Children’s work in West African cocoa production: drivers, contestations and critical reflections

Cocoa farming in West Africa has a long history of relying on family labour, including children’s labour. Increasingly, global concern is voiced about the hazardous nature of children’s work, without considering how it contributes to their social development. Using recent research, this paper maps out the tasks undertaken by boys and girls of different ages in Ghana and how their involvement in work considered hazardous has changed. We show that actions to decrease potential harm are increasingly difficult and identify new areas of inquiry.

References 21 Tables   Table 1. Estimated labour requirements in the average Ghanaian cocoa farm 8 Table 2. Labour standards for permitted work for different age groups 9 Table 3. Working hours in cocoa production for each age cohort, 2018-19 10 Table 4. Labour standards definitions of hazardous work 11 Table 5. Percentage of children exposed to hazardous activities in cocoa, 2018-19 12 Table 6 . Exposure to agro-chemicals in cocoa production in the past 12 months, disaggregated by activity, gender and age, 2018-19 13 Table 7 . Proportion of children involved in different types of work in the past 12 months, all agricultural households surveyed in 2018-19 (N=9,200) 15 Table 8. Changes in exposure to agro-chemicals 16 Table 9 . Prevalence of all children in cocoa-growing households exposed to the most frequent types of hazardous work 17 Acronyms ACHA Action on Children's Harmful Work in African Agriculture

Roy Maconachie is Professor of Natural Resources and Development in the
Department of Social and Policy Sciences at the University of Bath. Roy's research in sub-Saharan Africa explores the social, political and economic aspects of food production and natural resource management, and their relationships to wider societal change. Much of his recent work has been concerned with the politics of natural resource management in West Africa, with a particular focus on the extractive industries, livelihood change and social conflict. Roy's disciplinary background is in human geography, but his work has largely been interdisciplinary in its approach, drawing principally on anthropology, development studies and politics/political economy. Roy holds a PhD in Geography from the University of Sussex (2004)  1 Introduction 1 According to international labour standards set out by the International Labour Office, all work is considered harmful for children under the age of 12 years; work in excess of 13 hours per week or work considered hazardous is seen as harmful for children aged 12-14 years; and work in excess of 42 hours per week or engaging in hazardous activities is considered harmful for children aged 15-17 years (Sadhu et al. 2020).
2 The ILO Convention No.182 Concerning the Prohibition and Immediate Action for the Elimination of the Worst Forms of Child Labour targets all forms of slavery or practices similar to slavery; the use, procuring or offering of a child for prostitution and pornography; the use, procuring or offering of a child for illicit activities; and work that by its nature or the circumstances in which it is carried out is likely to cause harm to the child's health, safety or moral (article 3). In this paper, we are primarily concerned with work considered to be hazardous, as insufficient scientific data are available on work considered slavery or similar to slavery.
In 2000-01, widespread public concerns about children's work in cocoa production emerged in the United Kingdom (UK), following a series of high-profile media reports that referred to children in harmful work 1 in the supply chains of some of the world's largest and most well-known chocolate manufacturers (Blowfield 2003;Ould et al. 2004). While this reignited public debate about the prevalence of harm to children working in global commodity chains, it also spawned a wide range of initiatives to tackle exploitative labour practices in the West African cocoa sector. At about the same time, in the year 2000, Ghana -the world's second largest cocoa producer -ratified the International Labour Organization (ILO) Convention 182 on the worst forms of child labour 2 to complement the national Children's Act of 1998, which set out regulations concerning the minimum age for employment (Baah 2010). Subsequently, a series of national action plans were developed and directed specifically at the cocoa sector. This was followed by the implementation of the United States (US)-developed Harkin-Engel Protocol -a voluntary agreement with global scope, which was signed in September 2001. By bringing together governments, cocoa producers and labourers, the cocoa industry, and non-governmental organisations (NGOs), the protocol aimed to end the worst forms of child labour in cocoa production, and emerged as having far-reaching and long-term impacts on West African cocoa-producing communities.
