Already a member? Current customers are kindly asked to reset their passwords. Simply select LOGIN, then RESET PASSWORD.

Better Data Is Key to Bank Alternatives to Payday

Walk down your average street in this country, and you'll find it easier to take out a loan than buy a coffee. With 22,000 payday lending locations in the U.S., Starbucks would have to grow three times in size to compete. Since the 1990s, annual loan volume has bloated to an estimated $27 billion. That's a lot of coffee.

Despite their growth, payday lenders are obviously controversial. Perceived as unfair and even predatory, payday lenders have been targeted by regulators, consumer advocates and lawmakers who object to their pricing, which leaves borrowers in a debt spiral. However, most payday lenders act legally. And the Consumer Financial Protection Bureau's small-dollar loan proposal may not be the solution.

So what alternative to further regulation will make small-dollar lending safer while ensuring consumers can still get loans? My suggestion: Fintech firms and incumbents should collaborate on using alternative data sources to qualify more borrowers for bank-issued small-dollar loans. This collaboration would provide fair short-term loans to individuals, and would force payday lenders to become more competitive in their pricing.

The average payday loan borrower is largely misunderstood. Assumptions about those who need a small-dollar loan do not always hold.

It is too simple to describe payday borrowers as foolish for the financial choices they make. In some cases, they opt for a payday loan because they can't get a bank loan, and need an alternative. They didn't qualify for a bank loan because they fall outside the standard credit definitions. Structural problems in how creditworthiness is determined can disadvantage people from building good credit. The payday lender is the last resort.

Data from the Pew Charitable Trusts shows that payday loan borrowers are not necessarily chancers. They're responsible people who just fall outside credit structures. They're likely to be divorced, for example. But that shouldn't be a barrier to building good credit. They're also likely to come from ethnic minorities – again, not a barrier.

And the borrower can't necessarily be blamed for the consequences of taking out a payday loan. Virtually no payday lender demands a credit history. That might look appealing for individual clients, but due to the steep repayment rates (annual APRs average 391%) clients are almost certain to get into cycles of defaulting and reborrowing.

Creditworthiness measures the extent to which a financial provider can trust clients to repay the money it lends out. This is the broad objective of credit-scoring methods such as FICO. Credit criteria include payments on previous or existing loans, level of outstanding debt, and whether borrowers have met other commitments such as bills, among other factors.

These credit quality factors exclude not just people in developed markets, but many billions of people in emerging markets face the same problem. The global unbanked (2 billion) in emerging markets, and those living in poverty in industrialized nations (12% of the entire U.S. population) are already excluded from accessing financial services, and risk falling into cycles of bad credit. With little or no credit history, they cannot advance finances to build good credit history. And so the cycle goes on.

But there are more groundbreaking ways to assess creditworthiness. Fintech companies that lend to both businesses and individuals increasingly use alternative data sources and machine learning to gauge the likelihood that a borrower will repay.

For unbanked people in emerging markets, machine learning facilitates accurate measures of trustworthiness based on alternative data sources. Lending platforms can analyze smartphone data to assess risk using algorithms which extract data, providing a holistic picture of a person's riskiness. For example, if users wait until the evening to make phone calls, when rates are cheaper, they are more likely to be considered lower-risk borrowers. The U.S. startup inVenture, also operating in Kenya, provides loans using this kind of risk-analysis model. Other measures in its social data risk model include social media use and online behavior to score borrowers.

Social data used as part of the assessment include: Do applicants have reliable contacts, do they abide by acceptable measures and standards of social behavior, or are they erratic? Actually, machine learning can relay whether a potential borrower uses gambling sites; such users who do not delay paying debts and do not gamble excessively are more likely to be considered creditworthy. All of these measures can go into creating a composite picture of trustworthiness that enables fintech to lend to a greater range of businesses and people.

That said, regulation of any new credit-scoring model is still necessary, and that includes data security measures. Privacy is a concern, which is why proper regulation of the industry is required to ensure data management does not become problematic. Also, in terms of making sure the social data actually amounts to usable information, humans must always be involved to execute the final credit decision. Advanced analytics and machine learning build a picture of credit history, but often it requires human analytical skills to capture key learnings and insights from alternative data sources.

Banks collaborating with fintech in credit-scoring underserved customers has other advantages. Fintech lenders have leaner infrastructures than banks, and so spend much less on credit scoring, billing and overall compliance than traditional players do.

In the long run, competition will become collaboration, resulting in a more competitive and fairer alternative to payday loans.

Toby Triebel is the chief executive officer and co-founder of Spotcap, an online lender for small and medium-sized businesses. He can be contacted on Twitter @tjtriebel.

For reprint and licensing requests for this article, click here.