Kiva.org: The Dynamics of Group Lending

On March 24, Stanford Social Innovation Review columnist Sean Stannard-Stockton wrote a propitiously timed article entitled “Kiva.org: Made to Stick,” wherein he referred to Kiva.org as the gateway drug to social investing.  In his article, he asks how Kiva was able to impel individuals to become social lenders, and attributes some of their success to the concept of “stickiness.”  In a nutshell, simple ideas that are at once unexpected, concrete, credible, emotive, and draw on anecdotes, can have the most powerful effect on human engagement, and even addiction.

But at a deeper level, what it is about Kiva.org, and its site design, that has enabled it to successfully mobilize millions of crowd-sourced dollars from diverse global lenders, all for zero-percent returns?  On March 30, 2010 I released a working paper based on research done in the Cooperation Group at Harvard University’s Berkman Center for Internet & Society entitled “Kiva.org: Crowd-Sourced Microfinance and Cooperation in Group Lending,” that attempts to unpack the specific modalities in Kiva.org site design that foster unexpected cooperation and coordination between disparate individuals to provide zero-interest, online peer-to-peer loans.

Microfinance literature often highlights the benefits of borrowing groups.  However, such literature is focused on the advent and benefits of borrower-group organization, the uses of social ties and pressure for the purpose of lowering default rates.  Little, if any, literature comprehensively covers the dynamics of group lending within the context of cooperative Internet-based peer-to-peer microfinance.

Whereas traditional microfinance is one-to-one or one-to-many, models such as Kiva.org have popularized “crowd-sourced microfinance,” or many-to-one and many-to-many platforms for international capital allocation and distribution.

In late 2008, Kiva.org debuted a site feature called “Lending Teams,” allowing individuals to create teams, define teams as “open” or “closed” groups, and categorize them according to various forms of identity.  What emerged was an online lending environment in which individuals could assign loans to given groups, and a lending environment in which teams of varying size, openness, and categorization were performing better and worse along a number of metrics such as number of loans provided, per capita lending, and average loan amount.  An individual’s self-identifying choice within Kiva.org Lending Teams was apparently linked to lending participation and performance.

With the help of Kiva.org and Manal Dia, Aaron Shaw, and Tim Hwang, I accessed publicly available data for the top and bottom performing, open and closed access lending teams via the public Application Programming Interface (API), and tracked 120 Kiva.org Lending Teams throughout summer 2009.  The goal of such observation was to understand Kiva.org’s effective use of “Solidarity,” as a lever of site design, their ability to create –as Stannard-Stockton calls– a “sticky” site, and how Lending Teams effectively or ineffectively fostered greater cooperation in Internet-based peer-to-peer microfinance.  The windfalls of understanding had ramifications not only for crowd-sourced microfinance, and the ability of mobilize capital for visionary emerging market entrepreneurs, but also for any online organization hoping to foster greater on-site coordination and cooperation.

While the working paper certainly does not provide comprehensive answers, perhaps it asks questions which address some of the issues central to human cooperation.  In attempting to provide taxonomies of both the top- and bottom-performing Lending Teams, it happens across a number of empirical observations:

• Small communities with fewer than 10 members have greater per-capita lending averages when these groups are defined as “closed” access, perhaps because there is greater local accountability in lending.
• During critical periods of membership growth, such as between 10 and 20 members, again, groups classified as “closed” have roughly 300 percent greater per-capita lending than “open” teams of the same size, however beyond 20 members, large “open” teams lend over three times as much. At certain membership inflection points, “closed” access can perhaps cement solidarity, while at other junctions “open” access can bolster engagement.

Based on these, and other empirical observations made in this study, one is able to glean effective management practices that address dual goals of minimizing membership attrition, and maximizing participatory vitality in group coherence.  While, in the context of Kiva.org, these normative practices may enable greater Lending Team participation, per-capita lending, and aggregate capital mobilization to emerging market entrepreneurs, observations might yet be fungible to applications unknown.  I encourage your thoughts, your criticism, and your vision.

For more information on the study, “Kiva.org: Crowd-Sourced Microfinance and Cooperation in Group Lending” is available for download.  For more information how to create your own Lending Team, please see Kiva.org.

Read more stories by Scott E. Hartley.