Over the past two decades, considerable debate has continued to transpire over the role that children play in the cocoa sector. Critics are largely divided between those who view all children's labour inputs as 'harmful' and unacceptable (the 'abolitionists'), and those who locate children's work on a wider canvas, and favour a more nuanced, regulatory approach that is based on a contextual understanding of and response to work in children's lives (see Maconachie, Howard and Bock 2020;Thorsen 2012). With respect to the latter position, Berlan (2013) and Buono and Babo (2013) note the problematic nature of making judgements about the role that children play in African agriculture in isolation from their wider social, cultural and economic contexts. Cocoa production is embedded within indigenous social institutions and family relations, and historical reports have documented how family labour, including that of children, has played a significant role in the early development of the cocoa sector (Sutton 1983). A report commissioned by the Ghanaian government's National Programme for the Elimination of the Worst Forms of Child Labour in Cocoa (NPECLC) noted: Traditionally, working on family farms and with family enterprises is seen as part of the process by which children are trained towards adulthood… Children's involvement in the production of cocoa is an age-old tradition which, besides the immediate labour value, constitutes a… way of imparting cocoa farming skills to them and equipping them to take over from aging parents and relatives. (MMYE 2007: 6-7) The role that children's work plays in the cocoa sector is therefore not as easily constructed as being morally bad or undesirable, as some critics would like to argue. The abstraction of labour practices from their wider historical context is common in debates on cocoa production (Berlan 2013). Amanor (2019) further reminds us that it remains vital to examine the relationship between land tenure systems, labour markets and family relations in production, to gain a better understanding of the contemporary cocoa industry.
In this background paper, we seek to deepen understanding and provide a more dynamic assessment of children's work in the West African cocoa commodity chain, a development which would seem vital for designing tangible, empirically grounded child protection and pro-poor policies and interventions. Drawing predominantly on scholarship focusing on the Ghanaian cocoa sector, the paper aims to provide a synthesis of recent academic and policy debates in relation to children's work in cocoa production. In doing so, the discussion flags up some key areas for inquiry relating to the drivers and dynamics of harmful children's work, with the intention of providing future direction for planning and research design in Ghana for the Action on Children's Harmful Work in African Agriculture (ACHA) programme.
The paper is structured as follows. Following this introduction, we present a brief historical overview of cocoa production in Ghana, tracing the prevalence and importance of children's labour inputs in the cocoa sector from colonial times to the present. The next section then outlines in further detail the role that children have traditionally played in cocoa farming, demonstrating how children's work is embedded in the fabric of Ghanaian society and economy, and provides an important injection into the household economy at peak times in the agricultural calendar. Drawing predominantly on data from the extensive study carried out by the National Opinion Research Center (NORC) at the University of Chicago, the paper then proceeds to discuss the various tasks that boys and girls of different ages perform on cocoa farms, making it apparent that the distinction between benign work and work that carries different degrees of risk and harm is often blurred. The contentious relationship between schooling and children's work in cocoa communities is briefly discussed, followed by some reflection on how debates that concern children's work in the cocoa sector are shaped by policy and discourse. The paper concludes with some reflections on future directions for ACHA research in Ghana, advocating for a more holistic approach to harm in children's cocoa work, which seeks to incorporate both 'subjective' and 'objective' dimensions that place children at the centre of the analysis.

Ghanaian cocoa production and global incorporation in historical context
Historically, trade and barter have long been part of West African economies and social relations. But since the colonial era, farmers were incorporated into the production of commodities for export, often at the expense of food security and self-sufficiency. In many respects, Europeans viewed Africa as a giant farm that enabled them to produce a host of export crops such as cotton, cocoa, rubber, sisal, 3 Some reports suggest that between 1894 and 1908, a Ghanaian farmer could earn as much as ten times more income from cocoa than palm oil (Acquaah 1999). 4 In the early nineteenth century, the main supply of cocoa for British manufacturers came from the islands of São Tomé and Príncipe, where it was grown on European plantations using various forms of unfree labour. Public pressure put on Britain's leading chocolate manufacturer, Cadbury, forced the company to look for alternative sources of cocoa, with the Gold Coast becoming the preferred country for production (Amanor 2019). 5 During this period, there was an influx of migrants from Upper Volta (now Burkina Faso), Niger, and Mali, who were attracted by the relatively high wages that cocoa production offered in southern Ghana. 6 Most migrants lacked sufficient capital to purchase land, and so they sharecropped with earlier settlers under a system called abusa, in which labourers were paid one-third of the sales price of the cocoa they harvested (Kolavalli and Vigneri 2011). groundnuts, tea, coffee, sugar, and tobacco. In colonial Ghana (then known as the Gold Coast), the first export crop produced can be traced back to the nineteenth century, when vast quantities of palm oil were exported to Europe to fuel manufacturing during the industrial revolution. However, by the latter part of the nineteenth century, the insatiable demand for palm oil led to the establishment of new industrial plantations in Southeast Asia, saturating the market for palm oil and out-competing West African producers. Eventually, a fall in the international price of palm oil in 1885 (Kolavalli and Vigneri 2011) compelled farmers on the Gold Coast to plant alternative export crops, with cocoa becoming the preferred crop for income generation. 3 Commercial cocoa farming first began in the Gold Coast in the 1890s, when the colony emerged as the most promising new area for cocoa production. 4 By the 1920s, the Gold Coast was the dominant global cocoa producer, controlling over 50 per cent of all production (Green and Hymer 1966).
