Excluding people from the economy is bad for the victims and bad for the rest of the economy too. Why do less business than you could? Why leave people out?
The conventional credit scoring system has a serious problem with excluding people who could be good customers. Nearly 50 million Americans — disproportionately including the poor and minorities — lack a credit score and cannot obtain mortgages, credit cards or other lending products. Yet many people in this segment of society pay rent, utilities and other recurring financial obligations.
Fintech companies are reinventing credit scoring by measuring borrowers’ ability to pay (ATP) in new ways and by making it easier for consumers to match up with financial institutions that want their business.
The trouble with credit scores
Traditional credit scoring has been recognized as a barrier to financial inclusion since long before the OCC’s July launch of Project REACh. Many Americans — disproportionately including minorities — lack a credit score, and many more have a credit score that is misleadingly low.
Paradoxically, this problem arose out of an earlier attempt to promote inclusion. At one time, it was legal for credit underwriters to consider race, neighborhood and even religion. Serious concerns were raised in the 1970s that using such criteria for credit scoring, even if statistically defensible, would perpetuate injustice. There were also concerns that married women’s credit was insufficiently distinct from their husbands’ and that divorced people were being penalized for their marital history. The Equal Credit Opportunity Act of 1974 made it illegal for credit underwriters to consider applicants’ race, color, national origin, religion, sex, age, marital status and childbearing plans.
After a period of uncertainty about how to revamp credit scoring, in 1989 Fair Isaac introduced the FICO score. Based only on repayment history, credit history and utilization relative to credit limits set by lenders, FICO is blind to race and color. It is also blind to income and employment, and it cannot see payments for rent, utilities or other regular financial obligations.
The problem is, the FICO score knows too little. It is a circular measure whose goal is to predict future loan repayment based on past loan repayments. That is better than nothing, but it leaves out people who have never borrowed much money, and it cannot detect when someone’s financial situation has improved or deteriorated.
A look under the hood of one FICO alternative
Emerging approaches like Fair Isaac’s UltraFICO, Experian Boost and FormFree’s Passport differ from traditional credit scoring in two ways:
- They measure ATP by using information in the consumers’ bank accounts
- They put control of the process in the hands of the borrower (the consumer) rather than the lenders
Of these companies, it was FormFree that actually invented the concept of using direct-source asset data in credit decisioning, and in working with Fannie Mae and Freddie Mac, became the first ATP verification solution to be granted the governmental guarantee known as representation and warranties (R&W) relief in 2016. Experian introduced Boost in late 2018, and Fair Isaac rolled out UltraFICO in early 2019. Perhaps because it’s been around the longest, Passport also offers the most multi-dimensional analysis of these three.
Here’s a deeper look at how FormFree’s Passport analyzes the borrower’s whole financial situation, including assets, employment and income, using the bank and credit card transaction data provided by financial institutions at the request of the borrower.
First, transactions are classified. For example, paychecks and other income are classified based on their regularity and the names of the payers. Expenses are divided into discretionary, obligatory or mixed using the merchant name wherever possible (e.g., an expense from Kroger is groceries, while a line item from an AMC theater is entertainment).
If a payee name is not recognized, the expenditure is categorized via statistical analysis of other data points like transaction size, recurrence and roundness. For example, $1500 every month is likely to be rent, but $1513.26 every month looks more like a mortgage payment, and $1500 at irregular intervals is probably something discretionary.
Then the borrower’s cash flow is analyzed. A day-by-day time series of every bank and credit card balance is constructed. This makes it easy to look for rising and declining balances. Monthly residual income is calculated, taking into consideration not only money left over after actual spending, but also money that would be left over if discretionary spending were reduced or if the borrower were willing to slightly draw down savings during lean periods.
Unusual months are detected and can be pointed out in the report. Passport’s analysis even helps credit underwriters detect payments into or out of bank or credit card accounts that were not provided for analysis, indicating gaps in the information provided by borrowers.
Artificial intelligence technology is used on several levels. Crucially, financial conclusions are drawn by rule-based systems, not neural networks or numerical models, to ensure the system has not “learned” some indirect indicator of race or ethnicity — always a risk with machine-learning technology.
FormFree uses economic models to understand, on an even deeper level, the significance of a borrower’s ongoing and changing financial situation. This puts lenders in a position to be proactive with borrowers who are facing difficulty, perhaps even before the borrowers themselves realize it, and to distinguish nationwide or community-wide economic problems from personal ones.
How ATP scoring empowers the consumer
Traditional credit scoring is driven by the would-be lender. Today’s front-running ATP scores are driven by the borrower, who signs up, grants permission for his or her bank accounts to be analyzed and allows the results to be sent to specific lenders.
ATP scoring is poised to create millions of new financial opportunities for consumers with no credit history or subprime credit history, a group that disproportionately includes minorities and low-income earners. That’s because, for the first time, lenders will have the data they need to confidently extend credit to this historically underserved segment.
In the near future, FormFree predicts borrowers will be able to furnish their ATP scores to an open market of lenders and hear from the ones that want to make a loan. This will, for the first time, bring the efficiency of the free market to each individual borrower, not just to the banking system as a whole.
