Nutrition-Sensitive Food Distribution Amidst Inflationary Shock: Evidence from a Randomized Intervention in Egypt

The Egyptian Food Bank (EFB), an impact-driven non-governmental organization (NGO) established in 2006 to address food insecurity in Egypt, is redesigning its General Feeding Program (GFP). The GFP currently serves about 130,000 households across Egypt with access to monthly food packages.

This evaluation focuses on a new food package designed to meet the individual micro and macro-nutrients including carbohydrates, fats, proteins, vitamins and minerals. The new food items are expected to increase the monthly iron dietary reference intake (DRI) by 16.5 percent per family unit, as opposed to the old package that only met 5 percent of the family’s monthly iron DRI. Egypt’s malnutrition challenges are closely intertwined with poor dietary habits, lifestyle, and lack of nutritional awareness (e.g., Ecker et al., 2016; MoH, NNI and UNICEF, 2017). This justifies the need to address unhealthy dietary habits and practices to unlock the potential of achieving a healthy life (MoH, NNI and UNICEF, 2017). To address these challenges the EFB is complementing the GFP with a nutrition awareness campaign that aims at promoting and enhancing healthy eating habits and food preparation methods. Beneficiaries will receive printed materials, SMS messages, and voice messages, covering 12 key messages about nutrition and food safety over the course of one year.

2022-11-01

2023-02-28

Primary outcomes:
1. Household Dietary Diversity Score
2. Household level food insecurity
3. Nutrient intake for the female head of household
4. Amount consumed of milk, meat, legumes, and dried fruit in the past week

Primary outcomes explanation:
1. Household Dietary Diversity Score, which will be constructed based on 7-day recall following the guidelines by Food and Agriculture Organization (FAO).
2. Household level food insecurity. This will be constructed using the Food Insecurity Experience Scale (FIES), a metric developed by the Food and Agriculture Organization (FAO, 2014a). Using this module and the procedures described by FAO (2014b), we will generate binary indicators of moderate and severe food insecurity.
3. Nutrient intake for the female head of household. Respondents are asked to report all foods consumed in the last 24-hours along with all ingredients and quantity. The reported quantity will be used to construct nutrient intake using food tables developed by the National Nutrition Institute (NNI) of Egypt.

Nutrition Knowledge, Attitudes and Practices (KAP).

Secondary outcomes explanation
Nutrition Knowledge, Attitudes and Practices (KAP). Various outcomes will be defined based on questions eliciting respondent’s knowledge, attitudes and practices about the specific topics on which nutritional messaging is being shared, including:
• How to avoid anemia
• The importance of fruits and vegetables and proteins in diet
• Food safety
• Hypothetical willingness to pay for the new box
• Hypothetical spending priorities on food for an equivalent amount of cash

Experimental design:
The intervention will follow a clustered randomized approach. The randomization takes place at the village level while selection of eligible households in each village/community is performed by community-based organizations operating in each community. Communities will be randomly assigned into five groups/arms, based on the type of treatment and benefit they receive. These include:
• T1: New Boxes with nutrition messaging
• T2: New Boxes without nutrition messaging
• T3: Old Boxes without nutrition messaging
• T4: Nutrition messaging without any box
• T5: No Boxes and no nutrition messaging
To exploit power gains from repeated observations, we will collect three rounds of data, baseline, midline and endline data (e.g., McKenzie, 2012). A baseline household survey will be collected before the launch of the redesigned food packages and will collect information on pre-intervention outcomes, and key characteristics expected to generate potential heterogeneities in the impact of the program. A midline survey will be collected three to four months after the launch of the redesigned GFP and the above interventions while the randomization between receiving the old box, new box, and no box is maintained. After the midline survey, all participants will receive the new box and the treatment arms become:
• T1: New Boxes with nutrition messaging
• T2: New Boxes without nutrition messaging
• T3: New Boxes without nutrition messaging
• T4: New Boxes with nutrition messaging
• T5: New Boxes without nutrition messaging
An endline survey will re-survey the same households 6-9 months later after the midline to measure impacts of the full year of nutritional messaging. Given that our sample households are EFB beneficiaries or potential beneficiaries, we anticipate that attrition rate will be low or at least not systematic.

The randomization was done at the level of villages using a complete listing from the census ordered randomly using a random number generator in Excel after stratifying by urban vs. rural and governorate. Communities with no partner community-based organization for the Egyptian Food Bank or those with very few potential beneficiaries (as identified by the CBOs) are excluded, after which the first 50 listed communities are included in each treatment arm.

Village

Yes

250 clusters/villages:

5,000 of which at least 4,000 will be confirmed as eligible for the program

Treatment 1 : 50 villages
Treatment 2 : 50 villages
Treatment 3: 50 villages
Treatment 4: 50 villages
Treatment 5 : 50 villages

Based on a conservative assumption that 16 confirmed eligible households are reached per community, the MDE for household dietary diversity score (HDDS) for our sample size in each round between any two treatment arms is 0.22 standard deviations or 0.33 food groups (relative to a mean of 9.5 based on the summary statistics found in a recent evaluation in the same context). This seems appropriate given similar effect sizes found in Ecuador and Senegal. Hidrobo et al. (2014) find that the impact of the food vouchers in Ecuador was 0.61 food groups (average HDDS in control of 9.1) for a monthly voucher or food basket worth $40. Another similar food voucher program in Senegal is found to increase HDDS by 0.3 food groups (Savy et al., 2020). While expanding the sample size beyond this would reach smaller MDE, it is not clear that it would be worth discovering such a small impact as EFB expects the boxes to provide households directly with food groups that otherwise would not be consumed