Micro-lending sites like Kiva are an important development tool, but bias is affecting lending choices. Here's how to fix it.
Want to cash in on the surge in online peer-to-peer lending? As with many things in life, it helps to have a pretty smile. Sadly, it also helps to have lighter skin.
People who lend on Kiva.com tend to favor attractive, light-skinned females, according to an analysis set to be released by researchers at Nanyang Technological University in Singapore. To receive a loan online, it’s better to be pretty than ugly, female than male, skinny than fat and, yes, light-skinned than dark-skinned, the researchers found. “So, basically if you have any of these characteristics, you will get your money more quickly on Kiva.com,” says Walter Theseira, the lead author of the study.
Kiva allows internet users to browse candidates for charitable microloans around the world, sort by country or type of project, and then—while viewing a picture of the potential borrower—click a few buttons to lend to a microfinance agency supporting that individual. Kiva recently launched a pilot that allows users to lend directly to borrowers without the middleman, but the study focused exclusively on the main Kiva site.
Theseira conducted the study with Christina Jenq at the University of Chicago’s Booth School of Business and Jessica Pan at the National University of Singapore. The researchers used a skin color scale used in research on bias against new immigrants and found conclusively that lighter-skinned borrowers got loans faster. Being attractive was even more helpful.
“This is why [NGOs] spend so much time choosing just the right photographs to illicit donations,” Theseira says. People respond more, and with donations, to pretty, smiling faces. But “as for why people do that... it’s a bit hard to say,” he cautions. “Our hypothesis is this is probably more a form of implicit discrimination than people acting on explicit bias.”
Dark skin may slow the pace of receiving money through Kiva, but being from Africa helps enormously. Loans to Kenya fund several times faster than loans to Bulgaria on the platform. Geographic discrimination is far more important than skin tone. Users tend to believe there is more need in poorer countries, and African countries tend to be poorer. Lenders may also feel their dollars will go farther with a microloan to Africa than elsewhere. On Kiva, the skin tone bias only crops up strongly after other factors are accounted for.
While a litany of microfinance literature recommends lending to women over men—to counteract discrimination and because women tend to use money more prudently—the measurable presence of race bias serves as a warning to crowdfunding sites and donors alike. This sector is booming now, powered by promises to democratize lending and open doors to those excluded or discriminated against by traditional banking. Baked-in bias in peer-to-peer choices could undermine the very idea of a financial solution to unfairness.
Theseira’s research is part of the first wave of hard evidence that donors respond to subconscious bias in deciding where to drive their peer-to-peer charity and lending dollars.
It fits an all-too-common pattern on the internet: Stanford researchers found online shoppers were less trusting of someone selling an iPod online if the hand in the photo holding the iPod was black. A Wharton study found that black borrowers on peer-to-peer lending site Prosper.com were up to 35 percent less likely to get a loan at a similar interest rate as comparable white borrowers. In response, Prosper stopped including photos in user profiles.
Expect to see more academics examining online crowdfunding, microfinance, and lending sites. The question now is what the sites can do to fix the problem. Theseira suggests regular Kiva users could offer the key to a solution.
"What we found [on Kiva] are these patterns of discrimination are most evident in people who don’t lend much," he says. Habitual use of a site, he reasons, makes the lender more strategic, more businesslike, and less biased. Instead of just clicking on pretty faces, repeat users scan and search for details more relevant to loan-worthiness. "In the case of the Kiva website itself," Theseira says, "I think if we can try to establish more clearly whether it’s implicit discrimination or something else, it might be possible to use technology to address it." In other words, a lending site could show newer users a set of profiles designed to counteract measured biases.
There is also some evidence that race bias is more pronounced when we are overwhelmed with choice and when we make split-second decisions—a finding that came from a study of NBA refereeing but that Theseira suggests could apply to online lending, too. Offering fewer choices, or designing elements of a site to slow the decision-making process could help. Once you’ve used the site more often, the design could subtly change for you, he suggests.
Theseira, like any researcher at the tip of an iceberg, wants more data on why people prefer to fund light-skinned pretty people in charitable and lending decisions before he’ll commit to a prescription. He’s expanding his research to European lending site myc4.com, where investors earn interest, in contrast to Kiva’s zero-interest loans, to more fully understand the incentives behind lender decision-making.
For its part, Kiva is admirably open about the bias problem, and its leaders are eager to innovate their way around micro-discrimination. The company has observed geography, industry, and male-female bias. Kiva has opened up its data to other researchers to investigate further. More details on the nature of any bias will help point to a plan to counteract it.
Photo courtesy of Kiva.com