Big data vs. the credit gap
There’s a catch-22 at the core of the U.S. financial system: To get credit, you need to already have established a credit history. Millions of Americans never find a way around the contradiction, and as a result, are locked out of things like credit cards or student loans that the rest of the population can take for granted.
Banks and other financial companies usually rely on the three major credit reporting agencies to decide whether to let you have credit, using something called your FICO score, an algorithm based on your credit history. No credit history; no FICO score. (If you have a thin credit history or a bad score, you might be able to get a car or a loan, but you will pay higher interest rates and fees.)
But not having a credit history is not the same thing as being a credit risk. In fact, many people without credit histories may be very good credit risks; they’ve figured out ways to pay rent, buy groceries and keep the electricity on without the convenience of cards or other forms of credit.
A number of financial startups, recognizing that many of those transactions are recorded digitally, are exploring ways to open the door to such people with new credit-scoring methods, using a wider array of financial data to better identify who’s likely to repay a loan. If it works, it holds out the promise of helping more people start businesses, buy homes and get higher education. Businesses also see an opportunity to reach a new set of customers.
“People with little or no credit history, or who lack a credit score, have fewer opportunities to borrow money in order to build a future and any credit that is available usually costs more,” Richard Cordray, former director of the Consumer Financial Protection Bureau, said last year in announcing that the bureau would explore the idea. “That only deepens their economic vulnerability.”
But there’s a catch. The new companies use software and data sets that probe deeply into people’s personal and financial lives and raise concerns among privacy advocates. Other critics worry the new methods and data sets could repeat the same kinds of discrimination that have kept credit unavailable or expensive to minorities for generations.
Opening up credit could potentially benefit a huge swath of people: A 2015 study by the bureau found that 26 million adult Americans were “credit invisible,” meaning that for practical purposes, they don’t exist to the credit firms that collect financial history of borrowers. According to the study, another 19 million Americans had credit records of some kind, but they weren’t substantial enough to compile a traditional credit score.
Individuals with low incomes, as well as African-Americans and Hispanic-Americans, were more likely to be credit invisible or have financial files too thin for most lenders, the study found. Immigrants and younger adults who haven’t been in the workforce as long can also have a harder time getting a loan or are charged much higher interest rates when they do. New credit scoring methods hold out the promise of reducing those disparities.
“It elevates credit scores for communities of color in particular who have not yet benefited from home ownership,” said Aracely Panameño, a housing expert who also directs Latino Affairs for the Center for Responsible Lending.
Federal law prohibits any financial practices that lead to a pattern of discrimination in lending against low-income or minority communities. The CFPB is charged with making sure the alternative credit scorers don’t end up replicating patterns of discrimination or creating new ones.
The companies building these alternative credit models crunch different data points to try to identify people who are more likely to pay back loans than a traditional credit score might indicate. This past fall, one of those lenders—a company called Upstart—was granted permission by the CFPB to experiment with gathering a wider range of data to evaluate potential borrowers. Under the terms of the agreement, Upstart must share its lending and borrower data with the CFPB and continue to follow non-discrimination laws.
Upstart incorporates “non-traditional variables,” like education, employer information, and online behavioral patterns to extend credit to more borrowers with poor credit as well as lower the cost of loans to them. The company claims its "algorithmic underwriting" — that is, using proprietary software models to figure out whether a person might be a good risk — can identify a creditworthy individual who has scores below 700 on the traditional FICO credit score scale, a level at which it typically becomes more difficult to qualify for a loan without paying higher interest rates or fees.
One of the central questions around tweaking the algorithm, said Paul Gu, co-founder and lead data guru for the company, is, “How do you know if your model is fair if you’re doing something that no one else is doing?”
Another approach is being tried by another startup called Petal relies on a relatively simple data point to determine whether it should accept customer applications for the credit card it launched: Do you make and save more money than you spend? Cash flow analysis, often used in small business lending, factors heavily into its assessments of borrowers. Applicants voluntarily provide bank records that allow the company to analyze their spending patterns.
Petal started in part because of the experience of one of the company’s co-founders, Berk Ustun. Ustun immigrated from Turkey to study at the University of California-Berkeley for his undergraduate degree, and couldn’t qualify for a credit card, cellphone or apartment because he lacked a credit history in the United States. He’s now doing post-doctoral studies at Harvard, in addition to his work advising Petal.
Many established financial institutions, like banks, have taken longer to embrace alternative-data approach to creditworthiness in part because they remain skittish after the 2008 financial crash: Loose underwriting in the early 2000s contributed to the crisis as mortgages made to borrowers with poor credit went sour.
Consumer advocates, like CRL’s Panameño, don’t see incorporating alternative data in underwriting as a “panacea” for minority borrowers, and worry that if used in certain ways, it could merely shift credit gaps rather than solve them. For example, using a borrower’s address or ZIP code or location-related data, like shopping patterns, could reinforce existing disparities.
“Alternative data is not created equal. Alternative data means different things to different people,” said Panameño. “It can result in disparate impact, potential racial discrimination, [and] red-lining—meaning that they might end up being charged more for certain products and services on the basis of where they live.”
Some have concerns about privacy, as well. Social media makes it easier than ever to collect personal data on consumers, a tempting well of information for lenders to tap. But sensitive information that lenders are legally supposed to avoid in assessing creditworthiness, like race, gender and sexual orientation, can often be gleaned from a Facebook profile.
There are also broader concerns about the societal impact of using different data points in addition to traditional credit history. China has launched a pilot “social credit” program that some see as a threat to privacy, and even a means of social control. The system, which the Chinese government plans to take nationwide in 2020, reportedly includes behavioral factors, as well as the credit of your friends and contacts, in determining your social credit score. The aim is to make it easier for the government to track individuals, allocate privileges and encourage behaviors the government deems useful. It’s seen as a way
for the Communist Party to enforce social and political norms and head off potential political unrest.
For now, debate around alternative data in the U.S. remains mostly focused on traditional concerns about fair lending and access to credit. And proponents argue the benefits could extend beyond the recipients to help boost the economy as a whole.
“Thoughtful and responsible use of financial data about individuals could expand the credit available to underserved consumers,” Cordray said last year. “If it is possible to expand opportunity in this manner, it would benefit not only these consumers, but perhaps would buoy the economy in ways that benefit us all.”
Colin Wilhelm covers Congress for POLITICO Pro Financial Services.
via The Agenda
February 11, 2018 at 03:53AMNo tags for this post.