When Citibank refused to give UC Berkeley student Kennedy K. S. Yip a MasterCard last spring because he was a humanities major, consumers got rare insight into a controversial process many lenders use to determine how to award billions of dollars in credit and loans.
Traditionally, banks and other lenders relied on the five “Cs” in awarding credit: Collateral, cash flow, capitalization, condition and character, said Gail Popyk, assistant vice president at Comerica Bank in Detroit.
But, increasingly, lenders such as Citibank are turning to complex “credit scoring” systems to review consumer loan applications and make lending decisions based on statistical models that purport to identify people who are the best credit risks.
The systems often utilize computers to evaluate demographic and financial data such as an applicant’s level of indebtedness, income and even, on occasion, such factors as academic major, age, home ownership and residential ZIP code.
Student protests prompted Citibank this summer to stop using majors in its student scoring model. But more and more lenders--convinced that computer scoring is more efficient and fair than the old judgmental method of using only loan officers--are relying on other kinds of credit scoring systems to steer them away from potential deadbeats.
“In a high-volume credit operation, credit scoring is the only way to go,” said Don J. Sanders, director of credit policy for Commercial Credit Corp., a consumer finance company in Baltimore that uses a credit scoring program. “I’d say if you get 500 (loan) applications a week or greater, you can benefit from scoring.”
But critics say computerized credit scoring can lead to biases and mistakes more harmful than those exhibited by quirky human loan officers.
They say some scoring criteria have been shown to be racially discriminatory and that unsuccessful credit seekers are informed only in the most general terms why they were denied credit rather than which specific criteria they scored poorly on. What’s more, critics say, even the most bias-free scoring systems can sometimes overlook people, like Yip, who are credit worthy.
Yip, a 23-year-old rhetoric major, already had an American Express card and two Visa cards from banks that did not consider his major a credit factor. But Citibank deemed humanities students poorer risks than business or engineering majors. Citibank relied, in part, on such information because “it’s tough to judge a college student’s credit-worthiness when they don’t have much credit history,” said Citibank spokesman William J. Ahearn.
“The problem we have with any type of credit scoring system is that it is biased towards middle- and upper-middle-class type people,” countered Ken McEldowney, executive director of Consumer Action, a San Francisco-based consumer group. “When you are dealing with the aggregate, it unfairly discriminates against people at the extremes who may be perfectly good credit risks.”
Jerry Findley, a banking consultant and editor of a Brea-based industry trade publication bearing his name, calls claims of greater accuracy and impartiality with credit scoring “just a bunch of sales talk. The character of an individual is much more important than (a credit score.) You can’t decide who to lend to by using a computer.”
Berkeley student Yip agrees. “I asked them (Citibank) if they have the statistics to back up their claim that humanities students don’t pay their debts,” he said. “But they said they don’t have to show me anything.”
Issue Never Litigated
In credit scoring systems, computers analyze past credit applications and the payment histories of a selected pool of applicants and determine which are the best credit risks. The computer then establishes a formula under which new applicants accumulate points toward an ideal “model” based on answers to various demographic and financial inquiries about annual income, home ownership, etc. The number of points are then added up and applicants who fall below a certain cut-off point established by the lender are usually denied credit.
Jonathan Jerison, a lawyer with the Federal Trade Commission’s consumer credit division, said the federal Equal Credit Opportunity Act is unclear about which scoring criteria are discriminatory because the issue “has never been litigated.” But he said he believed that criteria such as age, a student’s major and criteria that evaluate applicants by location could be problematical: “If you have a practice that appears neutral on its face, but in fact has an adverse impact on a protected group, it can be illegal.”
Yet in the current climate of bank consolidation and cost cutting brought on by deregulation, computerized credit scoring has become an increasingly popular tool. It has allowed banks to increase profits through expanded lending while using fewer loan officers. Consumer installment credit nearly doubled to $628.9 billion at the end of the first quarter of 1988 from $344.8 billion in 1982.
Creative Business Decisions in Princeton, N.J., Credit Management Associates in Columbus, Ohio, and Management Decision Systems in Atlanta are among the firms that have capitalized on the growth by supplying credit scoring programs to major banks as well as finance companies and credit bureau reporting agencies such as Credit Bureau Inc. in Atlanta and TRW in Orange.
The granddaddy of the industry, San Mateo, Calif.-based Fair, Isaacs Co.--founded 30 years ago--had sales of $15 million last year from furnishing and designing credit scoring programs.
“Deregulation has definitely contributed to the growth of scoring,” said O. D. Nelson, senior vice president at Fair Isaacs. “It’s become a very active business in the last couple of years. Everybody from big credit unions, banks, captive auto finance companies and other lenders are now using” scoring.
Valley Bank & Trust Co. in Salt Lake City, for example, adopted a computerized credit scoring system after one of its loan officers granted a home improvement loan to a couple who fraudulently sought the money to fix up their rented apartment, according to Vaun Morrow, senior vice president. Morrow said there was a 25% reduction in loan losses in the first year after the system was adopted. In addition to income and indebtedness, Morrow said the bank’s scoring system looks at eight other factors, including the age of the applicant.
Yet even those who concede computers’ accuracy and speed say scoring can be skewed by the criteria chosen to evaluate applicants.
Multiple requests for credit by the same person over a short period of time, for example, are ordinarily considered negative because lots of inquiries usually indicate a bad credit risk flitting from lender to lender in a desperate search for credit.
