Theorising ‘harm’ in relation to children’s work

A central and implicit issue that shapes the present political and institutional consensus surrounding child labour is the notion of harm. Although efforts to address children’s work rest firmly on assumptions about what is harmful, no coherent theory of harm exists. In this paper, we critically explore ‘harm’ in the context of children’s work and call for a more situated and nuanced approach, incorporating ‘subjective’ and ‘objective’ dimensions. Such an approach has important implications for future research and policy action.

1 Introduction The International Labour Organization (ILO) estimates that, globally, 152 million children are in 'child labour'. Of these, almost 75 million are said to labour in conditions or circumstances that are hazardous. The ILO has made it an international political priority to eradicate all such labour by 2025, enshrining this objective as Target 8.7 of the Sustainable Development Goals. To that end, hundreds of millions of pounds are being invested by governments, non-governmental organisations and international agencies in support of the development and roll-out of policies and project interventions. In this, they are building on a now century-long tradition of political, legal and diplomatic efforts.
Undoubtedly, this points to an established political and institutional consensus. But a significant body of evidence suggests that there are major problems within that consensus. The dominant approach at its heart involves preventing children from working in conditions deemed unacceptable and removing them from those conditions where prevention has failed. Yet researchers from all continents and across disciplines, as well as movements of working children themselves, claim that this approach often fails and at times even makes life worse for the young people it is supposed to be serving. This is largely because the process by which unacceptability is defined is non-participatory and thus fails to take into account either the subjective or contextual realities of working children's lives. With this observation, critics identify a fault line within the field of actors working on child labour, between those who are understood as 'abolitionists' (because they seek blanket bans on types of work) and those who favour a more nuanced, regulatory approach that is based on a contextual understanding of, and response to, work in children's lives.
A central implicit concept in all of these debates is harm. The ILO formally defines child labour as work that is mentally, physically, socially or morally harmful to children. It also includes in this category work that interferes with children's schooling, because this is understood as harmful to future economic prospects. Critics of the ILO and the mainstream approach associated with it also frequently use the concept of harm, typically arguing two things. First, that the ILO and mainstream approaches are mistaken and overly simplistic in their identification of any given harm; and, second, that in this over-simplification they paradoxically end up harming the children they aim to protect. Nevertheless, even the critics are hard pressed to offer a formal definition of harm.
So what actually is harm? And how should we go about understanding it? Evidently this foundational concept is integral to understanding what have aptly been termed the 'rights and wrongs of children's work' (Bourdillon et al. 2010) and, evidently, its content is a site of contestation. This paper will attempt to move the debate forwards by sketching out some possible answers to the questions of what harm is and how it may be identified in the context of children's work. It will begin by outlining the formal, institutional understanding centred on the ILO and will explore how harm is discussed, explicitly and implicitly. It will examine how other major institutions address harm in the context of children's work and how this differs from the ILO's approach. Next, the paper will review literature from a variety of academic disciplines to assess how harm and crucial related concepts are referred to, understood and theorised. The data presented here are further supported by interviews conducted by the authors with expert figures across a number of disciplines. Third, the paper will synthesise the foregoing discussion, with the intention of pointing towards a more holistic approach to harm, which seeks to incorporate both 'subjective' and 'objective' dimensions, and shifts the focus of attention to the desired state that preventing harm seeks to achieve -wellbeing. The paper will conclude with implications for future research and policy action.
2 The mainstream institutional picture of harm 1 This was extensively confirmed in interviews with scholars who have worked at length with the ILO.
As the guardian of the world's labour standards, the ILO has the international mandate to define and differentiate acceptable from unacceptable work. It does this through tripartite negotiations between governments, representatives of employer associations, and representatives of organised labour. The definitions arrived at via these negotiations are formalised in conventions, which become the benchmark political norms for categorising different kinds of work. In relation to children, the two key conventions are 138 and 182, the Minimum Age and Worst Forms of Child Labour conventions, respectively (ILO 1973(ILO , 1999. These are supplemented by Recommendation 190 (ILO 1999). In essence, these three texts plus the ILO's many clarifying publications break children's economic activity down into the following categories: • Children's work, which is often described as 'a non-technical term for economic activities of children', where these activities are acceptable because they fall outside any of the following categories (e.g. ILO-IPEC 2012: 31 Embedded within the above are the concepts of 'hazard' and 'risk'. The ILO explains that 'a "hazard" is anything with the potential to do harm', while 'a "risk" is the likelihood of potential harm from that hazard being realised' (ibid.). Hazards can be physical, chemical, biological, environmental, ergonomic, and so on. Risks depend on the mechanisms put in place to mitigate or manage hazards. Importantly, ILO literature recognises that work can be hazardous and risks must be managed for adults as well as children, but it consistently makes a strong scientific case that children are different in important respects and thus more vulnerable. To cite one example, see Box 1 (ILO 2011a(ILO , 2011bILO-IPEC 2014: 47). Note, too, the absence of any relational analysis of vulnerability, a point to which we will return.
The ILO urges all governments to collaboratively elaborate lists of types of work and working conditions that constitute hazardous child labour. In its handbook, Determining Hazardous Child Labour (ILO-IPEC 2012), it suggests that this process should be led by relevant government agencies and take the form of a committee pulling together representatives of all key stakeholders, in particular the ILO's tripartite core. To its credit, it further suggests that committees consult with key social actors including child workers and their parents, so as to avoid the wholesale transposition of external norms onto local realities, as was the case during the period following formal decolonisation (ILO-IPEC 2012: 15). However, further reading reveals these processes to be limited -consultations are said to take place over a few hours and largely at national or regional levels, 1 which means they are far removed from the actual sites of most children's work and thus the lived contexts in which this work takes place.
