The role of consumer transaction data in increasing homeownership access
September 6, 2022, 12:57 pm By Brent Chandler
Share On
Click to share on Twitter (Opens in new window) Click to share on Facebook (Opens in new window) Click to email a link to a friend (Opens in new window)Click to share on LinkedIn (Opens in new window)
Over the past year, the Federal Housing Finance Agency (FHFA) and government-sponsored entities (GSEs) Fannie Mae and Freddie Mac have committed to exploring innovative ways to expand homeownership access safely and securely.
This charge couldn’t come at a better time.
The racial homeownership gap between Black and white Americans is wider than it was in 1960, and a combination of higher mortgage interest rates, rising home prices and an antiquated model for evaluating consumers’ ability to repay is exacerbating the problem. As mortgage professionals, it is our duty to support homeownership in the communities we serve with sustainable home financing opportunities. Luckily, advancements in evaluating consumers’ ability to repay have arrived and can enable lenders to adopt responsible and more inclusive lending practices.
The most straightforward method for understanding a consumer’s ability to make a mortgage payment is analyzing the funds going in and out of their bank each month. The long-standing model for determining ability to repay, a consumers’ credit score, only considers historic credit for determining credit risk. It excludes countless Americans who have no credit and thin credit, and it potentially miscalculates ability to repay for consumers with lower credit scores.
A holistic underwriting process that incorporates financial data from multiple vectors — disposable income, discretionary income and credit score — can offer greater protection from the risk associated with originating and manufacturing a loan while providing a more vivid picture of borrowers’ true Ability to Pay, which FormFree calls ATP.
Utilizing ATP gives lenders the best of both worlds. In addition to providing lenders with a powerful tool for responsibly expanding housing finance opportunities to underserved communities, ATP enables lenders to increase volume by capturing the business of creditworthy loan applicants that would have been declined under less sophisticated underwriting models. ATP helps lenders turn declined loans into closed loans, and it helps more consumers unlock the generational wealth building power of homeownership.
Understanding the language of cash-flow data
Though analyzing consumer transaction data offers a more nuanced understanding of a consumer’s ATP, manually reviewing bank statements is an enormous undertaking for busy mortgage professionals. Fortunately, modern advancements in data intelligence offer lenders a way to reincorporate that process without all of the hassle.
A powerful innovation of the 21st century, natural language processing incorporates aspects of linguistics, computer science and artificial intelligence (AI) to identify patterns between digital data and human language. For the purpose of mortgage lending, natural language processing can be used to extract consumer cash flow data from their bank statements to offer lenders a better understanding prospects’ disposable and discretionary income.
This wide-reaching data is helpful on its own but is even more useful when considered alongside traditional models of credit risk evaluation. Because lenders can corroborate the data gleaned by natural language processing with other external third-party data sources, such as the credit bureaus, it offers an extra layer of security when determining a homebuyer’s credit eligibility.
Natural language processing can also prevent unintentional biases that can come from alternate forms of data intelligence. While other AI- or machine learning-based technologies have been known to programmatically learn and apply human biases, natural language processing solutions use rules-based algorithms that only calculate income and cash flow. These algorithms intentionally omit information such as an applicant’s ethnicity to maintain objectivity. In this way, natural language processing is designed to find the truth — truth based on actual bank data — rather than support longstanding human biases that limit homeownership for minority borrowers.
Democratizing the credit decisioning process
This year, FHFA Director Sandra Thompson testified before Congress that rental payment history is equally as important to a consumer’s credit assessment as their mortgage payment, and she makes an excellent point. Rent is often the most significant payment that a consumer makes each month, but not all landlords submit consumers’ rent payment history to the credit bureaus. Small landlords with less sophisticated systems often cannot automatically report positive rental history, so their renters’ positive payment history is not reflected in their credit scores. This power dynamic, where renters must depend on their relationship with their landlord, is both unfair to the consumer and time-consuming for lenders.
By requesting landlords report a tenant’s payment history during the credit decisioning phase, lenders and borrowers become reliant on a third party for financial data. This not only slows the underwriting process but also adds another layer of risk, as data requested from landlords can be unreliable. When you send a rent verification request, a busy or uninvolved landlord may simply sign it and confirm a consumer paid their rent on time without the strict due diligence that lenders need.
In addition, natural language processing mitigates credit risk by analyzing cash flow data that lenders do not usually consider. For example, even though a person’s income may be sufficient enough to qualify for a loan, if they are paying thousands per month on their children’s daycare or private education, that expense will negatively affect the amount of money they can put toward a mortgage each month. With natural language processing mining a consumer’s bank data, lenders can take major monthly expenses like childcare into account when determining credit eligibility. This allows lenders to serve more homebuyers and keep pipelines flowing in a safer, more inclusive manner.
Using technology to support traditional credit risk evaluation methods puts more power in consumers’ hands during the credit decisioning process. While not every consumer has the ability to share their financial data to show lenders what makes them a reliable borrower, with natural language processing, their ATP can be computed as an average of a typical month’s income and expenses. Ultimately, this deeper insight into consumers’ financial habits expands equal affordable housing opportunities by allowing homebuyers to prove their credit risk with the transactions in their bank statements, not just their credit score.
Safeguard our economy by placing truth over trust
The housing finance industry is a central player in our economy, and a market of such enormous scale and influence inherently affects everything else around it. To preserve the health of our housing economy, lenders, investors, regulators, GSEs and fintechs take special care in the policies, procedures and data that they use to determine borrower risk. However, this commitment to safe and sustainable homeownership does not prevent our industry from evolving its processes. That is why the home finance industry’s biggest powerhouses, the FHFA and the GSEs, are committed to finding new, innovative ways to use consumer data to expand access to sustainable credit.
With deeper insight into direct source, consumer-permissioned bank data, lenders can determine much more about a prospective borrower’s financial situation — from how they pay their bills to their various sources of income, their assets and even their employment status. Natural language processing is able to analyze these various data points in a consistent, structured and organized way with rules-based algorithms that can identify a wide range of common monthly expenses quickly and easily.
It is important to note that natural language processing technologies for the mortgage industry did not appear overnight. It has taken computer scientists several years to build reliable algorithms that allow the GSEs’ automated underwriting systems to receive consumer data, identify transactions like rent payments and build the messaging that goes back to the lender letting them know a consumer’s ATP. These systems have been painstakingly tested and retested before being released to a wide lender audience. This due diligence ensures the technology available on the market is not only effective and accurate, but also has the proper regulatory safeguards and does not include some unknown form of bias.
A consumer’s own financial data is the most reliable for identifying the risk of a consumer, versus having it go through multiple hands and be evaluated by another type of risk model. Looking at actual financial transactions with natural language processing lets lenders consider both “truth data,” a consumer’s transaction history from their accounts, and more trust-based data, like third-party reports and credit scores, in the underwriting phase.
What’s next?
Effective natural language processing technologies extract deeper meaning from unstructured data to make a difference in the lives of countless would-be homebuyers who are credit invisible or have not had the ability to obtain access to affordable housing finance. By using this technology to compare cash flow against other credit risk evaluation models, lenders are better able to manage risk appropriately and inclusively.
The mortgage industry has made significant progress in the way it leverages direct-source data, and there’s more to come. Industry leaders are investing in research and development to expand the realm of financial data that consumers can offer permission for lenders to view and analyze. With support from the FHFA, the mortgage industry can anticipate policy changes that support more nuanced risk evaluations for the betterment of lenders, consumers and the housing economy as a whole.