The use of children's labour in Ghanaian cocoa production can also be traced back to the colonial period when the production of commercial crops was first established. Polly Hill's (1963) classic study of rural capitalism in Ghana provides important insights into the dynamics of labour that sustained cocoa production. Hill's detailed ethnographic work challenged orthodox thinking that cocoa was produced only by sedentary, small-scale, peasant farmers, by revealing that cocoa production in southern Ghana was dominated by entrepreneurial capitalist farmers who had initially accumulated wealth by investing in palm oil and rubber, before reinvesting in land and labour for cocoa production.
Although labour for cocoa cultivation was in high demand in the early 1900s, this capitalist class of farmers was able to draw on migrant labour from northern Ghana and the surrounding Sahelian colonies, in combination with supplementary family farm labour (Abdul-Korah 2007). 5 Migrant labourers were either employed under an annual labour arrangement when establishing a new farm, in which case they were paid at the end of their contract, or a sharecropping agreement when working on a mature farm. In both cases, plots for food production were usually provided to migrants so they could grow staple crops throughout their stay. 6 Extended family relations increasingly became involved in production, providing an important resource for the expansion of cocoa farms within new forest frontier areas. The significant expansion of cocoa production from 536 tonnes in 1900 to 176,000 tonnes in 1920 led to heavy demands for carriers, particularly for head-loading cocoa, and the labour of children became central to meeting this requirement (van Hear 1982).
By the mid-1930s, cocoa production had reached 300,000 tonnes, resulting in a 'frontier economy' in which cocoa cultivation continued to expand into the forest regions under the enterprise of migrant farmers with capital (Amanor 2019). Although labour migration was initially the domain of young, single, unemployed male adults seeking to accumulate savings, by 1935, increasing numbers of women and children began to accompany them (van Hear 1982). Migrants were often drawn towards the newer frontier areas where farmers had more capital to spend on labour, and where weeding was less demanding than on the more established mature farms (Amanor 2019). As explained by van Hear: Children worked on cocoa farms as members of sharecroppers' families and were thus not directly employed by farm-owners, who nevertheless benefitted from their labour.
Children also cultivated sharecroppers' food farms, thus freeing them for cocoa work, and helped to headload harvested cocoa to the marketing points. economy was emerging from the impact of the Great Depression in the 1930s. This often resulted in the direct recruitment of both children and youth to work on cocoa farms to meet rising labour demands. In 1947, the Cocoa Marketing Board (CMB) was established, which gave the colonial government a monopoly over the purchase of cocoa beans. Between 1957 and 1964 exports continued to climb, and overall production reached an annual peak of 430,000 tonnes, even in the face of a significant decline in world prices between 1960 and 1962 (Kolavalli and Vigneri 2011).
During the 1970s, new frontier lands became exhausted in Ghana, fuelling an ongoing shift in labour migration from the Sahelian regions to Côte d'Ivoire, where wages were more favourable and new land was available (Finnegan 1976). However, by the end of the decade, the global price of cocoa had dropped by two-thirds, and World Bank structural adjustment programmes were adopted in an attempt to revive the sector. Although Ghana's adjustment package did include a partial reform of the internal cocoa market, it had dramatic impacts on the lives of poor farmers, through increases in the cost of living and farming inputs. The liberalisation process included granting private companies licenses that allowed them to trade cocoa. Since this time, cocoa has largely been transformed globally into a more intensive farming activity that relies increasingly on the use of farming inputs such as insecticides and fungicides. However, to this day, family members -including children -continue to play an important role in the sector (Baah 2010 et al. (2020) at the NORC at the University of Chicago, to highlight the issues that call for more fine-grained research to understand shifts linked with awareness of hazards, need for labour, and educational aspirations for children.
In 2018-19, 83 per cent of the children aged 5-17 years in sampled agricultural households in cocoa-growing regions in Ghana had been economically active in the past 12 months: 73 per cent had participated in agricultural work, and 60 per cent in cocoa production in this period (Sadhu et al. 2020: 55-62). Children's work in cocoa must therefore be understood in the cultural context of rural Ghana, where contributing to family labour is a fundamental part of childhood, and a means by which children can gradually become economically active (Baah 2010;Berlan 2013;Hashim and Thorsen 2011).
