When the Consumer Financial Protection Bureau and the Department of Justice issued their joint consent order against Ally Financial in late December 2013, it set off a firestorm of anxiety in the indirect auto lending world.
Is this the end of dealer markup? What role do dealers play? What are the expectations of CFPB and other regulators regarding what indirect auto lenders must do to ensure compliance with fair-lending laws?
How about this one: Is this the only such enforcement action in fair-lending issues in indirect auto credit? Or are more on the way?
The answer here, at least, is straightforward.
More are coming—and soon— although when and against whom remain unknown. But it is evident a clear trend has developed toward intense regulatory scrutiny of this business model.
Recapping Ally case
For those unfamiliar with specifics, on Dec. 20, 2013, CFPB and the Department of Justice issued a joint consent order against Ally Financial, alleging “that more than 235,000 minority borrowers paid higher interest rates for their auto loans between April 2011 and December 2013 because of Ally’s discriminatory pricing system.”
The practice at issue is “dealer markup.” That is defined as lenders allowing dealers, who arrange vehicle financing for consumers at dealerships, the discretion to mark up a borrower’s interest rate above the lender’s established risk-based buy rate. In the bureau’s words from their press release announcing the order: “Markups can generate compensation for dealers while giving them the discretion to charge consumers different rates regardless of consumer creditworthiness.”
This carries fair-lending risk. While it is the dealer who sets the ultimate interest rate the consumer will pay, it is the lender’s responsibility to ensure that there is no disparate impact that negatively affects members of protected classes.
While there are many unanswered questions—I’ll present some of them at the end of this article—there are some clear points of guidance for indirect auto lenders.
Most importantly, lenders must take affirmative and proactive steps to monitor and control fair-lending risk at an enterprise level; the basis of the enforcement order was that Ally “failed to implement an effective compliance program to monitor its loan portfolio for discrimination.”
Building an indirect compliance program
So what might such a program look like? The following are critical elements that any indirect auto lender should consider:
1. Develop and execute an enterprise-wide indirect auto fair-lending compliance program.
Just about every lender out there has some sort of fair-lending compliance program. However, specific aspects of that program should be devoted to the particularities of indirect auto lending. Appropriate structural and governance processes include:
• Appropriate personnel charged with overseeing the program. This may be the lender’s fair-lending officer or someone else.
• Applicable committees and/or working groups within both the compliance and business line functions.
• Appropriate reporting and communication strategies directed to selected management and business line personnel.
An important statement in the consent order was that the lender “did not monitor whether discrimination on a prohibited basis occurred through the charging of markups across its portfolio of retail installment contracts.”
2. Develop and implement a dealer compensation policy.
A dealer compensation policy should establish and limit the maximum allowable spread between the lender’s buy rate and the retail installment contract rate (the dealer markup). Limits must exist at both the individual dealer and lender portfolio levels, and exceptions—if permitted and appropriate—should be addressed. The policy should be subject to normal policy review and approval processes.
What limits are appropriate? Many lenders have dealer markup maximums of 200 to 250 basis points, and there is no suggestion from the agencies that this level (or any other) is inappropriate. Rather, the issue in the consent order was differences in spreads between prohibited basis and control (non-minority) groups, all other factors being equal.
Related procedures and processes should be implemented to support the policy. The goal of both the policy and procedures must be to assure compliance with all applicable laws and regulations, especially Regulation B (the Equal Credit Opportunity Act). They should also address any differences in specific dealers’ operations, as well as handling markups versus fixed pricing structures, if any are utilized by the lender.
3. Perform periodic statistical analyses using proxy data.
Fair-lending statistical analyses have traditionally been performed only in the mortgage environment. The reason: Only these loans have associated applicant and borrower demographic information (called “government monitoring information,” or GMI) with which to work. For products (such as auto loans) where race, ethnicity, and gender aren’t available, how can reliable analyses be performed?
The answer is by utilizing “proxy” information. A proxy is a substitute; in this case it means making assumptions about a consumer’s demographic characteristics based on what the lender already knows, principally the borrower’s name (first and last) and address.
First names can be used as predictors of gender, and surnames for race and ethnicity as certain surnames are more common within particular demographic groups than in others. Geographic location can also be used as an indicator of race and ethnicity. Pinpointing the census tract where a consumer lives, then identifying the minority percentage of the population within that tract provides a proxy for the consumer’s race and ethnicity. For instance, if a consumer lives in an area identified by the Census Bureau as 80% or more minorities, chances are high that particular consumer is a minority.
But how predictive is this? For small portfolios or data sets, it’s hard to say. But for larger sample sizes, the law of large numbers increases the predictability (relative to a smaller sample) of the proxy data.
Of course this isn’t a perfect methodology. And some have a difficult time justifying in their minds the fact that enforcement may be predicated on assumptions.
But this is clearly the way the agencies are going, as stated in the consent order:
“The CFPB and the DOJ employed a proxy methodology that combines geography-based and name-based probabilities, based on public data published by the United States Census Bureau.” [Emphasis added.]
This was also stated in a letter from Richard Cordray in mid-2013 in response to questions from Congress on the bureau’s methodology, as well as in a CFPB blog posting.
What sources of demographic information can or should be used to assign proxies?
