ICI develops model to predict child labour risk at the household level
According to the most recent estimates from the NORC Report on child labour in cocoa there are 1.56 million children in child labour in Côte d’Ivoire and Ghana. This underlines the need to scale up effective systems to prevent and address child labour across the cocoa supply chain. The International Cocoa Initiative’s new 2021-2026 strategy, backed by all of our industry and civil society partners, is designed to support the sector do this, particularly through the design and refinement of more innovative approaches. ICI has recently developed a new model that predicts a household’s risk of child labour in cocoa-growing communities in Ghana. While the model has been designed using data on risk factors specific to Ghana, the same steps can be followed to make child labour risk models for other contexts, such as Côte d’Ivoire. It is the first model of its kind to be implemented in the cocoa sector and could help make child protection systems more efficient.
How could the risk model help scale up work to prevent and address child labour?
The household risk model could be used in any situation where up to date information exists about farmers and their households in a supply chain. Cooperatives often keep farmer registers with this information about their producers, while local Child Protection Committees can hold similar records about members of a community.
Child Labour Monitoring and Remediation Systems (CLMRS) are an increasingly common type of child protection system in the West African context, in which the use of risk models could be advantageous. Currently, these systems work by sending community-based agents to visit every cocoa-producing household in a supply chain, with the aim of identifying children involved in hazardous work and providing support tailored to their needs. While CLMRS have been shown to reduce child labour by up to half among identified children, the census-style approach involves many repeated household visits and is both time- and labour-intensive.
By applying a risk model, existing farmer records can be used to identify a subset of “higher-risk” households to prioritise monitoring visits. By doing this, ICI’s analysis estimates that the risk model could reduce monitoring costs by 20%, while still identifying the same number of children in child labour. Savings generated could be used to further increase the coverage of such systems, or to provide additional support to vulnerable households.
“Our goal is 100% coverage of the cocoa supply chains in Cote d’Ivoire and Ghana with child protection systems, such as CLMRS. To achieve this ambitious target, we need to see a massive expansion of effort as, today, only around one-fifth of cocoa households are covered. This is a huge undertaking, and we believe this risk model, with its ability to improve operational efficiency, could help us get there, also enabling resources to be directed where they are needed most, faster,” said Nick Weatherill, ICI’s Executive Director.
An important step in using a risk-based approach is to define what constitutes “higher risk” to ensure children who most need it are receiving direct support, and considering other types of prevention and assistance activities for children not prioritized for targeted support.
How does the model work?
The data-driven model uses several factors to predict child labour: information about the head of household, such as their level of education and age; about children themselves; the household’s access to basic services, such as water and electricity; and information about farming practices, such as the size of land under cultivation, crops grown, and the use of fertiliser and pesticides.
ICI found that two of the most important predictors of risk were children’s age and gender. Without this information, the initial model was unable to accurately predict hazardous and non-hazardous child labour. Cooperatives and others wishing to use a risk-based approach to predict child labour will therefore need to be strongly encouraged to collect demographic information on the age and gender of children living in cocoa-producing households.
The risk model produced in the context of this project is a first step in a longer process. ICI will continue to work on refining the model, using more up to date information and datasets, as well as building similar models for other contexts, including Côte d’Ivoire. Together with our partners, we are developing and testing new ways of integrating risk-modelling into different operational approaches: constantly working to improve the efficiency of child protection systems, while making sure that vulnerable children and their families receive timely support.
Read the full report here.