Expanding Opportunity: Testing asset grants and jobs programs for Egypt’s poor

FORSA is a program initiated by the Ministry of Social Solidarity building on Takaful and Karama that aims to graduate beneficiaries from reliance on Takaful and Karama (TKP) cash transfers by supporting the economic inclusion and empowerment of TKP beneficiaries and rejected applicants just above the PMT threshold. Forsa’s main objective is to reduce longer term dependency on Takaful transfers by linking beneficiaries to employment services, the provision of training, and asset transfers. The Forsa graduation program includes two modalities: (1) an asset-transfer modality and (2) a wage employment modality which focuses on job matching and training. 70 percent of the Forsa budget will be allocated to the self-employment modality activities in regions with limited formal labor market opportunities, while 30 percent will target wage employment modality activities in regions with more potential access to formal labor market opportunities.
The Forsa package, for both the wage and self-employment modalities, costs approximately US$1000 per beneficiary; these costs include the provision of assets to be transferred to beneficiaries, technical and financial trainings, wage subsidies, and administrative fees. The Forsa program activities are expected to last between 12-18 months, depending on the modality.
Geographical targeting:
The geographic targeting will focus on eight governorates in Upper Egypt and Lower Egypt: Fayoum, Beni Suef, Assiut, Sohag, Luxor, Menia, Sharkiya and Qalyoubia. 330 villages were identified as potential sites based on district and governorate level targeting criteria (poverty and economic opportunity) and having at least 500 current TKP beneficiaries in the village. 165 individual villages within these governorates have been selected as program sites.

Targeting based on administrative data on household characteristics:
Eligibility is limited to TKP beneficiary households and TKP registrant households that fall just above the poverty cut-off point (exact range of PMT scores to be determined), with 30% of beneficiaries and sample households drawn from the latter group.
Based on information from the TKP registration database, eligible households must:
• have at least one working age member (19-55 years old)
• have a household head and (if present) spouse without formal employment
• own 5 qerats or less of land that is used in agriculture
• not be receiving a transfer/asset from any other program
• own no more than 4 medium-sized livestock or 1 large animal
• have a house with no more than one floor
• not be receiving remittances from a member working out of the country

Self-selection and behavioral change “nudge”:
Within treatment communities, a door-to-door campaign for the eligible households will be organized by the implementing NGOs: Al-Koura and CARE Organization. Local MoSS and MoH social unit employees (“raedaat”) will be tasked individually invite eligible households to attend behavioral change sessions about the Forsa program in their community.
The main objectives of the behavioral change sessions are to introduce more details about the modalities offered by the Forsa program and to emphasize the benefits of financial independence and economic empowerment relative to financial reliance and aid dependency. The behavioral change sessions will be small (approximately 40 participants) and separate sessions will be offered by gender and age category. Additionally, attendees will be made aware that cash transfers will not be stopped upon registration for Forsa, they will be sustained for 6 months to allow a smooth transition into self-reliance. Program registration will be opened for the participants at the end of the session to express their decision about joining Forsa.

Profiling and final selection:
The roll-out of behavioral change sessions will take several months and will be followed by a period of profiling. Households that registered their interest at the behavioral change session will be interviewed to gather information about their work experience and preferences and based on this profile, 50,000 individuals will be selected by MoSS for inclusion in the Forsa program pilot.

Intervention services:
For the self-employment track: the service will be conducted by the contracted NGOs under the supervision of MoSS program staff. During the session, the NGOs staff will present to the beneficiaries the list of assets that could be transferred and the general and asset-specific trainings that will be offered such as financial literacy training, business development and entrepreneurship skills training, asset-specific/technical training and regular coaching. The asset costs and the associated trainings are valued at US$1000 per beneficiary.
FORSA participants will be given the option to either choose the asset on the spot, or to submit their choice after 2 to 3 days after discussing the options with other household members. In some cases, another physical verification visit may be prearranged to assess the suitability of the household for asset keeping and considering environmental factors.
Beneficiaries will also benefit from an ongoing mentoring relationship with a volunteer participating in a national service program.
For the wage employment track: the service will be conducted at the village level, through a Training and an Employment unit at affiliated premises to MoSS, or/and through collaborations with other partners when needed, such as the Job Search Clubs, established by the ILO in collaboration with Ministry of Youth, where sessions can take place to inform beneficiaries about the package components/services offered in the wage-employment modality and the expected timeline for delivering these components. The wage employment modality offers several components/services that fall under two main categories: 1) availing employment opportunities for unskilled, semi-Skilled and skilled labor in the formal private sector; 2) creation of new jobs categories such as assistant nurse, elderly care, home cleaning and babysitting services.
The wage employment package includes technical and soft skills trainings for employment; a benefit package offered in collaboration with the employer through which MoSS finances wage subsidies for up to six months (to be determined), health insurance, social security coverage, annual and sick leave, and life insurance; and a job-matching and mentoring program which is considerably more involved than a one-time support for the beneficiaries as it includes resolving issues and potentially finding a new match if the beneficiary loses or wants to leave their employment during the program period.
 

2021-10-01

2023-12-31

Both modalities:
1. Total Household Consumption
Employment modality only:
1. Unemployment rate including all household members ages 15-64
2. Underemployment rate for all household members ages 15-64
Asset transfer modality only:
1. Continued ownership of transferred assets
2. Weekly time spent by targeted member in productive activities

Consumption is based on 7 days recall for food items and 30 days and past year recall for non-food items. Items in the consumption survey were selected based on highest contributions to household budgets according to past survey experience for the Takaful evaluation with a similar target population in Egypt.

