject of course to the relevant Federal and State banking regulations. The most important inter-organizational work conducted in the NBI system was the authorization, clearing, and settlement of interchange transactions, and the oper- ating regulations contained numerous, explicit rules dictating exactly how each step should be accomplished. Having a clear and comprehensive set of rules was quite crucial; by the time NBI formed there were already thousands of banks participat- ing in the system in some form or another, and both the number of participants and the transaction volumes were growing by about thirty percent per year. In order to keep up, especially once interchange authorization and clearing became electronic (see the next two chapters), each bank needed clear directions as to how it should submit and receive transactional information sent through the system, how to handle abnormal cases, and how to rectify errors. One particularly important aspect of these rules was ascription of liability for fraud. As opposed to other forms of payment, NBI transactions were (and still are) guaranteed from the merchant’s perspective provided the merchant follows the rel- evant rules stipulated in the operating regulations. In the early 1970s, the rules dic- tated when the merchant was required to call for authorization, what transactional details must be captured on the sales draft, and how the merchant was to submit the drafts to the acquiring bank. The following of these rules transferred not only funds from the cardholder to the merchant, but also the liability for fraud from merchant to issuer. Cardholders and their issuers could still dispute the transactions (the process for which is described in detail in the next section), but unless they could show that the merchant or acquirer did not follow the rules, the issuer was responsible for the loss. In addition to the rules stipulating how inter-organization work was to be accom- plished, this section also contained a series of rules regarding the fees assessed by the system for various services. These fees constituted the basic economic dynamics of the system, most of which are still at play today. The Operating Regulations 59 When crafting these basic economic dynamics, the NBI staff had to answer a set of questions that are essential to any payment system: who pays, who benefits, and who gets to decide?8 Although it might seem a bit self-referential, payment systems cost money to operate; every exchange of money requires a bit of that money to facilitate the exchange itself. All payment systems have fixed costs, and most also have per-transaction costs that must be paid by someone if the system is to continue operating. The question is, who should cover these costs? The party on the buying end, the party on the selling end, or some combination of the two? Cost reimbursement is certainly required, but fledgling cooperative payment sys- tems like NBI also had to concern themselves with another, perhaps more important economic dynamic: balancing incentives in order to attract banks, merchants and consumers to participate in the system in the first place.9 Payment systems are use- ful only when buyers, sellers, and their financial agents (i.e., banks) all decide to participate in sufficient numbers, and one of the factors that determines their deci- sions are the fees they have to pay in order to participate. In the early 1970s, NBI was still the minor player in the credit card industry compared to Interbank, and credit cards accounted for only a small percentage of overall consumer payments, so NBI’s new prices had to be high enough to attract new card-issuing banks, but low enough to allow acquiring banks to offer competitive discount rates to new merchants. From a system competition perspective, the question was also: who is willing to pay and how much can we charge yet still attract new participants? There are no “natural” or perfect answers to these questions, and in most cases, the answers depend heavily on the systems’s historical context. They may also change over time as participants negotiate with each other. In the United States, the costs of cash and the national check clearing systems have traditionally been funded by taxpayers and administrated by the Federal Reserve, but NBI and other private payment networks typically do not have access to public funds, and thus must raise their own funding by charging some set of their participants for their services. In the case of NBI, they inherited the model whereby the merchant pays for the service through a transaction discount while the cardholder pays nothing. The BofA adopted this model because it had been well-established by the various travel and entertainment cards of the 1950s and 1960s, and the merchant-specific cards before them. By the 1970s, consumers had become accustomed to the idea that credit cards were free if balances were paid each cycle, and would likely resist additional usage fees, though several banks did attempt such charges as the cost of credit rose. Merchants were willing to pay the discount on credit cards primarily because they were used for a small percentage of their overall transactions, and the card systems provided studies showing that consumers purchased more on average when they were allowed to use a credit card instead of cash or check. As we shall 8For a discussion of this in the context of the UK’s planned point-of-sale electronic funds transfer system, see Howells and Hine (1993). For an excellent analysis of the way different value systems can influence answers to these questions, as well as design priorities, see Kling (1978). 9For an expression of this in the language of multisided platform economics, see Evans and Schmalensee (2005). 