Group Meeting Frequency and Borrowers’ Repayment Performance in Microfinance: Evidence from a Quasi-Natural Experiment in South Africa
37 Pages Posted: 2 Dec 2017
Date Written: November 30, 2017
A quasi-natural experiment has been carried out in which the Centre Meeting (CM) rules of some centres of a large Microfinance Institution (MFI) that offers microfinance services, in the form of group lending, were changed. The study has been carried out at the Small Enterprise Foundation (SEF), an MFI that operates in South Africa. The frequency of group meetings, organised as part of the “Monthly Centre Meetings” pilot programme, was reduced from bimonthly to monthly, and only one member per credit- group was asked to participate instead of all the group members. The purpose of these changes was to allow borrowers to save time to spend on their own economic activities. This new policy was put into force for one year, from May 2014 to the end of April 2015. After selecting a suitable control group of micro- borrowers, using Propensity Score Matching techniques, we ran difference-in-difference (DID) regressions to evaluate the impact of the policy changes on the client’s repayment and saving behaviour. The obtained estimates suggest that the change in the policy rules had led to a deterioration of the customers’ saving balances and had increased delayed repayments. Text mining techniques, applied to survey data, pointed towards a lack of trust within the groups, the members of which did not meet frequently outside the CMs, and this was found to be one of the main causes of failure of the pilot experiment. We have concluded that group meetings are an effective tool to stimulate the accumulation of social capital for some categories of clients, and that those MFIs that wish to implement similar interventions should be aware of the drawbacks pertaining to borrowers’ behaviour.
Keywords: Microfinance; Group lending; Group meetings; Quasi-natural experiment; Repayment delay; Savings.
JEL Classification: G21, O15, L31, I25
Suggested Citation: Suggested Citation