First Access, Cignifi Use Big Data to Bring Economic Security to the Unbanked
In October, First Access chief executive Nicole Stubbs took the stage at the PopTech conference in Camden, Maine, opened her wallet, and proceeded to cast her ATM card and other financial records to the floor. "If you live in a developing economy, chances are that your wallet looks a lot more like this one, it just has some cash in it," said Stubbs, holding up the empty billfold. "How do you build real economic security in a cash economy?"
First Access, a startup based in New York City, and Cignifi, an upstart firm based in Cambridge, Mass., are working to address that by analyzing the cellphone use of loan applicants in emerging economies and delivering the information to lenders as a predictor of creditworthiness.
Though their strategies differ, both companies see their algorithms as a way for the world's roughly 2.5 billion adults who have no bank account to build credit histories. "Ninety percent of adults in most markets have a mobile phone, so the data that is produced is a tremendous proxy for the way most people live," says Jonathan Hakim, Cignifi's chief executive.
First Access, which has introduced its service in Tanzania, has developed a cloud-based platform that pairs a lender's appetite for risk with analysis of how often prospects replenish airtime, buy bundles of data and interact with their social network.
The service suits Tanzania, where nearly 40% of people live in poverty and more than half have no access to any kind of financial services, according to Financial Sector Deepening Trust, which aims to broaden financial inclusion in the country. Fewer than half of Tanzania's 3.1 million businesses keep records while nearly two-thirds keep their savings in a secret hiding place, according to a survey conducted three years ago by the country's trade ministry.
"In developing markets, there's no instantaneous credit-scoring tool, so lenders need to go to people's homes and places of business and try to evaluate their net worth and ability to repay," says Jessica Carta, First Access's chief operating officer. "That can be difficult for someone who owns very little."
With First Access's service, a loan officer can text an applicant's mobile number to the company, which in turn asks the applicant, also via text, whether they authorize the lender to tap their mobile transaction history and other demographic, geographic and financial information.
If the applicant gives the go-ahead, First Access marries the applicant's data and the lender's risk profile to generate a recommended loan amount, which First Access texts back to the lender.
The company charges lenders about $1.25 per transaction, a fee that covers the cost of the data, which First Access buys from carriers. "We can keep our prices low by serving a wide variety of financial institutions and generating a high volume of loan recommendations," Carta says.
First Access has produced about 75,000 recommendations since it began operating 18 months ago and counts the global microfinance network FINCA and other lenders as customers. The company saves lenders, on average, between $12 and $16 per evaluation, according to Carta, who cautions that the data are preliminary.
"So far, we've scored loans ranging from under $100 to about $36,000 for a small but fast-growing business," Carta says. "But our analytical models are capable of doing just about anything."
Cignifi looks to do business in countries marked by an emerging middle-income population and a large number of people who have phones compared with those who have bank accounts. The company currently operates in Mexico, Chile, and Brazil, which has roughly 123 mobile phone subscriptions for every 100 inhabitants, while only 56% of the population has a bank account, according to the World Bank.
Like First Access, Cignifi gleans information about loan applicants' lifestyles from the airtime and data they buy, the texts they send and the times they call. In Latin America, the company is using the information to help banks and insurance companies identify qualified leads.
"We know who's likely to respond best to a text message and who's likely to respond best to a call from a telemarketer, as well as which time of day is the best time of day to call, and we know something about their risk profile," says Hakim, who adds that Cignifi has helped some lenders halve their selling expenses. "The objective is that I can bring down the cost of acquiring customers and screen out people who are unlikely to pass the underwriting test."
Cignifi also is preparing to help a bank in Ghana score customers for unsecured loans that the lender would deposit directly into applicants' mobile wallets. "Early next year one of our banking partners will begin making unsecured loans using our telephone-based scores," says Hakim, who declined to name the institution, citing confidentiality concerns.
The services come amid a push by varied companies to analyze the multitude of data that flows from digital channels and human behavior. Lenddo, a startup based in the Philippines, uses social media sources like Twitter and Facebook to help score the creditworthiness of borrowers in countries where a credit market is emerging. Another startup, Revolution Credit, scores borrowers based on whether they complete classes in financial education. A third firm, Entrepreneurial Finance Lab, rates loan applicants according to their performance on tests of character, abilities and attitude.
"Economics 101 says that if there's information out there that can help lenders assess risk better, this will mean more access to credit," says Dean Karlan, an economics professor at Yale and founder of Innovations for Poverty Action, which uses randomized trials to analyze social programs.
Despite the emphasis on big data, solutions can be surprisingly simple. Informing borrowers about the existence of a credit reporting service or obtaining their fingerprints can improve repayment rates, studies show. Examining how loan applicants pay their electricity or water bills also can augment the credit histories of people with so-called thin files, according to research published last year by the Political and Economic Research Council, a nonpartisan policy institute based in Durham, N.C.
"You're basically looking at how you can support low-income people to get them into a situation where they can take advantage of microcredit, which is one way for people to escape poverty," says Margaret Miller, a senior economist at the World Bank. "That's why there's so much interest in moving toward data analysis techniques that would enable that market to mirror and take advantage of the same advantages that higher-income countries have been able to take advantage of."
That's not to say the move to track mobile usage comes without concerns. The services could cause people to buy airtime to boost their credit scores, to become over-indebted or to misconstrue how their personal information is being used, according to economists. "The main regulatory challenge will be how to put the right friction into the system," Karlan says.
The companies acknowledge the unease but say it's virtually impossible for someone to game their systems. "The beauty of mobile behavior is that it's not just a simple one-to-one correlation between how much money you spend and your creditworthiness," says Hakim, who notes that consumers have long known that paying their bills on time can boost their credit score.
As Hakim sees it, analyzing mobile use presents a low risk of consumers becoming over-indebted because the credit histories that borrowers build should enable lenders to assess their willingness and ability to pay. "I think it will solve itself by having more traditional credit bureau-type data that will be created as people start to use financial products," Hakim adds. Executives at First Access note that the borrowers whom the company scores remain subject to authenticity checks and other safeguards that lenders use.
Both companies say their services complement existing tools for assessing creditworthiness. First Access's predictions have an accuracy ratio of 62%, according to Carta, who notes that scoring systems in the developing world hover below 50%, compared with a ratio of roughly 80% in the U.S. and other developed economies.
"Our ability to discriminate on a credit basis between good payers and bad payers is similar to what you would find on a traditional payments-based credit history," says Hakim, who notes that Cignifi has data for about 100 million mobile phone subscribers.
The companies also say collaboration factors heavily in their business plans. "Banks are becoming increasingly focused on this whole field of consumer analytics whether telephone data or social network information and it's not their core competency, so acknowledging that they have to partner with people likes us matters," says Hakim.
Cignifi, which last year raised $1.5 million from a group led by the investment firm of eBay founder Pierre Omidyar, plans to tap institutional investors to fund a plan for growth that would bring the company's service to additional countries throughout Latin America, Africa and Asia. "Our addressable market is very, very big," Hakim says.
First Access intends to operate in at least three regions in Africa and two in Asia within the next three years, according to Carta, who says the company expects to raise capital from so-called impact investors after financing its operations to date with a grant from Financial Sector Deepening Trust and investments from the company's board.
Though Carta says First Access's service might make less sense in developed economies, she can envision financial firms in the U.S. and elsewhere using the company to augment their current ratings.
Cignifi has begun talks with banks and mobile operators in the U.S., where Hakim says that financial services firms and insurance companies are asking whether its service may allow them to own more of their relationships with customers. "It's quite interesting," Hakim says. "We haven't really even begun to realize the full potential of this."