(Illustration by Jacob Stead) 

To spread the word about a new service, the Indian microfinance institution Bharatha Swamukti Samsthe (BSS) approached leaders in 43 villages—teachers, shopkeepers, savings group leaders, and the like. The goal was to publicize the program by encouraging these leaders to tell their friends about it. But in some villages, only 7 percent of households eventually took out a BSS loan, while in other villages up to 44 percent did so. What accounts for the difference?

According to a team of researchers who studied the BSS initiative, the choice of whom to tell first matters a lot. “There are very specific ways of measuring who the most influential people are in a society, in terms of spreading information,” says Matthew Jackson, an economist at Stanford University. He and his colleagues have developed a new measure of social influence that’s especially relevant in cases when word of mouth is an important medium of communication.

BSS entered the Indian state of Karnataka in 2007. Beforehand, Jackson and others on his team mapped the social network in each village that BSS intended to target. The researchers asked members of each household about their friends—the people whom they visit, pray with, lend rice and kerosene to, get advice from, and so on. Later, they used that information, along with data from BSS on which village leaders were initially introduced to the program and which households ultimately signed up for a loan, to model how the program spread.

Certain people had a greater impact on diffusion of the program than others. “In some villages, the teacher was very central” to a village network, Jackson says. “In other villages, the teacher wasn’t.” The researchers identified the characteristics of the most central village leaders—the ones who, in effect, were able to broadcast information farthest. And, as it turns out, those leaders don’t just know people; they know people who know people. It doesn’t matter “whether the first person has more friends or fewer friends,” Jackson says. “But how well connected that person’s friends are makes a big difference.”

By analyzing such diffusion patterns, the researchers were also able to tell whether any given villager’s decision to join the microlending program increased the likelihood that others would adopt it, too. In fact, it didn’t. The program did spread farther when more people took out loans—but only because those who participated in the program talked about it more than nonparticipants did.

There are problems to solve before policy makers and program implementers can use these findings. “The issue with implementing [the researchers’ model] is that it’s very expensive and difficult to map out a social network,” says Jake Kendall, a senior program officer at the Bill & Melinda Gates Foundation whose work focuses on financial services for the poor. By the time one has surveyed all residents of a village about their friends, Kendall notes, one may as well have marketed a program to them directly. “That sort of erases the value proposition of targeting selectively,” he says.

But knowing which villagers to approach first could soon help new programs spread faster and reach farther. Kendall, for instance has ideas on how he might adapt the researchers’ model to promote financial services programs in Africa. Jackson and his colleagues, meanwhile, are testing methods that would provide a shortcut around having to map an entire village network.

Getting information to people in remote villages often requires using word of mouth. This new research, in short, may yield a way to make word of mouth louder.

Abhijit Banerjee, Arun G. Chandrasekhar, Esther Duflo, and Matthew O. Jackson, “The Diffusion of Microfinance,” Science, 341, 2013.

Tracker Pixel for Entry