Strengthening Rural Food Systems through the UN World Food Program’s Fresh Food Voucher in Ethiopia’s Somali Region: An Evaluation of Nutritional and Market Impacts

Objectives of the Study

The core objectives of the study were to:

• Assess the impact of innovative community-supported FFV projects in improving the nutritional status of children under five and women.
• Evaluate the effectiveness of Social Behavior Change Communication (SBCC) in improving the nutritional knowledge of mothers participating in the FFV project.
• Examine the role of the FFV project in strengthening the local markets in the Somali region.

Study Location and Design

A quasi-experimental design was used to conduct the study from April to October 2023 in three rural woredas (districts) (Degahabor, Gode, and Kebridehar) of the Somali region in Ethiopia. The study aimed to examine the pre- and post-intervention impact of electronic fresh food vouchers (FFVs) on preventing wasting in children under five and pregnant and lactating women, improving their nutritional knowledge, and strengthening local market functionality.

Eligibility Criteria

Households enrolled in General Food Distribution (GFD) assistance with children under 5 years and pregnant and lactating women (PLW) were considered eligible for the study.

Target and Source of Population

Following the GFD-registered households, a total of 6200 households were enrolled for the FFV interventions based on the criteria mentioned above, across three targeted woredas (districts) as shown in Table 1; the sample size for the proposed study was calculated from the list of registered households accordingly.

Intervention

In the Somali region, the World Food Program (WFP) FFV program “topped up” the food basket of 6200 vulnerable households hosting pregnant, breastfeeding, and children aged 6–59 months to reduce wasting and promote growth by supplementing and improving the diet quality of these individuals. In the FFV program, a digital voucher is sent monthly via Short Message Service (SMS) to the beneficiary household’s mobile phone. The SMS message includes a code used to purchase fruits, vegetables, and eggs from small retailers in rural markets. Both the beneficiary and the retailer are enrolled in the program and need a simple analogue mobile phone. The beneficiary’s household sends the SMS-based code to the retailer’s mobile phone to process payment for the commercial transaction. No internet connection is needed for the transfer.

General Food Distributions (GFD)

GFD is an effective food-for-work (FFW) activity used to cover food requirements for 3–9 months per year in communities suffering from unexpected shocks resulting in unanticipated food shortages. Free general rations are permitted for those who are physically unable to work or unable to rely upon family or community assistance, or where capacity to organize FFW does not exist.

WFP Fresh Food Voucher (FFV) Top Up

The FFV households were registered from GFD households based on the eligibility criteria mentioned above. Restricted cash-based transfers (CBTs) are granted to redeem FFVs from retailers in local markets. The food voucher included locally available and locally produced foods, considering the nutritional requirements of PLW and children aged 6–59 months. The voucher provides a fixed cash amount that recipients can use to purchase food items from a specified list: cabbage, green pepper, onion, tomato, egg, potato, carrot, mango, orange, and banana, etc. The beneficiaries received SBCC messages about MIYNC, antenatal and postnatal care, and dietary diversity.

Sample Size Calculation

The sample size was calculated using the wasting prevalence from the recent national demographic survey,^[9] which reported 21% wasting in children under 5 years. We determined the sample size using OpenEpi,^[10] based on the parameters below. We adopted a 7% change in nutrition outcomes as the intervention effect size from the Research on Food Assistance for Nutritional Impact (REFANI) project,^[11] which examined the impact of food assistance (including cash, vouchers, and direct food transfers) on nutrition outcomes in crisis-affected contexts in Niger, Pakistan, and Somalia.

The sample size was calculated to detect a 7% effect size with 80% statistical power at a 5% level of significance (α = 0.05), assuming an exposed-to-unexposed ratio of 1:1. To account for potential non-response or loss to follow-up, a 10% adjustment was applied. Based on these assumptions, we determined a total sample size of 717 participants and allocated it across three woredas [Table 2].