But multiple inquiries can also be perfectly legitimate: An applicant may have had his credit rating checked while talking terms with several different car dealers before he makes a final purchase. Some systems don’t count multiple inquiries within, say, a week or two of a loan application. But not all credit scoring programs make such distinctions.
Likewise, length of residence is also an important predictor of credit worthiness.
John C. Douponce, a manager at Carlson Cos. in Los Angeles, found that out in August when he attempted to buy a new car from Universal City Nissan.
Although the 39-year-old Douponce says he has a clean credit record and has been steadily employed by the same firm, he was viewed as a poor credit risk by Universal because he has moved six times--from cities in North Carolina, Florida, Hawaii and California--in the last three years.
“He’s in the hotel business and he moves around a lot,” responded Chuck Castagna, vice president of Universal, which initially wanted to increase the interest rate on Douponce’s dealer-financed car purchase from 12% to a whopping 19%. “His credit record was fairly poor from what we saw. (But) he looked worse on paper than he was” in reality, Castagna said. Douponce got his car at the lower interest rate.
Under federal law, consumers are supposed to be told the reasons they were denied credit. But consumers groups say many credit seekers are usually given vague reasons for being turned down, such as “insufficient credit history” or “poor credit record.”
Ahearn of Citibank, says his institution uses several different scoring systems, including separate ones for gold MasterCards, regular MasterCards and personal loans.
He said Citibank doesn’t “reveal what criteria we use (to evaluate applicants) but when someone fails a scoring model we cite the reasons.” However, Yip complains that his rejection letter cited only his “field of study” as the basis for his rejection, even though Citibank claims that no student was rejected under the scoring model solely on the basis of their major.
TRW, the leader in the credit bureau reporting industry, this year introduced a new computerized scoring model that ranks an applicant’s potential for delinquency or bankruptcy on a scale of 0 to 1,300--the higher the score, the greater the chance of delinquency.
Robert T. Wheeler, TRW’s director of marketing business credit services, says the TRW model “results in costs savings because it saves time in the evaluation and processing of credit applications” and helps reduce bad debt loss.
However, some credit scoring systems have proved costly for their operators.
In 1980, Amoco Oil Co. paid a fine of $200,000 and signed a consent decree with the FTC agreeing not to use ZIP codes as a credit factor. The move came a year after Montgomery Ward paid a $175,000 fine in a similar case.
The FTC alleged that in awarding credit cards, the two companies used ZIP codes in a way that discriminated against black and Latino applicants because the firms gave a great number of points to areas that were predominantly white.
More recently, the respected Wharton Econometrics consulting firm quietly shelved a ZIP code-based credit information service this summer after it was attacked as discriminatory. The service broke down by ZIP code the size of consumers’ outstanding loans, the number of accounts and delinquency rates.
“There were a lot of high expectations for that (Infocredit program) but demand turned out to be very low, and getting the computer time from TRW was extremely costly,” said Mark Zandi, director of U.S. financial services for the firm.
Although Wharton had contended that lenders could legally use Infocredit in making loan decisions so long as ZIP code was not the only criterion relied on, Zandi acknowledged that controversy over ZIP code use cooled interest in the program, which at one time counted among its users more than two dozen savings and loans and banks.
Problems can also arise, Jerison said, in soliciting credit card and loan applications. While Jerison said it is legal to target specific groups, the elderly, say, or people in a certain ZIP code, it is not legal to judge any applications arising from such solicitations by an applicant’s age or place of residence. But critics say that solicitation and judgment amount to the same thing since undesirables are excluded from the pool through the application process.
Yet the experience of Wharton, Montgomery Ward and Amoco, has not deterred others from continuing to use ZIP codes in credit evaluation.
TRW uses ZIP code analysis in some of its credit reports, said Wheeler, because “location can be a factor in” credit worthiness. He said ZIP codes are used to evaluate a credit applicant’s likelihood for delinquency or bankruptcy in TRW’s new on-line computer credit report. For example, Wheeler said, “it’s better to be in the West ZIP code begining with 9 than in Texas.”
“To us, that smacks very much of redlining,” said McEldowney of Consumer Action. McEldowney is not alone in his concern.
“Using ZIP codes is something that has concerned the commission for some time,” said Jerison, the FTC lawyer. He said the commission “gets a steady stream of complaints” about alleged discriminatory credit practices. However, he declined to quantify the number of complaints, saying it may give a false impression of the scope of the problem, since the commission has no way of determining how many of the complaints are valid.
Proponents argue that even if scoring systems are biased, they are no more so than human loan officers.
“Almost all the things you can say about scoring pertains to the (human) judgment side as well,” said Nelson of Fair Isaacs. “You better believe the local loan officer knows what kind of neighborhoods he considers dubious. He won’t write it down in a memo. But you know that it’s always on his mind.”
“We can look at a lot more data than any individual and tend to be more accurate than a human being can be,” adds John Coffman, president of Mangement Decision Systems, the Atlanta credit scoring firm. “It’s never going to be the case where (lending decisions) are totally automated. There’s still going to be a gray area where you are going to want to personally review applications. But the positive thing is that there are more people getting credit today at a fairer price” because of credit scoring.
But the struggle goes on for Berkeley student Yip, who says he still has not received a personal apology or even an explanation from Citibank for being denied a MasterCard.
“They sent me a letter saying I have poor credit but my TRW credit report says my credit record is good,” said Yip. “I still don’t have a MasterCard. I guess they are telling me that rhetoric majors don’t pay their debts.”