The following is a guest post by Todd Thomas, who has been in financial services for more than 20 years.
According to data from the Urban Institute, the median FICO credit score for Hispanic consumers is about 75 points lower than the median white consumer’s. The median credit score for Black consumers is more than 100 points lower than the median white consumer’s. And the approximately 10% of American adults without usable credit files are disproportionately people of color.
These racial, demographic, and geographic disparities are rooted in “historical inequities that reduced wealth and limited economic choices for communities of color,” according to the Urban Institute. It notes that subprime borrowers can pay $3,000 more in interest and fees on a $10,000 car loan over four years.
This unequal status quo has advocates calling for a more objective approach to consumer risk modeling. Innovators have created new technologies and data sources that could do just that.
The following four consumer risk modeling innovations are poised to disrupt the current credit scoring regime to varying degrees. Collectively, they could change the concept beyond recognition and eventually render obsolete what we think of today as “credit scoring.”
1. Forward-Looking Changes to Current Scoring Models
The most recent changes to the FICO scoring model, collectively known as FICO 10 and FICO 10T, are iterative rather than transformational updates. In other words, they don’t radically alter the calculation of credit scores.
But FICO 10 and 10T hint at the direction traditional credit scoring models are moving — and what they might have to do to remain relevant in the future as more disruptive innovations take hold.
FICO 10 and 10T pay closer attention to consumers’ credit mix in the context of their overall debt loads. Specifically, they penalize consumers who take out new personal loans to consolidate existing debts, then continue racking up debt on those current trades (most often, credit cards).
Essentially, they aim to reward good credit behavior(paying down debt) and discourage risky habits (living beyond one’s means).
2. Cash Flow Modeling
Another recent FICO update attacks credit scoring discrepancies more directly. The UltraFICO score — a joint venture between Fair Isaac Corporation, Experian, and Finicity — pulls in noncredit data to provide a more accurate and fair picture of consumers’ credit risk. Cash flow monitoring includes cash flow in a bank account and payment history.
By incorporating banking information, such as account balances and account age, UltraFICO supports credit scoring for about 15 million people who don’t have enough credit history to have traditional credit scores. Unfortunately, those people are disproportionately lower-income and POC — those most likely to be left behind by the credit scoring status quo.
UltraFICO is an example of cash flow modeling. Long used by business lenders, cash flow modeling is working its way into the consumer credit mainstream thanks to adoption by fintech lenders like Accion, Brigit, and Petal.
According to an analysis by FinRegLab, “the predictiveness of the cash flow scores and attributes was generally at least as strong as the traditional credit scores and credit bureau attributes,” suggesting it’s a reliable complement to or replacement for traditional scoring. And cash flow modeling is more equitable than conventional scoring, according to FinRegLab’s data.
3. International Credit Scoring and Risk Modeling
The current credit scoring regime also explicitly discriminates based on nationality. Non-U.S. nationals who come to the United States don’t have the requisite credit history to qualify for FICO scores. They’re essentially invisible to lenders that rely on FICO scores to make lending decisions.
Fortunately, border-based barriers to international credit scoring are already crumbling, thanks to global consumer credit risk models like Nova Credit. As more U.S. lenders begin to trust and adopt these models, new arrivals to the U.S. and Americans relocating abroad could find it easier to obtain credit without traditional country-specific credit scores.
4. A Post-Credit-Score World
Finally, noncredit and not-only-credit scoring models like FICO XD hint at what’s possible in a truly “post-credit-score” world.
FICO XD does not rely entirely — or even principally — on credit bureau information. The model pulls data from property records, leasing databases, and noncredit contracts like utility agreements to develop a comprehensive picture of a consumer’s likelihood of default.
According to Fair Isaac Corporation, FICO XD can produce FICO scores for up to 70% of previously unscorable consumers, including many historically disadvantaged demographic groups.
This is an exciting time for consumer risk modeling. From incremental changes to existing models (FICO 10, UltraFICO) to more radical shifts (cash flow modeling, FICO XD) that could supplant credit scoring entirely, we’re seeing a wave of innovation unlike any since the early days of modern underwriting.
These innovations can reduce long-standing credit scoring disparities and produce more accurate consumer credit risk models. But they offer no guarantees. Their actual impact on consumer finance will depend on who wields them and how.
Todd Thomas has been in financial services for more than 20 years.
Photo by Dylan Gillis on Unsplash
The next is a visitor submit by Todd Thomas, who has been in monetary providers for greater than 20 years.