Sampling Procedure

The sample was drawn from the FFV-registered households. The households were sampled at the kebele (village) level, meaning the number of samples was weighted by the proportion of beneficiaries in each kebele using the probability proportional to size (PPS) approach, and households were then randomly selected from the sampled kebele using systematic random sampling. The researchers interviewed beneficiaries from the randomly selected households.

Sampling Flow

The selected woredas/districts constituted the primary geographical sampling units, from which eligible kebeles/villages were identified. Within the selected kebeles/villages, households targeted for General Food Distribution (GFD) were identified as the sampling frame. Among these households, those targeted for the Fresh Food Voucher (FFV) intervention were included in the study. FFV-targeted households were subsequently classified into three mutually defined beneficiary categories based on household composition i.e., households with children under five years of age (CU5), households with pregnant and breastfeeding women (PBW) and households with both CU5 and PBW.

Household Interview

Structured questionnaires programmed in MODA were used. Enumerators and local guides explained the objective of the study to the head of the selected household. They requested oral consent to conduct interviews and anthropometric measurements (Mid-Upper Arm Circumference (MUAC), edema, weight, and height/length).

In each household, we targeted children under 5 years and their mothers/caregivers for anthropometric measurements. Interviews were conducted only with the household head and spouse. If the household head and/or spouse were not available during the visit, an appointment was made for when they would be available to conduct the interviews.

Anthropometry

All anthropometric measurements followed the WHO-UNICEF 2016 guidelines.^[11] MUAC (cm) for both mother and child, as well as weight (kg) and height/length (cm) and nutritional bilateral edema for children, were measured to calculate nutritional indices. Measurements followed WHO/UNICEF and national guidelines.^[8]

Infield Data Quality

The anthropometric data were collected through ODK and hosted on MODA servers; the WHO-approved Z-score validation media files ensure anthropometric data quality at the enumerator/household level. Z-score validation prevents enumerators from continuing with incorrect measurements and prompts repeat measurements, improving data quality assurance and reducing data flags; it also supports anthropometric data and plausibility checks in Emergency Nutrition Assessment (ENA)/SMART^[12] software.

Data Quality Management

Data collectors checked for any inconsistencies and incomplete responses before leaving the household. In addition, supervisors followed up with data collectors to check data quality. The completeness, consistency, and appropriateness of the data were also checked as soon as it was uploaded to the central database of the online data collection system. Outliers were excluded if they could not be edited during analysis. Plausibility checks were conducted for anthropometric data using ENA for SMART software.^[13] Moreover, pre-tests and feedback sessions during training and data collection minimized potential errors.

Sensitization

Kebele (village) and woreda (district) government officials in the project areas were sensitized on the data collection process and protocols. Furthermore, the selected respondents were also briefed on the purpose of data collection and the confidentiality of the information they provided. Respondent-level sensitization was led by the enumerators as it was related to data collection.

Data Analysis

Statistical analyses were conducted using MS Excel, ENA for SMART, and SPSS. Household demographic data were analyzed in MS Excel, anthropometric data in ENA for SMART, and nutrition and food security indicators in SPSS. Descriptive statistics were used to summarize the data and examine variations in mean values, with statistical significance assessed at the 5% level (α = 0.05).

Ethical Clearance and Consent

The study was conducted as part of the program implementation. Therefore, no ethical clearance was required. However, study respondents were briefed on the purpose of the study and confidentiality of information. Interviews were conducted only with respondents’ consent after obtaining electronic consent.

Households’ Demography

The physiological status of women in the study population shifted over the study period. From baseline to endline, the proportion of pregnant women decreased from 27% to 22%, and the proportion of lactating women decreased from 48% to 35%. Conversely, the proportion of women who were neither pregnant nor lactating increased from 25% to 43%, indicating a natural progression through the reproductive cycle within the study population [Table 3]. The distribution of pregnant women across trimesters showed variation from baseline to endline. The proportion in the first and second trimesters declined from 21.8% to 17.1% and from 50.0% to 44.5%, respectively, while those in the third trimester increased from 28.2% to 38.4%. Additionally, the percentage of primigravida women significantly decreased from 8.0% at baseline to 3.1% at endline. The proportion of women breastfeeding during pregnancy declined from 5.0% at baseline to 1.7% at endline, suggesting improved child spacing practices or increased awareness and adoption of maternal and child health recommendations that discourage concurrent breastfeeding and pregnancy. The proportion of women breastfeeding during pregnancy declined from 5.0% at baseline to 1.7% at endline, indicating increased adherence to maternal and child health recommendations.

