Social Safety Net Interlinking Nutrition and Workfare Impact Evaluation 2014, Baseline Survey
Unemployment is strikingly high in Djibouti: only 30% of the adult population is classified as employed, according to the most recent household survey data (EDAM 2012). Since the time of the 2008 food and fuel crisis and in the lack of social safety nets that could be scaled up to respond to the crisis, the Government aimed at promoting temporary access to employment through workfare. Moreover, in spite of progress made towards meeting some health and education-related MDGs, Djibouti's health indicators remain among the poorest in the world. The national prevalence of moderate and severe stunting in the most recent MICS survey is of 33% among children aged 0-5 years old, a prevalence rate comparable to Sub-Saharan countries of a much lower income per capita.
In order to address both issues, the Government of Djibouti is piloting an innovative integrated public works and nutrition intervention. The intervention (i) actively involves the main caregiver in a number of ways (nutrition, workfare) to strengthen her role in the household and (ii) makes access to income (workfare) conditional on the caregiver attending regular nutrition promotion activities. The program targets households with pregnant women and children 0-2 years of age in poor areas (urban and rural) in Djibouti.
The first objective of this evaluation is to test the value added of combining a public works program targeted to women over and above the provision of information and promotion of behavioral change in child care practices. That is, the evaluation will specifically measure the impact of making an integrated nutrition and workfare intervention available compared to a nutrition program by itself. As access to the public works is given only to households where the women has registered to the community nutrition program, the program's goal is to leverage the additional cash income (net of the opportunity cost of participation) to enhance the adoption of improved nutrition practices. The planned evaluation will provide a test for the interaction effect between income and the information and promotion of behavioral change.
The second objective is to test whether these effects are only short term, or whether they extend beyond the fifty days of participation in the public works program. The safety net, by design, provides only short term income support during the first 1,000 days. This evaluation is interested in measuring the extent that the impact extends beyond the contemporaneous duration of the safety net program, after the women have stopped participating. Participants and the control group will be interviewed at endline five months after having exited the workfare program. The medium term effects might in part persist through short term savings, but also through labor supply activation and improved psychological wellbeing of the participants.
Kind of Data
Sample survey data [ssd]
Unit of Analysis
The units of observations are urban households eligible to a community-based nutrition intervention (i.e. households with pregnant women and/or children aged 0-2 years old at the time they joined the nutrition meetings).
v01 - edited dataset with direct identifiers removed.
Values in variables id4 (nom beneficaire), id5 (nom chef de menage), nom_mere (nom mre), id9 (nom enquteur), id9s (nom superviseur), id15_1 (numros de tlphone du mnage : #1), id15_2 (numros de tlphone du mnage : #2), c_id10 (coordonns gps), m1 (nous voudrons vous recontacter dans 6 mois. pourriez-vous nous donner le nom, ad), id6 (description foyer) were replaced with missing values to protect confidentiality of respondents.
- Woman Beneficiary Questionnaire: household socio-economic characteristics, health and nutrition practices, food security, durable assets, housing characteristics, household expenses, non-labor income, transfers, intra-household decision making, aspirations, personality traits, well-being and time use.
- Man Questionnaire: labor supply of household members and income from labor, household expenses on items usually bought by male members (khat, cigarettes, transport, etc.), intra-household decision making, aspirations, personality traits, well-being and time use.
The public works component of the intervention has been rolled out in two out of three eligible neighborhoods in the capital Djibouti ville shortly after launching the community growth promotion. In contrast, the public work component was only launched in May 2014 in the neighborhood called "Hayableh".
Producers and sponsors
The Abdul Latif Jameel Poverty Action Lab (J-PAL)
The World Bank
The World Bank
Strategic Impact Evaluation Trust Fund
- Evaluation design: To estimate the impact of offering the public works program to households that are participating in the community growth promotion activities, a randomized experiment was used. From a list containing all eligible households surveyed at baseline (1,011 in total), households were randomly assigned to two groups: i) a "treatment" group where households are given priority to participate in the public works and services offered by the program, and ii) a "control" group where households will get the offer later. The evaluation design was reviewed and cleared by the J-PAL Europe Institutional Review Board on February 2014.
The time lag in the exposure to the intervention between the two groups takes advantage of the phase-in design of the intervention itself: 250 public works positions are being set up every 5 months between May 2014 and December 2015. Thus, the program will make available 250 positions between May and September 2014, 250 between November 2014 and March 2015, 250 between April and July 2015 and 250 between August and December 2015.
