From HUD.

the charging party or plaintiff first bears the burden of proving its prima facie case that a practice results in, or would predictably result in, a discriminatory effect on the basis of a protected characteristic. If the charging party or plaintiff proves a prima facie case, the burden of proof shifts to the respondent or defendant to prove that the challenged practice is necessary to achieve one or more of its substantial, legitimate, nondiscriminatory interests. If the respondent or defendant satisfies this burden, then the charging party or plaintiff may still establish liability by proving that the substantial, legitimate, nondiscriminatory interest could be served by a practice that has a less discriminatory effect.

So, suppose that a lender uses a credit-scoring algorithm produces scores below the approval cutoff for blacks more often than whites (step one). Then, the lender shows that the credit scoring algorithm predicts default probabilities accurately for both blacks and whites. Does that satisfy step two? And then what sort of can of worms is opened by step 3? Suppose a community-action group claims that “If you pay us to set up a lending diversity program, we can bring you minority loans with acceptably low default rates,” does it have to prove its claim? If so, then this is actually harder on community-action groups than the current situation, in which all they have to do is threaten to sue and a bank will pay them protection money to make them go away.

I am only sort-of kidding. I would put the burden of proof to HUD to show that in recent years there has not been a lot more suffering caused by anti-discrimination regulations than by actual discrimination. And I am talking about suffering by people with “a protected characteristic.”