by Jeff Sovern
Chris Willis of Ballard Spahr's CFPB Monitor recently blogged about whether disparate impact can properly be termed discrimination and therefore prohibited by ECOA. For those who don't know, courts generally recognize three ways to prove discrimination in violation of ECOA: direct evidence, disparate treatment (except in the Seventh Circuit) and disparate impact, or as it is sometimes called, disparate effect. Direct evidence would exist, for example, if a lender said we don't lend to old people. For obvious reasons, that doesn't happen that often. Disparate treatment requires that the lender have discriminated intentionally. That turns out to be very difficult to prove, so difficult in fact that it almost doesn't matter that the Seventh Circuit doesn't use the disparate treatment test. That leaves disparate impact. Disparate impact requires the plaintiff to show that using a particular criterion has a disproportionate effect on a particular group. That can happen even if the lender did not intend to discriminate against anyone. If the plaintiff carries that burden, the lender has the burden of showing that the criterion has a manifest relationship to creditworthiness.
To take a somewhat fanciful example, but one that I think makes the rule fairly clear, suppose the cattle organization that sued Oprah Winfrey some years back for disparaging beef decides to open a bank and that it will lend only to people who eat meat. Some religions require their adherents to be vegetarians. Accordingly, the lender's policy would discriminate against members of those religions, which I believe includes Budhism, at least in some forms. The lender would not have intended to discriminate against Budhists, but its policy would have a disparate impact upon them. Consequently, the lender would have to show that its criterion enabled it to avoid lending to non-creditworthy borrowers.
So the disparate impact test prevents unintentional discrimination against members of protected groups. Why should we use such a test? I confess I haven't studied the literature on this as much as I would have liked, but I can think of several reasons (I think Chris knows more about this than I do, but I wanted to respond anyway). First, the legislative history of ECOA supports the use of the disparate impact test. Some lawyers and judges oppose the use of legislative history, but I don't intend to repeat that debate here; it happens that I believe in using legislative history, at least under some circumstances. Second, the administrative agencies that have interpreted ECOA, such as the Federal Reserve and HUD, have also endorsed using the disparate impact test. Courts usually defer to administrative agencies in interpreting statutes.
There also are policy justifications for using the disparate impact test. if Budhists can't borrow from a particular lender because of their religion, does it really matter that the lender didn't intend to discriminate against Budhists but just insists on lending to meat-eaters? Whether the lender intended to discriminate or not, the Budhists who refuse to eat meat can't borrow from that lender because of their religion. They are being treated as second-class citizens by that lender. And, again, if the lender can demonstrate that meat eaters are less likely to default, then the lender could continue insisting that its borrowers eat meat.
Another reason to recognize the disparate impact test is because the other tests have not been very effective. If courts don't block disparate impact, ECOA won't have very much impact at all. In other words, lenders could continue to treat people differently based on criteria that don't relate to creditworthiness, but that produce different effects on protected and non-protected groups. And just suppose the lender really did want to discriminate against Budhists, and is willing to accept as the price of doing so that it also discriminate against non-Budhist vegetarians, to avoid creating evidence of disparate treatment. If the lender managed to avoid providing direct evidence of animus towards Budhists, disparate impact would be the only avenue for stopping the lender. Take that away, and the lender could discriminate with impunity.
Chris provides an engaging example about two car loans he and his wife took out. I think his example is beside the point because, unless I am misinformed, disparate impact cases are not based on one or two examples, but on enough examples to create statistical evidence that different groups are being treated differently. If a car dealer systematically charges one group higher rates than another, it becomes much more difficult to say that the explanation is innocent (but maybe I'm not understanding Chris's example properly, because I don't see why the car price, as opposed to the loan terms, are relevant). Similarly, Chris's point that the different treatment is not that different also misses the mark, in my view. If the differences are not that great, then maybe there hasn't been a disparate impact. But that doesn't mean that in other cases, in which the differences between the treatment of two groups are statistically significant, we should ignore those differences.
I agree with Chris that unintentional discrimination doesn't carry the same moral objection as intentional discrimination. But treating people differently because of race or religion, etc. is still deeply troubling. In any event, I agree with the courts and administrative agencies that have concluded the disparate impact test is an appropriate test of whether ECOA has been violated.