A final issue of note is that, despite its voluminous writings, the ILO has itself never formally defined 'harm'. This means that harm tends to be implicitly constructed throughout ILO and related mainstream institutional literature as a negative impact that may be physical, psychosocial or economic. In turn, this implicitly posits that harm is morally 'bad' or 'undesirable', with its absence correspondingly 'good' and therefore desirable. Yet the actual content of this 'good' is itself never analysed or theorised. As we will go on to discuss, this is far too simplistic as a basis for developing well-grounded policies and interventions, tending instead to legitimate repressive policy approaches.
In short, at the mainstream institutional level, and despite the evidently extensive thinking and research that have taken place, there is a pressing need for deeper understanding of children's involvement in work that harms them -physically, mentally, socially and psychologically. A more dynamic and holistic picture of what harm entailsand whether, when and how one must intervene to prevent it -could benefit from being located within a variety of perspectives and disciplines, in particular those which are more responsive to local realities.
In the next section of the paper, we explore in greater detail how harm is referred to and understood, explicitly and implicitly, through different literature on children's work. While a universal definition of harm in relation to children's work does not exist, the social sciences can provide valuable insight -particularly with respect to concepts, tools and analytical processes -because of their solidly empirical foundations, which address real people in real contexts and situations (Bourdillon et al. 2010: 94).

Box 1: Children and vulnerability
Why are children so vulnerable to workplace dangers?
Many people assume that the work children do is not particularly dangerous. Others assume that it is not necessary to know about the effects of work on children's health, but simply to concentrate on getting them out of work (for those aged below 15 years) or finding them work if they are above that age. But in fact, until their late teen years, children are more vulnerable to workplace dangers than are adults, because they: • have thinner skin, so toxic substances are more easily absorbed; • breathe faster and more deeply, so can inhale more airborne pathogens and dusts; • dehydrate more easily due to their larger skin surface and faster breathing; • absorb and retain heavy metals (lead, mercury) in the brain more easily, which can disrupt the endocrine system that plays a key role in growth and development, retarding intellectual development, and affecting the whole nervous system; • use more energy in growing and so are at higher risk from ingested toxins; • require more sleep and rest for proper development; and • have less-developed thermoregulatory systems, rendering them more sensitive to • heat and cold.

Childhood studies
Childhood studies is an interdisciplinary field that brings together scholars from anthropology, sociology, development studies, psychology, history, education, law and economics. Typical histories trace its establishment to the work of Allison James and Alan Prout, and especially their book Constructing and Reconstructing Childhood (1997). Its central and foundational premise is that childhood is a complex social phenomenon that is constructed, contested and partially stabilised as an ongoing artefact of social practice. In this, the field rejects essentialism and argues that any given experience of childhood or of being a child can only ever be meaningfully understood in relation to the many criss-crossing contexts and structures that this involves. This, in turn, repudiates the idea that there is such a thing as a 'normal childhood', or indeed that any given practice of childhood is universal.
Although this may at first appear to be an unnecessary theoretical detour, it is in fact critical for understanding the social scientific pushback against the political consensus that 'child labour deprives children of their childhood', that certain kinds of work are inherently harmful, and that children should thus be protected by being removed from that work. For if childhood is a social construct that is never fully fixed and always contextually variable, then it is impossible to be certain that any particular type of work activity will always and everywhere have the same outcome, or that this outcome will be one of harm. This social scientific principle lies at the root of the anthropological and sociological critique and is found in work from psychology, politics, and law. In order to avoid repetition, we will address these disciplinary perspectives separately below.

Anthropological and sociological perspectives
A primary focus of the extensive anthropological literature looking at child labour and children's work has been on the negative side effects caused by abolitionist interventions seeking to protect children from harm by removing them from certain working circumstances. One of the seminal studies in this tradition is Boyden and Myers (1995), which documented the tragic case of children working in the Bangladeshi garment sector. In the early 1990s, as a result of a bill proposed by US Senator Tom Harkin to ban Bangladeshi textile imports to the United State unless employers could demonstrate that they were 'child-labour free', thousands of child workers found themselves unemployed overnight.
As a result, many ended up in demonstrably worse conditions, driven into the streets, sex work or factories operating even further under the radar.
Pakistani anthropologist Ali Khan tells a similar story of football stitchers in Sialkot (Khan 2007 At the heart of these studies is the idea of harm, and specifically that removing children from work that is difficult, dangerous, and at times even damaging, may not in fact be in their best interests because it causes them even more harm. Primarily (though far from exclusively), this harm is understood in terms of children being excluded from access to the resources that they and their families need to get by. This point has been forcefully underscored by scholars working in contexts as different as West Africa and South America. In research by one of the present authors, Neil Howard, with young migrant workers from Benin, for example, children who were removed from or denied access to work opportunities furiously accused those responsible of making their lives worse (Howard 2017 In South America, Taft (2013) and Liebel (2004) have documented the anger of working children's movements in relation to the ILO, expressing their concerns over the 'collateral damage of interventions that seek to eradicate child labour on the surface but create more harm by criminalising it'. 3 Misguided interventions designed to address harmful children's work by taking the institutional picture at face value are thus widely critiqued within the anthropological literature, with some claiming that they violate the core humanitarian principle of 'do no harm' (see Hart, forthcoming, for a discussion of institutional failure in this respect).
