Bonded Child Labour in South Asia: Building the Evidence Base for DFID Programming and Policy Engagement

Shavana Haythornthwaite , Wendy Olsen

Research output: Book/ReportCommissioned reportpeer-review

609 Downloads (Pure)


Executive Summary

This report was commissioned by the Department for International Development and funded by UK Aid to provide evidence for DFID to design a business case for a new Asia regional programme on child labour. The Asia Regional Team (ART) had identified bonded child labour as a particular problem in South Asian countries, which this report confirms. Research for the report was guided by three central questions:

1. What are the current trends and status of bonded child labour in South Asia?
2. What are the main drivers of bonded child labour in different country contexts?
3. What types of interventions have been most effective in reducing bonded child labour?
The research drew on evidence from several South Asian countries including Afghanistan, Bangladesh, Bhutan, India, Myanmar, Nepal, Pakistan and Sri Lanka.

Key Conclusions

1. The prevalence of bonded child labour in South Asia is unknown, reflecting a research gap in the context of the sensitivity and in some cases illegality of this form of employment relation. Evidence on children working shows a downward trend overall in Sri Lanka and broadly speaking across India. We do not equate ‘children working’ with child labour, nor with bonded child labour, however. The data for other countries in South Asia has gaps, and in all countries this issue could bear further re-examination with a view to gauging change over time, notably in Bangladesh and Nepal where multiple cross-sectional datasets are available.

2. This report offers a definition of bonded child labour focused on the child’s entrapment. This definition places most bonded child labour within the boundary of child labour estimates. Those children under age 18 who are in a forced labour situation are also inherently entrapped, and by virtue of their age they cannot give valid consent, so child forced labourers also constitute bonded child labour. Our definition takes into account ambiguous factors, such as the situation of early married children, who are not in an employment relation per se, but can still be deemed under bondage, undertaking child labour in the form of work for the family. In this sense, the definition of bonded child labour put forward in this report differs from ones recommended by country governments in the region. The use of different definitions of child labour itself by diverse international actors (UNICEF, ILO, Anti-Slavery International, Human Rights Watch) indicates both the importance and the research gaps around this phenomenon. The report is focused mainly on bonded child labour itself.

3. The prevalence of recorded child labour is highest in Nepal at 37% of children age 5-14 years. The Afghanistan prevalence is also high at 29% of children age 5-14. In Afghanistan this figure is gender-differentiated, with a lower prevalence among girls of 24%. In Pakistan, a variety of regional studies show different rates, and no single national figure can be obtained. Pakistani data from 2014 on children aged 10-14 indicates a 10% rate of child labour among those aged 10-14. The rate of child labour in India is very differentiated, both over space and by social group, with a national estimate of 12% of children age 5-14 in 2005/6. Recent national child labour estimates have not been published for India from 2011/12 and later, although solid survey data exists on child workers, and this area offers good opportunities for further research. Bangladesh has a good national survey dated 2013, showing a 4% child labour rate among children age 5-14. Myanmar’s rate is much higher at 9%. The rate of child labour for children aged 5-14 in Bhutan is 3%. The rate in Sri Lanka is 3%, excepting the far north Jaffna region which is omitted from official figures.

4. There are opportunities to extend knowledge by re-using the existing secondary data. For example early child marriage may be evident in survey data, but has not been included in bonded child labour estimates. Early child marriage when children are below the age of consent (typically 18 years) leads to engagement in domestic work without pay for long hours, but the surveys where wives are interviewed typically use only ages 15 to 49 (Demographic and Health Surveys; National Family and Health Survey), thus omitting child wives. The ILO has its own data on Bangladesh, known as SIMPOC data, which has married women of a broader range of ages. Indian NSS data also allow for child marriage. These could be re-analysed taking into account the age of marriage. Another research gap would be to use such data to examine the inertia of the occupations of parents and children; a new research topic is when and where children embark on an occupation new to their family. If access were granted, then a wider statistical modelling exercise could bring together the occupational details for India and Bangladesh alongside those from Sri Lanka which are not currently public, and the available UNICEF microdata from other countries, to gauge the whole South Asian situation.

5. Effectively the ILO in 2017 did a prevalence measurement exercise for child labour and hazardous child labour on a world scale, imputing evidence to countries for which they had no data. For South Asia they imputed data for all countries except Bangladesh. The ILO procedure was consistent with their agreed 2008 approach to child labour which allows no paid work among children under the age of 12, allows only non-hazardous work lasting less than 14 hours per week among age 12-14 (this is considered a level not disturbing school attendance), and allows non-hazardous work among those adolescents age 15-17.

