Shivani Siroya: I was working in India when I started lending people money in my spare time. Maybe $300 at a time, $500 max. I wanted to help out entrepreneurial people—a tilemaker, for example—who I thought would benefit from a little extra cash to invest in their businesses.
These individuals didn’t have credit scores. So I looked at other things. I lent to one woman, for instance, who I knew saved 30 percent of her income every month for her son’s computer class. I’d observed that she bought her inventory every third Thursday of the month, and that people in her community spent more time in her store than any other place on her block. A friend of mine noted that I was underwriting her by analyzing her daily life. The phrase stuck with me.
I realized all this daily life data is sitting right on our phones. When I pay my electricity bill, I get a text message confirming it. When I buy a bus ticket or my paycheck gets deposited, I get a notification. These messages amount to a record of my daily life.
Shivani Siroya
Dream job (after current one):
CIA agent
—
Weird alarm clock habit:
“I always set it for a time ending in an odd number.”
We built that observation into an Android app. With users’ permission, the app looks at their phone usage patterns, their behavior within our app, and other data. We scan texts for receipts and transactions, for example, and try to understand communication patterns and habits. This information allows us to establish trust and provide unsecured credit to customers we have never met.
We’re now in five countries on three continents, and our repayment rate is 92 percent. We’ve delivered more than 9 million loans of $100 each, on average, that are paid back in roughly a month. Our system is not all that different from a digital credit card or a working line of capital, but for an untapped population.
To ensure that we’re building an ethical system, we’re conducting a university study on fair lending and algorithmic bias. We don’t want to include things in our models that could bias us against customers unfairly. We don’t add gender. We don’t add location. We don’t add the number of languages someone speaks.
Data can hold enormous power. For our customers, it’s unlocking access to global markets. But it’s our responsibility to use it wisely. —As told to Jessi Hempel
https://www.wired.com/story/wired25-melinda-gates-shivani-siroya-credit-loans/