ICI launches new analysis of risk models to predict child labour
In a new study, the International Cocoa Initiative (ICI) analyses several risk models designed to predict the risk of child labour in cocoa-growing households. The study draws on case studies from Côte d’Ivoire and Ghana which show that it is possible to create highly effective models, which could reduce the time and cost of identifying children in child labour and help support to be targeted where it is needed most. However, up-to-date and good quality data from households is essential to use predictive models and remains a challenge.
Child labour risk models use existing information about households (such as the size of the farm and the age of children) to predict the likelihood of child labour. Accurate risk models could be used to identify vulnerable households and children more quickly and at a lower cost, enabling assistance to be targeted more efficiently to those who need it. This type of innovative approach could support the scale-up of effective interventions to prevent and address child labour.
ICI’s recent analysis shares key lessons learned and recommendations about when and how to develop risk models to predict child labour among households in cocoa-growing areas of West Africa. It draws on real-life examples shared by different cocoa-sector stakeholders who have been testing risk models to predict child labour. The learnings and recommendations are relevant to anyone wanting to learn about what child labour risk models are, why they could be useful, when to consider developing a risk-model (and when not to) and how to go about it.
Key findings from the study include:
- Accurate models can be developed to predict child labour using various methods and based on different indicators.
- High quality data on farming households is crucial to develop and use models to predict child labour. But in practice, available data about farming households is often outdated, of insufficient quality, or does not include the right information to develop or use a model.
- Information on the age and sex of children improves the ability of a risk model to accurately predict child labour.
- There is no “one-size-fits-all” model to predict child labour at household level – each model should be tailored to the context of its use.
Accurately targeting households at risk
Risk models can be used in a variety of ways to tackle child labour. By identifying households with a higher likelihood of using child labour, they can reduce the time and cost of identifying vulnerable children through traditional means (such as visiting each and every household to do a child labour survey). The most successful risk models can correctly predict child labour in more than 95% of cases. In practice, this means that fewer household visits are required to identify children at-risk. As a result, support and preventative measures can be targeted where they are needed most, more quickly.
Models can also be used to expand the number of children receiving support beyond identified cases of child labour. One of the cases studies in this report shows how a risk model was used in this way to identify vulnerable children who were not in child labour, but who were at risk of it, and to provide them with preventative support.
Though they can be very accurate, no risk model can perfectly predict every child at risk. It is therefore essential that data is kept up to date and that recurrent monitoring is conducted, to identify children whose risk status changes over time.
Quality data is key
Most cooperatives gather information on their members, their households and their farms. These records are kept for commercial purposes, as part of certification processes, and in the context of sustainability programmes. This existing information could potentially be used to run risk models to predict child labour. However, in practice, this data often has gaps or is outdated. ICI’s analysis underlines that it is crucial to have complete and accurate data first: without quality data, a risk model cannot be used to predict child labour. Improving the quality of available data will not only help prepare for the use risk models in future, but bring immediate benefits for certification, sustainability, and commercial purposes. Technical expertise is also required to manage data and to carry out the statistical analysis required to create a risk model.
“This study shows that it is possible to develop accurate risk models to predict child labour in cocoa” said Megan Passey, Head of Knowledge and Learning at ICI. “The recent NORC survey shows that around 60% of cocoa households in Côte d’Ivoire have at least one case of child labour, but that leaves 40% that do not. Risk models could help us identify the vulnerable households more efficiently – helping us to scale-up the provision of support, by ensuring resources are targeted towards the families who need it most.”
“The case studies examined in this report show that considerable progress has been made in developing effective risk models to predict child labour. While this is encouraging, the study also shows that there are some challenges, particularly that the sort of good quality data needed to use predictive models is often not available. Before we can use risk-based approaches at scale, we need to improve the availability of accurate data on all cocoa-producing households. Ongoing initiatives by the governments of Côte d’Ivoire and Ghana to develop and maintain national registers of farmers are very promising. Efforts are also needed to support farmer groups to own and manage the data on their members, and to encourage cooperation and data sharing between suppliers, buyers and authorities, so that these national databases stay up-to-date.”
ICI is currently supporting stakeholders in the cocoa sector to develop tools and training material to pave the way towards the creation of accurate risk models. If you are interested in learning more about risk models and their application, please contact email@example.com.
Cover image: awareness-raising session on child labour in Didoko, Côte d’Ivoire, ©Nestle, 2019.