But there was one bank in the country that had the resources, potential scale, and the corporate culture, to make it work: The Bank of America (BofA). BofA operated in California, a state which allowed banks to operate branches statewide.71 In the late 1950s, California was also one of the most populous and wealthiest of states, and BofA had a banking relationship with 60 percent of its residents and held more than 30 percent of its deposits. With assets of $5 billion, BofA was not only the nation’s largest bank, but also one of the largest in the world.72 As opposed to other large banks, BofA was also culturally predisposed to develop a credit card for the middle-class consumer. Most large banks of the 1950s did not engage in the extension of consumer credit, much less unsecured consumer credit, which was considered to be the domain of merchants and the less-than-reputable finance companies. Mandell writes that “If a bank had a consumer loan department, it was often found in the basement where no one could see the furtive borrower.”73 One critic of these early bank credit card systems complained that they were “low- ering banking’s image by engaging in an activity more properly associated with pawn shops.”74 Most bankers preferred to deal with safer, larger and more lucrative commercial loans. BofA, however, had a different organizational culture. Nocera explains that “It was a bank with the mentality of a finance company and proud of it.”75 It was started by A.P. Giannini, the son of an Italian immigrant, who prided himself on serving “the little fellow.”76 Ken Larkin, the BofA executive who would become synonymous with the BankAmericard program, saw it as a natural extension of the bank’s business: “We were always a leader in installment credit. Anything you could buy on time we financed . . . the credit card was just a natural extension of that.”77 But the BankAmericard, as their system was called, was a slight departure from the traditional consumer installment loan. Traditional loans were secured, meaning that the item being financed could be repossessed by the bank if the consumer de- faulted on the loan. The BankAmericard credit line was unsecured, and the bank had little recourse if a consumer could not pay. Traditional loans also required the consumer to apply for the loan, which would entail a review of the consumer’s credit position, and their intended use of the funds, by a bank loan officer. With the BankAmericard, cardholders could finance anything they wished without ever visiting the bank. It was a form of self-service credit that transferred the financing decision from the bank to the cardholder. In effect, it subtly blurred the line between buying and borrowing. The designer of the BankAmericard system, Joseph Williams, had friends at Sears and Mobil Oil, and he patterned his new system directly upon theirs.78 The card was offered to consumers without charge. Upon receiving their bill, cardhold- ers could pay the entire amount and not incur interest, or pay less than the total and finance the rest at the rate of 18 percent per year. Merchants were charged a six per- cent discount fee on their transactions, but would receive funds immediately upon deposit of the sales drafts without the need to bill or collect from the cardholders. Merchants also paid $25 a month to rent a card imprinter. The card itself was made of plastic with embossed account information, similar to the new American Express card. Just as in the case of the charga-plate, the embossed information was transferred to the sales draft using an imprinter, which reduced the potential of copy errors. Early imprinters did not have wheels for transferring the transaction date or amount, which had implications on machine processing that will be discussed in later chapters, so merchants hand-wrote these details on the sales slip, and customers added their signature to authorize the charge.79 The BankAmericard also had a formalized authorization process that was based upon the department store systems, but expanded to a multi-merchant environment. Each merchant was assigned a floor limit, over which the merchant was required to call for authorization.80 This authorization process will be discussed in more detail

in the following chapter, but it should be noted here that at this time it was entirely manual and exceedingly slow. The accounting side however was computerized from the very beginning, albeit in rather limited way. BofA was actually the first bank in America to use a computer, an IBM 702 installed in 1955, upon which SRI developed a program to automate BofA’s demand deposit accounts.81 To support the BankAmericard, BofA adapted this computer system to maintain cardholder accounts and process sales drafts. In- stead of using magnetic ink as they did with checks, the sales drafts themselves had a punch-card as the bottom layer, which would be punched with the transaction details upon deposit.82