60 3 Crafting the Social Dynamics: Staffing, Operating Regulations, and Advertising see, when the percentage of credit transactions rose, and when debit cards with discount fees were introduced to replace checks, the merchants attempted a revolt, challenging the fees in court. Although the basic “merchant-pays” model was inherited and not likely to change, NBI still had to determine how much merchants should pay, and how much of that should flow back to the card-issuing banks. This was ultimately determined by the interchange reimbursement fee (IRF), first discussed in Chap. 2, which was the percentage acquirers paid to issuers during the settlement of interchange transac- tions. This fee would effectively set a minimum amount for the merchant discount, as acquirers had to charge merchants something higher than the IRF in order to make a profit. Too high of a fee would discourage new merchants from accepting the card, but too low of a fee would discourage banks from issuing the cards. The interchange reimbursement fee has always been a source of controversy in the NBI/Visa system. At the core of the issue is a basic question: is it a form of anti- competitive price-fixing, or a necessary aspect of a cooperative payment system? Because this fee is agreed upon by a group of competitors for their own benefit, and because it effectively establishes a minimum for the merchant discount fee, many merchants have argued that this is indeed anti-competitive, and on occasion the US Department of Justice has agreed.10 Visa has typically responded that all cooperative payment systems require these kinds of fees, not only to help cover the participants’ operating costs, but also to provide the correct economic incentives for both issuers and acquirers. From Visa’s perspective, the fee is a method of obtaining an economic balance, providing both acquirers and issuers with sufficient profit while remaining competitive with other payment card systems.11 This shift in language from “cost reimbursement” to “economic balance” is of course not accidental; as automation reduced transaction costs as well as fraud, and as the cost of credit reduced in the 1980s and 90s, the rate could no longer be justified purely as a cost reimbursement. Over the years, this fee has also created controversy between the member banks, as each has tried to tip that balance in their own favor. On average, this fee con- stitutes ten percent of an issuer’s revenue, so banks that specialize in issuing often want to increase this fee.12 Not surprisingly, those that specialize in acquiring fight to reduce it. When NBI formed, most member banks performed both roles, but as the business matured, banks began to specialize in one role or another, creating a certain partisanship regarding the direction, and eventual segmentation, of the interchange fee. The primary issue with the interchange fee under the licensing system was that it was vague and unrealistic to audit, so NBI’s staff realized that they needed to establish a fixed interchange fee that was consistent and unambiguous. But they were faced with a difficult problem—how should they calculate the rate? Setting a proper rate depended on knowing not only the true costs of interchange transactions, 10For a review of Visa’s antitrust battles, see Mann (2006) and Evans and Schmalensee (2005). 11For a very detailed theoretical model of four-party payment systems and the need for an inter- change fee, see Baxter (1983). 12Schmidt interview. The Operating Regulations 61 but also the entire economics of the system, neither of which were well understood in 1971. The industry was still very young, and most bankcard programs did not delineate their costs by function, much less by transaction type. Many of the small rural banks barely had accounting systems at all.13 It was clear that more information was needed to understand the economics of the business. To begin gathering it, NBI established a requirement in late 1970 that all member banks must submit a “certificate of sales” each quarter, which reported their basic operational information such as number of accounts, sales volume, delin- quencies, charge-offs, and the like. This information not only helped NBI calculate member fees, but also provided them with the data they needed to build a basic eco- nomic model of the business. This information was eventually compiled and sum- marized by Ron Schmidt into a quarterly “profit analysis report” that was then made available to all the membership starting in third quarter of 1971.14 Thus, for the first time each member could now see how their program compared to others in the sys- tem, as well as how the system was performing as a whole.15 Furthermore, Hock could use these reports at the Board meetings to prod directors of under-performing banks to improve their programs. As with any network industry, improvements in one element of the system often brings about a benefit to all. The quarterly reports also served another important function at the time. Throughout 1970 and 1971, industry papers such as the American Banker were publishing articles about the huge losses incurred by bankcard programs, question- ing whether this endeavor could ever be profitable.16 According to these articles, many banks were considering giving up on the card business altogether. The quar- terly reports provided a concrete rebuttal, showing the member banks that some of them were indeed beginning to turn a profit.17 Furthermore, each bank could now see how their programs compared to others, creating an incentive for improvement. But most importantly, the reports encouraged the member banks to ask NBI to take an active role in helping the individual programs achieve profitability; if a bank was struggling, they could call NBI and a team would visit them to study their proce- dures and costs, and to share best-practices learned from other banks in the system. NBI was thus helping banks learn from one another for the benefit of the entire system. 13Schmidt interview. 14Schmidt interview. See also Brooke (6 August 1971), p. 1. 