“‘I’m an extremely good risk,’ said Ms. Stuart, 60, a lawyer. She had a 30-year work history and a credit rating above 800…. But her mortgage company denied her application. ‘I didn’t qualify for a refinance because I didn’t have enough income,’ she said. ‘It was extremely frustrating.’”
“‘We just couldn’t get anybody to accept our offer,’ said Shawn O’Farrell, who tried for more than 50 homes this summer before eventually closing on a three-bedroom, three-bath house in Appleton, Wis… Mr. O’Farrell served in the U.S. Army for seven years, deploying to Iraq three times. But getting an offer accepted on a home while moving his family from Illinois to Wisconsin turned out to be another battle.”
These narratives from The New York Times capture an issue that both older Americans and servicemembers are increasingly complaining about: challenges in refinancing and getting mortgages. Older Americans often do not meet income requirements while servicemembers face issues because of their mandatory frequent moves.
Mortgage and refinance application denials cause challenges for these two groups, and can lead to institutional and industry-wide hazards: in underwriting, regulatory and reputational risk, as well as negative P&L impact.
Juxtaposing the Need for Human Judgment and Artificial Intelligence
Today’s powerful AI-driven tools often reinforce traditional policies and standard methods which negatively impact certain groups of customers. Customers are individuals, each with unique circumstances — and we need both human judgment and AI to work together.
Ingesting our customers’ stories makes it possible for us to listen to them, identify High-Risk Interactions, and determine what needs to change. Their stories are the intelligence we need to improve our models and enable these segments more access to financial products.
What are our customers telling us when they are rejected for a mortgage or refinance opportunity and believe the outcome is not justified?
The use of customers’ narratives, by segment, as data input to underwriting methods can expand the approved universe for mortgage and refinance applications. Additionally, these innovations will reduce historical biases from accepted practices.
Using PositivityTech®, an intelligent platform that combines human insights and advanced technology, let’s take a look at issues that older Americans and servicemembers face, based on the stories they share:
In Their Own Words: How Automated Mortgage Underwriting Impacts Retirees
According to a recent study by the Federal Reserve Bank of Philadelphia, overall, the results suggest that older individuals systematically face higher barriers to mortgage access.
According to The Wall Street Journal:
“Lenders typically base their mortgage decisions on an applicant’s income, assets, debts and credit score. Discrimination against credit applicants on the basis of age is prohibited by the Equal Credit Opportunity Act. However, while lenders may not consider age per se when qualifying an applicant, they can look at age-related factors such as whether that applicant’s income might drop because they are about to retire. That is why lenders scrutinize not only the source of a borrower’s income, but the likelihood that it will continue as well.”
In Their Own Words: How Automated Mortgage Underwriting Impacts Servicemembers
According to Online Mortgage Advisor:
“Lenders try to take a more flexible approach to [military personnel] applications. They are aware that the nuances of military life can make it difficult to get a mortgage. Forces personnel may have a poor credit rating, for example, due to moving around so often. Or, they could receive a lot of their income through allowances, like a combat bonus, which may not typically be an earnings source that providers accept.”
What the Data Reveals: Retirees and Servicemembers Share their Stories Publicly
When customers submit their complaints to the Consumer Financial Protection Bureau (CFPB), they have the option of making their complaints public or keeping them private. In looking at the behaviors of retirees and servicemembers, we note the following:
Older Americans have the highest percentage of narratives that were made public. They are choosing to share their complaints publicly.
Specific to mortgage complaints, servicemembers, as a segment, have the highest percentage of narratives that were made public. Their level of frustration with mortgages is very high.
Using PositivityTech’s proprietary Severity Score, a domain-specific lexicon that identifies severe complaints and future risk based on customer narratives, we see that older Americans and older servicemembers have the largest concentration of complaints in the high Severity Score range.
Using PositivityTech’s Bias Index™, we see that servicemembers feel greater levels of discrimination than other populations, especially pronounced in the mortgage product.
Within mortgage products: percentage with the highest Bias Index range for servicemembers is almost 17%.
The Human Aspect is Critical to AI-Powered Underwriting
As our industry adopts more and more AI underwriting capabilities, we must ensure that the human aspect is not forgotten. By inputting customer narratives into lending models, we can ensure that each underwriting decision is appropriate, fair, and mitigates risk.