“The Drop” Nocera writes that BofA approached the rollout of their system cautiously. They chose to test it in the relatively isolated town of Fresno, California, partly because

they had a banking relationship with 45 percent of the families there, but also be- cause they were not entirely sure it would work. They reasoned that if the card failed there, the bank’s reputation would be damaged less than if the test were conducted in San Francisco.83 Just as in the case of Diners Club, the BankAmericard system faced the clas- sic chicken-and-egg dilemma. Convincing merchants to give up six percent of the transaction and rent an imprinter for $25 a month would be possible only if there were a significant number of cardholders wanting to use the card. BofA solved the dilemma in much the same way all the previous systems did. They referred to it as “The Drop.” In the weeks leading up to September 18, 1958, they simply mailed 65,000 unsolicited cards to households in Fresno. They followed this up with ag- gressive advertising to educate those consumers about their new card.84 As in the other systems, merchants faced a central tension when deciding whether to participate. Accepting the card meant possibly losing the customer loyalty built by the merchant’s private card system, but not accepting the card put the merchant at risk of losing customers to a competitor that did. In general, the larger merchants with established card systems, such as Sears, JC Penney and Wards, did not ac- cept the BankAmericard, but the smaller merchants did. These smaller merchants considered the loyalty to be less valuable than the costs of maintaining, billing and collecting on credit accounts. Ken Larkin recalled one merchant he visited: He had three girls working on Burroughs bookkeeping machines, each handling 1,000 to 1,500 accounts. I looked at the size of the accounts: $4.58. $12.82. And he was sending out monthly bills on these accounts. Then the customers paid him maybe three or four months later. Think of what this man was spending on postage, labor, envelopes, stationery! His accounts receivables were dragging him under.85 A similar set of reasons was discovered by Stallwitz in his survey of Bay Area merchants in 1968. Interestingly, the reasons given were striking similar to how the cards were marketed to merchants: it would increase their sales; reduce their risks and costs of extending credit; for competitive reasons; and for customer con- venience. He noticed, however, that suburban merchants also tended to accept the card more as a personal favor to their bankers, primarily because those bankers had given them their initial business loan. He continued: “The most unusual reason for joining was put forth by one merchant who had just moved into a new store (not a related type of business) and a . . . sticker was already on the door. Rather than scrape it off, he signed up on his next trip to his primary bank.”86 The BofA managed to get merchants interested, but they also needed to solve one more problem to make the system truly work: how would cardholders recognize merchants that accepted the card? Diners Club had simply provided a list of estab- lishments to cardholders, but the number of merchants accepting the BankAmeri- card would potentially be far greater and more diverse. Cardholders needed a simple

way to identify merchants that accepted the card without consulting a pre-printed list. Their solution was to create a mark for the system, which would be printed on both the cards as well as signs that hung in the windows of participating merchants. The mark looked a bit like a flag. It consisted of three colored bands—blue, white, and gold—running across the background, with the word BankAmericard written in white within the blue band. This was a new twist on the function of identity discussed earlier. Not only would the card identify the cardholder’s account to the system, the mark on the card would also identify a participating merchant to the cardholder. The Fresno drop went smoothly, though it attracted little attention, and con- sumers began to use the cards. Today it seems normal for consumers to use credit cards to pay for goods and services, but one should question why the people of 1958 Fresno would bother. Although the Diners Club card provided an obvious benefit to the traveling business person, the general consumer shopping locally would actually have little need for such a device. Checks were commonly accepted at local shops, most merchants already extended credit to their frequent customers, and anyone who could qualify for a BankAmericard could no doubt qualify for a traditional in- stallment loan. Considering the manual authorization process of the time, using the BankAmericard for purchases over the floor limit would have been more time con- suming, awkward and embarrassing for the cardholder than writing a check. Nocera offers three possible reasons why the card was adopted. First the BankAmericard fit with the general trend in America toward impersonal self-service. Applying for a traditional installment loan meant looking a loan officer in the eye and promis- ing to repay the loan. The BankAmericard arrived unsolicited and the decision to finance a purchase was entirely left up to the consumer. Second, the card did offer a level of convenience through consolidation. Instead of maintaining accounts and carrying cards from multiple merchants, the consumer could carry one BankAmeri- card and see all their charges on one bill. But his third reason is perhaps the most convincing—consumers used the card because it was a novelty in an age and culture of novelty. Nocera writes: The card was a novelty at first; just as Americans spent hours staring at the test pattern of their new TV, so did the citizens of Fresno gather around the checkout counter to watch someone pay with a BankAmericard. This was the 1950s, after all, a time of wonder at the miraculous march of progress. BankAmericard was part of that march.87 Shortly after the Fresno drop, BofA learned that their competitors were planning to launch a card system of their own in San Francisco.88 Nocera writes that upon leaning this, all caution was put aside. BofA began sending cards to nearly all their depositors and consumer loan customers in California, and using their extensive network of branches to enlist as many merchants as possible. Larkin later remarked that this was “a calculated risk, done just the one time to get the plan off the ground,”