Others note that smaller child advocacy groups face significant pressure to conform to the stance that powerful organisations such as the ILO adhere to, given the financial resources and international clout that it possesses. This has the tragic effect of ensuring that damaging policy tends to be recreated, with feedback and learning loops insufficiently powerful to affect meaningful change (Howard 2017;Bourdillon and Myers 2020). The situation is succinctly summarised by Michael Bourdillon, who has dedicated his career to understanding the place of work in children's lives and their development:

• by sustaining them and their families materially;
• by enabling them to continue their schooling or other education as a result of their earnings; • by providing them with skills, including those which may offer them a future livelihood; • psychosocially, by fostering a sense of competence and self-esteem; • by developing their social skills and relations; • by enabling their socialisation into maturity and collective responsibility; • by enabling their social transition into adulthood, including through providing the resources necessary for marriage.
Point 1 has been addressed above so there is no need to repeat it, other than to note that the literature on this point is very large.
Point 2 is interesting and, from a mainstream policy perspective, often counter-intuitive. A common policymaker's assumption around children's work is that it sits within an 'either/or' relationship with their education. Yet a large body of ethnographic data now show that, in many cases, children are able to prolong their schooling in contexts where it carries hidden or heavy opportunity costs, precisely by working, including in circumstances deemed harmful. Okyere (2017a) has demonstrated this in Ghana with a compelling case study of children working in quarries. His findings are echoed by Maconachie and Hilson (2016) in the context of artisanal mining in Sierra Leone, as well as in case studies from across the African continent assembled by Hashim and Thorsen (2011).
With regard to point 3, it is important to bear in mind that much formal education in the global South is of extremely poor quality, and very few formal employment opportunities may exist following schooling. As such, school is not always a guarantee of better future employment, which makes learning practical, marketable skills attractive for children and parents alike (e.g. Morrow and Boyden 2018).
In relation to points 4-7, it is worth remembering that the United Nations Convention on the Rights of the Child (UNCRC) understands education broadly as being about developing 'the child's personality, talents and mental and physical abilities to their fullest potential', with a view to supporting children's growth into and flourishing as citizens able to partake fully in the human community (UNCRC, Article 29.1a). Much of the anthropological literature on children's work suggests that it can be beneficial for them in precisely this educational fashion. Data from every continent, for example, suggest that young workers feel proud and experience heightened self-esteem when they are able to contribute to their families' wellbeing through their labour (see, especially, the examples from the celebrated Young Lives study, Crivello, Vennam and Komanduri 2012).
This, in turn, gives them confidence, and indeed fosters resilience -which is vital in contexts of socioeconomic vulnerability and can only be obtained through exposure to hazards that one learns then to manage (Boyden and  A key point in all of the above is that work is understood and experienced in much of the world as a pathway by which children attain maturity and social responsibility, before eventually becoming adults. The present authors have found this extensively in their own research in West Africa, and it has been documented widely elsewhere. Heissler (2012), for example, found female Bangladeshi migrant labourers increased their respect and status in their families as a result of their work; while Pankhurst, Crivello and Tiumelissan (2015) note that in contexts across East Africa, children's work is regarded and experienced positively both by children and adults as a mechanism by which they can become integrated into the reciprocal fabric of family and community life.
Embedded in all of the foregoing analyses, and indeed in literature across anthropology and sociology, is the concept of wellbeing as the 'true' benchmark by which we should evaluate the pros 5 There is an entire strand of primarily anthropological and sociological literature that echoes the critiques discussed in this section of the abolitionist approach to child labour within the field of migration studies. Migration studies, like development studies, is an inter-disciplinary domain that brings together scholars of different stripes around a central phenomenon -human mobility -and the ways in which this is understood, lived and governed. Over the past 15-20 years, the focus on children's mobility has increased exponentially, particularly within policymaker communities concerned with preventing 'child trafficking' and, in turn, within anthropological and sociological circles in migration studies that are worried about how policymakers go about policing what they see as 'trafficking'. The analyses put forth within this body of scholarship overlap fundamentally with those in the scholarship looking more specifically at child labour, in part because 'child trafficking' is a political and conceptual offshoot of child labour and in part because the scholars looking at it are often the same as or closely related to the scholars looking at child labour. For reasons of space, therefore, we choose not to include a section specific to migration studies; doing so would largely repeat the arguments laid out here. However, interested readers might wish to consult these important texts for further information: Howard (2017); Howard and Okyere (2015); Huijsmans and Baker (2012); Okyere (2017b); Thorsen (2014); Thorsen and Hashim (2011). and cons of children's work, and, by extension, the policies that seek to limit it. Although rarely made explicit by most writers, wellbeing is implicit in almost all commentary, while Bourdillon et al. (2010) stand out in making it the centrepiece of their analysis. Indeed, in the conclusion of their seminal treatise on children's work, they argue that policymakers should ground their efforts in a rigorous, social scientific attempt to understand and then advance child wellbeing (which they see as the cornerstone of decisions around what the UNCRC calls 'the best interests of the child').
For them, although any individual experience of harm will necessarily diminish aggregate wellbeing, the assessment as to how that harm should be navigated can only be made contextually and with reference to the overall bundle of inputs contributing to a child's wellbeing or illbeing. Naturally, this points to a different policymaking approach to harm and its place in children's labouring lives than is presently the norm. We will return to this below. 5

The view from human geography
Scholarship within the discipline of human geography also makes a valuable contribution to understanding the process of how harmful children's work is conceptualised, most notably by shedding light on the importance of locating 'harm' on a broader spatial canvas. Such an approach can offer a useful lens for exploring variations in the notion of harm, as they are defined and redefined in the shifting landscapes of children's work in a globalised world. A common analytical thread shared by many scholars of human geography concerns the need for a situated analysis that poses questions about the spatial patterns of economic development across various scales. Situating knowledge and practice about children's work in time and space can reveal the geographically and culturally specific nature of what may be considered harmful work, and what may not be considered so.