Secondary Outcomes (endpoint):
Both modalities:
1. Total Household Food Consumption by broad category
2. Total Household Non-Food Consumption by broad category
3. Implied Consumption Share for Children, Men, and Women
4. Net Debt and Savings including informal saving and borrowing methods such as gamaiya and installment plans
5. Self-Efficacy of Targeted Member
6. Women’s Decision-Making
7. Subjective measure of life satisfaction
8. Financial literacy
9. Financial inclusion
Employment modality only:
1. Targeted member works in formal sector
2. Targeted member current wage
3. Targeted member monthly income variance over past 6 months
4. Targeted member has work certification
5. Targeted member valuation of formality of employment
Asset transfer modality only:
6. Total weekly time spent by all household members over age 5 in productive activities
7. Total weekly time spent by all household members over age 5 on child and elder care

We will explore heterogeneity of program impact on the following household and individual characteristics that we believe are likely to influence ability to benefit from the intervention. In each case, we will test for heterogeneity by interacting the treatment dummy with a demeaned baseline value of the characteristic.
1. PMT score
2. Risk aversion
3. Uncertainty aversion
4. Score based on self-reported literacy and numeracy
5. Digit span and reverse digit span scores
6. Stated preference at baseline between self-employment and wage-employment (possibly also interacting with rural and urban if these are strong predictors of wage-employment or self-employment modality being offered in a community)

Self-efficacy index will be an average based on responses to a series of questions based on standard scales but revised for cultural appropriateness based on feedback from local partners and field-testing.
Women’s decision-making outcome indicator will be further specified before analysis starts based on qualitative work that is still on-going.
Life satisfaction indicator will be the average response of likert scale questions on satisfaction with home, food, income, children’s clothing and selected questions from a standard happiness index.
Formal sector defined by whether has a written contract and is enrolled in social-security
Valuation of formality to be measured by difference between lowest monthly pay willing to accept for a formal sector job and lowest pay willing to accept for informal sector job
Financial literacy measured by the score recorded from the financial numeracy questions and number of “don’t know” responses to questions about cash flow.
Financial inclusion measured by the share of participants with a bank account or who have taken formal credit.

The study relies on a cluster randomized controlled trial to estimate the causal impacts of the offer of access to the Forsa program on outcomes, with two rounds of data collection. A baseline household survey will collect information on outcomes, and key characteristics expected to be predictive of future outcomes prior to the start of the Forsa program. An endline survey will re-survey the same households two years later to measure outcomes and characteristics that enable tests of different potential mechanisms in the causal chain.
The sample will include 7,920 households, 3,960 each from treatment and control villages with 24 households sampled per village.

Not available

The randomization will take place at the sub-village level, stratified at the district level and by the size of the population eligible to participate in the Forsa program, using randtreat Stata package, with a seed of 1142021 and using the wglobal option for misfits. 165 villages were allocated to the treatment arm and another 165 villages were allocated to the control arm.

sub-village

Yes

330

7,920

165 subvillages control (3,960 hhs)
165 subvillage treatment (3,960 hhs)

We conducted power calculations to ensure that the study sample size would enable us to identify impacts on household expenditure per capita at least as strong as those from the Takaful program itself (8%) and the likelihood of employment for the targeted individual. To do so, we used data on Takaful beneficiaries to calculate the mean and standard deviation of household expenditure per capita and the share of employment among beneficiaries. We additionally calculated the intracluster correlation for both outcomes at the sub-village level. We assumed a baseline correlation between observable characteristics (including the lagged outcome) and the endline values of the outcomes of 0.3 for consumption—based on past research that collected multiple rounds of household consumption data in low or middle-income countries (see McKenzie 2012)—and 0.1 for employment. Based on fieldwork budget and feasibility constraints, it was determined that we would be able to include approximately 25 observations per sub-village at baseline and that there were 322 sub-villages with enough Takaful beneficiaries to be attractive program locations for the implementing organization; we opted to assume that these would be split evenly across the two treatment modalities and we assumed that one household per study cluster would attrit during the study period. At endline, this leaves us with 10 observations per cluster, per treatment modality, on average. A complicating factor for the impact evaluation, discussed at length above, was how to select the sample in treatment and control arms to maximize the expected difference in program take-up across arms using an identical sampling method in both arms. We do so by using administrative data on all Takaful beneficiaries in study sub-villages together with the responses to a short screening questionnaire of 125 Takaful beneficiary households to predict program take-up; including the 20 households in each village with the highest predicted probability of program participation, and randomly sampling 4 additional households per sub-village from the rest of the probability distribution. To account for this sampling procedure in the power calculations, we assumed that the probability of Forsa participation among Takaful beneficiaries was drawn from a normal distribution with mean 0.2 and standard deviation 0.23. We simulated the distribution of the probability of program take-up for 125 observations in each of the 161 treatment communities, retained the 20 highest probabilities in each village, and then calculated the mean take-up rate as the expected probability of participation across all 161 treatment sub-villages. The resulting power calculations give us estimates of the minimum detectable effect size for the consumption per capita and employment outcomes, under the assumptions outlined above. We expect to be able to identify a difference in consumption per capita of 102.6 Egyptian Pounds per capita, or around a 13.7% change in consumption per capita; similarly, we expect to be able to detect an increase in employment as small as 11.9%.