Households interviewed at baseline, 1,011 in total, were randomly assigned to 4 groups:
• Group A: public works and services offered between May and September 2014
• Group B: public works and services will be offered between November 2014 and March 2015
• Group C: public works and services will be offered between April and July 2015
• Group D: public works and services will be offered between August and December 2015
The evaluation exploits the gradual rollout of the public works within the neighborhood with a randomized assignment of the timing of offer in the program. The 500 hundred households that will be given the opportunity to work between May 2014 and March 2015 will constitute the treatment group. The remaining 500 households will constitute the control group, randomized to receive. They will receive the intervention on average nine months later than in the treatment group (or, equivalently, seven months after the intervention in the corresponding treatment group ends).
Each group will receive a baseline survey immediately before the start of the program, a first follow-up survey, collected during the public works intervention, and an endline survey, after the program had ended. While all groups were administered a baseline survey between January and March 2014, the different groups are interviewed in a staggered fashion, so that 'each treatment' group will be interviewed with its corresponding randomized control group, both during the intervention as well as at endline. Group A will be interviewed with group C, and group B will be interviewed with group D. The endline surveys for the treatment groups and their corresponding randomized 'control' groups will take place before the latter get offered the intervention.
- Overview of Sampling for the Baseline: the sample consists of all eligible households resident in Hayableh District of Djibouti City. Eligibility is defined as follows: a household with a pregnant or lactating woman (i.e. with children younger than 2 years old) that registered in the program and had been assigned to a nutrition session group. A total of 1,055 eligible households were identified based on program administrative data. Out of it, 1,011 households were interviewed (96% response rate). One third of the non-responses are due to the absence of the household and another third to the refusal of the beneficiary and her husband to be interviewed.
Dates of Data Collection
Data Collection Mode
Computer Assisted Personal Interview [capi]
Three supervisors and fifteen surveyors were trained during 7 days by a consultant hired by the Bank and a field coordinators hired by the ADDS. The training was organized in three phases: (i) questionnaire content (using paper questionnaires), (ii) use of CAPI and (iii) a field test. Surveyors went through several applied exercises all along the training and obtained daily feedback from trainers. Data collection quality was assured directly by the presence during interviews of supervisors and the field coordinator.
Data Collection Notes
Data is being collected using the Computer Assisted Personal Interviewing (CAPI) Solution developed by the World Bank. This study was the second project to pilot CAPI within the Bank and the first one to do it with tablets initialized in French. Numerous programming issues were faced during the administration of the baseline survey, which required intensive support from the CAPI team at the Bank. All problems encountered during this first survey have been solved since then. The launching of the follow-up surveys required much less support from headquarters than the initial survey.
The data collection has been carefully planned to match the timing of the intervention for the different groups of eligible households in Hayableh.
Phase 1: Baseline Household Survey
A baseline household survey was administered between January and March 2014 to 1,011 households (belonging to the test and control groups) selected for the study.
Phase 2: Follow-up Survey
A short follow-up survey will be conducted while the public works are taking place. This survey will allow identifying short-term effects on intermediate and final indicators while the household is receiving income from the government work. The questionnaire will be administered to the beneficiary woman and her husband.
Phase 3: Endline Household Survey
A final survey will be conducted once the household has already finalized the 50 days of work with the aim of measuring the effects of public works in the short term. On average, households will be surveyed five months after the public works have finalized. A household questionnaire similar to the one administered at baseline will be administered to the beneficiary woman (and her husband).
A baseline household survey was administered to beneficiary women, and another shorter survey to the husbands of these women.
Data was downloaded from CAPI server on regular basis and systematic daily checks on the quality of collected data were ensured by another consultant hired by the Bank. This allowed giving frequent feedback to surveyors on data quality. The field coordinator was in charge to share this feedback with the whole group and with specific surveyors if necessary.
The Abdul Latif Jameel Poverty Action Lab (J-PAL)
The World Bank
Use of the dataset must be acknowledged using a citation which would include:
- the Identification of the Primary Investigator
- the title of the survey (including country, acronym and year of implementation)
- the survey reference number
- the source and date of download
Florencia Devoto, J-PAL; Emanuela Galasso and Stefanie Koettl-Brodmann, The World Bank. Djibouti Social Safety Net Interlinking Nutrition and Workfare Impact Evaluation 2014, Baseline Survey (SSNNWIE-BL). Ref. DJI_2014_SSNNWIE-BL_v01_M_v01_A_PUF. Dataset downloaded from [URL] on [date].
Disclaimer and copyrights
The user of the data acknowledges that the original collector of the data, the authorized distributor of the data, and the relevant funding agency bear no responsibility for use of the data or for interpretations or inferences based upon such uses.