Drawing inspiration from anthropology, many geographers have undertaken valuable studies of non-Western childhoods, challenging Eurocentric ideas of childhood, work and harm; see, for example, the work of Katz (1991Katz ( , 1993 in rural Sudan or studies by Punch (2000Punch ( , 2002 of village children in Bolivia. Indeed, the study of the diverse conditions under which children live in non-Western contexts is an important sub-area of geography research, revealing the impacts of changing landscapes for children and work around the world in response to a wide range of drivers, including globalisation, neo-liberalisation and economic restructuring. See, for example, work by Abebe (2007)  By locating children at the centre of analysis, rather than on the periphery, and situating their perspectives and experiences within the contours of a broader landscape that surrounds the work they undertake, geographers have contributed to rethinking the process of uneven development and the way in which global capitalism has bearing on local realities (see : Bessell 2011;Dobson and Stillwell 2000;Dyson 2008 A further important focus of geographers that is relevant to the present discussion concerns the issue of representation and how our perceptions are shaped by the construction of knowledge. Nieuwenhuys (1996) reminds us that the concepts of child labour (and by extension, harmful children's work) are socially constructed -they embody a set of generalised representations of childhood (that are often founded in the global North), and are shaped by political and moral undertones. Many children's geographers have expressed concern about the widespread tendency for academics and policymakers to generalise experiences and to naturalise constructions of childhood, both of which are common in orthodox framings of child labour. Such constructions can play a role in both excluding and marginalising children, and denying their agency (Holt and Holloway 2006).
In sum, the work of geographers underscores the significance of the work that children and young people do, their importance as social actors, and their agency within studies of globalised production (Punch 2002(Punch , 2007Robson 2004). As Robson (2004) clearly demonstrates, children at work are active social and economic agents, making real contributions in rural societies. What is considered harmful, and what is not, must be seen in this context. Grounded geographies of working children in a globalised world demand an understanding of the contexts within which children work, the origins of framings of childhood and work, and the voices of children and young people as they are incorporated into the global economy.

Political science
The political science literature concerned with harm in relation to children's work can be broken down into two strands. First, that which studies working children's movements and perspectives, and thus repudiates abolitionist efforts to protect children from harm by preventing them from working, since this often causes them even more harm. Of this strand of work written in English, that of Liebel (2004) and Taft (2013) is especially powerful, with both authors documenting at length the many positive benefits that organised working children claim to derive from their work and organising around it.
In particular, Taft has discussed the concept of protagonismo, which is a complex notion that refers to the enactment of individual and collective agency and power. 'For the movement of working children', she notes, 'this has meant discussing how children are an oppressed social group that has been excluded from power, often with paternalistic justifications' (Taft, forthcoming). In turn, this sees children organising their collective power in ways that radically enhance their self-esteem, confidence, life skills and relationships -each of which is evidently beneficial (ibid.).
Importantly, neither Liebel, Taft nor working children themselves suggest that 'anything goes' in relation to children's work or work conditions. Working children's movements in Latin America, Africa and South Asia all strongly advocate for 'dignified work' and for a definition of 'harm' that is neither exclusively age-or activity-based but instead takes account of 'exploitation, violence or abuse'. 7 Crucially, each of these concepts is inherently relational and deeply tied to the meta-concept of wellbeing, which points to the importance of relational experiences in children's conceptualisations of what is or is not acceptable and unacceptable work.
The second strand of political science literature echoes the first and emphasises children's right to participate in conversations over their wellbeing. This strand of literature is heavily influenced by anthropological and feminist thinking around participation as a question of substantive citizenship, and sees children as full rather than partial citizens whose preferences and perspectives must be taken seriously as their democratic right. As may be expected, writers in this tradition (e.g. Swift 1997) are especially concerned with children's exclusion from conversations over how to define harm done to them, arguing that such exclusion is itself a power-based form of harm. 8

Legal studies
The legal literature on harm in children's work is relatively limited and primarily centres on reinterpreting and challenging the hegemony of the ILO's foundational texts, as a way to contest the hegemony of the abolitionist approach to children's work (e.g. Cullen 2007; van Daalen and Hanson 2019). The key concept in this effort is that of 'living rights', developed primarily by Hanson and Nieuwenhuys (2012). According to them, the idea of 'living rights' denotes that, far from being abstract or universally concrete, rights can be, and in practice always are, exercised in different ways and in different settings.
For instance, although 'the right to quality education' may always be written using the same words in the same order, what actually comprises 'quality' or 'education' and how practically to deliver either will necessarily vary. 'Living rights' thus attempts to inject dynamism into the textual stasis of law and to do so by emphasising the importance of paying close attention to the contexts in which young people live and grow, as well as to those young people themselves. These legal arguments are heavily influenced by those working in the various disciplines discussed above and by debates over what actually harms children and how.
A second, related strand of legal work relevant to this discussion focuses on the UNCRC and the extent to which it either supersedes the ILO framework or can be used to challenge its hegemony. In brief, as mentioned above, the ILO framework is built around Conventions 138 and 182. These texts have been interpreted as pushing for the complete exclusion of children from certain sectors or tasks, even where such exclusion goes against the express preferences of children themselves and may, arguably, go against their interests.