In keeping with knowledge from the City Institute, the median FICO credit score rating for Hispanic customers is about 75 factors decrease than the median white shopper’s. The median credit score rating for Black customers is greater than 100 factors decrease than the median white shopper’s. And the roughly 10% of American adults with out usable credit score information are disproportionately individuals of colour.
These racial, demographic, and geographic disparities are rooted in “historic inequities that lowered wealth and restricted financial decisions for communities of colour,” based on the City Institute. It notes that subprime debtors pays $3,000 extra in curiosity and charges on a $10,000 automobile mortgage over 4 years.
This unequal established order has advocates calling for a extra goal method to shopper danger modeling. Innovators have created new applied sciences and knowledge sources that might just do that.
The next 4 shopper danger modeling improvements are poised to disrupt the present credit score scoring regime to various levels. Collectively, they might change the idea past recognition and ultimately render out of date what we consider at present as “credit score scoring.”
1. Ahead-Wanting Modifications to Present Scoring Fashions
The latest adjustments to the FICO scoring mannequin, collectively referred to as FICO 10 and FICO 10T, are iterative somewhat than transformational updates. In different phrases, they don’t radically alter the calculation of credit score scores.
However FICO 10 and 10T trace on the course conventional credit score scoring fashions are shifting — and what they may should do to stay related sooner or later as extra disruptive improvements take maintain.
FICO 10 and 10T pay nearer consideration to customers’ credit score combine within the context of their general debt masses. Particularly, they penalize customers who take out new private loans to consolidate present money owed, then proceed racking up debt on these present trades (most frequently, bank cards).
Primarily, they goal to reward good credit score habits(paying down debt) and discourage dangerous habits (dwelling past one’s means).
2. Money Movement Modeling
One other current FICO replace assaults credit score scoring discrepancies extra immediately. The UltraFICO rating — a three way partnership between Truthful Isaac Company, Experian, and Finicity — pulls in noncredit knowledge to offer a extra correct and honest image of customers’ credit score danger. Money circulate monitoring consists of money circulate in a checking account and cost historical past.
By incorporating banking info, akin to account balances and account age, UltraFICO helps credit score scoring for about 15 million individuals who don’t have sufficient credit score historical past to have conventional credit score scores. Sadly, these individuals are disproportionately lower-income and POC — these more than likely to be left behind by the credit score scoring established order.
UltraFICO is an instance of money circulate modeling. Lengthy utilized by enterprise lenders, money circulate modeling is working its method into the buyer credit score mainstream because of adoption by fintech lenders like Accion, Brigit, and Petal.
In keeping with an evaluation by FinRegLab, “the predictiveness of the money circulate scores and attributes was typically not less than as sturdy as the standard credit score scores and credit score bureau attributes,” suggesting it’s a dependable complement to or substitute for conventional scoring. And money circulate modeling is extra equitable than typical scoring, based on FinRegLab’s knowledge.
3. Worldwide Credit score Scoring and Threat Modeling
The present credit score scoring regime additionally explicitly discriminates based mostly on nationality. Non-U.S. nationals who come to america don’t have the requisite credit score historical past to qualify for FICO scores. They’re primarily invisible to lenders that depend on FICO scores to make lending choices.
Thankfully, border-based boundaries to worldwide credit score scoring are already crumbling, because of international shopper credit score danger fashions like Nova Credit score. As extra U.S. lenders start to belief and undertake these fashions, new arrivals to the U.S. and Individuals relocating overseas might discover it simpler to acquire credit score with out conventional country-specific credit score scores.
4. A Publish-Credit score-Rating World
Lastly, noncredit and not-only-credit scoring fashions like FICO XD trace at what’s doable in a really “post-credit-score” world.
FICO XD doesn’t rely completely — and even principally — on credit score bureau info. The mannequin pulls knowledge from property data, leasing databases, and noncredit contracts like utility agreements to develop a complete image of a shopper’s probability of default.
In keeping with Truthful Isaac Company, FICO XD can produce FICO scores for as much as 70% of beforehand unscorable customers, together with many traditionally deprived demographic teams.
That is an thrilling time for shopper danger modeling. From incremental adjustments to present fashions (FICO 10, UltraFICO) to extra radical shifts (money circulate modeling, FICO XD) that might supplant credit score scoring completely, we’re seeing a wave of innovation in contrast to any for the reason that early days of recent underwriting.
These improvements can cut back long-standing credit score scoring disparities and produce extra correct shopper credit score danger fashions. However they provide no ensures. Their precise impression on shopper finance will rely on who wields them and the way.
Todd Thomas has been in monetary providers for greater than 20 years.
Picture by Dylan Gillis on Unsplash