December 28, 2012

Sharing Central

New research reveals how connectivity in local social networks promotes cooperation.
Print this story

In rural India, poor citizens face difficulty accessing loans and other financial products from the formal financial sector, and few banks offer products designed for those with so little in the way of financial resources. Even basic contracts are difficult to negotiate: a land dispute can take 10 years to move through the court system.

These small Indian villages fill the void with informal social structures that can make managing meager resources easier while also reinforcing community expectations about sharing and cooperation. Professor Emily Breza has undertaken a series of projects aimed at understanding how local social structures influence informal contracting in these small, under-resourced communities.

In a recent project designed to learn how social networks impact peers’ willingness to share resources and invest, Breza worked with Arun Chandrasekhar of Microsoft Research and Stanford University and Horacio Larreguy of MIT. “Our experiment transcends simple sharing,” Breza says. “People may have access to profitable investments, but they might not have the resources to exploit them.”

The function and role of social networks in rural India is very different and arguably more consequential on the whole than relationships among people in online social networks like Facebook. A key angle of the researchers’ inquiry was to learn more about what peer effects occur in a laboratory setting when people have meaningful relationships in the real world.

The researchers recruited about 1,000 participants from 41 small villages — about 25 people from each village — in Karnataka, India, for which they had extensive social network maps and related household data. Two participants from the same village played a series of standard trust games with each other, one acting as a sender and another as receiver, each starting with an endowment of 60 rupees. (The actual stakes were substantial for those who agreed to participate: each person had the opportunity to earn one-half- to a full-day’s wage for three hours of their time.)

The sender’s sole task was to decide how much of her 60 rupees to transfer to the receiver, knowing that whatever amount she sent would triple by the time it reached the receiver. The receiver then decided how much of her total allocation — her original 60 rupees plus any rupees from the sender — to transfer back to the sender. The allocation that maximizes the pie available to both recipients occurs when the sender transfers all 60 rupees and the receiver transfers back half of that 240-rupee largess, so that each participant ends up with 120 rupees.

Why would either player send any allocation to the other? The researchers expected that players with existing social relationships to maintain would be more likely to share more than would strangers, with whom trust experiments are typically conducted. The most efficient outcome — efficient in the sense that participants’ exchanges left them both with the maximum possible amount — happened about 3 percent of the time. Small as it may seem, that frequency is significantly higher than when the game is played between strangers. Overall, the closer the relationship between the two players, the more efficient the outcome.

Next, the researchers wanted to learn how the presence of a third player — sometimes a member of the community, sometimes a stranger from another village — impacted the game. In some of these experiments the judge — the third person — was allowed to spend his or her own allocation of 100 rupees to punish other players by reducing their total winnings if he deemed their allocations unfair.

“The setup simulates local contract enforcement or local arbitration by bringing in a third party with the power to monitor and, in some cases, punish the other participants for behavior that he deems unfair,” Breza says. “The judge’s visibility and identity are key because past studies done with anonymous players find that third-party punishment helps to increase the efficiency of the game.” As before, the closer the relationship in the social network the more efficient the outcome is likely to be. “But what really increases efficiency is going from having a judge with unimportant social status to a judge of important social status,” Breza says.

But the way in which the judge is important matters: having a judge who was widely connected across the network yielded different results than having high social or political status. Using Eigenvector centrality, a network measure of importance and connectedness, the researchers identified villagers who act as information hubs, transferring information easily around the network. When one of these centrally connected villagers took the role of judge, the outcomes were the most efficient. In contrast, judges who were politically important but not central to the network tended to play in a way that maximized only their own allocations rather than ensuring that resources were maximized between the sender and receiver. The games were also less efficient when judges were close friends with one of the other players. In other words, network importance rather than closeness, wealth, or titles most often yielded the best benefits.

Breza’s larger goal in pursuing this line of research is to use what economists have learned about the powerful influence of social networks and how peer dynamics within them can be adapted to complement formal financial products for the poor. One such project addresses savings. “Banks in India are very unwilling to provide new tailored products to the poor, but they do have no-frills accounts that let users maintain lower balances. But the accounts are not well used. Savers have a hard time meeting saving goals. This may be because they have to walk the three kilometers to the bank; or it may be due to other obstacles, such as procrastination.”

In the savings project, Breza and her co-researchers are studying whether reporting savings activity to a third-party monitor in the same village can help would-be savers surmount those obstacles. “And should that person be the highly central person or the close friend? Our aim is to identify a solution that leverages the local social network and the formal banking system to help people achieve their goals.”

Emily Breza is assistant professor of finance and economics at Columbia Business School.

Read the Research

Emily Breza, Arun Chandrasekhar, Horacio Larreguy

"Mobilizing Investment Through Social Networks: Evidence from a Lab Experiment in the Field"

Download PDF 
View abstract/citation