Certain scholars have therefore called for international agencies to give primacy to the UNCRC, and in particular Article 3, which states that all decisions related to children must be aimed at advancing their best interests and explicitly links this to their wellbeing (Bourdillon et al. 2010). In addition, others have called for greater attention to Article 12, which states that authorities must give 'the child who is capable of forming his or her own views the right to express those views freely in all matters affecting the child', and to ensure that those views are 'given due weight' in eventual decisions concerning them (Hanson 2012 In Hanson's view, 'this could provide an interpretation of children's rights law regarding child labour that allows (and even calls) to regulate child labour much more vigorously than has been done until now'. 10 This, in turn, would operationalise children's 'living right to work', the central underlying assumption of which is that if children are recognised as legal subjects, then their work-related rights in the International Covenant on Economic, Social and Cultural Rights also have to be acknowledged and enforced, in the same contextually relevant way as obtained for adults (Hanson and Vandaele 2013). Implicit in this effort is the notion that abolitionism harms children and their interests, while contextually relevant regulation would help.

Economics
The economics literature on children's work and child labour is reasonably extensive, though it generally does not interrogate either the concept of harm or existing, mainstream institutional approaches to it. The discipline tends to work with datasets built on the back of ILO estimates of child labour, hazardous child labour and so on, often within the context of national surveys. But by exploring children's work exclusively using the data and proxies that are established as normal by the ILO, much is missed. For example, it is impossible to assess the overall level of impact that any specific work activity may have on wellbeing or illbeing. Likewise, although some research in economics explores potential income and non-income related benefits of children's work, such as the development of skills (Basu 1999), or ability to combine work and schooling (Edmonds et al. 2008), often studies tend to view child development narrowly in terms of how income enhances measurable variables that contribute to national economic development (e.g. levels of education or health; see Bourdillon et al. 2010). This is evidently limited.
However, there is one sub-field within the discipline of economics that has shed considerable light on the concept and experience of harm -feminist economics and particularly its focus on unpaid care. 11 How the notion of 'care' is understood by society is intertwined with structures of inequality, gender, race, ethnicity, and social class. Worldwide, austerity measures and neoliberal reforms beginning in the 1980s exacerbated a growing gender bias in unpaid care work, with particular impacts especially for girl children. As pre-neoliberal social protection structures were eroded under the impact of structural adjustment, increasing numbers of poor mothers in the global South were compelled to enter the workforce to survive on poverty wages. This led more and more girls to have to take on the social reproductive burden, with significant impacts on their wellbeing and human capital formation.
Relatedly, a key consequence of neoliberalisation in the global North was the emergence of global care chains involving women and adolescent girls from the global South moving to richer parts of the world to provide care services that the women of the North could no longer provide. This too led to care deficits in the migrants' home communities, again to be filled by young girls (Anderson 2000;Folbre 2006;Razavi 2007). From the perspective of harm, these structural dynamics matter because evidence suggests both that care work is routinely undervalued and that it can take place in circumstances that are severely abusive. Considerable evidence suggests that girls in particular at times suffer from physical and verbal abuse, isolation, seclusion and less time for school, and are vulnerable to sexual abuse in care work in the home and domestic work outside of it (Murray, Amorim and Bland 2004;Sturrock and Hodes 2016). As will be explored later in the paper, such exploitative working conditions also lead to severe harmful psychological impacts on (mostly female) children.
Feminist economists thus remind us both that the gendered nature of the global division of caring labour has political economic roots and that its impacts are not equally shared. Given this inequitable burden, it seems reasonable to assert, with Nieuwenhuys (2007) and others, that children, especially in the global South, represent a vast reservoir of cheap and flexible labour that contributes towards circuits of global capital: 'Children's everyday work that is done unpaid is even more "for free" than women's and can therefore be tapped into ad libitum' (Nieuwenhuys 2020: 130). This means that children, particularly girls, provide essential yet unremunerated 'services' which produce the value that drives local and global economic growth. The fact that this labour -like most caring labour -is unpaid makes it arguably a form of structural, economic harm.
Importantly, ILO regulations are clear as to the potentially harmful impacts of unpaid care, with Convention 189, the Convention on Domestic Workers, establishing guidelines for what is and is not acceptable. But in reality, the core of the ILO and mainstream approach to children's work centres on Conventions 138 and 182, both of which focus primarily on paid work outside a domestic context, typically done by boys, which arguably reproduces gender bias by drawing attention away from the vital and at times damaging work that girl children often do (Cullen 2007: 156).

Developmental psychology
Beyond the social sciences, the two disciplines with the most to say about harm and how it can be understood are developmental psychology and health. We will address each in turn.
Developmental psychology has made important contributions to research on children's work and the notion of harm. The discipline focuses on processes through which individuals grow and develop cognitively, emotionally and socially. In particular, studies emphasise the importance of human relationships in advancing children's wellbeing (or illbeing), while some have offered valuable longitudinal analyses, providing a picture of how children's lives are impacted over time. Such research has been valuable in assessing both the cumulative impacts of 'invisible' harms, and how the occurrence of the harm/benefit balance may shift over time due to changing conditions. While children themselves may place greater emphasis on harm that is experienced in the short term, where there are immediate consequences, an understanding that takes on board the long-term effects of work activities is also vital.
The effect of work on children's psychological functioning was first raised as an issue by the World Health Organization (WHO) in 1972.
In the 1980s, a joint ILO/WHO Committee on Occupational Health commissioned the first series of studies on the psychological factors associated with child work in four occupational sectors.
However, little further research was carried out in the decades prior to 2000 (Gunn et al. 2015). One of the main reasons for this was the lack of appropriate tools (Dorman 2008). In addition, the impacts of psychological harm, such as mental illness, addiction and stress can be slow to manifest as they are often 'invisible', making them difficult for researchers to identify without training (ILO 2011a).