Cocoa is typically cultivated on small, familymanaged farms between 2 and 5 hectares in size, and food crops are often first planted, along with cocoa seeds or seedlings, on newly cleared land, and later between cocoa trees once they are mature. More commonly, once the cocoa is established, food crops are produced on separate plots. Younger children can be found 'working' alongside parents on young and mature farms, regardless of whether the farms are in the hands of farm owners or their sharecroppers. Schoolchildren will also work on farms, especially but not only during peak harvesting periods when neighbours and relatives might work together, jointly completing harvesting and bean extraction from pods, one farm at a time. Younger children's 'work' often involves caring for younger siblings, time for play, and gradual learning of farming skills, while older children feel obliged and are expected to do farm work. Those not attending school might combine farm work with other work and may not always work under supervision. Access to labour is gendered and where men can rely on the labour of their wife/wives, female farmers depend more on their children, relatives and neighbours.
In the Ghanaian cocoa belt, one-quarter of the farming households were headed by a woman in 2018-19 (Sadhu et al. 2020 and this can be an important determinant as to whether or not a child ends up working on a cocoa farm (Berlan 2013).
When discussing children's work, hazards, and potential harm in cocoa production, it is critical to bear in mind that the distinction is often blurred between benign work and work that carries different degrees of risk of harm. As a result, it is exceptionally difficult to estimate and compare children's work across surveys and qualitative studies of different depths and rigour, particularly if the social characteristics of labour are not deconstructed to consider age-appropriate work as laid out in Ghanaian legislation, and as understood in rural Ghanaian communities. Some studies use ILO standards to distinguish between 'permissible work', 'child labour' and 'hazardous work', while other studies use Ghanaian standards whose age restrictions are different and attuned to Ghanaian education policies (see Table 2 below). Both sets of standards stipulate that no individual under the age of 18 should carry out work defined as hazardous (Amoo 2008(Amoo , 2016Sadhu et al. 2020).

Tasks boys and girls perform
The major labour demand in cocoa production occurs in the harvest season from October to December. During this period, the cocoa pods are harvested, carried to points where the pods are broken open, and the wet beans are removed and allowed to ferment for four to seven days, before they are transported back to the homestead and sundried for five to fourteen days (estimated time use, see Table 1). The other labour-demanding tasks in the production cycle involve clearing and weeding the undergrowth beneath cocoa trees in May/June and just before the main harvest in September/October, and spraying pesticides (fungicides are applied two to three times annually and insecticides three to four times annually).
The most common activities in which children in cocoa-growing areas of Ghana were involved over a 12-month period in 2018-19 were (Sadhu et al. 2020: 181-2): • Maintenance activities such as weeding (28 per cent of children working in cocoa) and carrying water for applying agro-chemicals (24 per cent); • All harvest activities from plucking the cocoa pods (16 per cent), to gathering and heaping them (53 per cent), to breaking the pods and preparing the seeds for fermenting (27 per cent); • Post-harvest activities, in particular carting the fermented cocoa beans (25 per cent) and drying them (24 per cent).  Abenyega and Gockowski (2003).
While boys and girls participate in many of the same activities and shoulder a similar workload in tasks involving transportation, an earlier survey undertaken by the Ghana Ministry of Manpower, Youth and Employment in 2007-08 revealed that boys participated significantly more than girls in tasks such as applying pesticides, mistletoe control, and felling trees and burning -all of which had a male/female ratio over 5 (MMYE 2008). Age also shaped the types of work in which children participated. According to the farmers, the youngest children, aged 5-12 years, mainly engaged in weeding and gathering and carrying cocoa pods a short distance to the sites where the beans were separated from the pods for fermentation. It was also reported that they were involved in carrying water for spraying pesticides, and turning the beans during fermentation. The oldest cohort of children, aged 15-17 years, contributed significantly to the harvesting of pods, pod-breaking and mistletoecutting. These activities, the communities agreed, were all within the capabilities of the children (MMYE 2008).

Prevalence of children's work and harmful work
As mentioned, there are slight differences between international ILO and Ghanaian labour standards, particularly regarding the minimum age for undertaking light work, and the preservation of time 8 The survey offers statistical analysis of changes in relation to the earlier surveys in 2008-09 and 2013-14, including significance of difference values, but does not provide these values for differences across gender and age cohorts in the report. It is possible that the values could be calculated if access to the original data set could be obtained.
for schooling (Table 2). Likewise, the definitions of hazardous work vary, with the Ghanaian standards being more elaborate regarding the organisation of work. This difference makes it more difficult to assess across different studies whether the prevalence of work categorised as 'child labour' exceeds the minimum age standards only or also involves hazardous work.
Ghanaian labour standards set out daily limits for permitted work that are closely aligned with national strategies for school education and therefore distinguish between school days and non-school days. These standards also apply to children over the minimum age for admission into employment.