Nutrition and Infant Feeding

The study found a significant (p = 0.001) improvement in maternal awareness of key infant and young child feeding (IYCF) practices and maternal nutrition, reflecting enhanced reach and effectiveness of social behavior change interventions [Table 5].

Awareness of the recommended duration for breastfeeding (24 months and above) increased significantly from 28% at baseline to 44% at endline. This improvement is critical, as sustained breastfeeding is linked to lower risks of infection, improved cognitive outcomes, and better nutritional status in early childhood.^[28] Knowledge gains were also observed in complementary feeding practices, though improvements varied by food type. The proportion of mothers aware that animal-source foods should be introduced at 6 months increased from 10% to 21%, while awareness of the appropriate age to introduce fruits and vegetables increased significantly from 1% to 16% and 17%, respectively.

Awareness regarding maternal nutrition also improved. The proportion of mothers who understood the need to eat more frequently during pregnancy increased from 37% to 51%, and during lactation from 59% to 74%. These findings suggest improved understanding of increased nutrient requirements during these physiologically demanding periods.^[20] Knowledge of appropriate feeding during child illness is a critical but often overlooked practice and increased from 34% to 42%. The proportion of mothers who reported receiving any nutrition message in the last month more than doubled, from 28% to 65%, reflecting strengthened outreach and communication channels. The main sources of nutrition information were home visits (increased from 40.3% to 53.5%) and health posts (increased from 49.5% to 74.8%). These channels played a critical role in delivering IYCF counseling and maternal nutrition education, in line with global recommendations for community-based nutrition interventions.^[29,30]

Accordingly, a significantly greater proportion of mothers reported receiving messages about the importance of giving colostrum, increasing from 57% at baseline to 65% at endline (p = 0.006). Similarly, awareness around early initiation of breastfeeding within one hour of birth improved from 35% to 64%, aligning with WHO recommendations for optimal neonatal health. Knowledge of exclusive breastfeeding (EBF) for the first 6 months also increased from 49% to 64%, reflecting strengthened promotion of foundational infant feeding behaviors. Colostrum feeding, early initiation, and EBF are critical for the child’s survival and are often emphasized in maternal counseling and community-based programs.^[31]

While the proportion of mothers receiving messages about initiating complementary feeding at 6 months remained unchanged at 36%, there was improvement in specific complementary feeding practices and diversity of complementary feeding practices [Table 5]. Furthermore, the proportion of mothers receiving messages about feeding sick children increased significantly from 7% to 27%, which is a positive change that may support improved recovery and nutritional resilience during childhood illnesses. The proportion of mothers exposed to messages on water and sanitation also increased from 10% to 52%. The improvement reflects better integration of WASH messaging into nutrition programs, a critical factor given the strong links between poor hygiene, enteric infections, and malnutrition.^[32]

Nutrition Perceptions and Practices

The perception that mothers should eat more frequently during pregnancy increased from 49% to 66%, nearing statistical significance (p = 0.06). Similarly, belief that mothers should eat more during lactation remained high, increasing from 78% to 83% [Table 6].

