New Study: Fintech Reduces Discrimination in Lending
Although discrimination in lending remains a problem in America today, discrimination is declining, largely as a result of fintech firms’ use of algorithms. According to the latest research, fintech algorithms discriminate 40 percent less than traditional, face-to-face lenders.
The study’s authors, from the University of California, Berkeley, reviewed data from thousands of the largest mortgage lenders from 2012 to 2018. About half of them offer online or app-based mortgage applications. Whereas fintech firms employ algorithms to determine the price each borrower must pay for a particular loan, traditional lenders determine the price through face-to-face interaction.
As a result, the report claims that Latinx and African American “borrowers pay 7.9 and 3.6 basis points more for purchase and refinance mortgages, respectively, costing them $765 million in aggregate per year in extra interest.” Although fintech firms still discriminate through the use of faulty algorithms, they discriminate 40 percent less than traditional lenders. Moreover, the study finds that fintech firms do not discriminate against borrowers in the loan approval process.
In an interesting twist, Democratic Senators Elizabeth Warren and Doug Jones appear to have expressed concern over the study’s findings in a letter sent to the Federal Reserve, the Office of the Comptroller of the Currency, the Federal Deposit Insurance Corporation, and the Consumer Financial Protection Bureau.
Dated June 10, 2019, the letter references a study from the University of California, Berkeley, that finds, “that ‘face-to-face and FinTech lenders charge Latinx/African American borrowers 6-9 basis points higher interest rates than comparable white or Asian-ethnicity borrowers – which is almost identical to the interest rate premium that minority borrowers face with traditional underwriting.” As a result of the study’s findings, the Senators asked the federal regulators how they plan to identify and combat the discrimination by fintech algorithms.
The confusion stems from the Senators mixing up two different studies. The study finding that fintech firms discriminated 40 percent less than traditional lenders was published in May 2019. The authors of that study released an earlier version in October 2018, which found that “fintech algorithms have not removed discrimination, but may have shifted the mode.” The Senators’ letter is referencing the earlier October 2018 study and not the updated version from May 2019.
Despite the confusion, the latest research clearly indicates that fintech firms and algorithmic models reduce discrimination in lending by large margins, but have not eliminated the practice altogether.