Using the ILO standards for children's work, the 2018-19 survey by Sadhu et al. (2020) suggests that the majority of children worked much less in cocoa than the hours permitted for their age cohort (Table 3). 8 It is also clear from the survey that the proportion of children who worked more than the permitted hours was highest for the youngest cohort who, according to both the ILO and Ghanaian labour standards, are too young to work. However, an assessment of whether such work is potentially harmful due to the hours they work needs careful consideration. It is important to know more about what constitutes 'work' for children within this age group who, as mentioned above, often accompany parents to the farm but only gradually are obliged to work. Rather, in this age group children are encouraged to learn the tasks involved in their family's livelihood activities and to participate in the chats, banter and social life that are also part of work and being present on the farm, particularly when working in groups with neighbours and relatives (Buono and Babo 2013). Work, in other words, is part of normal childhood (Bourdillon 2014). For the other age cohorts, an assessment of potential harm linked uniquely to the working hours would require more information about temporal aspects of their work related to seasonality and the timing of their labour input (e.g. school day vs non-school day) in accordance with the Ghanaian labour standards. Finally, in addition to work in cocoa, children also work in other agricultural activities as well as in off-farm activities, so a narrow focus on one part of the livelihood portfolio is insufficient from a child protection perspective.
While the standards make a firm distinction between permitted and unacceptable work timewise, the delineation of hazardous work is more fuzzy ( Table 4). The international standards label all work exceeding the permitted working hours as hazardous work, as well as activities such as: land clearing; the carrying of heavy loads; exposure to agro-chemicals; the use of sharp tools; working long hours and doing night work. The Ghanaian standards are more elaborate, but also more aligned with rural childhoods than the international standards. Accordingly, longer working hours are seen as unfavourable, but only considered as hazardous work if impeding schooling. In addition to the physical hazards of particular types of work shared with the international standards, the Ghanaian standards emphasise social aspects and deem as 'hazardous' work that is done outside the normal social context, or without protective clothing.
The 2018-19 survey analysed the exposure to 'child labour' in terms of exceeding the permitted working hours and the exposure to 'hazardous child labour' in terms of engaging in one or more hazardous activities. The data show that both types of exposure were highly gendered and linked with age. Of the children exposed to child labour, 58 per cent were boys and 42 per cent were girls.
Regarding age, 47 per cent were aged 5-11 years, 30 per cent were 12-14 years, and 23 per cent were 15-17 years. A similar pattern is revealed for children exposed to hazardous child labour: 59 per cent were boys and 41 per cent girls, while the distribution across age cohorts was 43 per cent among those aged 5-11, 32 per cent among those aged 12-14, and 25 per cent among those aged 15-17 (Sadhu et al. 2020: 183). These data suggest a consistent pattern of boys being more involved than girls in cocoa work that exceeds the permitted hours and in activities deemed hazardous.
Had the data been analysed to the Ghanaian standards, which take into account daily working hours, a more nuanced understanding might have been revealed of how labour allocation and educational aspirations intersect. Due to the different age cohorts, more children in the youngest cohort (5-12 years) would have been deemed to be engaged in child labour, while fewer children aged 13-14 years would have been labelled as exposed to child labour if their work in cocoa did not interfere with their schooling. The proportions for those aged 15-17 years would likely differ, to account for the standards aimed at protecting those who attend school.
Scrutinising the 2018-19 survey data on hazardous activities vis-à-vis the Ghanaian standards, our analysis reveals that no children aged 5-17 years were working full-time instead of attending school. However, 56 per cent of children were exposed to one or more activities considered to be hazardous; 9 per cent missed school in order to work in cocoa production; 3 per cent undertook night work, either finishing after 6pm or starting before 6am; 5 per cent experienced working in isolation; and 16 per cent lacked protective clothing. Unsurprisingly, children's participation in hazardous activities was largely determined by a pattern of work allocation and labour bottlenecks. Thus, the high participation of children in harvesting activities was reflected in their exposure to harm from using sharp tools (50 per cent), carrying heavy loads (10 per cent), and climbing trees (6 per cent). Involvement in maintenance activities resulted in exposure to the heavy work of land clearing (14 per cent) and exposure to agro-chemicals (14 per cent). Post-harvest activities, primarily relating to the fermenting and drying of cocoa beans, were not considered hazardous (Sadhu et al. 2020: 110).
Notions of gender-and age-appropriate work are reflected in the proportion of children participating in the different tasks and thus in their exposure to hazards (Table 5). For example, land clearing is often regarded as a male activity as it involves heavy work, such as cutting down trees and bushes, as is reflected in the higher proportion of older boys being exposed to such work (19 per cent boys compared to 8 per cent girls). The three hazardous activities in which children participate the most -carrying heavy loads, exposure to agro-chemicals, and using sharp tools -are particularly interesting because they relate to the gender division of work and possibly also to the gender differences in adults' access to labour within the family. They are also subject to evaluations of which types of work are appropriate for a child to do in relation to age and maturity. Yet, with its focus on the prevalence of child labour and hazardous work, the survey by Sadhu et al. falls short of examining if there is a link between who makes claims on, or recruits, children's labour and the conditions in which they work. Moreover, it is important to explore the degree of choice that children have regarding their labour contributions within and outside the family.
