The statistical method the Consumer Financial Protection Bureau uses to identify which consumers belong to minority groups correctly identified African-American borrowers only 24 percent of the time.
That was one of the findings of a nearly yearlong study commissioned by the American Financial Services Association, a lender group with offices in Washington. AFSA has been at odds with the CFPB over the bureau's methodology.
"From start to finish, this thing is flawed," AFSA Executive Vice President Bill Himpler told Automotive News.
The CFPB's statistical analysis is so inaccurate that the CFPB's conclusions about alleged discrimination in auto loans are also invalid, the lender group said.
Chris Stinebert, AFSA CEO, said the research was an independent study by financial consulting firm Charles River Associates. He said the study looked at around 8.2 million auto loans for both new and used vehicles. The loans were originated in 2012 and 2013, he said.
"The conclusion reached by the study is the CFPB proxy methodology to determine alleged discrimination is significantly biased," Stinebert told Automotive News. Besides a low success rate in identifying minorities, the CFPB methodology also produced a lot of "false positives," exaggerating the number of minorities, the study said.
Auto lenders are legally prohibited from collecting race or ethnic data from loan applicants. In contrast, mortgage applicants self-report race and ethnicity.
To analyze auto-loan contracts for potential discrimination, statisticians use so-called proxies -- such as names, addresses and census data -- to assign a probability as to whether a given consumer with a particular name at a particular address belongs to a legally protected class of borrowers, such as minorities or women.
To check accuracy, statisticians can compare those results with mortgage data from the same individuals, where race and ethnicity are reported.
In an industry "guidance" released in March 2013, the CFPB had concluded, based on its analysis of auto loan data, that lenders permit discrimination when they allow dealerships to set their own compensation for acting as a middleman to set up auto loans.
That compensation comes in the form of a small amount of interest, known as dealer reserve, that lenders allow dealerships to tack onto the car-loan interest rate charged by the lender.
The CFPB wants lenders to switch to flat fees or some other way to compensate dealers, where dealers would have no discretion in setting the consumer's final rate.
The bureau said it plans to "carefully" review "the AFSA-commissioned" report. "The CFPB is always interested in relevant data regarding important issues like discrimination in auto lending," the bureau said.
The National Automobile Dealers Association strongly endorsed the latest study results.
"This study shows that CFPB's attempt to upend the auto lending process is insufficiently informed and the victim of flawed assumptions and inadequate peer review," Peter Welch, NADA president, said in a statement. "Allegations of potential discrimination are explosive -- and certainly should not be made without a reliable foundation in data."