15These reports were actually quite detailed. Banks were grouped by the sizes of their portfolios, whether they did acquiring, issuing or both, and whether they used a third-party processor or not. 16For example, see the series of articles that begins with Brooke (18 May 1971), p. 1. In the 14 June edition (p. 5), there are a few letters to the editor in response to this series. In one, a bankcard manager wrote “My very candid opinion is that, after having operated a credit card operation, I can see nothing in store for the future of this operation but disappointment. Evidently, many banks are reluctant to admit their mistakes and prefer to continue to lose money for their institutions rather than admit they were wrong.” 17According to the American Banker, one third of the member banks were profitable by 1971. See McKenna (19 June 1971), p. 3. 62 3 Crafting the Social Dynamics: Staffing, Operating Regulations, and Advertising The quarterly reporting data helped NBI understand the general economics of the system, but they still did not contain enough detail about costs to determine an optimal interchange fee rate. Thus, NBI began a detailed research project that would eventually produce what was known as the “Functional Cost Study,” first published in the fourth quarter of 1971. While, the quarterly profit analysis reports provided a high-level overview, the functional cost study was a very detailed examination of the functions performed, and costs incurred, by a typical program. Most banks had never examined their programs this closely, so a team comprised of NBI staff and Arthur Andersen consultants visited the bank, observed their operations, and studied their accounting records.18 This process required about two weeks of research in each bank, plus two or three weeks of additional analysis, but the results provided the bank and NBI with detailed information about costs and problem areas. The bank then knew where to target their efforts, and NBI could construct a more accurate economic model of the system. All of this information would eventually help NBI set an optimal rate for the IRF, but unfortunately NBI could not wait for all the studies to be compiled. The rules governing the IRF under the licensing system needed to be fixed, and that meant specifying a new fixed rate in the operating regulations, but Schmidt was still in the field researching the member banks. He was asked to estimate a new rate based on his work so far, and the model he and his Arthur Andersen consultants were building calculated a rate around 2.6 percent, but this seemed too high to Hock. At this time acquirers were sending issuers about 2 percent on the average, and according to Schmidt, Hock felt that the new rate could not be any higher than that, so he proposed a rate of 1.95 percent instead. Interestingly, Schmidt remarked that in his opinion, 1.95 percent was suggested instead of an even 2 because it would appear as if it had been arrived at through sophisticated calculation, and not simply decided upon through instinct.19 Dispute Resolution The last major area of the operating regulations to discuss concerned how disputes over interchange transactions were to be resolved within the system. Dispute res- olution is a critical feature of any cooperative network of competitors; without it, the competitive members would have no way to resolve disagreements that invari- ably arise, especially when money is involved. From the beginning, Visa defined a chargeback and arbitration process that would provide a reasonably fair and equi- table method for resolving these disputes.20 18These were called “Profit Improvement Teams (PITs)” (Honey interview). 19Schmidt interview. Of course, NBI continued to adjust the fee based on the results of the studies once they were available. 20Information about dispute resolution comes from interviews with Tindal, Baum, and Kollmann, all of whom managed the process and served as arbitrators. The Operating Regulations 63 The process has remained essentially the same throughout the years, though it has been modified slightly based on experience. The process typically begins when a cardholder complains about a particular transaction. This complaint can be raised for a number of reasons: the cardholder did not conduct the transaction (or does not wish to admit to it21); the transaction was keyed incorrectly or was processed multiple times; the merchandise was “not as advertised”; the merchandise was never received; was defective or damaged in shipment; etc.22 If the issuer thinks the com- plaint is valid, the issuer then submits a chargeback transaction through interchange. Because the original transaction had already been cleared and settled, the issuer es- sentially puts through a compensating transaction, “charging back” the acquirer for the original amount plus a penalty fee. The acquirer then has a certain number of days to research the transaction on their side, after which they can either close the case or “represent” the transaction through interchange. Originally, the operating regulations allowed the transaction to bounce back and forth a few times, but today, the issuer and acquirer must resolve the dispute after one cycle, or submit it to Visa for arbitration. Initially, the group that handled chargeback disputes within NBI was the same group responsible for maintaining the operating regulations. In the early 1970s, in- terchange volumes were still rather low, and disputes were less common, so the arbitration process was simply a side job of those maintaining the regulations. As interchange volumes increased, so did the disputes, and Visa reacted by developing a more complex, formalized arbitration process, and staffing positions within the operating regulations area to manage it.23 This group’s primary responsibility is to shepherd cases through NBI’s formal ar- bitration process. When two members cannot resolve a chargeback dispute amongst themselves, they submit it to