I want to thank Jeff Sovern over at the Public Citizen Consumer Law & Policy Blog for having an interesting back-and-forth with me over the last week about the application of the disparate impact theory of liability to dealer finance charge participation in the auto sales finance context.  We’ve been debating whether the CFPB’s announced intention to bring enforcement actions against auto sales finance companies will cause consumer prices for auto financing to increase.  In his latest post on the issue, Jeff makes the point that, even if consumer costs increase, consumers should be willing to bear that expense “as the price to pay for blocking odious discrimination.” 

The question that raises in my mind is, when can we fairly call something “discrimination”?  In my mind, the word “discrimination” has always meant intentionally treating someone differently on a prohibited basis.  When we take Jeff’s question and evaluate it in the context of that meaning of the word “discrimination,” it’s easy to agree that we should be willing to pay more for something if we can eliminate that kind of behavior.  Intentional discrimination is wrong, and it is unquestionably proscribed by ECOA.  I don’t think anyone would disagree with Jeff’s argument if we all shared that meaning of the word “discrimination.”

But using the word “discrimination” to describe a disparate impact analysis in the auto finance context doesn’t have quite the same moral force behind it.  To start off with, disparate impact is, by definition, not intentional discrimination – it is a result that occurs by happenstance from the application of a neutral, non-discriminatory practice.  Moreover, as applied to dealer finance charge participation in the auto sales finance context, there are several critical reasons why inferring even a disparate impact from the available data is highly problematic.  I detailed these reasons in a recent post on this blog, but to summarize a few of them: (1) the use of proxies to “assume” protected class status involves significant error rates; (2) a portfolio-level analysis can create the false impression of a disparate impact, even when no discrimination actually has occurred at the dealership level; and (3) analyzing retail installment contract APRs without taking into account the purchase price of the car and any ancillary products can also create a false appearance of disparate impact where none exists. 

Here’s an example that illustrates my point.  Let’s say that my wife and I (who have similar credit scores) both go to the same dealership and buy the same car with an MSRP of $50,000.  My wife (who is Hispanic, from Colombia) has gone on the internet and researched the wholesale price of this particular car, and negotiates with the dealer to pay a cash sale price of $46,000 for the car, but she pays less attention to her contract APR, which lands at 3.49%.  Under a 60-month retail installment sale contract with zero down payment, her monthly payment will be $836.61, for a total of payments of $50,196.60. 

Now let’s say I go into the same dealership and buy the same car.  I haven’t researched the wholesale price of the car on the internet, and I end up paying a cash sale price equal to the MSRP of $50,000.  But before walking into the dealership, I called my credit union and found out that they would finance my purchase at a 2% APR.  So, when it comes time to determine my contract APR, I tell the dealer that it must come in lower than the rate offered by my credit union, or I’ll finance the purchase through my credit union.  The dealer does better, and gives me a 1.49% APR, which is the “buy rate” for one of the banks or sales finance companies to which it assigns its retail installment contracts.  Under this scenario, my monthly payment is $865.28, for a total of payments of $51,916.80. 

Who got the better overall deal here?  Obviously, my wife did.  If we both make our payments on time, she will have paid $1,720.20 less for her car than I did.  Should we conclude from these facts that she was subjected to “discrimination,” either because she is a woman or because she is Hispanic?  I think it would be difficult to reach that conclusion when she ended up paying less overall than I did. 

But, if we run our two transactions through a regression analysis, looking at the contract APR and the buy rate in isolation, the conclusion would be that she suffered clear “discrimination,” because her APR was 2% higher than mine, even though we have the same credit qualifications.  Under such an analysis, we would conclude that she has been “damaged” in the amount of $2,433, which is the difference in finance charge between a 1.49% APR and the 3.49% she actually paid.  Thus, if the CFPB brought an enforcement action against the bank or sales finance company that was unlucky enough to buy both of our contracts, that company would end up “compensating” her over $2,400 for the “harm” she suffered because of “discrimination.” 

When we understand disparate impact in this way, and especially with the potential inaccuracies that attend statistical analysis of automobile retail installment sale contracts, it’s hard to justify the application of the theory at all, much less to endorse the notion that consumers should be forced to pay higher cash prices to fix this kind of  asserted “discrimination.”  The case for this result becomes even weaker when we realize that the retail credit pricing “disparities” we are talking about here are historically in ranges as low as 10 to 30 basis points in previous government fair lending enforcement actions.  At this minimal level of “disparity,” the dollar difference on an automobile retail installment sale contract is tiny – in my wife’s hypothetical example above, a 20 basis point difference in APR on a $46,000 car over 5 years ends up being $246.50. 

So, although Jeff’s Consumer Law & Policy post suggests that consumers should be willing to pay higher cash prices to eradicate “discrimination,” the question really is whether consumers should have to do so in order to eliminate very small pricing differences that may or may not indicate any real difference in economic outcomes because of the inaccuracy of the use of proxies and the omission from the equation of the cash sale price of the vehicle and any ancillary products.  I seriously doubt that consumers would agree that this is a cause noble enough to cause them to dig further into their pocketbooks when they buy a car.  And this discussion illustrates the fact that applying the disparate impact theory of liability to dealer finance charge participation in the auto sales finance context involves using the term “discrimination” to refer to something that most people would never equate with that term.  In my view, Congress didn’t legislate that result.  The text of ECOA prohibits only discrimination on a prohibited basis and, thus, proscribes only what we all agree is wrong and odious: intentional discrimination.