Skip to content
Link copied to clipboard

Fintech lenders are tracking and judging your digital trail

Your mobile phone, online shopping, social media, and search engine visits leave a trail that corporate data scientists can use to judge you.

Gurvinder Ahluwalia, a former IBM executive who founded and runs Digital Twin Labs, LLC, lectures to Philadelphia Federal Reserve officials and guests about fintech decision making and blockchain at a conference in 2017
Gurvinder Ahluwalia, a former IBM executive who founded and runs Digital Twin Labs, LLC, lectures to Philadelphia Federal Reserve officials and guests about fintech decision making and blockchain at a conference in 2017Read moreFederal Reserve Bank of Philadelphia

Your mobile phone, online shopping, social media, and search engine visits leave a trail that corporate data scientists can use to judge you.

That's the premise behind "fintech" lending, a new industry that uses technology to make credit and customer decisions and set loan prices by scraping data from your digital profile, so firms can see quickly if you look like a good risk for a home mortgage, credit card, personal loan, or retail purchase.

Fintech start-ups have attracted tens of billions in loan financing and business investments the last few years from private-equity investors, who expect that nimble digital companies can win borrowers away from traditional banks.

The industry has also won academic validation from a study by a team of U.S. and German finance professors, who reviewed records for 270,000 purchases and loans at an online furniture seller in Europe.

The study, "On the Rise of FinTechs: Credit Scoring Using Digital Footprints," written by a team led by Tobias Berg of the Frankfurt School of Finance and Management and published by the National Bureau of Economic Research, found that a dozen broad personal tech clues, used together, did a better job predicting who would pay for what they bought over time than traditional FICO credit scores and other common credit measures banks use to grant or refuse loans.

They found, for example:

  1. iPhone users are more likely to pay their bills than Android users.

  2. T-Mobile customers are less of a credit risk than Gmail users, who are less than Yahoo! email users.

  3. Shoppers who click paid online ads before ordering are twice as likely not to pay for what they buy as those who spent time on price-comparison sites first.

  4. People who use shopping sites from desktop computers in the morning are more likely to pay for products they order than people who shop from mobile phones after midnight.

  5. Folks who use their true names in their email addresses are better bets than those who jumble common nouns and numbers.

"The simple act of accessing or registering on a web page leaves valuable information" that businesses can grab to check you out, Berg and his colleagues wrote. "Every website can effortlessly track whether a customer is using an iOS or an Android device; or track whether a customer comes to the website via a search engine or a click on a paid ad." That information, combined with loan and default-pattern data, "can be used for the prediction of consumer payments and defaults," they added. That's especially important to online retailers, "because goods are shipped first and paid later."

But is this really cutting-edge data science or just a new way to identify who's rich and who's poor, a digital version of the "redlining" neighborhood-avoidance policies made illegal by U.S. fair-lending laws?

Gabriel Weinberg, as founder of Paoli-based trackless search website DuckDuckGo, knows the power and promise of online data accumulation. But he warns that government agencies as well as consumer lenders can use online data to target citizens in ways that sometimes amount to unfair discrimination. "The algorithms are often not complicated. But they can still be inherently biased. And every time there are unintended consequences," Weinberg told me last spring. The more we know about the algorithms companies and governments use to check on members of the public, the easier it is for members of the public to spot problems.

It's easy to see how digitized personal data attracts lenders. For most of banking history, "a personal loan was a super-manual process. It involved coming into a bank branch, talking to a loan officer, and relying on that loan officer to make a decision. Well, we've eliminated the uncomfortable part of the process. Now it's nonjudgmental, and we know if you qualify in five seconds," said Paul O'Donnell, chief credit officer at Marlette Funding, a Wilmington fintech firm that speeds personal loans through its Best Egg rapid-lending affiliate.

Marlette, which employs more than 120 at its Wilmington headquarters and other offices, has raised more than $1.7 billion from investors recruited by Goldman Sachs and other Wall Street banks in the last two years.

