What’s wrong with the Global Estimates on Child Labour?
Big numbers make headlines, but they must also be treated with extreme caution
It is difficult to overstate how important the Global Estimates on Child Labour are for the International Labour Organization’s (ILO) campaign against child labour. For more than two decades they have stood at the centre of both mainstream discourse and the global policy agenda on this topic. It’s easy to understand why: the headline number is enormous. The declaration that 160 million, or around one in 10, children were working in 2020 is a powerful tool for mobilising political will and resources. But the ways in which the global estimates are acquired, presented and instrumentalised unfortunately mislead at least as much as they enlighten. The picture they present is warped. Here’s why.
The allure of outrage
The first publicly promoted estimates of child labour, produced by the ILO in 1996, were substantially higher than they are today. Part of the report ‘Targeting the Intolerable’, they put the figure at a quarter billion. They were also produced for a purpose, namely to help provide a rationale for the adoption of the 1999 Worst Forms of Child Labour Convention (C182). C182 qualifies practices such as child slavery, bondage, trafficking, soldiering, and prostitution as the worst forms of child labour and prioritises action against them.
To bring this ‘dry’ number of 250 million to life, campaigners also released shocking images and narratives of children in these kinds of work. The result was a hugely effective media campaign. As Frans Röselaers, the director of the ILO’s International Programme on Child Labour (IPEC) at the time, noted:
The number drew international attention to the magnitude and scope of the child labour problem worldwide. It was widely publicised; hardly any article on child labour failed to mention it.
The strategy hasn’t changed since. The ILO publishes new numbers approximately every four years: 211 million in 2002, a number which gradually dropped to 152 million in 2017, before rising again to 160 million in 2021. It has also doubled down on its global advocacy campaign. Kailash Satyarthi of the Global March Against Child Labour (now a Nobel Peace Prize laureate) was recruited to be the public face and voice of this campaign. His focus on child slavery, servitude, and sexual abuse has been instrumental in shaping how the public now understands and speaks about child labour. For instance, during his Nobel Prize acceptance speech, he spoke of child labour in the following terms:
Twenty years ago, in the foothills of the Himalayas, I met a small, skinny child labourer. He asked me: “Is the world so poor that it cannot give me a toy and a book, instead of forcing me to take a gun or a tool?” I met with a Sudanese child-soldier he was kidnapped by an extremist militia. As his first training lesson, he was forced to kill his friends and family. He asked me: “What is my fault?” Twelve years ago, a child-mother from the streets of Colombia – trafficked, raped, enslaved – asked me this: “I have never had a dream. Can my child have one?”
These are the kind of situations the ILO and other campaigners against child labour want the public to think of when we try to wrap our heads around the mind-boggling global estimates, be it 250 or 160 million children.
Is the egregious also representative?
Seemingly clear-cut statistics as well as heart-wrenching vignettes both serve the same function: they collapse a huge diversity of experiences into a few quick takeaways. This makes the story being told around child labour easy to digest, but it also makes it severely incomplete. If we allow that diversity back in, we see that the situation many ‘child labourers’ face is more complex and often much less dramatic than the ILO makes it out to be.
The data underpinning the global estimates is based on national household surveys that question families about the work children do during a particular reference week (usually the week before the survey). Children are considered to be in child labour when:
- They are aged 5-11 years and have worked for one hour or more in any form of work except for unpaid household activities.
- They are aged 12-14 years and have worked for 14 hours or more, including after school or during holidays.
- They are aged 12-17 years and have worked for one hour or more in predefined hazardous industries or hazardous occupations (e.g., mining, quarrying, construction).
- They are 15-17 years and have worked 43 hours or more per week.
This means that an 11-year-old child that goes to school full-time but has helped their parents at the market or in the fields after school or on the weekend for an hour or two, in that reference week, is considered to be one of the 160 million child labourers whose work needs to be eradicated. Needless to say, this is far removed from the images and stories of children stuck in slave-like conditions that accompany the global estimates in media and advocacy campaigns.