The effects of psychological harm on working children are still therefore largely under-recorded (Fassa, Parker and Scanlon 2010;Fekadu, Hagglof and Alem 2010;Sturrock and Hodes 2016). However, a burgeoning literature does exist, especially in societies of the global North. But generalising this to the study of child work in the global South is not straightforward. The export of observational, attitudinal and other diagnostic instruments is especially problematic unless their content is thoroughly revised to ensure relevance to child work situations (Woodhead 2004).
In the psychological literature, harmful work is most often associated with the sister concept of abuse (Fekadu et al. 2010), a pairing that is also shared by WHO (Woodhead 2004: 340). Children's experiences of abuse are known to be strongly affected by their gender, in terms of the likelihood of boys and girls becoming victims of sexual, physical and emotional abuse. Girls, in particular, are vulnerable to sexual abuse at work, with the impacts of this made worse when associated with shame and stigma (ibid.). 12 However, both boys and girls equally may suffer emotional abuse at work, which can be manifested through: unreasonable expectations of work productivity and work standards; lack of encouragement and support to ensure children are able to complete a task; scolding and punishment for failures, including ridicule and humiliation, harassment or shaming; and isolating the child and denying his or her needs or requests for help (ibid.).
Most occupation-specific psychological studies have focused on the relatively small proportion of child workers engaged in sex-related occupations and armed conflict (ILO 2014). Domestic work outside of a child's home has also been a focus. Findings suggest that, although potentially benign, this can be rendered harmful by the conditions and social relations in which it takes place (Bourdillon et al. 2010). Some research suggests that domestic work under poor working conditions can be among the most harmful types of work (Sturrock and Hodes 2016). In such situations, girls in particular can suffer physical and verbal abuse, isolation, seclusion and less time for school, and are vulnerable to sex abuses (Murray et al. 2004).
It is therefore equally important to factor in the issue of gender-based violence when defining and assessing harm. Gender-based violence not only occurs in the context of domestic work, but also within the family or inside educational institutions. Knowledge and understanding of violence in schools has only recently emerged. Although normally considered to be a safe space, there is growing evidence to suggest that 'gender violence [has] become institutionalised and accepted as part of the landscape of schooling' (Dunne, Humphreys and Leach 2006;Fassa et al. 2010;Fekadu et al. 2010;Seth and Buhr 2012;Sturrock and Hodes 2016). The abuse of power by school authorities also increases the occurrence of transactional sex between adolescent girls and school authorities, which in sub-Saharan Africa is highly linked with high infection and attrition rates of HIV (UNESCO 2006;Ijumba 2011).
Martin Woodhead and Barbara Rogoff are perhaps the most celebrated psychologists looking at the relationship between children's work, development and harm. An important observation made by Woodhead (2004) is that well-recognised physical hazards can have strong psychosocial consequences. Such hazards may include: • Toxic substances, which may impact on the developing nervous system and in turn on children's psychosocial functioning, as for example in the evidence on lead.
• Unhealthy, noisy, poorly lit and poorly ventilated environments, which can affect children's general health, increase stress and fatigue and cause demoralisation. If children find it difficult to work in these circumstances, stress levels may increase.
• Dangerous tools, which without adequate safety precautions may induce stress and fear of accidents. Children may be traumatised by suffering or witnessing serious incidents. Children working in extreme conditions (e.g. mining, fishing) are especially vulnerable.
• If children do suffer an accident in which they are disfigured or disabled, this may increase the risk of social rejection, isolation and stigmatisation.
13 Personal communication with Jo Boyden, 4 March 2020. 14 Postcolonial scholars from Africa go even further with this line of argument. Building on the work of Frantz Fanon and Aimé Césaire, they advocate for overcoming the psychological effects of centuries of slavery and colonisation. In doing so, they deconstruct the dominant system of global capitalism and its objective of accumulating individual wealth. Grounded in indigenous cosmologies and modes of knowledge, postcolonial scholars define labour not only as an economic category, but as a cultural artefact that forms a bridge between culture, society, spirituality, politics and the economy; and thus as something that the individual worker, including the child worker, must experience as a means of weaving a connection with wider life (e.g. Sarr 2016).
However, Woodhead's work also strongly suggests that whether or not work is considered harmful is more closely connected to relationships and the social context in which it takes place, than to the nature of the work itself. His six-country comparative field study is considered seminal in this respect (Woodhead 1998) and it has been supported by a variety of further studies that emphasise that cultural 'meaning-scapes' are vital for understanding how any given experience can be understood and processed by the individual in question as harmful or beneficial (Korbin 2002;Rogoff et al. 2017). In turn, Woodhead (1998) suggests that children often value their work because it provides them with a sense of self-esteem and pride, and because it can play an important role in their personal development by helping them build a sense of efficacy. As Jo Boyden remarks, 'children prefer to work in factories instead of home: they earn more money and at home, parents have more power over them '. 13 Likewise, the findings of Rogoff (2014) suggest that work situations may provide developmental benefits to children's competence in more specific, culturally valued cognitive skills such as alertness, and skills in collaboration, perspective-taking, self-regulation and planning, in addition to their gaining information and skills. 14 As mentioned above, numerous anthropological studies have confirmed these observations, with some also showing how the personal value that children attach to their work can foster a sense of resilience (Boyden and Mann 2005;Werner and Smith 1992).
In short, this literature suggests that studies which build in an appreciation of culture and an understanding of its context are necessary components of any analysis in relation to psychosocial child maltreatment, harm and work. Psychological harm in children's working lives must be assessed in relation to the many other influences in their lives; and placed within a broader picture of the different contexts in which they occur and the aggregate wellbeing or illbeing to which they contribute. This should be carried out at different levels, to include individual or micro-systems (e.g. family, school), mesosystems (e.g. neighbourhoods) and macrosystems (e.g. religious institutions) (ILO 2014).