Fintech lenders say they adapt quickly to changing markets and signals. At traditional banks, updates in underwriting or approval signals might take place "once in a quarter — get everyone to agree to it, test it, roll it out. We can do it in as little as 24 hours," said Brian Conneen, chief information officer at Marlette. "We are constantly collecting more data and doing A-B testing, champion-challenger testing, and capturing the value."

In addition to information collected from public websites and traditional credit tools like FICO scores and credit-agency ratings, many applicants agree to open their personal asset and spending records to lenders — sometimes using and other personal-finance software — when they are told that will speed loan approvals, O'Donnell said.

The lender can see quickly, for example, if an applicant is still working and has direct-deposited a paycheck. That eliminates the time and expense of calling human resources to verify employment or having to pull copies of canceled checks.

"We don't make it mandatory, but we do say it makes the process faster, to pull this one snapshot of the customer's data," O'Donnell added. "We were surprised with the customer opt-in rate. It's more than half."

Approvals are fully automated: After 20 minutes of applying online, when all goes well, "they are a customer and we can get the loan deposited into their checking account," O'Donnell said. The company still checks data against credit-bureau records and human efforts go into spotting fraud trends, testing new credit markers, and data analysis.

Marlette markets loans by direct mail, places online ads triggered by Google searchers looking for "personal loans," and pays borrower research sites like Credit Karma for referrals. To keep track of all those personal credit inputs, "we have a flexible, cloud-based architecture that allows us to store thousands of attributes per customer," Conneen added.

The loans aren't cheap. Fintech lenders don't so much target poor people as they do college-educated people in a hurry. And young people in a hurry aren't always stopping to count costs: "When we launched, we had a feeling that price was going to be the most important thing. But as we've gotten to know customers, convenience and speed are more important. We were a little surprised," Conneen said.

Delaware's evolving fintech node grows from its role as the nation's credit-card lending center. Other start-ups include Fair Square Financial LLC, the two-year-old, 50-worker Wilmington company that markets the Ollo credit card. Fair Square has raised $300 million since 2015 from investors including Orogen Group, the New York investment firm headed by former Citigroup CEO Vikram Pandit; Atairos Group, headed by former Comcast chief financial officer Michael Angelakis; and Swift Financial, a First Round Capital- and Sutter Hill Ventures-backed fintech small-business lender founded by Delaware banking-tech guru Ed Harycki and purchased last year by Silicon Valley electronic-payments leader PayPal. There's also College Ave., the private student lender founded by Sallie Mae veteran Joe DePaulo and backed by longtime Sallie Mae boss Al Lord and Comcast Ventures.

America has a history of innovative finance leading to over-rapid credit growth followed by costly losses. O'Donnell said his company's lending "has been performing in line with expectations." He said the economy is due for a slowdown in the next few years, which will test fintech lenders' underwriting.

O'Donnell, a veteran of business software giant Oracle, said big companies find it cheaper to update their systems by buying small tech companies than building their own. "The inertia of a large company makes it hard to innovate," he told me. "Apple is now acquiring machine-learning companies left and right." That's an inspiration to fintech start-ups and their backers, who hope to get bought at a premium to what they invested.

(Added 7/9/18:) Not everyone's ready to ditch old-fashioned ways of measuring credit. "FICO remains the 800-pound gorilla," says Allan Stevens, chief credit officer at $1.1 billion asset, Chadds Ford-based Franklin Mint Federal Credit Union. FICO scores "are validated, and re-validated, on consumer populations numbering in the hundreds of thousands. They do not take into account gender, ethnicity, age," or other groups protected by anti-discrimination banking rules, he added.

Fintech lenders "are making a large impression on the consumer market," especially on national consumer-finance lenders, Stevens said. But in the few years they've been active, "it is hard to say whether the digital footprint" factors "are valid, and-or sustainable."

And how soon until Americans get to work figuring out how to game fintech by masking our less-attractive digital traits? It's expensive to buy higher-class smartphones and paid internet and email services so you look solvent, the Berg study notes. But the authors worry digital profiling could inhibit personal behavior: The more people realize the tech we use can get us branded and affect our opportunities in life, the more "consumers [will] fear to express their individual personality online."