The nuance involved in originating mortgages to borrowers with nontraditional credit profiles often forces lenders to rely on manual underwriting. But these methods cost more and lack efficiency, creating the need for a more inclusive automated solution.
At New American Funding, which lends to many Hispanic borrowers, the company's president Patricia Arvielo noticed that loan officers serving this demographic — many of them Latino or Latina themselves — were encountering major hurdles when their loans went to underwriting.
"They were getting caught in the underwriting by an underwriter who wasn't culturally aware," Arvielo said.
To remedy this, Arvielo "halted the way we were doing business" and created a separate team of 20 experienced underwriters. These individuals were familiar with the nuances of Hispanic borrowers and now manually underwrite these loans.
"They're more seasoned underwriters that have seen these cultural nuances," Arvielo said. "I rarely now get a Latino loan agent stuck in underwriting."
In many ways, Arvielo's strategy resembles the process at many larger lenders. At Fifth Third Bancorp, loans that cannot be processed through an automatic underwriting system, often due to a borrower's nontraditional credit background, are elevated to an exception desk, said Edward Robinson, senior vice president and head of mortgage at Fifth Third.
This group of underwriters evaluates these loans to determine whether Fifth Third is in a position to take on the risk involved. In some cases, loans even make their way to Robinson's office if the exception desk cannot reach a decision.
"We have an exception desk that is committed 100% to making sure that as we get folks that are on the fringe or below where our automated tools would allow, there's a desk and a policy in which we can underwrite them," Robinson said.
"And if it doesn't even get approved by the exception desk, they can elevate it to me and I can handle it with my capital markets team," he added.
New American services its loans and holds nearly all of its originations in portfolio, allowing it to monitor pressure points and better understand potential warning signs for default.
"I'm able to look at the delinquencies in our book and begin to start peeling back the onion," Arvielo said. "I can see which borrowers are in default and why they are in default."
By holding onto the loans it originates, New American — which Arvielo said boasts a 0.06% 90-day delinquency rate in its $17 billion servicing portfolio — can more easily expand its credit box.
While manual underwriting and balance sheet lending alleviate many of the issues regarding lending to underserved borrowers, they are far from a panacea. As a result, the number of lenders able and willing to tap into these segments remains limited.
For starters, manual underwriting is a costly and time-consuming process. At New American Funding, underwriters devoted to borrowers with nontraditional credit background only get through two loans per day, compared to a workload of four to six loans per day that an underwriter can manage using automation.
"We have to go back to borrowers several times for receipts, trailing the funds," Arvielo said. "It's difficult."
Many lenders also lack staff with the talent to handle the complexities of manual underwriting, particularly given that many younger underwriters have only ever operated in an automated environment.
"There are not a lot of good, solid manual underwriters left" following the financial crisis, Altavera Mortgage Services President and CEO Brian Simons said.
Where portfolio lending is concerned, there is a higher amount of risk involved. Banks are limited by regulations in the amount of such risk they can take on, and nonbank lenders must have the funds to be able to keep loans on their balance sheet.
Nonbank lenders are "afforded the ability to do things that I can't," Robinson said.
Steps are being made to alleviate the issues involved with both manual underwriting and portfolio lending. Fannie Mae, which has long offered manual underwriting for nontraditional borrowers, recently introduced a new process for underwriting these loans through its automated underwriting system, Desktop Underwriter.
Beginning with the latest edition of Desktop Underwriter released in September, lenders can now underwrite these loans through a nominally automated process.
"Although we have a minimum credit score requirement, DU completes a direct assessment of the borrower's credit history," said Jude Landis, vice president of single family credit policy and risk management at Fannie Mae. "This allows us to be more flexible and directly assess credit in a way that makes the whole conversation about credit scores less urgent for us."
Lenders first submit information to meet Desktop Underwriter eligibility criteria, including loan limits, a loan-to-value ratio of no more than 90% and a debt-to-income ratio of less than 40%. If an approval recommendation is received, the lender must document a payment history for the borrower with at least two nontraditional credit sources, including at least one that is housing-related, such as rental payments. If a borrower does not meet the eligibility criteria, the loan can still be manually underwritten to receive approval.
"The innovation of being able to submit loans through Desktop Underwriter is so important," said Deborah Momsen-Hudson, director of secondary market lending at Self-Help Credit Union in Durham, N.C.
But Fannie Mae's offering isn't perfect. Indeed, it is not completely automated, and adoption could be an issue.
"People still talk about nontraditional mortgages with 2007 and 2008 sitting in their minds," Simons said.
And the program still presents restraints or requirements that may prove problematic for some borrowers, Arvielo said.
