The world's 1.7 billion people without access to banks do not have credit scores, but a Santa Monica, California-based start-up has developed a way to lend money to the unbanked based on smartphone data.
Tala has raised nearly $220 million from private investors to offer loans of up to $500 to people without credit histories or bank accounts. Its app uses a scoring model based on 250 data points, from cellphone bill payments to online behaviors.
The digital microlending company has grown to 600 employees and a valuation of nearly $1 billion. It operates in Kenya, Mexico, Philippines and India.
"Having proven the model works, the challenge now is to continue to scale the business and expand into new markets," said Jeff Richards, managing partner at GGV Capital, which is one of the investors backing Tala.
A world without credit scores
Only 31% of the adult population worldwide is covered by a credit bureau, making them the only ones who can access credit in a traditional banking environment. The World Bank estimates that about 1.7 billion people are unbanked, but two-thirds of them now have a mobile device.
Enter the fintech disruptors who have found a way to monetize the regulatory hurdles that have kept traditional banks from lending to the poor — or people who have had to mostly rely on nontraditional sources including relatives, family, loan sharks, pawnshops and in a few select countries, microfinance institutions.
"The idea of lending purely based on the data available on a consumer's mobile device was completely unproven — no one had ever done it," Richards said.
But it's working now.
Richards first met Tala founder and CEO Shivani Siroya in 2015, almost a year after Tala had launched its first mobile app prototype in Kenya.
"Emerging markets are traditionally looked at as risky, and thus not holistically served by traditional banks, and that's the opportunity we're chasing," says Siroya.
Kenya already had a well-established mobile financing culture, thanks to M-Pesa, that country's version of Apple pay.
"The effort-to-opportunity ratio was low," said Siroya, who has a master's in microfinance and has worked as an investment banking analyst at both UBS and Citi. Her work at these firms exposed her to the credit vacuum left by big banks around the world.
"From a competitive differentiation standpoint, Shivani was one of the first to build a 'mobile credit score' – without traditional data like income, banking history or a credit score," said Richard, who invested in Tala in 2015.
How the app works
Tala's scoring model relies on about 250 data points. "We determine the weights of individual data points using state-of-the-art machine learning techniques, trained on historic user data," the company says.
The app from Tala analyzes two categories of data: Android device data, in a nod to Google's Android global dominance, and behavioral data.
Tala's app asks for customer inputs including name, date of birth, gender, text messages showing utility or phone bill payments, reason for the loan and the amount sought.
The first thing the app determines is fraud. Are you who you say you are? The app matches customers' input with records of their phone bills or utilities bills, accessed via text messages on the customer's phone. When a customer passes the fraud test, the model runs a credit test to assess the user's capacity and likelihood to repay.
Siroya said the credit model does not use specific identifying information such as age or gender in credit decisions to prevent algorithmic biases. The credit assessment stage utilizes a mix of device and behavioral data.
Device data such as type and year of the operating system help assess borrowing capacity. Mobile bill payment history and certain behavioral data all help assess the likelihood of repayment. Behavioral data includes how customers interacted with the app, time spent while moving through the pages to see if they actually read the terms and conditions, any mistakes while typing basic biographical information, and other apps on their phone.
Tala insists it continually assesses the user data and does not share any personally identifiable information with third parties. Siroya says all customer data is deleted after repayment.
Terms of the loans
On average, the repayment period for a Tala customer is 30 days. The most it can be is about 90 days, with 20 days being the shortest. Users have the flexibility to repay early with no penalty. In fact, unlike traditional financial institutions they even offer incentives for early repayment.
Once a customer has taken a loan, repayment behavior is the most important factor for future lending decisions, according to the company.
First-time users are eligible for loans of $20 to $100, while repeat borrowers may be eligible for up to $500. Siroya said about 90% of eligible customers come back for repeat loans.
So who is a Tala customer?
Aimee Dingle, an out-of-work cook in Manila, the Philippines, was denied credit from traditional banks because she didn't have a credit history.
"The system is not set up to meet the needs of the millions like her," Siroya said.
After getting approved for her $20 loan through Tala, Dingle bought the ingredients to make tocino, a Philippine-style bacon and local breakfast delicacy.
She sold them for the profit she needed, or $4, in one week. Realizing the potential, Dingle set out to grow her loan limit to grow her business.
"From a starting loan of 1,000 pesos, I've now achieved gold status with Tala," which is reserved for Tala's best borrowers, Dingle said. In two years she's raised her loan limit to over $200.
How the microlender is financed
Tala recently secured $110 million in Series D funding along with a $100 million debt round, a typical funding round for a digital lender. To date, Tala has raised $219.4 million in funding from investors like GGV, Institutional Venture Partners and PayPal Ventures. The money raised via these debt and equity offerings is lent to consumers in countries where it operates.
"Tala empowers the unbanked through innovative mobile solutions to improve their overall financial health, and we are proud to help them achieve this goal," said Mark Britto, executive vice president of global sales and credit at PayPal.
Tala earns a flat fee of 5% to 15% on every loan it makes. All first-time loans sit on Tala's own books, while loans made to customers in more mature markets like Kenya are off-balance-sheet, asset-backed loans sold to large large institutional investors such as Colchis Capital."
Dealing with defaults
Smartphone data helps the company assess early on whether customers will default on their loans. One good indicator is if the borrower deletes the app. App referrals are another gauge of the intent to repay.
Most customers repay early, Siroya said, adding that about 15% to 20% of loans made may go past due. They are able to recover about half those loans that are overdue through a mix of "educational" messaging and informational calls.
In some cases, Tala uses outside vendors "but they're all monitored by our team in-house and have to adhere to our strict policies," Siroya said.
When it enters a new market, most loans it makes are "blind loans," says Siroya, because it has no data on the market and no parallels to draw from, so no models to truly assess a customer's ability and intent to repay.
It's only after it hits a critical lending mass that it has some measurable statistics to make more guided lending decisions. A mix of digital ad buys on Facebook, Twitter and Google adwords helps drive initial brand awareness and recognition. While in app referrals, organic growth drives future growth.
And five years after starting off in Kenya, Siroya has expanded her market to Mexico, India and the Philippines.
Sizing up the competition
Homegrown competitors from emerging market countries have woken up to the opportunity that Tala has tapped.
Siroya points to Chinese players that are now lending in the Philippines, or the numerous Tala-like apps in Kenya, and even Plaid, a U.S.-based microfinance app. But having surveyed the competitive landscape, Richards said he isn't too worried.
"We believe she a) has the best technology and longest track record in this space (thus her credit models are more proven) and b) access to capital via a strong investor base that some other emerging players don't have," he said.
But what about Facebook and its plans with the digital currency "libra?"
"Facebook has the resources but perhaps not the trust of the customer, even if it does create a personalized digital wallet," Siroya said.
Richards said the competition will not be Siroya's most significant concern.
"Today, her biggest challenge is likely just managing scale — her team, her business across multiple geographies around the world," he said. "When I first met Shivani in 2015, she had around 10 people in the company. ... The challenge now is to continue to scale the business and expand into new markets while maintaining a high bar with the operation."