• Race and ethnicity: CFPB’s blog post opined that “responsible lenders” could use publicly available data to proxy surnames, and a link to the Census Bureau’s last name database was provided. Geographic information, including minority percentages within census tracts, has been available and used for years by banks in HMDA and similar fair-lending analyses.
CFPB’s letter to Congress stated “when we utilize proxy data, we use both surnames and geographic location.” This information is then utilized, according to the enforcement order, “to form a joint probability using the Bayesian Improved Surname Geocoding (BISG) method.” The details of this method are beyond the space allowed here, but it involves combining probabilities to formulate a more statistically reliable proxy.
• Gender: While the enforcement order does not allege any gender-based differences, the bureau’s related blog post discusses whether “a responsible lender [would want] to make sure that their female customers are not paying more for a loan than similarly situated men.”
It goes on to state that gender could be proxied by relying on the Social Security Administration’s first name database, and again a link was provided. Unfortunately this is not a user-friendly database and it is not presented in a form that would allow a user to easily determine whether a particular name is more likely to be male or female.
Once a set of loans is assigned proxy information, standard fair-lending analyses (disparity ratios, regression, matched pairs, and so forth) can be performed. This should be done periodically, although how often will depend on a lender’s portfolio size, previously-identified issues, and so on.
Analyses should be conducted at both individual dealer and lender portfolio levels to look for statistically significant disparities in pricing (dealer reserve) on prohibited bases. Specifically, differences in interest rates “based on race [or other prohibited basis] and not based on creditworthiness or other objective criteria related to borrower,” to quote the consent order, should be identified.
4. Implement appropriate corrective action.
What levels of differences are problematic? In the order, basis point differences between protected class and nonminority groups were in the twenties, but that should not be used as a guide to suggest a minimum difference that might be an issue. Differences in the teens, or even less, could be meaningful if the results are statistically significant.
In any event, appropriate corrective action should be directed to dealers identified in the periodic analysis of pricing data for disparities on a prohibited basis. Corrective action could be progressive, from initial warnings, whether verbal or written, to restriction or elimination of a dealer’s ability to exercise discretion in setting a consumer's contract rate (flat-rate pricing), to even eventual termination of the dealer-lender relationship.
What about remuneration to affected consumers?
That’s a particularly sensitive topic, and one best left to discussions between a lender and its legal counsel. While some might argue that proactive remuneration is a curative measure, others have doubts regarding whether it makes sense to cut checks to consumers the lender is assuming (rather than knowing) are members of a protected class.
Plus, if refunds are sent, does that constitute some sort of admission of guilt on the part of the lender?
5. Implement dealer fair-lending awareness efforts.
The objective of this step is to make dealers acutely aware of the lender’s fair- lending responsibilities and the dealer’s role in compliance and fairness. Information should be provided to all dealers on a regular basis (no less frequently than annually), including:
• An explanation of fair-lending laws and regulations, stating the lender’s expectations with respect to compliance.
• The dealer's obligation to price retail installment contracts in a non-discriminatory manner, especially when discretion is permitted.
• An explanation of the corrective action steps that may be taken by the lender, up to and including termination, if issues are discovered and not properly addressed.
You don’t have to change your pricing practices
Nowhere in the joint consent order, nor in any other rule, regulation, bulletin, or otherwise for that matter, was there any mandate or requirement that the practice of dealer markup be abandoned.
The lender was directed to develop a more robust fair-lending program to identify and remediate the risks inherent in indirect auto lending, but it was not a call for flat-fee pricing.
Clearly, however, should any lender choose to continue with dealer markups, much more attention must be paid to which consumer is charged what contract rate, by what dealer(s), and what this shows at both the dealer and lender portfolio level.
The argument that the discriminatory practices were perpetuated by dealers rather than lenders is a nonstarter with the agencies. The consent order makes it clear the lender is ultimately responsible, as it states:
“[The lender] regularly participates in the decision to extend credit by taking responsibility for underwriting, setting the terms of credit by establishing the risk-based buy rate, and communicating those terms to automobile dealers. [The lender] influences the credit decision by indicating to automobile dealers whether or not [the lender] will purchase retail installment contracts on the terms specified by [the lender].”
Note that even though this discussion deals with indirect auto lending, the proxy data principles extend to any non-mortgage loan product where GMI is not collected.
CFPB’s letter to Congress makes the point that “we have found frequent instances where lenders had robust fair-lending compliance programs for mortgage lending, but weak or non-existent fair-lending compliance programs for other types of consumer lending.”
Monitoring and corrective action processes should be implemented to control and mitigate fair lending risk for any product type.
Questions to keep in mind going forward
As I stated earlier, this is an evolving area, the matter of indirect auto financing. As this goes forward, lenders should remember these questions:
1. Is it the CFPB's opinion that dealer markup is an unacceptable pricing methodology? (In other words, will future cases change the answer to this question.)
2. What level of dealer markup presents an unacceptable level of fair-lending risk? Is there a minimum number of basis points that is too high?
3. Are there acceptable pricing mechanisms (aside from flat-fee pricing) that would minimize fair-lending risk more than dealer markup?
4. Will the CFPB share with the industry its precise methodology of assigning proxy data to consumers?
5. What is a simpler way to proxy gender information?
6. At what point would it be advisable to terminate a dealer relationship?
7. When (if ever) should a lender consider proactive remuneration to affected consumers?