Usually, a fight with Congress is not a fair fight. That’s especially true for federal agencies. Their directors must be aware that disrespecting the legislature can have consequences come budget time. But a few agencies are not reliant on Congress for their annual budgets, and the Consumer Financial Protection Bureau (CFPB) is one of them.
This summer, the CFPB and the House of Representatives had a tussle over an issue very important to dealers and the indirect auto finance industry. Round 1 went to the CFPB.
The issue was the CFPB’s concern regarding how indirect finance sources compensate dealers for the contracts they buy. The agency believes that permitting dealers to exercise discretion in pricing, and paying the dealer based on the interest rate “markup,” often leads to higher financing costs for black and Hispanic car buyers.
In March, the bureau announced its intention to hold indirect auto creditors responsible under the Equal Credit Opportunity Act (ECOA) for pricing decisions made by the dealers from whom they buy contracts.
In separate letters, Democrats and Republicans on the House Committee on Financial Services asked CFPB Director Richard Cordray some specific questions that are very important to dealers and financing sources. Let’s see how the bureau responded.
1. Where Are You Getting Your Information?
How is an indirect auto creditor who never sees the consumer supposed to know the consumer’s race, ethnicity or gender? The CFPB said it uses “proxy” data to make these assignments. Lacking real data, the bureau substitutes other data, including geographic location and surnames.
Trying to infer a borrower’s race based on geographic location is a dicey task. The exact methodology used greatly affects the results. Do we call a person a minority if he or she lives in a neighborhood that is 70 percent African American? How are surnames used to tell a person’s race or whether they are Hispanic? Is a woman with a Hispanic surname Hispanic, or did she — or one of her female ancestors along the way — simply marry someone with a Hispanic last name?
In a recent webinar, the Federal Reserve Board was more forthcoming with its methodology, which is quite different from the CFPB’s methodology. But the CFPB, in the same webinar, remained mum.
2. How Do You Compare Price Differences?
It is important for the CFPB to ensure that any average pricing differences are based on illegal factors and not on legitimate differences between consumers or deal structures. The CFPB believes that some banks have engaged in pricing discrimination based on its analysis of markup rates. But the bureau has frustrated banks by refusing to use controls for many factors that accurately predict the dealer’s interest rate quote and markup.
The outcome of a pricing analysis often turns on the controls used to identify borrowers who are similarly situated. A failure to control for relevant factors can give a false picture, leading to a charge of illegal discrimination where none has occurred.
The issue of acceptable controls should be the subject of a roundtable discussion with the CFPB, which should be open to every interested party. Barring that, it would at least be nice to know what the CFPB thinks are acceptable controls. For instance, credit score, loan-to-value ratio, model year and payment-to-income ratio are all controls that often predict markup, but the CFPB has resisted the use of some of them.
So we were eager to know how the CFPB answered the congressional committee’s question. Here’s the opaque response: “Our analysis considers appropriate analytical controls in reviewing data.”
3. Where’s Your Threshold?
The third issue is related to the numerical threshold for making a “federal case” out of average differences in markup rates between groups. Using statistics to determine whether illegal discrimination has occurred has always been controversial and terribly hard. In the area of indirect auto finance, the undertaking is more difficult still.
The margin of error for assigning race and ethnicity by using proxies is often large. Is it reasonable to conclude that discrimination has occurred based on a difference of, say, 15 basis points, if the margin of error for the race assignments is 30 percent?
The CFPB tells creditors to monitor their data and find “solutions” for differences in pricing, including “prompt remuneration to affected consumers” when “unexplained” pricing disparities are identified. But the CFPB’s response to the House Democrats leaves us still in the dark about the important answers the lawmakers sought.
The CFPB makes regular references to its commitment to transparency. I know many people at the CFPB; some of them I have known for many years. They are smart, thoughtful and hardworking. I have no doubt their commitment to transparency is genuine. But actual transparency would be preferred.
I suspect it was the CFPB’s commitment to transparency that prompted its decision to participate in the FRB’s Indirect Auto webinar on Aug. 6. Yet these important questions remain unanswered. And if lawmakers felt stiff-armed by the bureau’s response to their letter, one can understand why.
As long as pricing includes discretionary markups, the bureau’s concern will continue and the industry must deal with the issue. It is unclear why the CFPB is so reluctant to be candid with Congress and the auto finance industry about its methods. One guess would be that the CFPB staff is not confident that it has all the right answers to these important methodology questions. If so, that uncertainty is probably well-founded. There is a solution, however.
The CFPB should host a series of day-long working group sessions to discuss these questions. Let the CFPB benefit from the careful work the industry has done on these issues for many years. Perhaps a consensus can be built around the following:
• The best way to use proxies for race, ethnicity and sex: Proxies are far from perfect, but we would all benefit from using the techniques that match self-identification as closely as possible.
• The control factors that are acceptable to ensure comparison of similarly situated borrowers: The CFPB staff needs to hear why certain factors are meaningful and legitimate controls.
• Determine how much of a difference in pricing warrants corrective action: This is, in part, a policy question, but the discussion is informed by technical issues.
The bureau and industry may not ultimately agree on each point, but I would bet we’d come closer, especially in a forum that promotes thoughtful exchanges — outside the context of a looming enforcement action.
Round 1 was a polite, surface-level scuffle with Congress. Round 2 should not be in the courts. Is a collaborative approach worth a try? I think it is.