Banks are mining data from smartphones to assess lending risks to customers


May 12, 2014
MUMBAI—Indian banks have started mining data on customers’ smartphones for fast loan approval, testing out cutting-edge but controversial technology in what is potentially a huge market for such products.

Long hampered from lending to the hundreds of millions of Indians without credit histories, banks are hoping to slash risk-assessment costs and trigger a new wave of consumer lending with apps that look at everything from Facebook connections to online shopping habits to rate potential borrowers.

India’s most sophisticated banks are working with local and international fintech startups to develop, test and launch a version of a technology used by microlenders in Africa, China and elsewhere.

“This is not an exact science, but actually it is a much better science than looking at static data,” on long loan application forms, said Rajeev Ahuja, executive director at RBL Bank Ltd., which is working with partners to use social media to score customers without credit histories.

America’s largest credit-ratings firm, FICO, last month began offering credit scores in India for the first time using these methods. FICO used so-called alternative data sources to “evaluate millions more consumers who were previously unscorable,” said Sally Taylor-Shoff, a FICO vice president.

In the U.S., credit agencies have been cautious in digging deep into customers’ phones due to regulatory hurdles and consumer concerns. There is also less of a need because of a robust nationwide database of information on borrowers. In India, banks say they are cautiously moving ahead as they await regulatory guidance on privacy and develop systems to get permission from applicants for access to their phones.

Mainstream Indian banks using the methods on a large scale could bring the technology to millions of people who were too expensive to lend to before. Until now, primarily financial startups and microlenders have dabbled in the phone probing methods.

“If we manage to do it right as a country, potential for application is the largest,” in the world, said Rajesh Kumar, co-head of retail risk at HDFC Bank Ltd. , India’s second-largest private-sector lender which could start using the alternative data as early as next year.

Commercial banks had around $1.09 trillion of loans outstanding in September, according to the Reserve Bank of India. Of that, about $270 billion were personal loans, a portion whose growth is outpacing the overall loan market.

But about 40% of HDFC’s 8 million and 12 million loan applications a month are from people without credit histories. Most Indians have never had a credit card or taken a home loan.

“People are invisible to the bank,” because it is too expensive to nail down how much debt they can handle, said Monish Anand, chief executive of fintech startup Datasigns Technologies.

To work with such consumers, Indian lenders have traditionally visited borrowers’ homes to see where they live, talk to their families and neighbors and inspect documents. This process is costly for small loans.

Working together with fintech companies, many private Indian banks, including Kotak Mahindra Bank and HDFC, are exploring how the phone-scanning fintech apps can reduce risk assessment costs.

Bangalore-based payments startup Simpl is sharing a risk score it developed from looking at its users’ online shopping habits with its financial partners for clues on their credit risk. Datasigns Technologies, another Bangalore startup, is working with IDFC Bank and triangulates family and friend connections on Facebook to ensure loan applicants are telling the truth.

Singapore-based Lenddo, which is working with FICO as well as banks in India, looks at the number of contacts a person has in their phone book and how often they interact with them, among 12,000 other data points. Regular interaction with the same group of people is a sign of a good borrower.

In trials, HDFC is using the phone location data of their agents to confirm the addresses of potential customers whose exact locations are hard to place with a ZIP Code, such as farmers. Kotak Bank is testing technology that lets it scan applicants’ SMS messages, to get an idea of spending habits, since bank transactions are recorded there. It also can see their mobile connection speed and type of phone, another spending gauge.

ICICI Bank, India’s largest private-sector bank, said this month that it has joined forces with Paytm, an Indian e-wallet company, to give loans based on their payment history on the Paytm app.

“On the fly we will be able to do a credit rating of them and we’ll be able to give them a limit,” for borrowing, said Anup Bagchi, executive director of the bank.

There are global implications. India’s large scale provides a testing ground to hone algorithms. Once FICO has perfected this type of rating in India, for example, it says it could roll it out in other emerging markets.

Figures on the number of loans given this way aren’t yet available. But optimists think the eventual potential is in the hundreds of millions of new borrowers and billions of dollars.

“India, by sheer numbers, is a fantastic market,” said Abhinav Haldia, India director at Lenddo.

With corporate borrowing stuck in the doldrums, Indian banks are hoping to get more growth out of retail lending with these apps. Retail loans at commercial banks rose 15% so far this year while overall lending in India is up only 5%. Banks know there is even more consumer business out there and the new phone-lending apps will help them serve that untapped market.

“It’s not that the banks didn’t want to lend, it’s just that they didn’t have too much information” on borrowers, said Vivek Belgavi, a partner at PwC in Mumbai.