Determinants of Take-Up of a Poverty Graduation Program among Cash Transfer Recipients

The intervention is a video message produced by the Ministry of Social Solidarity (MoSS) providing information relevant to households’ decision to participate in the Forsa poverty graduation program. The objective of the video messages is to provide credible information that targets specific concerns that respondents report having about participating in Forsa, as determined through the aforementioned baseline survey and various reports from the project implementers. Treatment assignment will determine which video segments are viewed by the respondents in an edited version of the video.

2023-10-22

2023-11-30

Interest in joining Forsa

Interest in joining Forsa: Following the video intervention, we will ask respondents: “Are you or someone from your household interested in joining Forsa’s self-employment modality?” and also “Are you or someone from your household interested in joining Forsa’s wage-employment modality?”. Our primary outcome will be an indicator equal to one if a respondent answers “Yes” to either of these questions, and zero otherwise. Thus, our primary outcome is the household’s interest in joining Forsa through either modality.

Intensity of interest; mechanism-related outcomes; take-up of Forsa in administrative data

INTENSTIY OF INTEREST: Responses to the following post-intervention questions measure the intensity of interest in Forsa and will serve as secondary outcomes:
1. On a scale of 1-5, how much do you want to join Forsa? Your answer will not affect your Forsa benefits. (1=DID NOT WANT TO, 2=A LITTLE, 3=SOMEWHAT, 4=MODERTELY, 5=VERY MUCH)
2. How likely are you or someone from your household to approach your local NGO about joining Forsa in the next two weeks? (1=EXTREMELY UNLIKELY, 2= UNLIKELY, 3=NEUTRAL, 4=LIKELY, 5=EXTREMELY LIKELY)
3. How likely are you to recommend Forsa to a friend or neighbor? Your answer will not affect your Takaful or Forsa benefits. (1=EXTREMELY UNLIKELY, 2= UNLIKELY, 3=NEUTRAL, 4=LIKELY, 5=EXTREMELY LIKELY)

MECHANISM-RELATED OUTCOMES: Responses to post-intervention follow-up questions on the following topics are designed to measure perceptions of how Forsa functions and can be used to test what information was communicated through the video message and what most affected household interest in Forsa.
• Duration of Takaful cash transfers either with or without joining Forsa
• Value of and expected monthly income from the asset transferred through Forsa’s self-employment modality
• Time cost, expected monthly income, and expected longevity of the employment opportunity offered through Forsa’s wage employment modality.
• Perceived potential monthly income from asset-based or wage employment - (gender specific?)
• Expectation of non-monetary benefits either with or without joining Forsa

TAKE-UP OF FORSA IN ADMINISTRATIVE DATA: Working with the Ministry of Social Solidarity, we aim to acquire administrative data on whether households actually join Forsa following the intervention and its rollout in eligible communities. Should we obtain these data, “enrollment in Forsa” will be a key secondary outcome. However, we refrain from pre-specifying this as a primary outcome for a concern that data acquisition may be difficult.

The interventions will be tested using a randomized controlled trial (RCT) study design. The interventions will be randomly assigned to participants as different treatment arms (including a control group). Expecting near 100% compliance, we consider the average treatment effect (ATE) of the randomized interventions on our outcomes.

This study is embedded within the midline data collection of a cluster randomized controlled trial to measure the impact of the Forsa program. The video message intervention will be at the end of the midline survey to allow for comparison across treatments. To study the impacts of the video messaging intervention, the sample consists of the 3,838 households surveyed previously in Forsa-treatment communities (i.e., communities in the cluster RCT who are presently eligible to receive Forsa). Meanwhile, those households in Forsa-control communities—where Forsa is not currently active—will not receive a video message. Since the intervention is embedded in the survey, we expect nearly 100% compliance among treated respondents.

Not available

Randomization was done at the household level and assigned a 25% probability to a household being assigned to any treatment arm, stratifying by the 160 community designations and baseline Takaful status (i.e., enrollment in the pre-existing cash transfer program). By design, there are 24 study households in the typical community, 16 of which were Takaful beneficiaries at baseline and 8 of which were not. Therefore, within most Forsa-treatment communities, this randomization strategy assigned 4 Takaful-beneficiary households and 2 non-beneficiary households to each treatment arm. Randomization was implemented using Stata/MP 18.

Household

No

None.

3,838 households

Sample size in each treatment arm among the 3,838 target households is as follows:
• C - Control: 968 households (25.22%)
• T1 - Treatment 1: 957 households (24.93%)
• T2 - Treatment 2: 965 households (25.14%)
• T3 – Treatment 3: 948 households (24.70%)

In a prior survey, willingness to join Forsa among the sample was 95.8%. So, if we assume a power of 0.8, an alpha of 0.05, and that current willingness to join Forsa among the sample is 95.8%, the minimum detectable effect (MDE) is 2.2 percentage points of any treatment arm relative to the control group. However, we believe that this is an optimistic scenario and suspect that interest to join Forsa has declined significantly as evidence by current reluctance to join the program. Therefore, if we conservatively assume that current willingness to join Forsa among the sample is 50%, the minimum detectable effect (MDE) is 6.5 percentage points. Both are considered acceptable MDEs for the study, showing that we have the power to detect policy-relevant outcomes.