Project Everest

Proposed Experiment

[Proposed Experiment]: SoCon Malawi - Replicating social pressure to reduce default risk - July 2018

Kurt Michl
Kurt Michl | Jul 5, 2018 | in Knowledge Base

Over the course of July we have been able to gather a lot of information on how commercial lending is currently administered, but also a lot of information on more informal lending mechanisms. There are recurring themes in both of these two systems which is to utilise social pressure to ensure that the principal is repaid. 

The idea is to distribute a loan to a group of people, rather than to any one individual. Under this mechanism, if any single member of the group is struggling with their share of the repayment, then the other members of the group will have a huge incentive to ensure that the loan is repaid. By the same token, village banks operate financing in a way that the members' pooled savings are loaned to individuals within the group. It is therefore in the best interest of all of these members that the loan is repaid to these communal pools of funds. The key in this is social pressure on the loan repayment.


The next step for us would be to try and replicate these practices, and then to adapt it such that it becomes naturally scaleable.


Replicating these behaviours.

We may be able to replicate these behaviours of social pressure through the grouping together of individuals who take out loans with us. Imagine in a village there are 10 people who want to adopt a payment plan. One of the conditions of the payment plan could be that, if all 10 loans are repaid in full on time, then all 10 individuals could receive a partial refund on the interest that they have paid. As such, the social pressure would be replicated, because if one person is struggling to make repayments, then it is in the best interest of the 9 others that they find a way to finish making the repayments. 

I would suggest that we could trial this out with the sale of the solar products, and it could work in conjunction with the commission of the chief and their recommendations. 


Proposed Experiment

Hyp: We can reduce the default rate and thus the default risk premium with systems which replicate the already-present group repayment systems mentioned above. 

Objective: We want to test whether grouped loans perform better than individual loans. We want to test if the social pressure systems are reliable and effective. If yes, then we want to test if a retro-active cash-back system is incentive enough to enforce social pressures.

Method: We must extend our criteria for credit-worthiness to encompass this aspect. 

  • Individuals will apply for loans as a group that they choose (minimum three people). Alternatively, the groups could be allocated randomly, such that there can be no communication from members 
  • We will set out a clear payment plan for them all to follow together
  • We will clearly state that it is in their best interests to repay the loans as agreed, offering a cash back, and such reducing the risk premium if the default is zero.

We will wait until all repayments have been made by all individuals in each group before providing the cash back. This method can be compared to Groupon discount, except it is repaid retro-actively. 

Ideal Scenario: The Energy Team is in the process of selling their product with a payment plan, which is effectively a form of micro credit. They have been selling their proxy solar product through Nancholi, which is proving to be a success. When they continue this expansion in December, we would ideally offer this new payment plan in conjunction with the individual payment plan. this would also give us an indication of individuals propensity to choose a grouped loan for a retro-active discount rate. These payment plans would follow the same structure, but in the village where we would conduct this test, we would offer a cash back. 

For example, we have a set payment plan of 1200 MK over 26 weeks, rounding to         MK 31 200, on a MK 20 000 principal.

I would propose that if the first 24 weeks are paid in full by all parties, then we could neglect the need for the last two repayment in the 25th and 26th weeks. This provides them with an effective saving of MK 2 400. This would save on transfer costs, but it may allow for people not paying the last payments if they know they wont be eligible for the cash back. If the energy team deem it appropriate to change the payment plan, which may be the case, then the grouped payment plan would also follow that change as well.

Alternatively, we could wait until all  payments had been made and once the whole sum had been paid, we could offer a cash back which effectively educes the interest paid retro-actively.

Findings: We would hope to find that this is an effective avenue to pursue.

  • that grouped loans are effective
  • that grouped loans are attractive
  • that social pressure alone is enough to incite repayment  


How can we scale this?

Longer term, I can imagine this becoming scaleable if there is tech adoption. I envisage a platform, perhaps app-based, in which individuals can make an account. they key features of this platform would be that it:

a)connects borrowers with lenders,

b)connects borrowers with other similar borrowers (financially) in close geographical proximity,

c) it acts as a ledger and accounting method for all transactions,

d) that there would be a design in which the lenders would rate each other, thus creating an in-app credit-rating system. The rating would determine the amount one can borrow and at what interest, in what group size.

In this model, the lender could be a PEV subsidiary, or it could  integrate a design into this platform which would connect investors to loan-takers, as suggested in the post "Can we entirely hedge default risk in micro-finance markets?"

