|
Do minorities and women get a fair shake from credit scoring? That's hard to say, because bankers have resisted collecting credit data on
small-business loans. The question arises, though, because consumer credit-scoring systems have come under attack for hidden, unintentional biases
built into their formulas--and small-business credit scores put heavy weight on a business owner's personal credit history.
Bankers say scoring prevents bias because it minimizes the human touch. In the past, if a loan officer didn't like that you were a man and wore
an earring, it could affect your ability to get a loan, says Jim Carswell, vice-president for credit scoring at Canadian Imperial Bank of Commerce
in Toronto. Now, he says, "I can guarantee you that guys with earrings is not a factor in the model."
Skeptics say the system isn't as bloodlessly objective as it seems. Consumer credit scoring is based on practices that are common in the white
community such as home-equity loans, they say, which are more likely to be recorded in databases and easily accessed. Little or no weight is given
to deals common in minority neighborhoods, such as loans from community groups, sub-prime lenders, and local finance companies. These lenders often
don't report to credit bureaus, so verification would require extra work and expense. Worse, some predatory lenders withhold good payment records
to prevent customers from refinancing at lower rates, says Matthew Lee, a prominent banking activist at Inner City Press in New York. The result:
no credit history for such entrepreneurs, and thus no loan from a bank. That's if they can find one. Lee says the dearth of branches in minority
neighborhoods makes it hard for entrepreneurs to build relationships with bankers and create a conventional credit history.
No less than Eugene Ludwig, the U.S. Comptroller of the Currency, warned in 1997 that the system might be flawed. He pointed to misuse of
"overrides"--single pieces of negative information that can cancel out a passing score. The agency's examiners found signs of bias built into
scoring systems and data that had a disproportionate impact on minorities.
Women face similar problems, says Kathryn S. Keeley, an executive at Count-Me-In.org in Washington, which helps women entrepreneurs get capital.
The nonprofit group is developing its own "women-friendly" scoring model with support from gilt-edged backers such as American Express Co., which
also harbors doubts about conventional models.
Keeley, whose background in women's finance dates from 1983, notes that if she applied for business credit, her consumer-credit record would
seem too thin because "I am married, and all of our joint accounts are on my husband's file. My score is lower right off the bat." Women are more
prone to gaps in employment--also a red flag--because they often take time off for child-rearing. Then there are innocuous-sounding questions such
as: How long have you owned your business? Most women start out less formally than men before incorporating, she says, so they're more likely to
answer "less than two years"--another "ding" in credit scoring.
Fritz Elmendorf, a spokesman for the Consumer Bankers Assn., says his group supports the use of credit scoring, but notes that its annual survey
found that 69% of members have added programs to help women get loans, and almost half have inaugurated programs for minorities or inner cities.
Still, he acknowledges that the secrecy surrounding the scoring formulas breeds trouble. "People don't understand it," Elmendorf says. "It's a
black box, and that makes them suspicious."
How does Fair, Isaac & Co., the biggest seller of credit-scoring models, respond? "Independent research has been done to make sure that the Fair
Isaac credit-bureau risk score is fair to minorities and people with little credit history," the company told frontier. "Scoring has proven
to be an accurate and consistent measure of repayment for all people who have some credit history." Sounds unequivocal, except for one thing: Fair
Isaac is also helping Keeley build her women-friendly system.
By
Rick Green
|