Health
Within academic and institutional literature on health and child labour, harm is most often understood in relation to injury or illness. The ILO classifies children's occupational health as an area of analysis at the interface between: (a) work and child illnesses/injuries; (b) work and children's psychological functioning; and (c) work and children's physical and emotional development (ILO 2011a). The WHO definition of 'child health', on the other hand, encompasses the complete physical, mental and social wellbeing of a child and not merely the absence of disease or infirmity (ibid.). However, in both cases, the ILO and WHO do not use the term 'harm' but rather focus on hazards, as identified in section 2 of this paper. So, once again, we are confronted with the challenge of coming to terms with how we should go about understanding harm in relation to children's health and wellbeing.
From a health perspective, one might assume that the identification of the forms of child labour that are potentially the most harmful would appear to be straightforward. Children working under appalling conditions in construction, mining and manufacturing can face immediate and obvious threats to their health. However, evidence on the health consequences of children's work activity can be both limited and misleading. Indeed, the vast majority of studies in this field make conclusions about the injuries and illnesses that could occur, basing their assessments on general observations or on known risks to adults (ibid.). Illness and injury hazard rates by sector of employment tend to describe the risks faced by working children, but in the absence of comparison with a 'no work' counterfactual, they do not provide an effective basis for evaluating the impact of work on health (O'Donnell, van Doorslaer and Rosati 2002).
In addition, from a health perspective, the definition of hazard as stated in ILO Convention 182 becomes both challenging and complex. Interpretation of all current estimates of the relationship between child labour and health is difficult given the absence of analyses that account for the potential endogeneity of child work activity to health outcomes. For example, if individuals born with a predisposition to poor health are also those who are most likely to engage in work as a child, correlations between children's work and health will overstate the impact of the former on the latter. Due to the potential intergenerational effects of toxic substances, some children who work may also begin their life with lower intelligence or neurocognitive problems (Ide and Parker 2005). On the other hand, if healthy individuals are selected into work early as children, the true health impact of their work will be understated. Additionally, comparisons between the growth rates of working and non-working children in rural settings provide mixed results (O'Donnell et al. 2002).
Despite the absence of rigorous research focusing on the impact of work on the health of children in the global South, over the past decade there has been some progress with child labour data becoming more comprehensive and accurate (Fassa et al. 2010). However, there is a pressing need for a clearer understanding of how child work relates to health, and how the notion of harm fits into this equation. Without the availability of larger, longitudinal follow-up studies, including those which use proxies that are more contextually meaningful than those of ILO surveys, the long-term health implications and potential gender disparities of children's work-except for missed school days-cannot be accurately characterised.
Many recent studies on the impact of health on child workers are of poor methodological quality, lack gender disaggregated data and do not build intersectionality into their research designs. In such cases, sub-Saharan Africa is particularly under researched (Kuimi et al. 2018), with little analysis of the relationship between child labour and wider contextual factors such as high tuberculosis and HIV rates. This is surprising given that AIDS orphans belong to one of the most vulnerable groups that rely on work as a means of survival (Hurst 2007).
Consequently, the overall global health burden of illness or injury related to children's work may remain underestimated, and perhaps even worse, poorly understood (Fassa et al. 2010;Ide and Parker 2005;Kuimi et al. 2018;Shendell et al. 2016). ILO publications (e.g. see ILO 2011a) rely on data from studies undertaken in developed countries -mostly from work-related injuries of children and young people in the US and Canada where routine data collection is much more complete (Fassa et al. 2010). However, these data cannot fully reflect the situation in non-industrialised countries (FPRW-IPEC 2014).
To date, there has been no pattern across countries in the raw correlation between whether a child works and whether he or she reports health problems (O'Donnell et al. 2002). Until the literature considers wider family and community effects, only part of the health story will be told (Fassa et al. 2010).
In short, it is clear that a more systematic and situated focus on health among child workers is necessary. Moreover, equally problematic, is the fact that harm is most often seen solely in relation to injury or illness that could occur, with very little research being carried out on links between workrelated harm and children's emotional and relational wellbeing. As Dorman concluded over a decade ago, much progress could be made by situating child labour health research on a much wider canvas. He notes: while adult occupations have been studied in great detail in developed countries, corresponding data are limited for the rest of the world. This is probably due to the tendency of development economists and other specialists to overlook the contribution of health to economic outcomes except in a few prominent areas.
15 Personal communication with Michael Bourdillon, 19 February 2020. With a broader outlook on human capital factors, research into work and health should be intensified, and our understanding of the health consequences of children's work should benefit correspondingly. (Dorman 2008: 46) 4 Towards a reconceptualisation of children's work and harm?
From the preceding discussion, it is clear that the concept of harm in the context of children's work is understood and theorised in diverse and farreaching ways. At the heart of the critical literature reviewed -particularly the anthropological and sociological critique, supported by the psychological -is the idea that the concept of harm is ambiguous, relative and contextual. It may be unhelpful (and even problematic) to present harm as an 'objective' concept that can be defined, measured and assessed with discrete criteria.
This literature underscores that there is a poor understanding of the contours of this broader 'harm-scape' and the relative importance of children's work within it. The potential for harm arises from a complex combination of factors, including: (1) the situation of the child; (2) the specific nature of the work; and (3) the conditions or social relations that surround the work being undertaken. Also very apparent in the anthropological critique is the position that removing children from work that is arduous, dangerous or even potentially harmful may not always be the best option for them.
