In 2014, the New York Times covered a small event which spoke volumes about a large problem – a man in Washington D.C. was unable to refinance his home mortgage. This man had just changed jobs, from a stable government job of 11 years to a sporadically paid speaking and writing gig. This change flagged his credit score and made it appear that he was an increased credit risk. The loan was denied.
A man who had just signed a $1 million+ book deal, makes over $200,000 per speech, and just led the world through the largest financial crisis since the Great Depression was not able to refinance his $839,000 house. The man was Ben Bernanke, former Chairman of the United States Federal Reserve Bank.
Seems a bit absurd, right? Several fintech companies think so and they have started building new credit scoring models using a wide variety of personal data. Before we get to the future of credit scoring, let’s take a look at the current state of the legacy scoring industry.
Since the economic crash in 2008, lending and credit had significantly tightened. Banks have always needed to lend money to make money, but new capital and liquidity ratios forced banks to hold money and minimize lending. Relationship based lending suffered but the use of FICO scores (as the core metric of debt repayment) became central. FICO scores have long been the key determining factor in making lending decisions; this is how they are calculated:
- Payment history (35%)
- Amounts owed (30%)
- Length of credit history (15%)
- Credit mix in use (10%)
- New credit (10%)
As you can see from the list, FICO scoring is focused on few specific factors. In the case of Dr. Bernanke, these factors were a little too specific and restrictive. If the person running the largest bank in the history of the world can’t get a loan, how can you? Has the traditional credit scoring system failed us?
FICO factors have some merit in predicting credit worthiness and we don’t think they’ll disappear anytime soon. However, these factors certainly do not give lenders a very wide snapshot of a borrower’s likelihood to repay. Let’s take a look at the several “new” factors being tested in the marketplace and where those ideas came from.
Before the “new” app-centric Silicon Valley credit scoring metrics were all the rage, Ireland did some fancy innovating of its own. In the 1970s, Ireland experienced a banking crisis as bank worker strikes closed the sector. How did the people respond? Mass chaos and riots? Fighting in bread lines for table scraps? Not at all. Bartenders became character based lenders and pubs became the “banks.”
As the Harvard Business Review notes, “the Irish economy was characterized by intense, frequent, conversational personal contact: tight, dense, solid local knowledge circulating at high velocity within and across communities. Result? Borrowers and lenders could build solid microfoundations of trust. In other words, when you’ve been chatting with Bill every night at the local pub for twenty years, you probably know whether his note is a good bet or not (and further, just how much to discount it to earn a sustainable and fair return, that neither fleeces Bill, nor robs you). Furthermore, if you’re the publican, and you’ve been chatting with me and with Bill, then you’re even better positioned to become a de facto arbitrator of notes — a bank. And that’s exactly the role that pubs began to play.”
The Irish essentially created a character based peer-to-peer lending platform and locally circulated currency notes…all before Bitcoin. They analyzed the traits, actions, and history of each community member at a micro level to build a picture of an individual’s credit worthiness. At a small scale, this system worked well. However, timely clearance and settlement proved to be an issue at scale. Not only that, who wants only one person to be the arbiter of access to both beer and credit?
The Fintech Lending Revolution
Fast forward to the post 2008 world. As the world emerged from the global financial crisis, fintech firms realized that outdated credit scoring models needed to change. Firms began experimenting with different underwriting schemes based on consumer data, social media actions, and personal data analysis.
Experian and TransUnion incorporated rent payment tracking onto credit reports. Social networking companies (i.e. Facebook, Lendup, and Wonga) are verifying identity through social media actions and are also looking at connections to try and determine how risky an individual is based on their friend network. Louis Beryl of Earnest considers these factors, “Contributing to your 401(k), regularly investing in a safety net fund, and spending less than you earn are key indicators of financial responsibility. Your educational career, your past income growth, and your projected earning trajectory are important predictors of stability.”
Feeling like you are being watched at every move? You are. Companies are tracking cell phone and utility bill payments, behavioral analytics such as mouse scrolling movement, and your e-commerce shopping data. Jonathan Hakim of Cignifi breaks it down even further, "The way you use your phone is a proxy for your lifestyle. It's not random. So we're looking at things like the length of calls, the time of day, and the location you make them from. Also things like whether you top up [a pre-paid SIM card] regularly. We want to see how stable the patterns are. When you look at that, you can create these behavioral clusters that give you information about users' appetite for new [financial] products, and their ability to repay a debt ."
In Asia, Africa, and South America, psychometric lending practices are being tested. Similar to the Irish parable above, psychometric lending is an underwriting schema that uses borrower’s personal characteristics, like honesty, ethics, drive, motivation, optimism, intelligence and business skills to determine credit worthiness.
The operationalization of these new metrics has created a marketplace which has delivered billions of dollars of new credit to consumers, businesses, and students in a time where other credit options are restricted. However, the deep dive into such private and personal data has raised significant concerns. While some borrowers are quite willing to trade privacy for credit access, others fear what some people or companies may do with the data. Regulators and consumer protection agencies have no idea how to monitor these new factors as no standardization exists yet.
Up until the last few weeks, the fintech lending future looked pretty rosy. Most companies were on solid growth trajectories. “We're seeing an unprecedented interest in banks wanting to work with us, embrace us, partner with us, invest in us, and buy loans from us,” said Ron Suber, President at Prosper Marketplace. That all changed when it was recently discovered that Lending Club made $22 million worth of loans that did not meet investor criteria. The CEO was fired, the shares tumbled, and regulators are now swarming around the entire sector, stoking fears that sector funding will be pulled back. Fallout from this event will most likely instigate other probes and investigations which surely net a few other bad actors.
Will this event take down the entire industry? Probably not, but it is too early to tell. Many large VC firms are deeply invested in the sector and the credit scoring industry still demands innovation. The most likely outcome is that innovators will face more regulatory scrutiny and will be forced to tighten internal quality controls. In turn, this should benefit everyone. As regulators gain comfort with fintech processes, traditional banks will regain comfort as well, leading to streamlined integration. Stay tuned for more fireworks over the next few weeks.
Header image from Cafe Credit.