Credit scores not biased, but data types could be broadened

WASHINGTON -- In a report to Congress that is certain to generate controversy, the Federal Reserve Board says that credit scores vary "substantially" among racial and ethnic groups, but have made credit more available for major consumer purchases such as buying a home.

The Fed's study, encompassing credit bureau records and demographic data from a national statistical sample of 301,536 individuals, was mandated by Congress in 2003. Credit scores are now heavily used not only in home mortgage underwriting and pricing, but in credit cards, auto loans, employment and rental application screening, and the insurance industry.

Critics have questioned the accuracy and fairness of credit-score models, charging that in some cases they are inherently biased against minority groups such as African-Americans and Hispanics.

After a massive research effort over several years that focused on three different credit-scoring models -- including one created by Federal Reserve staff economists -- the agency concluded that:

•Credit-score statistical factors and models are not biased against any particular demographic group, but are highly predictive of future payment performance. Lower scores correlate strongly with future delinquencies; higher scores are associated with good payment performance.

•African-Americans and Hispanics, on average, "have lower credit scores than non-Hispanic whites and Asians."

•Younger individuals of all demographic groups tend to have lower credit scores on average than older individuals, in part because credit-scoring models focus on past payment histories and length of credit accounts. Younger consumers generally have fewer accounts and shorter payment histories.

•The payment performances of some demographic groups are somewhat worse -- or better -- than their numerical scores might suggest. For example, according to the Fed, "blacks, single individuals, individuals residing in lower-income or predominantly minority census tracts show consistently higher incidences of bad performance than would be predicted" by their credit scores. On the other hand, "Asians, married individuals, foreign-born (particularly, recent immigrants), and those residing in higher-income census tracts consistently perform better than predicted" by their credit scores.

•Recent immigrants' scores might be improved by expanding the range of credit-like accounts reported to the national credit bureaus to include rent payments and other recurring accounts. Once in the bureaus' files, these accounts could then be used to help compute immigrant scores, enhancing their ability to predict future payment behavior.

•The types of credit used by various demographic groups "do not appear to be the source of differences in (payment) performance once (the) credit score is taken into account." That finding runs counter to criticism by some consumer advocates that mortgage lenders and brokers tend to steer African-American and Hispanic borrowers into higher-risk, higher-cost loans -- subprime adjustable-rate home loans with hefty payment jumps, for example -- that increase the likelihood of default and foreclosure.

Researchers confirmed that African-Americans and Hispanic borrowers make heavier use of finance-company installment loans, and that interest rates paid by African-Americans appear to be higher in general than those paid by whites.

Though most fair lending groups have not yet reviewed the Fed's study fully, at least one consumer advocate lambasted the report. Calvin Bradford, formerly of the University of Minnesota's Hubert Humphrey Institute of Public Affairs and now a fair housing consultant, challenged the study's statistical sampling methodology, selection of data, and failure to probe more deeply into issues of marketing of credit products to minorities. Bradford said the Fed used "an old database" dating to 2003 provided by credit bureau TransUnion that underrepresented minorities such as African-Americans and Hispanic homeowners with mortgages, while overrepresenting whites, based on their actual proportions in the general population.

"I don't think anybody in the (statistical) modeling business would accept such small samples," said Bradford.

Bradford also rejected any conclusion that types of credit products marketed to minorities have no impact on payment performance. "If blacks are paying more for access to the mortgage market than they really need to," he said, "then that is going to increase their household debt burdens" and contribute to higher delinquency rates.

Paul Hancock, former director of the Justice Department's fair lending unit, now in private practice with the law firm of K&L Gates, called the study's results "startling."

"It is very troubling for me as someone who has been an advocate for fair lending for many years" to find such credit-score divergences among racial and ethnic groups. "It shows some very serious social inequities attributable to our history" -- the lingering impacts of segregation, unequal educational opportunities and severe economic differences.

Mortgage lenders are often accused of charging minorities more, said Hancock, but the Fed study suggests that credit scores accurately gauge future payment performances, "so the burden here can't fall entirely on the shoulders of lenders."

© 2007, The Washington Post