Report Finds Racial Bias in AI-Powered Mortgage Decisions, MBA Objects



Even decades after the Fair Housing and Equal Credit Opportunity Act passed, The Markup study found that loan refusal rates are shifting heavily towards non-white borrowers.

Compared to white mortgage applicants with similar financial characteristics, black consumers faced an average of 80% higher probability of failure, according to the analysis of the data of the Law on Disclosure of Information on Housing Mortgages for 2019 by the non-profit organization. The probability of rejection has reached 70% higher for Native Americans, 50% for Asian or Pacific Islans, and 40% for Hispanics.

The report analyzed 2 million home mortgage applications nationwide. The analysis did not include loans from the Federal Housing Administration or the Department of Veterans Affairs, which have lower qualification standards and a lower borrower profile than conventional loans.

The Mortgage Bankers’ Association challenged the assumption that the industry perpetuates racial discrimination. He said the report was garbled inequality problem and his results did not give a complete picture.

FICO scores were not used in the report as they are not published. However, traditional credit ratings can be inherently disadvantageous to people of color, as they have historically not had equal access to the banks or assets that create credit. However, the lease payments were not taken into account in the credit estimates. Fannie Mae will start using a positive rental history. as an underwriting factor starting September 18 (for those with an inconsistent record of timely lease payments, this record will not count as such payments may be made in cash that will not have an electronic record).

“[The Markup’s] HMDA data analysis and its predetermined conclusions regarding mortgage lending do not take into account several key components that form the basis of lending decisions, including the borrower’s credit rating and credit history, ”the MBA said in a press release. “As we told the authors, the Federal Reserve, CFPB and other regulators have made it clear that negative differences in HMDA data alone cannot be used to measure fair lending.”

Today, most decisions on loans and the underwriting process is automated… The algorithms governing this technology are designed to be “color blind” and do not take into account race or possible human bias. However, the problem may be related not so much to the honesty of the algorithms themselves, but to the underlying data on which they are based.

“This is how structural racism works,” Chi Chi Wu, a staff attorney for the National Consumer Rights Advocacy Center, said in his report. “This is how racism takes root in institutions, politics and practice without any hostility.”

Credit history was the reason for mortgage denials for 36.9% of Native American applicants, 32.5% of Blacks, 21.3% of Hispanics, 20.9% of Whites, and 12.1% of Asians, according to HMDA data.

Some vendors have sought to offer a more comprehensive rating for those who do not have an extensive credit history. VantageScore, a third party credit score provider, estimates that according to his model, 37 million borrowers who do not have a FICO score will become creditworthy, and a third of that number will be either black or Hispanic.

“They’ve been testing alternative grades for years and I don’t know why the process is taking so long.” said Lisa Rice, President and Chief Executive Officer of the National Fair Housing Alliance, a consortium of hundreds of fair housing organizations. “Deserving customers are left behind.”


Source link