"I like to remain positive about all changes or enhancements that may help LMI borrowers access homeownership, [but] this particular program is too strict with the down payment requirements and DTI," Arvielo said.
In addition to the 3% down payment loan programs that Fannie Mae and Freddie Mac began offering in 2015, another opportunity has emerged in the partnerships being formed among banks, smaller lenders and the government-sponsored enterprises. Self-Help Ventures Fund, which manages the secondary market programs for Self-Help Credit Union and Self-Help Federal Credit Union, has entered into two partnerships, one with Bank of America and Freddie Mac and another with Wells Fargo and Fannie.
Through these partnerships, the banks originate, process and underwrite affordable, low down payment loans to traditionally underserved borrowers. In some cases, Self-Help offers risk coverage for the loans, which are all sold to the GSEs.
Self-Help does its own retail lending, too, focusing on underserved communities. But because of its size, its reach is limited.
"What the secondary market partnerships with Wells Fargo and Bank of America allow us to do is take learning we have from our own direct lending in low-income communities and leverage that with the large distribution networks of national banks," Momsen-Hudson said.
Fifth Third has developed its own strategy for addressing the needs of borrowers in underserved communities. The bank is forming community economic development "pods" in these communities, Robinson said, "that will include mortgage loan officers devoted solely to unit production. These loan officers will be married with select processors and underwriters to handle the complexities the loans may present if there at the edge of the credit curve."
While the program is expected to benefit these communities, it may not provide as much marginal benefit to Fifth Third, Robinson said.
"On the margin, it's not going to be as profitable as other parts of my business," he said. "But if I'm able to provide credit to folks who may not be able to get it elsewhere and I'm able to do it in a more efficient way, then I can move the needle and get a lift from a productivity standpoint."
While such partnerships and strategies certainly alleviate the credit crunch in underserved communities, they, too, are limited in the industrywide impact they can have — barring other companies follow suit.
When it comes to finding solutions for the problems posed by manually underwriting loans to borrowers with nontraditional credit backgrounds, lenders can look to the examples of startups in the small business lending space such as Camino Financial, a company that Arvielo sees as a potential "good marriage" with a mortgage lender on a strategic basis.
Formed in 2014 by twin brothers Kenneth and Sean Salas, Camino Financial provides small loans to Hispanic-owned small businesses. The company provides business loans, excluding commercial real estate, of up to $500,000.
To serve this niche, the company developed a technology-based underwriting model. Having a largely automated process was a necessity for the company given the small size of the loans it originates. The average size of a Hispanic-owned business, Kenneth Salas said, is less than $200,000, and the average loan size for Camino Financial is roughly $30,000.
"You definitely need to have the technology from a process standpoint to turn volume quickly," he said. "It lowers transaction costs."
Since many of the businesses did not have strong credit histories, Camino Financial utilized big data in developing its underwriting platform. The system runs regressions with data the company has collected on Hispanic businesses along with a credit pull for the prospective borrower to ascertain creditworthiness.
"You do need to take an alternative view on credit," Salas said. "Our super-FICO model, leveraging a lot of the data we've gotten from different Hispanic business owners, allows us to take a different view."
Salas argues that a similar approach to underwriting would work well within the mortgage industry, particularly because of the glut of data lenders have at their disposal, albeit challenges would remain for mortgage lenders that Camino Financial has not faced since it works with private capital.
"I think they can totally apply a similar model — there's a lot more Hispanic consumer data out there than there is Hispanic business data" Salas said. "There are certain standards that mortgage companies need to abide by, and that limits innovation."
Another high-concept solution to these issues could come from the use of artificial intelligence. Integrating such technology could allow lenders to move beyond matrix-based automation and infuse human-like intuition into a computer-based process by learning how underwriters make their decisions in a manual environment.
The technology has "the computer mirror and mimic decision-making processes that humans would have, which is not necessarily matrix based," Simons said. "They could track and have underwriters underwrite files, and create an underwriter's brain about how and why they made decisions."
Another high-level possibly that Robinson suggests is adding geocoding overlays to determine where a borrower lives.
"If a loan comes in and is outside of box, maybe you have an alternative where you have a geocoding for a low-to-moderate tract. That may be a way to move the needle with some automated tools, without risk to RESPA," he said.
Such technology is still a long way from being implemented. But there are other steps the industry can take in the interim to adapt automated underwriting that's better suited for the changing face of its borrowers.
"Everybody is still living with this stigma that this type of lending has some sort of negative association," Arvielo said. "It's about using our voice to advocate for opening the credit box sustainably."