The Use of Control Groups in Impact Assessments for Microfinance
Investigates problems encountered in the use of impact assessments of microfinance programs. Focuses particularly on the control group methodology, employed to compare a population that had benefited from a microcredit scheme to another group that had not, which is increasingly being used.
Outlines in detail methodological problems of using control groups such as:
- Sample selection bias;
- Misspecification of causal relationships;
- Motivational problems.
Makes recommendations that:
- Where possible, microfinance programmes should be assessed by means of a quasi-experimental method which holds constant all causes of variation in the chosen target variable apart from microfinance itself. For practical purposes, this means either using control groups or regression analysis;
- As potential distortions arise with the use of control groups, contervailing measures should be applied, but how far it is appropriate to go with these depends on the budget and information needs of the analyst;
- Evaluations of programme impact should be conducted long after the end of disbursement to take note of downstream effects, occurring over time, and indirect effects;
- Programme managers who are interested to know whether a microfinance programme works better if specific features are added to or subtracted from it should apply the "pilot project" method of evaluation, in which the participants in the pilot project are the target group;
- Caution should be undertaken in evaluating poverty impact using control groups as particular problems will occur given the multi- dimensionality of poverty.