How Behavioral Data Can Help Lenders Assess Consumers’ Credit
(Originally published on Equities.com)
Traditionally speaking, credit reports define consumers. Good credit signals trustworthy borrowers; bad or little credit is defined as too high-risk. But these reports don’t tell the whole story.
A once-reliable borrower might become seriously ill, incurring massive medical debt that forces him to default on a loan. He might get divorced, putting a strain on his finances and causing him to fall behind on credit payments. A traditional credit report doesn’t take these external factors into account. There are also millions of “credit invisibles”: people who don’t have enough credit history to secure a line of credit. It’s a catch-22: If you can’t get credit, you can’t build credit.
And that’s the root of the problem: Credit history provides useful information about borrowers, but it isn’t an accurate indicator of creditworthiness on its own.
The Fintech Revolution
To halt this cycle, more financial technology companies are using behavioral data to supplement credit reports and compile more informed pictures of borrowers. Behavioral data pulls information from other payments and transactions to find patterns in borrowers’ financial habits, providing powerful indicators of payment and default. Alternative data — like public records, social media accounts, or banking profiles — can even predict when external shocks may occur.
The traditional lending space’s overreliance on credit scores leaves few options for addressing “thin files” (borrowers who have too little credit activity to merit a score). My firm actively markets to people who use credit, but 15 percent of our applicants don’t have credit scores.
This isn’t a philosophical problem; it’s a real-world challenge that needs to be addressed. Credit cards drive the U.S. consumer economy: In 2014, Visa (V) credit cards alone generated $1.2 trillion in purchase volume. Basing lending decisions solely on credit scores excludes these thin files from home and car ownership, and it prevents financial companies from accurately assessing borrowers who do have credit histories.
That’s why new fintech companies are looking to bring more consumers into the credit fold, basing loan and credit card approval on several factors instead of just one. For example, companies like Upstart look at education and job history to determine creditworthiness. From bank account records to “credit school,” these organizations are creating a more dynamic and multifaceted lending industry.
Banks that don’t use alternative and behavioral data are throwing business away. My company did a retroactive analysis of our portfolio once we’d incorporated behavioral and alternative data into our underwriting algorithms. The findings were eye-opening: We had rejected thousands of customers who likely weren’t as risky as their credit scores suggested. Each rejection was a lost opportunity.
By incorporating behavioral and alternative data into our approval and pricing algorithms, we’ve significantly cut our projected loss rate. We can better identify customers who are riskier than their scores suggest and pick out those who are safe bets to pay us back — even if we increase their approval amounts.Executives at traditional firms will remain reluctant to use this data until they realize that without it,they’re lending too much to certain individuals or not charging them rates commensurate with the risk they truly represent. Their competitors will eventually exploit this and become more aggressive with their margins.
Changing an underwriting system is a lengthy process. Start by taking stock of your current system, and look for options that work for your organization. Once you’re ready to transition, these steps will help you incorporate new data sets:
1. Reconfigure your infrastructure.
The easiest way to incorporate behavioral data is by using an alternative scoring product such as TransUnion’s (TRU) L2C or Experian’s Extended View. Companies like Mint (INTU) and Yodlee pull information directly from users’ bank accounts, providing excellent insights into borrowers’ habits and reliability.
However, you need the infrastructure to run new data and analytics systems, which presents a challenge for legacy organizations. Wells Fargo (WFC) and Citibank (C) are more reluctant to overhaul their traditional underwriting systems, which is a massive undertaking and defies their experience. Innovation is occurring in the lending space as newer Silicon Valley fintech companies take a broader approach to assessing creditworthiness.
2. Develop a multivariate approach.
Alternative scoring data helps you identify outliers. Bank transactions, utilities payments, retail activity, and social media tell borrowers’ stories beyond their credit reports. For example, someone with a great credit score might have risky behavioral patterns. A person with a low credit score may pay his bills on time and save consistently — and be a safer borrower despite previous credit issues.
Before implementing new lending standards, study the relationship between different data points on your current portfolio. Understand what drives your previous customers’ behavior so you can roll that information into the new underwriting system and build a comprehensive evaluation process.
3. Follow up with continual monitoring.
Once you’ve made a credit decision, behavioral data becomes progressively more important, so you need a solid technology stack in place. Analyze every individual’s payment cadence and methods. Identify common denominators in default or prepayment using algorithms like k-means clustering to spot patterns and modal credit and payment personalities.
These strategies will enable you to make informed decisions when customers ask for credit line increases or new loans. You can tie ongoing data points to the original factors that influenced the loan decision to predict future behavior.
Technology empowers lenders to make smarter decisions and compile comprehensive data sets on their customers. By accounting for behavioral data over time, you can identify trends that help you adjust your terms and offerings according to consumer repayment patterns. Credit scores alone can’t help you do that. Alternative data will.