Did federal regulators use reliable data analysis to conclude that dealers may be discriminating in their lending practices?
A new study commissioned by the American Financial Services Association, a lender group, challenges the statistical method used by the Consumer Financial Protection Bureau. The process correctly identified African-American borrowers only 24 percent of the time and had a high rate of "false positives" that overstated the number of minorities, AFSA said.
That makes the CFPB's conclusions about alleged discrimination in auto loans invalid, the lender group said.
In a 2013 industry "guidance," the CFPB concluded that lenders permit discrimination when they let dealerships set their own compensation for acting as middlemen on auto loans. Instead of allowing dealers to tack on points to car-loan interest rates charged by the lender, the CFPB wants lenders to switch to flat fees or other nondiscretionary compensation.
The National Automobile Dealers Association disputes the CFPB conclusions and has asked to see the data and methodology the CFPB used. The CFPB has been less than fully forthcoming in its replies.
Solid data are slippery. Auto lenders cannot legally collect race or ethnic data from loan applicants. To review auto lending, the CFPB used indirect analysis, calculating the probability a person is a minority or female based on name, address and census information. That analysis was the basis for the CFPB's charging discrimination in auto lending.
To test that, the consultants that AFSA hired did likewise, then cross-checked the results against mortgage data for the same individuals, who had self-reported race and ethnicity.
The CFPB said it would "carefully" review the AFSA study.
Ensuring equal access to credit for all Americans is too important to get trapped in a frenzy of finger-pointing. Lenders, dealers and regulators must share data and their concerns.
The CFPB must vigorously enforce anti-discrimination lending laws, but only with reliable and thoroughly vetted data.