Experimental Design Details
The RCT is based on a randomized assignment process at the locality level. The randomization allocates localities that were eligible to be included in the program to either one of the treatments or to the control group. The protocol was as follows.
For each sous-prefecture, an exhaustive listing of localities was realized by IRC. The second wave of the program, on which the RCT focuses, involves 11 clusters of sous-prefecture which represent 415 localities (villages and districts), out of which 354 were considered eligible to the program.
The PRISE program organized public lotteries involving all the localities and sous-prefecture representatives. The randomization was clustered by sous-prefecture and stratified by types of locality, namely village or city district.
Two different public lotteries were organized.
A first randomization (“sampling lottery”) randomly selected localities into the study sample. A targeting data collection exercise was then completed in all localities selected into the sample. The data collection was conducted by IRC in each 207 localities selected after the first sampling lottery. The steps of the targeting data collection were:
- Organize an assembly to present the project and inform the population of the enrollment and the targeted population of the program.
- Create a validation committee in charge of the validation of enrollment list in their respective localities, in order to check the veracity of the given information and the real belonging to the village.
- Open the enrollment and record information about the potential participants within 1 to 3 days.
The targeting data collection was implemented by IRC. The final enrolled population was 14 880 individuals in 207 localities. Among all registered individuals, eligibility criteria were applied. The eligible criteria required applicant to be aged of 18 to 40 years old; exemption were made for single mother (from 15 years old on) and widow and handicapped (up to 60 years old). Applicants should not be involved in any other economics assistance program or registered at the ADDR (another program that help disarmed individuals). The last exclusion criteria is to be a student at the time of enrollment. In total, 13 217 individuals were confirmed eligible.
The second lottery randomly assigned the 207 eligible localities to one of the three treatment groups or to the control group, divided as follows:
- 60 localities were assigned to the control group
- 64 localities were assigned to receive the training and the semi credit option, with 50% of the credit to be reimbursed within 6 months, and conditional on writing a viable business plan.
- 53 localities were assigned to receive the training and support to set-up VLSA
- 30 localities were assigned to receive the training and a cash grant conditional on writing a business plan.
The final selection of candidates was based on an individual vulnerability score. Baseline data, collected during enrollment, were used to compute a vulnerability score for each individual. It is a weighted score based on 4 broad criteria including incapacity, marital status, formal education, employment, economic status economic responsibility towards household and assets. Weights were chosen to maximize the dispersion of the score. Selected individuals are those above the vulnerability cut-off established in each locality, ordering vulnerability score in descending order.
An end line survey will conducted after a year and half. It will include individuals above the vulnerability cut-off in treatment and control groups, in order to estimate treatment effects for each treatment, and between treatments. In addition, a sample of eligible but non-selected individuals (with vulnerability scores below the cut-off) will also be included in order to provide information on whether the program generated economic and/or social externalities through spill-overs within localities.