Please comment your thoughts and criticisms,


edited on 8th January 2019, 09:01 by Ella Grier

William Lee Jul 5, 2018

Hey Kurt,
Definitely would be awesome to test this with the solar product sales.
I've had a go of thinking about the same issue, but instead of pooling people together into loan groups, I've tried to replicate social pressure/trust to score each individual's credit score independently (see link below). Again, the scalability of this model is also dependent on tech adoption. Let me know your thoughts.
What do you think would be the best way to test the viability of social pressure/trust in lending mechanisms in order to achieve this future state?

Reply 0

Kurt Michl Jul 6, 2018

I've had a look, and you do make some very good points, but i also think we have some very common ideas. I would like to come up with a test we can implement to test it. This is where the idea of grouping people together came to mind, simply from a feasibility perspective.

Perhaps i was not clear, and I will adapt the post as my idea becomes clearer, but i was imagining a system like "groupon", where if there are more successful sales there is a group discount.
In the context of a loan, the success of a "sale" can only be measured if all of the loans in the group are repaid in full on time. Then there would be a retroactive discount (cashback) if the sale is successful.
Evidently, the more loan in the group, the larger the "discount" incentive can be. It may also mean that we will be able to start with a higher rate of interest (whilst still affordable), with it clearly stated that if all parties repay their share, then they would get a cash back.
This may also encourage saving, because the cash back amount will be lump sum, providing the with a payoff at the end of the period.

As I said, I will have a bit more a think about how we can test this, this month.

Cheers :)

Users tagged:

Reply 1

William Lee Jul 6, 2018

Yeah, very common indeed. I think your idea is far more feasible and sensible in terms of experimentation. I quite like the groupon idea - i think it will form a good incentive to repayment. The key I think is to ensure that the loss from not paying back fully (i.e. loss of potential cash back) is greater than the gain from not repaying. I look forward to reading your proposed experiment!

Reply 0

Kurt Michl Jul 9, 2018

Hi Will,
I've updated the post to include a proposed experiment design. I'd love to hear what you think, and if you think this is a good avenue to explore, I'll happily adapt our current payment templates to include a proposed feature.
I have notified Andrew and Tink in the hope of getting some feedback from them as well.

If this is something we would want to pursue there is an element of time constraint.

Users tagged:

Reply 1

William Lee Jul 9, 2018

Hi Kurt, the experiment looks good from my perspective. Definitely would be great to kick this off this month. I'm building the CRM currently, so I'll look to incorporate this element into it once the structure of the group lending has been finalised.

Reply 0

View all replies (4)

Ella Grier Jul 20, 2018

I really rate this idea Kurt. It would be interesting to see in what contexts you would experiment with the concept of 'social pressure' as is quite abstract and not easily measurable.

The cashback model provides a good incentive and catalyst for the social pressure to occur, but what would be interesting is to see how it unfolds in different social settings. You could test a range of 10 people groups across both different demographics and social settings, eg. 10 random individuals from Blantyre Markets, a group of 10 stall owners in close proximity at Limbe Markets, 10 neighbours in a rural village, 10 members of the same village bank, 10 members of different village banks in close proximity. This would allow to see what factors influence pressure.
E.g. NAYO allows groups undertaking loans to establish their own set of rules to ensure the loan is repayed. How would these rules vary if you changed the context?

Given that the energy product was originally a 6 month loan, but feedback has indicated that 3 months would be a viable time frame, you could set this experiment up to be started in early December and monitored over the course of Dec/Jan/Feb such that you could observe with consistent checkins the way that these different groups interact.

Reply 1

Kurt Michl Jul 24, 2018

Thanks for the feedback Ella. As discussed the other day, I will elaborate the experiment proposal and try and set it up such that it cold be adopted day one of December project. I will keep you informed :)

Reply 0

Jordan McLoughlin Jul 26, 2018

I like your idea too Kurt. From the energy teams perspective, we aim to target schools as a sales channel starting early December. Replicating the same script and operating procedure already in use for community meetings. In terms of social pressure a closely-knit schools environment consisting of customers who are also family, friends, peers, close neighbours etc would a great environment to pilot this experiment. The hope would be that all customers would come from a similar demographic, so that as we move from village to village, we could monitor the relative wealth, financial literacy and repayment habits. This could be mutually beneficial for SOCON and Energy team as it will give insights that could help you guys design future project and also help the Energy team in positioning their product/pitch on a community by community basis. Just a thought!

Reply 0

Jordan McLoughlin Jul 26, 2018


Users tagged:

Reply 0

Kurt Michl Jul 27, 2018

Thanks Jordan, this gives indication of validation from the Energy team as well as from SoCon.

Reply 0

View all replies (4)

Justin Hakeem 11 months ago

Status labels added: Proposed Experiment, Proposed Solution

Reply 0

Justin Hakeem 11 months ago

Status label removed: Proposed Solution

Reply 0