Eight dealer, financial services and pro-business organizations delivered another strong message to the Consumer Financial Protection Bureau to address the bias and error found in the method the CFPB uses to determine whether unintentional disparate impact exists in an indirect auto finance company’s portfolio.

In asking for thee specific modifications in a letter to CFPB director Richard Cordray, these organizations insisted the bureau has used — and continues to use — this methodology to support allegations of discrimination, despite its flaws.

Since 2013, letter authors explained CFPB has urged financial institutions to change how they compensate dealers for arranging financing, based on the bureau’s allegation that dealer reserve poses a risk of disparate impact. However, a November study by Charles River Associates (CRA) found that the CFPB’s analysis overstates the impact on minorities, and cast doubt on many of the bureau’s findings.

In light of these findings, the coalition called on the CFPB in a letter to revisit its enforcement approach.

“The associations request that the bureau conduct a thorough review of the CRA study, provide a public response to its findings and recommendations, and correct any bias in its testing methodology, before pursuing further dealer mark-up discrimination claims through supervisory or enforcement action.”

Signatories of the letter include the American Financial Services Association, American Bankers Association, Consumer Bankers Association, Financial Services Roundtable and U.S. Chamber of Commerce.

Offering their full support of what was shared in the letter were the National Automobile Dealers Association, the American International Automobile Dealers Association and the National Association of Minority Automobile Dealers.

In the letter to Cordray, officials recapped that CRA found the bureau’s application of the Bayesian Improved Surname Geocoding (BISG) proxy methodology creates significant measurement error, which results in overestimations of minorities in the population by as much as 41 percent.

In its own white paper on the method it uses to proxy for race — published prior to the CRA study — the CFPB acknowledged the overestimation (which it found to be 21 percent), but never indicated how, if at all, it has corrected for this discrepancy.

“Given such high error rates, proxy analysis has limited application to compliance monitoring and is unjustifiable for use when asserting legal violations or liabilities,” officials said in the letter to Cordray.

In light of the situation, AFSA, ABA, CBA, FSR, the U.S. Chamber as well as the dealer associations asked the CFPB to make a public response, addressing:

1. Portfolio level analysis of aggregated contracts sourced from dealers with different operating models, cost structures, pricing policies, locations and competitive landscapes. These factors, along with the seasonality of auto sales and financing, are major factors that must be corrected for in an analysis at the portfolio level.

2. Implementation of economic controls to adjust for legitimate business factors such as new, used, trade-in, options, insurance and warranties. Failure to consider business factors for observed disparities increases the potential of reaching erroneous conclusions.

3. The bureau must address and adjust for the bias within the BISG methodology and its overestimation of individuals within protected classes.

By law, auto finance companies are prohibited from inquiring into or considering a consumer’s race or ethnicity. In order to estimate the background of consumers for pricing comparison purposes, the CFPB uses a proxy that is based on a statistical analysis of the consumer’s last name and residence.

“The CFPB’s own study of its proxy methodology revealed that it is subject to significant error,” officials said in their letter.

The CRA study, based on 8.2 million vehicle contracts originated in 2012 and 2013, showed that the CFPB’s method overestimates minorities by as much as 41 percent – further calling into question the reliability of the CFPB’s testing method. The CFPB has not indicated that it has made any corrections for these error rates.

“We share the bureau’s commitment to combating illegal discriminatory treatment in the vehicle finance market,” the letter said. “This common goal is best achieved when fair lending standards are evidence-based, applied using analytically sound and transparent methods and predicated on accepted legal foundations.”

The dealer associations pointed out what they called the “flaws” in the CFPB’s methodology have not stopped dealers from developing procedures to address the CFPB’s concerns about the risk of discrimination in auto financing.

Last year, NADA, AIADA, and NAMAD jointly released the NADA Fair Credit Compliance Policy & Program, which provides a dealer with an optional mechanism for ensuring that its credit pricing is established in a consistent manner and is based exclusively on legitimate business considerations.

NADA president Peter Welch noted the program is modeled on — and fully adopts — a Department of Justice fair credit compliance program that the agency required dealers to implement in prior enforcement actions.  

With that plan available, Welch spoke for the dealer associations in backing up the latest message the financial services organizations delivered. 

“Discrimination in the market simply cannot be tolerated,” Welch said.  “However, in light of the rigorous peer-review that has cast significant doubt on the CFPB’s findings, the bureau should change course — or at least hit the pause button — and address these new concerns.  We applaud the courage of these organizations for speaking up.”