Maximizing Big Data to Minimize Risk in Small-Dollar Lending
A number of lenders are betting that creating a closer tie between employers and loans to their employees will lead to lower default rates and cheaper consumer credit.July 3
Part three of four.
There are already a lot of factors that lenders consider when they're deciding whether to extend credit, but ZestFinance founder Douglas Merrill says he's identified a new one.
Namely, does the applicant type his or her loan application in all caps, in all lowercase letters, or a combination of the two? In other words, does the prospective borrower use the shift key correctly?
Los Angeles-based ZestFinance has concluded that people who type in all caps are more likely to default on their loans. "All caps is bad," says Merrill, a former Google chief information officer. "All lowercase is slightly better. Correct case is much better."
It may seem trivial, but this uppercase vs. lowercase variable is one of tens of thousands of data points that ZestFinance uses to make small-dollar consumer loan decisions. Individually, none of the factors matters very much, but taken together they're more meaningful, according to Merrill.
"When you take thousands of small signals, you get a really big signal," he says. "You're resilient to something not being correct."
ZestFinance is one of several startups seeking to use "big data" to lower the risk of lending a few hundred bucks to borrowers who don't qualify for mainstream credit.
If firms like ZestFinance are successful, such households will wind up with more access to credit, and lenders will be able to build their portfolios by reaching deeper down the credit spectrum.
Consumer advocates and some of the high-tech lenders have another hope, which is that lower default rates will lead to cheaper loans for consumers who currently rely on payday lenders, auto title lenders and other expensive sources of credit. So far, though, it's not clear that savings are being passed on to consumers.
The innovations being born at consumer finance startups are part of a larger, heavily hyped big data movement, which seeks to upend long-standing business models in industries from retailing to real estate.
Using big data to underwrite loans provides new tools for evaluating risk, but it also presents lenders with a new set of challenges. Because the loans are made online, the data-driven lenders are encountering the risk that applicants are not who they claim to be. This is less of an issue in traditional payday storefronts.
"In a face-to-face transaction, you're able to see: Are they acting nervous? Are they doing anything unusual?" says Ken Rees, president and chief executive of ThinkFinance, an online lender that is leaning heavily on data analytics. "Most online lenders charge more than brick-and-mortar [lenders] do, and largely because of the challenges of the identity and the underwriting."
Still, Rees argues that big data is helpful in lowering the chances of fraud and default. One piece of data that his company considers is whether the prospective customer goes to the online loan application immediately, rather than first checking out the loan's terms. Customers who take the former route may not be so interested in repaying the loan, he says.
"The customer that looks at the loan price and terms that's more like what the normal customer might do," he says.
ThinkFinance is one of the biggest companies among the new data-driven online lenders, and the company says that it's dedicated to lowering the cost of borrowing among Americans who don't have access to mainstream credit. Its defaults have fallen by 50% in the last few years, according to Rees, who adds that the savings have been used to lower costs to borrowers.
Another challenge the high-tech lenders face is striking a balance between acquiring enough information to make a sound decision and asking the borrower to reveal so much that they decide the loan application is not worth the effort.
Consumers seeking to borrow a few hundred dollars online are expecting the same easy application process and instantaneous response they get when they make a purchase through Amazon, says Ryan Gilbert, chief executive of BillFloat, another data-driven lender.
"There is a very limited attention span that most consumers are applying to these transactions," he says.
San Francisco-based BillFloat acquires data directly from the loan applicant and also from third-party companies. It's looking to answer several questions about the applicant, including: Is the person who he says he is? How consistent are his deposits? What are his current financial obligations? Does he need the cash, or simply want it?
The trend that everyone in the industry is currently talking about, according to Gilbert, is the use of social media data. "I think we're seeing an increased use of that data at various stages of the customer relationship," he says.
Lenders are employing social media data to evaluate an applicant's current life situation, says Eric King, chief executive officer of Sociogramics, a Palo Alto-based data provider. By comparison, credit scores tend to be more backward-looking, he notes.
"It's a matter of finding out who has the right character and who has the right life stability," King says.
But King also acknowledges that the value of social media data has not yet proven its worth in relation to the hype it's received. "It's still early days," he says.
Billfloat's Gilbert says that while social media is proving useful as an aid in collecting unpaid debts, it's unclear whether it has real value as an underwriting tool.
"It's unproven yet what a Facebook profile really says about a consumer in a credit environment," he says.
Lenders are primarily using big data to evaluate a borrower's creditworthiness, but the ultimate aim for some is to reduce the cost of small-dollar loans.
LendUp, a San Francisco-based startup, promises, over time, to graduate borrowers into cheaper loans with the help of big data. "LendUp loans are designed to help you build credit, not keep you trapped in debt," the company's website states.
A LendUp customer's first $200 loan costs $35.20, assuming the borrower repays in 30 days. That translates to a 214% APR. The company promises APRs as low as 29% for customers who establish a proven track record.
At this point, though, the cost of such small-dollar loans remain stubbornly high for consumers.
ThinkFinance partners with a Native American tribe, which allows the company to avoid state interest rate caps on its Plain Green Loans product.
A Plain Green Loans customer who borrows $250 will owe $440 by the time the term ends 16 weeks later, an APR of 379%. Someone who borrows $2,000 will owe $5,206 by the end of the 76-week loan term; that's an APR of 160%.
ZestFinance also uses a partnership with a tribe in its case it's the Turtle Mountain Band of Chippewa Indians of North Dakota to get around interest rate caps in many states.
Back in 2010, when Merrill launched ZestFinance , he was highly critical of payday lenders. "The payday loan business is fundamentally abusive," he told VentureBeat. "You pay $60, and you have $300 for 14 days. If you annualize that, it's a lot of money."
The annual percentage rate on the low-tech payday loan that Merrill described is 520%. But that rate is not so much higher than the 390% APR on ZestFinance's product, called Spotloan.
Merrill is unapologetic, saying that the loan product should be evaluated on the basis of its value to consumers.
Spotloans range from $300 to $800, with terms of three to eight months. The loans have won praise for the transparency of their price disclosures, but Spotloans are expensive in comparison to mainstream forms of credit. A $500 loan would cost $1,058 to pay off over five months.
Next: Why small-dollar consumer loans remain so expensive.