and they did target their efforts a bit by using lists of existing charga-plate holders provided by the retail service bureaus.89 Over the next 13 months, BofA issued 2 million BankAmericards and signed up over 20,000 merchants.90 It is important to note that in one move, BofA created more cardholders just in California than Diners Club had ever had nationwide. As the cards reached the urban center of Los Angeles, BofA began to experience the inevitable effects of sending out millions on unsolicited, unsecured credit cards. The problems of fraud in the BankAmericard system will be discussed in more de- tail in the next chapter, but the initial effects were staggering. Delinquencies were 22 percent, compared to 4 percent on traditional installment loans. Both thieves and merchants were creating numerous fraudulent transactions under the floor limits to avoid detection. Within 15 months of the Fresno drop, the BankAmericard system had officially lost $8.8 million, but Nocera estimates that the real losses were actu- ally closer to $20 million.91 Despite these early losses, BofA chose to continue the system, reducing the num- ber of outstanding cards and weeding out dishonest merchants. By May 1961, the BankAmericard system was generating a profit, though this was initially kept quiet as the publicity from the early losses was still helping to keep other banks from developing competing systems of their own.92

As noted earlier, by the mid 1960s the BankAmericard system had overcome its initial difficulties and was generating increasing profits, but it was still restricted to the state of California. BofA realized that both consumers and commerce were increasingly traveling across state lines, and for their card to be truly useful, it had to be accepted nationwide. American banking regulations at the time prohibited BofA from opening branches in other states, so they decided the best way to expand the system was to license the program to banks in other states. Although BofA could have legally solicited cardholders outside of California, it would have made little sense to do so. There were no centralized credit reporting agencies at this time, so BofA had no way to establish the credit worthiness of prospective cardholders in other states without the help of a local bank. Furthermore, directly signing up merchants would have been extremely difficult, as the merchant would have had to maintain an account with a BofA branch in California. BofA created a subsidiary organization known as BankAmericard Service Corporation (BASC) that was tasked with signing up licensees and administering the entire system. Licensee banks paid BofA $25,000 for the franchise, plus a percentage of their transaction revenues as a royalty. In return, they received the accounting software developed for the BankAmericard system, as well as an invitation to a training ses-

sion in San Francisco.99 Much to the dismay of the licensee banks, however, this training session was given by the marketing department, and much of the discus- sion revolved around the marketing aspects of the program. Many of the licensees discovered that they could obtain more helpful and accurate information on how to run their programs by visiting the BankAmericard processing centers, directly observing and talking with their operations people.100 Initially BofA licensed only one bank in any particular geographic area, essen- tially providing it with a local monopoly. These licenses were mostly domestic, but a few were located in other countries; Barclays Bank became the first international li- censee in 1966, and the sole BankAmericard issuer in the UK. These licensee banks typically had correspondent relationships with BofA, and thus were “loyal” or at least tied to BofA in some sense. Although this practice might seem a bit exclusive or restrictive, it was likely a necessary consolation in order to entice banks not only to pay the license fee and royalties, but also to give up their chance to issue a card with their own brand.101 The licensing system also created a new function never before seen in payment card systems: interchange. Because cardholders from one bank could now use their card to make purchases at merchants represented by a different bank, the two banks needed a way to clear and settle those transactions. In all previous payment card sys- tems, the same organization both issued the cards and represented the merchants, so all settlement and clearing was done within the same organization. Check payment systems had always experienced this scenario, and as discussed earlier, the Federal Reserve had established a national clearinghouse in 1915 for just this purpose. But in a move that may have perhaps sown the seeds of its own destruction, BASC chose not to create a centralized clearinghouse for the BankAmericard system. Instead, acquiring banks were required to mail their drafts directly to the issuing bank for payment, less a discount fee, now called the interchange reimbursement fee. This will be discussed in more detail in the next chapter, as it became one of the central reasons for the creation of a new independent organization. The banks that competed with the BankAmericard licensee banks quickly re- acted by forming regional, non-profit cooperative associations of their own. In many cases, these regional associations were also centralized processors. The members still issued cards, signed merchants and held the receivables, but the regional associ- ation provided the more mundane operational functions such as authorization, sales draft processing, accounting, and billing. By centralizing their operations, these as- sociations could also achieve an economy of scale, which reduced the operating costs for each of the members. These regional associations then joined together into a national, non-profit coop- erative association known as Interbank in order to allow their cards to be used across the country. National merchant acceptance was hampered, however, by their lack of a common name on the card. Although all Interbank cards contained a common mark, it was only a small “i” in one corner of the card, barely noticeable compared to the regional association’s name and marks, which varied from region to region. In contrast to the BankAmericard system, which used and promoted a consistent name and mark, the Interbank system did not actively promote their common mark, and thus cards were not as readily accepted outside their issuing region. In 1969, Interbank began to address this problem by purchasing the rights to the name first developed by First National Bank of Louisville, Kentucky, and the mark popularized by the Western States Bankcard Association (WSBA).102 The name was “Master Charge” and the mark was the overlapping yellow and orange balls, and these were eventually used on all cards issued by Interbank members. In the 1980s, they changed names again to MasterCard and in 2006 became an independent, for- profit stock corporation.