Ironically, the global estimates do not actually provide us with any data on child slavery, bondage, trafficking, soldiering, or prostitution. These practices go beyond the scope of the information that can be collected through standard household surveys. In short, the global estimates are instrumentalised to forward a narrow representation of child labour that shocks and ‘sells’, but that representation isn’t found in the data being gathered. On the contrary, the numbers aren’t even attempting to capture it.
The ILO has fully exploited the dramatic effect of pairing huge overall numbers with a narrow representation of working children as victims of modern slavery.
A second major problem is that the numbers themselves are highly unreliable. Numerous studies have shown that much of the national data underpinning the global estimates is biased and inaccurate. This has much to do with who we ask, how we ask, and when we ask about child labour.
Who we ask has shown to be crucial. While ideally the ILO wants children themselves to be asked, in practice this almost never happens. Parents or other household members are the ones answering questions about the work children do. A study conducted in Tanzania suggests that prevalence of child labour increases by 35% to 65% when children are asked themselves. Research in Ethiopia shows a gender dimension in the collection of data as well. The work of girls in agriculture is systematically underreported by adult male proxy respondents when compared to the reporting by girls themselves.
The same study shows that when households are asked about child labour greatly affects survey outcomes as well. Numbers vary 45% to 75% depending on the season (harvest or rainy) during which the survey is taken. As does how we ask. A study from the Ivory Coast shows that asking questions indirectly makes farmers feel less compelled to give a socially desirable response, and as a result the number reporting that they engage children actually doubles.
For all these reasons we need to remember that the global estimates are no more than that: estimates. And for that matter, they are estimates that are highly prone to bias and inaccuracy. In this sense, not much has changed since Francis Blanchard, the former director-general of the ILO, wrote in 1983:
Global figures purporting to demonstrate the extent of child labour are not very meaningful. They may have dramatic effects but they do not offer a basis for policy. In view of these reservations, I hesitate even to advance any figures. In themselves, they tell us nothing about the nature of the work children are doing or the circumstances and conditions under which it is being done.
The ILO has done exactly that which Blanchard had warned against ever since it tasted the success of its 1996 global estimates. It has fully exploited the dramatic effect of pairing huge overall numbers with a narrow representation of working children as victims of modern slavery and other worst forms. The result is a loss of all shades of grey, as well as the assumption that when we’re talking about a working child, we are in all likelihood talking about somebody who is living through their own worst nightmare. The data does not back this up.
Inadequate data, inadequate conclusions
Perhaps most worrying is the way that the estimates (and their steady decline) are used as evidence for the effectiveness of the ILO’s ‘abolition through legal prohibition’ approach. This is a prime example of how the estimates are treated as more than they really are. Remember: these numbers are highly speculative, and it is impossible to control for factors such as economic policies, educational programmes, environmental changes, etc. This makes them a poor measure of progress, as much as the policymakers pursuing Target 8.7 of the Sustainable Development Goals (SDGs), which requires UN member states to “end child labour in all its forms” by 2025, would like them to be.
To sum up, the global estimates are used: to sell a certain idea of child labour based on narrow representations of working children as victims of modern slavery; as proof that the abolition-through-legislation approach works; and as the prime indicator of progress towards the goal of eradicating all forms of child labour. In other words, they are a central component in the global advocacy and policy machines that affect millions of families and children worldwide, despite their unreliable nature and the fact that, to use Blanchard’s words again, “they tell us nothing about the nature of the work children are doing or the circumstances and conditions under which it is being done.”
Instead of spending its scarce resources on extrapolating and promoting global estimates to legitimise a global regime that revolves around an idealistic, unrealistic and potentially harmful goal, the ILO would be better off funding national programmes to improve the overall quality and nuance of the data they are working with. Only a full and balanced picture of what any particular country is dealing with will lead to realistic and relevant programmes adapted to local circumstances and the needs of working children.
Get our weekly email