A Secret Bias Hidden in Mortgage-Approval Algorithms
An investigation found lenders still strongly favor white borrowers, but it raised a new question: What if a lender isn’t biased but its data, notably credit scores, is?
NEW YORK – An investigation by The Markup determined that lenders in 2019 were more likely to refuse home loans to people of color than to white people with similar financial characteristics, even when adjusted for newly available financial factors that the mortgage industry previously said would explain racial disparities in lending.
In Markup’s study, lenders were 80% more likely to reject Black applicants and 70% more likely to reject Native American applicants, while Asian/Pacific Islander applicants were 50% more likely to be denied loans and Latino applicants were 40% more likely.
The bias varied by metro area. Finer analysis found that lenders were 150% more likely to reject Black applicants in Chicago than similar white applicants, over 200% more likely to reject Latino applicants in Waco, Texas, and more likely to deny Asian and Pacific Islander applicants than whites in Port St. Lucie, Florida.
Underpinning these trends are biases baked into software mandated by Freddie Mac and Fannie Mae, specifically the Classic FICO scoring algorithm. The credit score determines whether an applicant meets a minimum threshold to be considered for a conventional mortgage in the first place, and traditionally, it’s been considered biased against non-whites because it rewards types of credit that are less accessible to people of color.
The loan approval process must also be okayed by Fannie or Freddie’s automated underwriting software, and research found that some variables within the programs weigh can impact people differently based on race or ethnicity.
“If the data that you’re putting in is based on historical discrimination, then you’re basically cementing the discrimination at the other end,” says Aracely Panameño at the Center for Responsible Lending.
Source: Associated Press (08/25/21) Martinez, Emmanuel; Kirchner, Lauren
© Copyright 2021 INFORMATION, INC. Bethesda, MD (301) 215-4688