Who Do You Trust? Building a “TrustScore” for Savings Groups
It’s a question to which we at DreamStart Labs would like to know the answer. For while we use demographic and transactional data generated through our DreamSave savings app to build credit scores for members, there’s an extra dimension we have long wanted to explore: member-driven assessment of the trustworthiness of fellow Savings Group members.
We see value in member assessments for two reasons. First, Micro Equity investments in Savings Groups.Second, we are interested in understanding the potential to develop this TrustScore as a separate credit scoring system in the long run, either for external lenders who wish to lend to Savings Groups or for ourselves when looking at
Overall, a TrustScore could help Savings Group members build a credit score faster, allowing them to access external credit sooner.
How to measure trust?
Our first challenge was to find a reliable and credible measure of trust.
We started by conducting a literature review to discover the work others have done in this field. We found broad agreement on the importance of trust in group formation, operation and success. In particular, personal trust between specific pairs of group members is the bedrock for self-selection and screening in microfinance groups. Social connections, especially proximity – such as living in the same village (but not having too many relatives in the same group!), have a strong positive impact on repayment and savings.
So, if interpersonal relations are important to group success, how could we measure the strength of those bonds or, in other words, the degree of trust between members?
Stage 1 – Physical interview
The first step to figuring this out was to conduct face-to-face surveys in Tanzania and Benin. Representatives from DreamStart Labs went to Savings Group meetings in both countries to test a variety of trust-based questions.
Members were willing to answer questions on whom they trust, and we explored the concepts of “trust,” “sponsorship” and “lending” with them to figure out how best to get at the information we were seeking. We found that members judged trust on the basis of the personal relationship with another group member, sponsorship on the collateral the member possessed, and lending money based on familiarity and proximity. Therefore, a question on “who would you lend to?” seemed to capture the concept of trust that we were trying to uncover, i.e., creditworthiness. Moreover, members were willing to specify the amount they would lend to other members, leading us to believe that we could use this as the “trust metric.”
Stage 2 – The telephone survey
The objective of the next stage of the research was to reduce complexity by employing the simplified “trust” question we had developed in Stage 1 (i.e., who would you lend to and how much), and to test it using a phone-based approach, as opposed to the in-person survey method in stage one.
We found that members were just as willing to answer questions on trust remotely (via a telephone survey) as they were face to face. We also discovered that members were comfortable sharing a phone between members (for those who did not have their own phones) and in answering a trust survey delivered on that phone. (Note that we confirmed with each member at the start that they felt sufficiently removed from other members to answer the questions honestly.)
Stage 3 – The app-based approach
The findings from the telephone interview stage gave us confidence to move to a more scalable app-based approach. The willingness of members to share a smartphone at Stage 2 convinced us that we could deliver an app-based survey to one smartphone per group (such as the Savings Group phone that uses our DreamSave app) and that all members would be able to take turns answering the survey on that phone. Even using the remote app-based approach, we found that members were prepared to name those they trust in the group and to rate members by attaching a notional personal loan total to each.
Using a survey app also meant that we could test a picture polling approach (using profile pictures we have already) to make it easier for less literate members. Members select the photos of those they trust in the group and enter a chosen loan amount against each photo. This approach proved popular with members.
Based on these results, our next step is to start rolling out the TrustScore survey to thousands of additional groups using the DreamSave App around the world. Our plan is to poll groups regularly to gauge changes in member ranking and to update each member’s TrustScore over time.
If this means access to external credit sooner, or to a bigger external credit limit, this will be a big win for all our Savings Group members.