A healthcare research algorithm has suddenly taken hold elsewhere: the Consumer Financial Protection Bureau.
Marc Elliott, a statistician at Rand Corp. in Santa Monica, developed the formula to estimate a person’s race depending on their address and last name. While many health insurers do not ask patients about their race, they may want to find out if black or latin patients are receiving the same treatment as white patients.
The Rand algorithm is a proxy method that estimates an unknown variable with the assistance of known variables. The company began developing the formula 16 years ago, and Elliott completed it.
Geocoding, or looking at a person’s place or residence, and surname analysis are both imperfect ways of determining race but have been used by companies in the past. While it’s likely that someone whose last name is Gonzalez is probably of Hispanic decent, last names like Smith or Jackson are less distinguishable.
The Rand formula, officially known as Bayesian Improved Surname Geocoding, or BISG, combines information about the ethnicity of neighborhoods and a compilation of last names, divided by how common they rank among six ethnic groups: Asian, Hispanic, black, white, multiracial or American Indian/Alaska Native.
Since it’s completion, it has fallen into the hands of financial regulators. The CFPB has used BISG to pinpoint racial discrimination among auto lenders.
Auto loans’ interest rates average around 4.16%, but car dealerships will often increase the rates set by the lender to pay itself back. This is known as markup.
However, the CFPB has found that dealerships and lenders often offer higher markup rates to minority buyers.
Auto lenders believe that this practice is unfair, considering that the algorithm is not foolproof — it’s only an estimate.
Scott Pearson, an attorney who works with the Los Angeles firm Ballard Spahr, believes that using BISG could be detrimental to auto lenders and produce unfavorable views of them.
Rep. Jeb Hensarling, a Republican from Texas, agrees.
“It is irresponsibly branding companies with the stigma of racial discrimination based on nothing more than junk science,” said Hensarling.
Regardless, Elliott’s status in the financial sector was completely unprecedented and unexpected. He was unaware of his formula’s contribution until he received an email from a friend that read: “Did you know you just cost Ally Financial $80 million?”
The CFPB used BISG to file against the GM lender in 2013. Ally Financial was forced to settle with an $80 million fine.