When considering the prevalence of carrying heavy loads, the survey suggests that this was most typically associated with gathering and heaping cocoa pods (34 per cent), moving wood and other loads carried when clearing land (26 per cent), transporting fermented beans (25 per cent), and carrying water for the application of agro-chemicals (15 per cent). These types of loads were all carried frequently (Sadhu et al. 2020: 204), but the data do not provide information on the distance travelled or the gender or age of the child carrying the load. Such an appreciation would seem vital, given the importance of norms surrounding gender-and age-appropriate work, which might explain why adults call on their children to do specific work, or the tendency for children to engage in these tasks to demonstrate their maturity and inclination to work. The carrying of heavy loads, for example, is an integral part of reproductive work, and girls frequently carry water, crops, and firewood for long distances. In relation to cocoa production, there is hardly any gender difference in the carrying of heavy loads -34 per cent of boys are involved in such work and 30 per cent of girls (Table 5).
The study further reveals that a child is exposed to agro-chemicals if participating in the spraying of fungicides or insecticides, but this also includes preparing the mixture and cleaning the equipment. 9 Additionally, children are seen to be involved in hazardous work if they carry water for spraying, are present or work in the vicinity of a farm during spraying, or if they re-enter a sprayed farm within less than 12 hours of spraying (Sadhu et al. 2020: 33-4). Table 6 shows the differences in exposure related to notions of gender-and age-appropriate work. Girls and the youngest cohort of children are predominantly exposed to agro-chemicals when carrying water to the cocoa farm and through being present in the vicinity of the farm during spraying or shortly after spraying has been done. Spraying is mostly carried out by boys over 15 years of age and adults, while boys aged 12 years and above are involved in work with agro-chemicals (MMYE 2008). Thus they are at risk of coming into direct contact with the chemicals, while those working in the vicinity of a farm being sprayed or re-entering the farm too soon after spraying are exposed to fumes. As previously noted, fungicides are applied two to three times annually and insecticides three to four times annually, but it is not known whether the same children undertake these tasks, or whether they are present in the vicinity of the cocoa farm each time spraying takes place. As such, it is impossible to assess the potential harm caused by exposure to agro-chemicals using the available data.
The use of sharp tools is the most commonly reported hazardous activity to which children are Table 5. Percentage of children exposed to hazardous activities in cocoa, 2018-19
exposed in cocoa production. According to Sadhu et al. (2020: 34), this includes the use of machetes and long cutlasses for weeding, handling motorised equipment or machines, using a knapsack sprayer and/or a chainsaw, harvesting with a machete or sickle, harvesting overhead cocoa pods with a harvesting hook, or breaking cocoa pods with a knife or a sharp object/tool. However, it is important to note that the prevalence of children using sharp tools is not confined to cocoa farms alone. Children in rural households are socialised into a variety of tasks, both on the farm and in the domestic sphere, that require the use of machetes, sickles and knives. Both boys and girls therefore become skilled in using sharp tools from a relatively young age (Berlan 2009). The awareness of harm is mostly related to physical injuries and, as different surveys have shown, children experience injuries ranging from machete wounds, tree stump injuries, slips and falls, fingernail pricks, thorn pricks, snake bites, leg/neck pains, small objects entering the eyes, skin rashes, and itchy backs. However, children do not judge these afflictions to have long-term effects on their health and wellbeing (MMYE 2008;Sadhu et al. 2020). In 2018-19, the most common injuries and health complaints related to agriculture among children aged 5-17 years were: wounds and cuts (28 per cent), back pain (1 per cent), muscle pain (no incidents), and skin itches and scratches (10 per cent), with not much difference between work in cocoa production and work in the production of other crops (Sadhu et al. 2020: 80).
Little information is available about the circumstances in which children engage in different types of activities on the cocoa farm. For example, available research does not generally indicate whether children primarily work with one or both parents, whether they help out a relative or someone else, or whether they do casual work in their community or elsewhere for a wage. It is important to understand the specific conditions or social relations which surround the work being undertaken. Indeed, research in other contexts has shown that whether or not work is experienced as hazardous or harmful may be more closely connected to its social context and the relationships in which it takes place, than to the nature of the work itself (Bourdillon et al. 2010).