The proportion of pregnant women who consumed an extra meal during pregnancy increased significantly from 19% to 61%, and consuming two extra meals daily during lactation increased from 33% to 74% (p = 0.004). Adequate maternal nutrition during these periods is associated with better birth outcomes and improved lactation performance.^[33,34] Furthermore, the proportion of mothers feeding thick porridge to children increased from 47% to 59% (p = 0.004). Mothers also increased the use of different foods to diversify their children’s complementary foods. Enriching porridge with eggs, meat, or greens introduces essential nutrients like protein, iron, and vitamin A, which are often lacking in children’s diets in low-resource settings.^[28]

Health and Hygiene Practices

The study revealed significant improvements in the utilization of antenatal care (ANC) services and hygiene practices among mothers in the target population. The proportion of women who had ever attended ANC during pregnancy increased significantly from 28% at baseline to 69% at endline (p = 0.01), indicating a major improvement in awareness and utilization of maternal health services. However, for current or most recent pregnancies, ANC attendance increased from 32% to 43% [Table 6]. The proportion of women initiating their first ANC visit at the appropriate time (first trimester) increased significantly from 5% to 29%, suggesting improved understanding of the importance of early maternal care.

Despite improvements in service attendance, the proportion of mothers receiving nutrition counseling during ANC increased from 28% to 40%, and use of iron and folic acid (IFA) tablets increased from 29% to 47%. For hygiene, the proportion of mothers who regularly washed their hands before eating and after using the toilet increased from 69% to 84%, reflecting improved health awareness and behavior. Such hygiene practices are essential for reducing exposure to pathogens that contribute to diarrhea and undernutrition in children.^[32]

Program Snapshot: Dietary Diversity, Market Functionality and Fresh Food Voucher Interventions

Before implementing the FFV program in the woredas of Gode, Kebridehar, and Degahabor districts of the Somali Region in Ethiopia, access to and consumption of eggs, fruits, and assorted vegetables were critically low. Local markets rarely stocked these nutritious food items due to low consumer demand, which stemmed from limited awareness and traditional dietary habits and presented a significant barrier to improving household dietary diversity and nutrition, particularly for vulnerable populations such as pregnant and lactating women and young children.

To address these challenges, the WFP and its implementing partner introduced a multi-component FFV intervention that combined market engagement with community-level SBCC strategies.

• Retailer engagement: Retailers in local markets were initially reluctant to stock eggs, fruits, and vegetables, fearing they would not sell. WFP and partners sensitized and encouraged these retailers to diversify their stock in anticipation of increased demand from FFV beneficiaries.
• SBCC campaigns: In parallel, the FFV team collaborated with woreda health offices to roll out mass SBCC activities aimed at increasing awareness and demand for nutritious foods. These campaigns highlighted the nutritional benefits of egg consumption and dietary diversity through community mobilization and interactive sessions.
• Capacity building: To support sustained behavior change, health extension workers and community volunteers were trained on the benefits of diversified diets and demonstration-based cooking practices. They were equipped with SBCC tools such as flip charts, counseling cards, posters, and leaflets to support community outreach and nutrition education.

Outcomes and Observed Changes

Monitoring observed a trickle-down effect on protein-rich (eggs) consumption in Gode, Kebridehar, and Degahabor districts, spurring the growth of poultry farming. In addition, rural households near urban centers began keeping small flocks of poultry to supplement their incomes. A local poultry farm owner noted, “Chicken meat and eggs are new to Kebridehar residents, but my customers are adopting these foods.” Similarly, a participant in the FFV program, Surer Ahmed, shared that she initially didn’t know how to cook eggs. After attending a cooking demonstration, she learned how to cook eggs properly and diversify her family’s diet. WFP enhanced its retailer engagement strategies by strengthening capacity, monitoring food quality, diversity, and price, and addressing the concerns of people in need. The local retailers (80% female) were enrolled in the FFV program, stimulating local markets and increasing the availability of nutrient-dense foods in rural areas. Female retailers were encouraged to participate to build their business and economic capacity while encouraging healthier consumption within households.

The case study demonstrates the effectiveness of a combined approach that leverages SBCC, economic access, and market engagement to improve food availability and dietary diversity. By transforming both demand and supply systems, the intervention catalyzed a shift in nutrition practices, empowered women economically, and encouraged localized poultry farming. The model offers a scalable solution for improving food and nutrition security in other low-resource, market-constrained environments.