65According to Mandell, The L. Bamberger and Company department store was the first to develop a system with revolving credit in 1947 and it was quickly adopted by the major New York stores such as Gimbels and Bloomingdales. In 1956, J. L. Hudson’s of Detroit added the idea of an interest-free period. See Mandell (1990), pp. 24–25. 67Mandell (1990), p. 26, Struble (1969), p. 4. Note that many sources claim that Franklin National Bank was the first “credit card,” but a close reading of Struble reveals that although he used the that term, the Franklin plan did not offer revolving credit and was thus similar to the T&E cards. Struble wrote in 1969, and the terms defined above were not yet used consistently. Other sources have suggested that Franklin first offered a non-revolving card, and then later added the revolving credit feature, thereby becoming the first bank credit card. 71Nineteen states and the District of Columbia allowed statewide branching, 16 allowed limited branching, and 15 enforced unit banking. See Goldberg (1975). 79Merchants were also supposed to fill in details about the goods purchased, but most found this too time consuming and neglected to do so. 80The term floor limit comes from the department stores, where it literally meant the amount un- der which the “floor” could authorize. Any amount above require a telephone call to the finance department (Powar interview). 81The system was called Electronic Recording Machine-Accounting (ERMA). That system auto- mated the processing of checks through the use of magnetic ink printed on the back of the checks, a technique that would eventually be used in the Magnetic Ink Character Recognition (MICR) standard in 1957. According to O’Brien, the first ERMA installation was demonstrated in San Jose in early 1956, though the BofA did not fully convert all their accounts to ERMA until 1962. See O’Brien (1968), pp. 2–5, and Campbell-Kelly (2003), p. 49. 88According to Business Week, this was the First Western Bank and Trust Company of Los Angeles. See ‘The charge-it plan that really took off’, Business Week (27 February 1965), p. 58. 89‘The charge-it plan that really took off’, Business Week (27 February 1965), p. 58. 90Nocera (1994), p. 30. 91Nocera (1994), p. 31. 92‘The charge-it plan that really took off’, Business Week (27 February 1965), p. 58. For the ac- cusation that it was purposely kept quiet, see Galanoy (1980). Mandell actually questions this profitability, claiming that BofA did not include the cost of funds or advertising in the accounting of the card system, thus making it look profitable when it actually was not (Mandell 1990, p. 58). 93Mandell (1990), p. 30, Jutilla (1973), p. 44. 94‘The charge-it plan that really took off’, Business Week (27 February 1965), p. 58. 95Mandell (1990), p. 40. Mandell actually claims that Visa “acquired Uni-Card,” implying that Visa purchased the portfolio, but this would be out of character for that organization, which was an association of card issuers and acquirers. Visa’s chief of operations at the time confirmed that NBI did not purchase anything from Chase, and that Chase merely joined Visa and offered a Visa product (Russell interview). 96Struble (1969), p. 5. 97Struble (1969), p. 5. 98O’Brien remarks that the 360s were so much more powerful that they easily justified the cost of porting applications. In 1966 BofA replaced their 32 GE mainframes used for the ERMA system with just two IBM 360/65s (O’Brien 1968, p. 14). 100This was mentioned in interviews with Don Jutilla, director of one of the first BankAmericard licensee programs. Jutilla created flowcharts from his observations that were then used to train his staff. 101Initially banks were not allowed to add their brands to the card. See the next chapter for more details.