Children may also work as part of their schooling, as was recorded in an earlier child labour survey by the Ministry of Manpower, Youth and Employment in the cocoa areas of Ghana (MMYE 2008). In the 2006-07 farming season, this report suggests that 13.3 per cent of surveyed school children worked for teachers and 11.5 per cent for a school contract farm (ibid.:11). Berlan's (2009) 15-month study in Ghana's Ashanti region further documents how children were required to do farm work for their teachers at least once a week, typically being allocated the task of clearing or weeding school plots with machetes under little supervision. In comparison, children in the study noted that weeding their family farms involved lighter work, such as maintaining rather than clearing land, and in particular, work on the family cocoa farm offered shade and protection from the sun, which was absent when clearing the school plot. Importantly, the study suggests that child labour and education are not always mutually exclusive, as is often assumed to be the case. Moreover, the children reported that work carried out on family-owned cocoa farms was both safer and less strenuous

Work and school
The theme of education, and in particular schooling, has been prominent in debates about child labour, both because schooling is one of the pillars in the universalised notion of childhood and because the need to, or lure of, work is perceived to hinder educational achievement (Bass 2004;Boyden 1997). Yeboah (2019: 162) notes that despite the view in Ghana that universal basic education is a means to improving the future of marginalised rural children, policies have not gone sufficiently far to overcome the financial strain and gender disadvantage to allow poor children to remain in education. Some scholars argue that the value ascribed to formal education among adults and children can obstruct children's retention in school, either because they do not find the curriculum suitable for rural livelihoods, or the provision and quality of teaching is substandard (Berlan 2009;Hashim 2007;Yeboah 2019). Moreover, parents may consider the age of a child when striking a balance between schooling and work, allowing younger children to attend school and withdrawing older children who are of an age when they can do more work (Yeboah 2019: 165-6). However, withdrawing a child from school to work some days during the agricultural peak seasons does not necessarily imply that they drop out of school altogether. Many poor children combine work and education in order to remain in school. They work to contribute to their own school-related expenses and sometimes also to demonstrate their commitment to the family to obtain the necessary financial assistance to continue their education (Howard, Jacquemin and Thorsen 2018).
School enrolment in cocoa-growing areas is generally high and a study carried out in 2007 revealed that many household heads had been to school for an average of 8 years (Abenyega and Gockowski 2003: 7). In 2007, 87.9 per cent of the children in the study aged 5-12 years were enrolled in school, as were 95.5 per cent of the 13-14-yearolds and 83.9 per cent of the 15-17-year-olds (ibid.). The 2018-19 survey by Sadhu et al. (2020: 93-8) found that, between 2008-09 and 2018-19, school attendance in the previous 12 months among children in cocoa-growing areas increased from 89 per cent to 96 per cent generally, and from 91 per cent to 96 per cent among children in cocoa-producing households. There was no discernible difference in school attendance between working children, children exposed to child labour, and children exposed to hazardous child labour in any of the three age cohorts. The 15-17-year-olds were slightly less likely than the younger cohorts to attend school and their attendance only increased from 88 per cent in 2008-09 to 89 per cent in 2018-19.
However, as the enumeration of educational data is based on a 12-month recall period, attendance data show whether a child was enrolled at school but fall short of illuminating the frequency of school attendance. It is thus impossible to examine whether children's work impacts on their learning and performance at school, or whether there are differences between households producing cocoa and those who do not. Moreover, as school attendance reflects a range of issues, of which the need for children's labour on the farm is only one, it is important to explore the reasons why children do or do not go to school. Finally, the survey does not examine whether the children working in cocoa work only with members of their own household, whether they take on casual work or whether they migrate temporarily to other cocoa-growing areas or elsewhere. These issues are all important for understanding the links between children's work in cocoa activities and their schooling.

Trends in children's work
An examination of the changing nature of children's work in farming households in the cocoa region provides an interesting picture. According to statistical analysis carried out by Sadhu et al. (2020), in the ten-year period between 2008-09 and 2018-19, the proportion of economically active children increased by 5 per cent, but this rise was mainly reflected in non-agricultural work ( Table 7).
The study further suggests that although children's work in agricultural activities has remained stable, there has been a major shift from work in other crops towards work in cocoa production, which Sadhu et al. (2020: 50) relates to the 36 per cent increase in cocoa production during the same period.
The increase in the proportion of children working in the cocoa sector over the ten-year period begs a closer examination of how gender and age influence children's labour contributions.