In many respects, it is misleading to discuss the notion of harm in isolation, without some consideration of the related benefits that may also be derived, or the wider contextual environment. In the context of children's agricultural work, hazards, risks and benefits are often intermingled. Judgements about harm and thus acceptability therefore require weighing up both costs and benefits in a situational context, while making some comparison with the realistic alternatives that exist. Such judgements must involve subjective and cultural values being combined with objective criteria. Here, there are undoubtedly lessons that can be learnt from the literature on relational wellbeing since wellbeing remains arguably the implicit benchmark context in any discussion of harm or work that is harmful to children.
Although the concept is itself the subject of much debate (e.g. Doyal and Gough 1991;Gough, McGregor and Camfield 2007), the growing body of research on relational wellbeing could be vital here, since it recognises the importance of context (White 2015), and could also help to prioritise the need for children to be appreciated as subjects rather than objects, refocusing analysis on understanding their lifeworlds in their own terms (De Berry et al. 2003). This contrasts with most mainstream approaches to understanding harmful children's work, which tend to take a more positivist and objective stance, and fail to appreciate the social and relational aspects of what is or is not harmful, to whom, when and why, and which alternative states are possible and worth considering.
At its heart, as noted by Atkinson et al. (2012), relational wellbeing holds that notions of wellbeing and illbeing are socially and culturally constructed, and rooted in a particular time and place. Shifting the focus to locate harm on a continuum of wellbeing and illbeing may prove to be a more appropriate and coherent way of understanding why children work, including in harmful situations; which situations are actually experienced as harmful; and how this varies across a variety of different contexts.
Without a more holistic outlook, a dominating fear of harm can result in policies that are more damaging to children than the harm they have been designed to avert, in the process unjustly overriding children's perceptions, goals and wishes. 15 In short, a more balanced and conceptually sophisticated approach to policymaking is needed -one that also considers the impact of work on a child's development, wellbeing and needs from a multiplicity of perspectives. This position is well summarised by Richard Carothers, founder of the Children and Work Network, which brings together scholars and activists, including working children themselves, from over 100 countries and includes many of the most recognised scholarly authorities on children's work: I think an important step in dealing with 'harmful work' is the idea that we are trying to get rid of 'harm' and not necessarily get rid of work. The definition of harm will vary with situation and context and it may be possible to eliminate the harm within work or help children move from harmful work to less harmful work. In doing this, businesses and business support programs can play a useful role and the Occupational Health and Safety program of the ILO can become an ally with a different approach and contribution than the IPEC people. 16 The belief that children's work can be dichotomised into typologies of 'acceptable and unacceptable' or 'good and bad' can therefore be problematic, and this in turn problematises administrative categories such as 'child labour' or 'harmful child labour' themselves, since these tend to be equated with good and bad and abstracted from all context or relationality. The distinction between harm and benefit can be fuzzy and variable, depending on the specific situation. It can also change over time. Studies have demonstrated that work many observers would consider to be benign can be harmful due to the social relations and conditions that surround it. Alternatively, potentially harmful work can be beneficial to children if social conditions are supportive. Ultimately, this all needs to be analysed in relation to the benchmark concepts of wellbeing and illbeing.
It may therefore be more useful to think of children's work as lying on a spectrum of harm and benefit.
In assessing harm, we suggest that a variety of factors must be taken into consideration, including: (a) the temporal nature of harm (e.g. the cumulative or 'invisible' aspects of harm); and (b) the trade-offs which must be assessed to determine if potential benefits outweigh potential risks. A set of key questions thus arise: who is assessing the relative nature of harm, and how does this sit with other perspectives? How are different perspectives on harm reconciled? Likewise, is one instance of a hazardous activity enough to describe the entire work experience as 'harmful'? And, in the end, how does all of this relate to wellbeing, which is the implicit counterfactual state against which harm and benefit are being evaluated?
Models designed to explore different children's work environment scenarios, such as the state-andtransition model proposed by Sabates-Wheeler and Sumberg (2020) harm actually entails, and when and how it needs addressing. A definition of harmful children's work remains challenging, even if the work of Woodhead (2004) provides an excellent starting point, at least for examining its subjective dimensions. On balance, the evidence suggests that it is incredibly difficult, and probably of little value, to develop a clear-cut, 'one-size-fits-all' set of criteria to distinguish between harmful and tolerable children's work.
Although in recent years there have been some important changes in the way that international bodies such as the ILO have translated understandings of child labour into international standards, discourses and policies, the tension between abolition vs regulation seems as strong as ever. Legal scholars continue to stress the need for an alternative translation of international labour law that is compatible with children's rights and the 'living right to work'. Work needs to be understood as bringing both benefits and harm to children and it is necessary to balance benefits against harm, including in policy considerations.
While the extreme aspects of harmful children's work may be relatively unambiguous, this may not be the case in less extreme situations. The risk of harm associated with work, and the actual experience of harm through work, will vary significantly among children. A starting point for understanding different contexts and getting policies right involves participatory consultative processes that prioritise the perspectives and voices of children themselves, as well as their communities. Such accounts must make reference to both physical and psychosocial factors, while also drawing on the extensive literature from health and related sciences about the impacts of specific activities or substances. Grounding policies more solidly in properly researched evidence, and less in conventional wisdom and institutional traditions, will thus be key.
At the same time, policymaking institutions have an in-built imperative to work at scale, and for this they tend to require metrics or processes that can function meaningfully across contexts. This, therefore, is as much a political as a conceptualtheoretical tension, and means that diplomatic endeavours may be equally as impactful as scholarly pursuits. For such endeavours to be successful, they will need to find common ground to integrate both the best of contemporary research and major institutional imperatives. This is not impossible -see Howard, Jacquemin and Thorsen (2018) for an example of how such diplomacy may work in practice -but it will be a challenge, and it is one that ACHA is likely to have to confront.