Reducing child labour and improving education in cocoa-growing communities: experimental evidence on the impact of school kits from Ghana

Children in primary 5 and 6 in the treatment group will receive a comprehensive school kit at the beginning of the new school year in February 2023. Children in the control group will receive a school kit after the end of the experiment in February 2024 (i.e., post-endline data collection). School kits will contain the following items: one school uniform, one set of shoes, one school bag, 10 exercise books, five notebooks, 10 pens and one mathematical set at a value of 600 GHS/kit.

2023-01-15

2023-12-15

Work:

- Child labour (binary variable): child accomplished at least one hazardous activity or experienced one hazardous situation in the last 7 days, 6 months or 12 months. The list of hazardous activities or situations is defined by the national legislation in Ghana, and in line with relevant ILO conventions No. 138 and No. 182.
- Hazardous activity (binary variable): child accomplished at least one hazardous activity in the last 7 days, 6 months or 12 months.
- Number of hazardous activities accomplished by the child in the last 7 days, 6 months or 12 months.
- Hazardous situation (binary variable): child experienced at least one hazardous situation in the last 7 days, 6 months or 12 months.
- Number of hazardous situations experienced by the child in the last 7 days, 6 months or 12 months.
- Number of non-hazardous household tasks accomplished by the child in the last 7 days, 6 months or 12 months.

Education:

- Child school attendance
- Number of times the child missed school in the last school week.
- Number of times the child arrived late/left early in the last school week.
- Child learning: computed from literacy and numeracy questions/tests.

Well-being:

- Child well-being: index of questions on self-reported measures of children’s psychological well-being.

Injury:

- Injury (binary variable): the child was ill or injured in the last 7 days, 6 months or 12 months.

We apply a clustered RCT, where 60 primary schools are randomly assigned to a treatment group (receiving school kits) or a control group (receiving nothing), stratified at the regional level (Ashanti or Eastern). Half are selected from four districts in the Eastern region (Abwakwa South, Ayensuano, Suhum, West Akim) and the other half from two districts in the Ashanti region (Asante Akim Central and Asante Akim South). The primary schools are randomly selected from cocoa-growing communities in which, to our knowledge, no programmes are actively present to combat child labour and improve education. To limit spillovers and for ethical reasons, we restrict our sample to communities where only one primary school is present and treat all children in the selected class(es) irrespective of whether their families report receiving their main income from cacao farming. We will target children at the end of grades 4-5 (approximately aged 12 years old) at baseline who are, according to the literature, most at risk of transitioning into child labour and forgoing formal education to overcome financial challenges.

The research is conducted in three research phases. In Phase 1, we collect baseline data on children, their caregivers, and headteachers in November-December 2022, when children are more at risk of being involved in child labour. In Phase 2, we take the body and footwear measurements of children in the treatment group in January 2023 and distribute the school kits in February 2023, during the start of the new school year. In Phase 3, endline data is collected one year after the baseline in November-December 2023.

Not available

Randomisation done in office by a computer using Stata.

School

Yes

60 schools

1800 pupils + their respective caregivers (1800 caregivers) + 60 headteachers (one per school) = 3660 respondents.

30 schools (900 pupils, 900 caregivers, 30 headteachers) in the control group and 30 schools (900 pupils, 900 caregivers, 30 headteachers) in the treatment group.

The sample size is calculated for a power of 80% at a 5% significance level, for an intra-cluster correlation of 0.085 and an effect size of 10 percentage points reduction in child labour from a current level of 77% in cocoa communities in the Ashanti and Eastern regions for children aged at least nine years old in primary school in Ghana (defined as being in child labour in the last 12 months). The power calculations were done without any control variables that could potentially reduce heterogeneity, hence the minimum detectable effect size with the sample size.