This piece was originally published for Forbes.
Have fintechs really bridged the inequality gap? This was the question presented in last week’s inaugural post, in an effort critically examine the performance of the industry. Real estate technology, and mortgage lending in particular, is a key area where fintechs have not managed to bridge the inequality gap as successfully as anticipated. This second in a three-part series examines the limitations of artificial intelligence, and highlights the ramifications on the rapidly heating US lending and refinancing market.
Inequality in Mortgage Lending
Although it is illegal for lenders to discriminate on credit and loan decisions on the basis of race, in reality, this has proven to not be the case. A LendingTree study found that African American borrowers had the highest denial rates at 17.4%. Their white counterparts, on the other hand, had denial rates of only 7.9%.
A study conducted by UC Berkeley proved that when solely the income and credit score of previously rejected borrowers was used, their mortgage applications were accepted. It should stand to reason, then, that using an algorithm would remove the probability of such unfair practices occurring, and level the mortgage lending playing field. After all, the analysis would be driven by data and computers, not people.
Not so fast.
As Nathan Kallus, assistant professor of operations research and information engineering at Cornell Tech, explains, “How can a computer be racist if you’re not inputting race? Well, it can, and one of the biggest challenges we’re going to face in the coming years is humans using machine learning with unintentional bad consequences that might lead us to increased polarization and inequality.”
One notable example of the “bad consequence” that Kallus alluded to was in the results of a risk assessment software called COMPAS, which was supposed to help predict which criminals were more or less likely to re-offend. As Mark Sears wrote in “AI Bias And The People ‘People Factor’ In AI Development,” “When the algorithm was wrong, people of color were almost twice as likely to be labeled a higher risk, yet they did not re-offend.”
Clearly, AI is not immune from the challenges of systemic biases. As Sarah Myers West, a postdoctoral researcher at New York University’s AI Now Institute explained to CBS News, “”We turn to machine learning in the hopes that they’ll be more objective, but really what they’re doing is reflecting and amplifying historical patterns of discrimination and often in ways that are harder to see.” An inadvertent vicious cycle has been created, whereby tainted data is being utilized to inform future decisions.
The impending refinancing storm
The consequences of this are far-reaching, particularly in the mortgage lending industry. Rocket Companies, the parent company to Quicken Loans, went public this year, and has fundamentally disrupted the housing industry. According to data from the Wall Street Journal, “Quicken was the largest lender during the first 6 months of 2020, ahead of perennial front-runners such as Wells Fargo & Co.WFC and JP Morgan ChaseJPM.” Rocket is positioning itself as a technology front-runner in everything from personal loans, with their their Rocket Loans platform, to home-buying, via Quicken Homes.
Furthermore, Rocket is poised to capitalize on the current historically low interest rate environment. Weekly refinance applications hit record highs earlier this year, and many lenders struggled to keep pace with the deluge of applications. Kelly King, CEO of Truist Financial Corp.TFC, lamented to the Wall Street Journal, “The industry does not have the capacity to handle it.” Firms that were more technologically advanced weathered the storm better than those who were not.
Putting it all together
Ironically, as the mortgage lending industry becomes increasingly reliant upon data analysis to automate and expedite decision-making, more human diligence is required to ensure that outcomes are fair. As the COMPAS study results highlight, biased data leads to biased outcomes. Diversity in recruiting is more important than ever, to ensure that objective data analysts and scientists are in charge of handling such delicate figures. For now, it is too soon to say whether fintechs have meaningfully reduced inequality in mortgage lending. Challenges persist.
In the meantime, buyer beware. If you are a black or LatinX home buyer, be sure to shop around for your mortgage.