In a story released last night, Carter Dougherty at Bloomberg reports that the CFPB has warned “at least four banks” that it may take enforcement action against them under ECOA, alleging that a “policy” of “allowing” dealers to negotiate contract APRs with retail buyers had a disparate impact in the pricing of retail installment sale contracts purchased by the banks.  If the story is accurate, this signals that the CFPB is planning multiple enforcement actions on this flawed legal theory. 

In a sense, this report is not a surprise, since the CFPB has been signaling for quite some time that it intends to apply the disparate impact theory of liability to auto finance, and dealer finance charge participation has been mentioned as a practice that the CFPB would target.  The Bureau seems to assume that there is no legitimate business justification for dealer rate participation.  But even beyond this, there are several things about potential enforcement actions in this area that make them profoundly unfair, and which should cause the CFPB to refrain from pursuing enforcement based on this flawed theory. 

Here are some:

(1)   Dealer Finance Charge Participation is a Ubiquitous Market Practice That Banks are Powerless to Eliminate.  Auto dealers sell cars and enter into retail installment sale contracts with consumers  They have the choice to sell those contracts to a variety of assignees – both banks and non-banks.  Even if a particular bank decided to cease “allowing” dealer finance charge participation, the only practical impact of that decision would be for the dealers to sell their contracts to another assignee.  The bank doing the “right thing” in the CFPB’s eyes would be rewarded with a dramatic reduction in its market share, and the underlying “practice” would continue unabated.  The flow of retail installment sale contracts merely would have been diverted to other assignees.  In a situation like this, rather than proceeding against a select few assignees, the CFPB should adopt a regulation that binds the entire industry in a fair manner, and gives industry participants an opportunity to avoid liability by complying with the rule.  Using enforcement in this situation “punishes” banks for a market dynamic that they cannot control.

(2)   The CFPB’s Theory Would Hold Assignees Accountable for Dealer Behavior.  If discrimination on a prohibited basis occurs with respect to dealer finance charge participation, it necessarily would have to arise from the conduct of the auto dealer, who actually negotiates the contract APR with the buyer and enters into the contract with the buyer.  Auto dealers are explicitly excluded from the CFPB’s jurisdiction under Dodd-Frank, and they are not “service providers” to an assignee – they are merely the sellers of retail installment sale contracts that the assignee purchases.  An enforcement action against an assignee for the dealers’ conduct would have to ignore both of these points.  (It is also likely to trigger a political reaction from the auto dealers based on the perception that the CFPB is circumventing the auto dealer exclusion in Dodd-Frank).

(3)   An Assignee-Level Analysis Can Show a False “Disparate Impact” Even if No Consumer Ever Experiences Discrimination.  The CFPB presumably is analyzing retail pricing for portfolios of retail installment sale contracts on an assignee level, rather than on a dealer-by-dealer basis.  Perhaps this approach is being taken because an individual dealer typically does not assign enough of its retail installment sale contracts to allow a meaningful statistical analysis of its pricing of contract APRs.  But whatever the reason for this approach, it is fundamentally flawed.  By aggregating all of the contracts in an assignee’s portfolio together regardless of which dealership assigned them, an assignee-level analysis can create the appearance of a disparate impact, even when no consumer has ever been discriminated against by any particular dealer. 

Take this hypothetical example: an assignee purchases retail installment sale contracts from two dealers: one of them always sets the contract APR 2% above the assignee’s buy rate, and the other always sets the contract APR at the wholesale buy rate without any rate spread increment.  If both dealers apply these retail pricing policies consistently, no consumer is ever treated differently by either dealer.  But if there is any difference in the demographic makeup of the dealers’ customers, a regression analysis performed at the assignee level will reflect a false “disparate impact,” purely as a result of the accident of the nature of each dealer’s customer mix. 

To suggest that ECOA makes such an accident a violation of the law stretches the language of the statute beyond any semblance of credibility.  It also underscores one of the many errors inherent in attempting to use a disparate impact theory of liability against assignees to regulate dealership conduct.   The CFPB could easily end up alleging a disparate impact claim against assignees without any evidence of underlying disparate treatment by the dealerships who supposedly have been “allowed” to engage in subjective decision making with respect to their contract APRs.

(4)   Reliance on Proxy Data Makes Proof of a Real “Disparate Impact” Highly Uncertain.  In the context of mortgage-related fair lending cases, there is a ready source of data to ascertain the race, gender or ethnicity of borrowers – they are asked to self-identify, and the lender is required to collect and report the data under HMDA.  No such data collection occurs in auto finance, and the assignee therefore has no information about the demographic characteristics of retail buyers. 

In order to bring a fair lending enforcement action against an assignee of retail installment sale contracts, the CFPB would have to use “proxies” to make an assumption about the buyer’s race, ethnicity or gender.  The problem is that these proxies may have significant error rates.  For example, the typical proxy used for Hispanic ethnicity is surname – if you have a Hispanic-sounding last name, you are counted as Hispanic for the purposes of the regression analysis.  But it is obvious that this proxy can be inaccurate – particularly for buyers whose last name changed because of marriage.  Yet in order to bring an enforcement action, the CFPB would have to ignore this inaccuracy and indulge the false assumption that the proxy analysis is correct and accurate.  Since it is known to have inaccuracies, and because the pricing “differences” typically relied upon in fair lending enforcement actions are quite small (in the ranges of 25-30 basis points, or ¼ of a percentage point), the combination of using inaccurate demographic data and the very small pricing “differences” raises the very real prospect that an assignee of retail contracts could be subject to an enforcement action when, in reality, no disparate impact exists in the first place.

Even if one accepts the validity of the disparate impact theory under ECOA (which I do not, since it is plainly inconsistent with the statutory text), the considerations discussed above should cause the CFPB to address auto retail installment contract pricing through rulemaking, and not enforcement.  If the Bureau follows through with the warnings reported in the Bloomberg article, it will have undertaken an action that is: (i) legally unsound and fundamentally unfair to the targets of its enforcement actions; and (ii) inconsistent with the auto dealer exclusion in Dodd-Frank.