Do Two Half-Victims Make a Whole Case? Banks pay for allegedly discriminating against fractions of humans. April 13, 2015 7:12 p.m. ET 150 COMMENTS Welcome to the new frontier of progressive law enforcement: extorting damage awards from businesses without naming anyone who’s been damaged. More than a year after persuading Ally Bank to pay $80 million to allegedly abused borrowers, the Consumer Financial Protection Bureau (CFPB) still hasn’t distributed a nickel to the alleged victims. Is it possible that victims aren’t getting paid because there are no victims? Recently we told you about the bizarre federal campaign against auto lenders in which bureaucrats guess the ethnicity of borrowers based on their last names and addresses. The feds then claim discrimination in interest rates if the people they assume are minorities on average pay more than similar borrowers that the feds assume are white. This is not a joke. By law, auto dealers who offer financing to car buyers are not allowed to record a borrower’s race. But bureau staff and their colleagues at the Obama Justice Department still want to sue the banks that provide these loans. So they assign probabilities for the race of the borrowers based on their names and where they live. It’s good enough for government work. But what if one day the government has to identify the victims, verify that they really are members of minority groups and confirm that they suffered discrimination? That task is proving to be difficult. Ally agreed to fork over the $80 million in damages (and $18 million in penalties) in December 2013. The bureau said that “discriminatory pricing differences resulted from Ally giving dealers the ability and incentive to mark up interest rates” and that this pricing and compensation structure “injured more than 235,000 minority borrowers.” But remember, 235,000 was a guess, and 16 months later the number of victims who have been identified and paid is—zero. The problem isn’t Ally, which says it deposited $80 million into an escrow account in January 2014. The CFPB won’t say how it arrived at the 235,000 figure, other than that it used a statistical method known as Bayesian Improved Surname Geocoding. It’s possible the feds counted only people that they think are highly likely to be minorities. But people familiar with their process say the bureau doesn’t like that approach. The victim totals are too low. Economist Marsha Courchane of Charles River Associates recently co-authored a critique of the bureau’s methods for spotting discrimination in auto lending. She tells us that if the bureau used its current methodology in the 2013 Ally case, fractions of borrowers would be added together to generate the total. In other words, if two borrowers each have a 50% chance of being black, they would count as one black borrower. In reality, both could be white, black, Asian, or members of any other racial category. But at the CFPB two fractions can add up to one victim. This is supposed to qualify as a great leap forward in the politics of racial grievance—not just protected classes of people but protected fractions of people, or perhaps aggrieved percentages of people. But how do you distribute money to fractions of victims? A bureau spokesman tells us that “payments cannot be sent until all affected consumers have been given a full opportunity to participate in the settlement, which requires extensive preparation and outreach.” We can only imagine. The bureau says it’s working closely with Justice, Ally and a settlement administrator paid by Ally “to effectively distribute the settlement fund. The consumer-facing materials, which are being prepared in multiple languages, will be provided by the administrator when finalized.” Since bureau staff already have the names and addresses, perhaps they could call the fractions and ask them what they look like. But besides the work this would require of the bureaucrats—not to mention the risk of offending potential victims who may not enjoy a government agency cold-calling to inquire about their race—this outreach could also threaten to undo the whole cockamamie political enterprise. If this search for evidence (conducted years after the case was settled) were to reveal enough mistaken guesses, it could blow up the statistical argument that borrowers paid more or less based on their race. Alternatively, the government could decide that it will only send checks to people who are perhaps 80% or 90% likely to be minorities. But that all but guarantees some checks will go to whites while some minority borrowers get bupkis. How to solve this bureaucratic conundrum? Here’s an idea: Don’t bring cases without any evidence that anyone has done anything wrong.