this group, along with all the supporting documenta- tion, and a fee to discourage frivolous cases. In the early years, the cases were re- viewed and decided by an ad-hoc committee composed of mid-to-upper-level man- agers, but today the dispute resolution group itself acts as the “court.” In most cases, the applicable rules are fairly clear and unambiguous, but the cardholder/issuer and merchant/acquirer testimonies conflict. The court then does their best to review all the available evidence and decide which party is correct. In some cases, however, the dispute is based on differences of interpretation; that is, the evidence might be clear, but the two parties have different opinions as to 21Several interviewees indicated that male cardholders would occasionally conduct morally ques- tionable transactions that they later regretted, and instead of admitting to them, would claim their card had been lost or stolen. Unfortunately for these cardholders, their signatures on the original sales drafts would be retrieved, making the transactions more difficult to deny. 22Consumer protection laws in the United States allow cardholders to refuse payment for goods that are “not as advertised” or defective if they are purchased within a certain, limited geographic range from the cardholder’s residence. In other countries this may not be a legitimate complaint. 23Unfortunately, statistics on the number of disputes raised for arbitration are not available to the public, but interview sources indicated that they were relatively few, perhaps a dozen per week, less than a tenth of a percent of all chargebacks. The dispute resolution staff actually spends more time educating and fielding general questions from the membership. 64 3 Crafting the Social Dynamics: Staffing, Operating Regulations, and Advertising whether their actions adhered to the rules. For example, the rules require merchants to compare the signature on the sales draft with the signature on the card, but what constitutes a match? Two signatures that might seem similar-enough to a merchant might seem completely different to an issuer or cardholder. Initially, the rules did not clarify what it meant for two signatures to “match,” and the members of the arbitration committee were forced to make their own judgments on a case-by-case basis. As the number and dollar value of these kinds of disputes increased, this quickly became untenable, and the rules were eventually modified to require only that the content of the signatures matched, and not the hand in which they were written.24 Although they may be rare, these differences in interpretation are always a pos- sibility in rule-based systems. As Wittgenstein argued, a rule in itself cannot fully specify what it means to follow or not follow the rule.25 The interpretation of a rule is necessarily bound up with participants’ attempts to follow it, disputes over those attempts, and judgments made by a commonly-recognized authority as to whether those attempts were correct or not. In this case, Visa acts as the commonly- recognized authority, establishing a canonical interpretation of the rules through their arbitration decisions. As discussed in the signature example, NBI/Visa also has the luxury of modifying rules that are seen as too vague or problematic. Visa’s ultimate goal is to minimize disputes and arbitration cases, so if particular rules cause too many interpretation disputes, Visa suggests changes to clarify them. Although the operations committee must approve those changes, they have little incentive to resist, as disputes reduce the overall efficiency of the system, and in turn, the profitability of the member’s card programs. Additionally, the dispute resolution group takes an active role in educating the membership about the chargeback and arbitration rules. The operating regulations are actually quite detailed, and currently fill multiple printed volumes, so it is not un- common for a member to be unfamiliar with certain sections. The resolution group helps to educate the members on the entire process, including their rights and obli- gations. They also send out regular communications, informing the members of new rules and their implications, or recent arbitration decisions that might help clarify how the rules should be interpreted. The decisions handed down by Visa’s “court” are typically final, but under certain circumstances, such as high-value transactions, the loser is allowed to appeal the decision. There are currently several layers through which the case can be appealed, with the Board of Directors acting as the final, supreme arbiter. Interestingly, Visa’s arbitration process is not entirely binding; the members still retain the right to bring suit against one other in their country’s legal system if they do not like the result of the arbitration. 24Honey interview. Note that this rule typically applies to fraudulent transactions involving a lost or stolen card that has since been recovered. If the contents of the signatures on the sales drafts do not match the card, the merchant or acquirer must absorb the chargebacks. If the contents do match, even if they are in a completely different hand, the issuer must absorb the loss. 25Wittgenstein (1958), §201–202. The Operating Regulations 65 It should be noted here that although Visa’s chargeback and arbitration process is well designed, it is by no means perfect. Powerful members can and do abuse the mechanism for their own gain. Smaller banks have been known to pay questionable chargebacks from intimidating issuers without a fight, primarily because they lacked the resources to investigate the matter and collect the necessary documentation. Those reviewing the cases are humans, and as such, can never be completely impar- tial. Lastly, members have occasionally felt that the arbitration process did not pro- duce the correct result, and have chosen to continue fighting in the civil court system.