Fostering human capital development through school feeding and teacher incentives

School feeding intervention - Schools assigned to the school feeding program receive resources from the WFP to provide meals to students five days a week. Each school food management committee is in charge of selecting the suppliers and preparing the daily meal for the students. The target is to provide children a meal with a daily average of 580 Kcal. The meals include five typical Gambian dishes. Following the WFP school procurement manual, the schools purchase food items from local communities. The WFP monitors the schools’ procurement practices to ensure they adhere to the manual. The first meals were provided to students in January 2022.
Teacher Incentives intervention - The teacher incentive program provides grade 3 teachers in selected schools with a bonus to perform a daily data collection task that requires them to be present at the school. Teachers are asked to keep daily records of which kids received a school meal (in schools receiving school feeding) or brought/bought breakfast (in schools that do not receive school feeding). A meal ledger was prepared for teachers to record this information in a consistent manner. Teachers assigned to receive the incentive are provided a monthly bonus equal to a maximum of 500 Dalasi, roughly 10 percent of their monthly salary, for filling the meal ledger; the payment is proportional to the number of days the ledger is filled conditional on having filled a minimum number of days.

2023-01-01

2023-07-31

First Stage Outcomes: child dietary diversity, child food insecurity experience scale, teacher attendance.
Second Stage Outcomes: child mental health/psychological wellbeing (life satisfaction, stress, depression, agency), child physical health (health status, nutritional status), child school attendance, child school progression (dropout, repetition, grade progression), learning (reading skills, numeracy and mathematics skills), child cognitive ability (attention span, verbal working memory, fluid intelligence).

We leverage the expansion of school meal coverage supported by the World Food Programme’s (WFP) in certain regions. Due to financial constraints, the program expansion was limited to 46 additional schools. We randomly assign 92 schools to a treatment or control group: in school feeding treatment schools, all students receive a warm nutritious lunch; students in control schools do not.
Additionally, we investigate supply-side constraints to school attendance and learning by cross-randomizing a teacher incentives arm designed to increase teacher attendance. We ask teachers to complete a daily meal ledger for the classroom and in return, offer a payment equivalent to 10% of their salary. The daily meal ledger indicates whether each student received a meal that day. To fill the ledger, teachers must be present on school, and multiple checks are put in place to ensure that the incentives are conditional on teacher attendance. Cross-randomizing this incentive arm with the school feeding arm allows us to investigate potential complementarity between demand (school-feeding) and supply (teacher attendance) side interventions.

Not available

Randomization done in office by a computer

Primary Schools (Clusters)

Yes

92 schools

2175 children

School feeding and teacher incentives: 23 schools, 608 students, 31 teachers
School feeding no teacher incentives: 23 schools, 539 students, 26 teachers
No school feeding and teacher incentives: 23 schools, 470 students, 30 teachers
No school feeding and no teacher incentives: 23 schools, 558 students, 27 teachers

We base our power calculations on five measures: numeracy test scores, household dietary diversity, household food expenditures, weight-for-age, and height-for-age. These measures were chosen for their similarity with the main outcomes of our analysis (data on our exact main outcomes are not available). Assuming 46 schools per treatment arm, 20 children per school, and power of the test of 80% yields a minimum detectable effect of 0.19 standard deviations for the standardized numeracy test scores and weight-for-age z-score, and of 0.18 standard deviations for the height-for-age z-score; these are outcomes for which we plan one follow-up measurement (Table 4). With five follow-up measurements for dietary diversity and food expenditures (Table 4), assuming a correlation of 0.2 between follow-up measurements, our experiment is powered to detect a 0.11 standard deviations or 0.24 units increase in dietary diversity (with mean of 8.41 and standard deviation of 2.14 in the control group) and a 0.11 standard deviations or 2.96 dollar increase in weekly household food expenditures (with mean of 41.43 dollars and standard deviation of 27.04 in the control group). Increasing the correlation between followup rounds to 0.5 yields a minimum detectable effect of 0.15 and 0.14 for dietary diversity and food expenditures respectively. Test scores data come from the 2011 wave of the Africa Program for Education Impact Evaluation (APEIE). Household dietary diversity and food expenditures come from the 2015 wave of the Gambia Integrated Household Survey on Consumption Expenditure and Poverty Level Assessment (IHS). Weight-for-age and height-for-age data come from the 2019-2020 wave of the Gambia Demographic and Health survey (DHS).