6. Looking at the prevalence and causes of bonded child labour country by country, differences of political economy were notable, such as civil war, refugee camps providing cheap mobile labour, and highly differentiated legal systems. The country evidence also brings to light many case studies showing which sectors have bonded labour.

7. A survey of the causes of child labour showed a mixture of economic, social and cultural causes in all cases. The economic causes include the economic demand for child labour as cheap labour in enterprises where labour is informally engaged. These are not necessarily informal enterprises overall; they often offer their outputs as supply on formal subcontracts. Further, aneconomic basis for child labour arises when inequality and poverty together lead parents to feel desperate to gain employment for one or more children. This syndrome of poverty and desperation labour supply is also associated with the child failing school, failing to attend, or being withdrawn from school. These children often do considerable household work. The poverty syndrome can be seen as implying that higher average incomes will trickle down to a reduced level of child labour. However, raising economic growth will not work as a solution when and where social exclusion and cultural marginalisation means that employers can bond child labourers with impunity. Recent growth in child labour in Nepal could for example have arisen in the context of the Nepal civil war, and previous waves of refugees who arrived in Nepal from Bhutan. Similarly, at present, a huge wave of refugee families coming into Bangladesh from Myanmar place children at risk of future bonded labour.

8. Social and cultural causes are not only structural determinants, but also contextual conditions for debates about bonded child labour. Thus a self-reflexive mode of analysis is useful, and reflexivity via dialogue among the stakeholders is a principle of research that improves the ethics of discussing bonded child labour. Given sensitivities around the definition of ethnicities, and the role of gender norms within and between ethnic groups, it is important that future research take up bonded child labour not merely as an economic or poverty-related phenomenon, but also as a cultural phenomenon with a complex history subject to many different currents of opinion in the present day.

9. Interventions fall into two main categories, legal and socio-economic: the two can coincide in function. Legal interventions include the international conventions and the national laws that work in tandem. The law is used to create a criminalization of the employer’s act of bonding a worker to them (in India) or a landlord bonding children to themselves (e.g. in Afghanistan). Other countries are now moving toward having laws against the abuse of children (perhaps Nepal) and some countries have laws which help to rescue and rehabilitate bonded child labourers (India).

10. On the socio-economic side, there are numerous initiatives and programmes created by government, NGOs and the private sector. These include cash transfers, social protection programmes and social labelling, to name a few. Whilst governmental interventions are increasing in frequency, the means and impact are inconsistent and not always effective This is due to the lack of an accurate definition and other problems arising from ‘bondedness’ itself, which are not accounted for and often placed within a more general child labour approach itself. A bottom-up approach has been more promising, with the meaningful involvement of local communities, as well as empowerment of women through self-management and educational initiatives, having more impact. Nonetheless, there are very few impact assessments available to analyse interventions. Social labelling in particular involves all three groups of stakeholders positively contributing to the issue of child labour. The success of social labelling rests upon a large firm, or firms, agreeing to a normative convention of good treatment of labour, followed by the award of a kitemark. One example is the Rugmark logo used for carpets made without forced labour. This has potential, but requires much more coordination for greater impact.

11. One noticeable point on interventions is that most focus on child labour more generally and not bonded child labour specifically. Future interventions must bear this specific focus in mind, given the unique nature of bondedness. Given that most interventions seem to have only a short term effect, interventions with longer term impact must be designed. One such long term plan could involve self-management and income generation strategy to prevent families and children falling back into bondage. Work by Save the Children has shown successful results for this type of intervention.
Original languageEnglish
Place of PublicationManchester
PublisherDepartment for International Development
Commissioning bodyDepartment For International Development (DFID)
Number of pages80
Publication statusPublished - 31 Mar 2018


  • Childhood
  • Employment
  • South Asia
  • Child Labour
  • Gender
  • Norms
  • Financialisation
  • Development
  • India
  • Bangladesh
  • Nepal
  • Afghanistan
  • Bhutan
  • Myanmar
  • Pakistan


Dive into the research topics of 'Bonded Child Labour in South Asia: Building the Evidence Base for DFID Programming and Policy Engagement'. Together they form a unique fingerprint.

Cite this