According to Sadhu et al.'s (2020: 189) analysis, the prevalence of children in cocoa-growing households who worked more than the permitted hours dropped by 3 per cent for boys in the youngest cohort (5-11 years) and 9 per cent for boys aged 12-14 years, while it remained the same for boys aged 15-17 years between 2013-14 and 2018-19. For girls, the prevalence remained the same for the youngest cohort but decreased by 9 per cent for the 12-14 years cohort and just 1 per cent for the 15-17 years cohort. No significant change was recorded in the overall prevalence of child labour or hazardous work during the same period (Sadhu et al. 2020: 190). However, a comparison across high-, medium-and low-producing areas reveals regional differences. In the low-producing areas, the prevalence of child labour among children from all agricultural households increased by 26 per cent between 2008-09 and 2018-19. In the same period, the prevalence of children doing hazardous work increased by 25 per cent. In comparison, the prevalence of child labour increased by 16 per cent in medium-producing areas and decreased by 5 per cent in high-producing areas. Similar trends were found in the prevalence of hazardous work, which over the ten-year period increased by 12 per cent in the medium-producing areas and decreased by 7 per cent in high-producing areas. The authors of the study note that this indicates some success of interventions addressing children's work, which have primarily targeted the high-producing areas (ibid.: 103).
The same study reported that when examining the range of hazardous activities, the use of agro-chemicals stood out. The proportion of children exposed to agro-chemicals increased significantly in all agricultural households between 2008-09 and 2018-19 (Table 8), from 7 per cent to 36 per cent for boys, and from 6 per cent to 27 per cent for girls. The increase was higher for older children than for the youngest cohort. Data from cocoa-producing households reveal that the proportion of households using pesticides increased from 66 per cent in 2013-14 to 82 per cent in 2018-19, while the use of herbicides rose from 51 per cent to 75 per cent between the two surveys (Sadhu et al. 2020: 53). The proportion of children exposed to agro-chemicals had already increased in 2013-14. By 2018-19, exposure in cocoa-producing and non-cocoa-producing households was very similar. Children's exposure to agro-chemicals is thus perhaps best explained not as a direct consequence of their participation in the cocoa value chain, but rather due to the effects of broader transformations in the agricultural sector.
Noteworthy changes have also happened across gender and age categories. Overall, boys are more involved in the work dealing with agro-chemicals, and although the prevalence of involvement in spraying decreased by 17 per cent between 2013-14 and 2018-19, the gender balance was still skewed, with 80 per cent of boys and 20 per cent of girls having sprayed in 2018-19. Similarly, for involvement in handling agro-chemicals (purchase, transport, storage, mixing, loading, and cleaning of equipment), the prevalence among boys decreased by 21 per cent, with a gender balance of 72 per cent of boys and 28 per cent of girls in 2018-19. The relative decrease among boys was absorbed by the involvement of girls, whose participation in spraying and handling agro-chemicals rose by 17 per cent and 21 per cent respectively. With the exception of re-entering a sprayed field within 12 hours, the same pattern was apparent as in other activities relating to agro-chemicals. Looking at which age groups were more likely to engage in the work revealed that there was a decrease in the involvement of 15-17-year-olds in spraying (-15 per cent). This was because of a drop in those in this age cohort who reported working in the vicinity of the farm sprayed (-25 per cent) and handling of agro-chemicals (-32 per cent) between 2013-14 and 2018-19. This work had shifted to younger children, and often disproportionally to those aged 5-11 years (Sadhu et al. 2020: 195-6).
The report does not offer insights into why these shifts have happened and it would be important to have a better understanding of why and how tasks are allocated to children of different ages.
Exposure to land clearing, the use of sharp tools, and carrying heavy loads also changed between 2013-14 and 2018-19, but differently for different age groups (Table 9). The prevalence of children in cocoa-growing households who participated in land clearing increased significantly, especially among Notes: † Significance of difference *** p<0.01.
* The comparison of changes within all farming households and within cocoa-producing households is hampered by the sampling methods used in the three surveys, which only allow certain comparisons.
the 15-17-year-olds (29 per cent increase) and the 12-14-year-olds (21 per cent increase). However, changes in prevalence of children using sharp tools remained more or less the same, except for the 12-14-year-olds, for whom prevalence increased by 10 per cent. Notably, children were less likely to carry heavy loads across all age groups (Sadhu et al. 2020: 154).

Drivers of children's work and harm in cocoa production
In many cocoa-farming communities in Ghana, the predominant view among adults is that children are drawn into work on the farm to learn the essential skills of farming, but also because their labour is appreciated and needed. Yeboah (2019) argues that from parents' perspectives, this social context is vital to impart moral, cultural and social values to the younger generation. Children also learn practical skills in this sphere, often through interactions with their siblings and peers. Farm work does sometimes get in the way of schooling, especially when work also serves to incorporate children into the network of kin. But the reliance on maternal relatives, for example, to help in times of hardship or more regularly with school expenses, drives reciprocities in terms of supplying labour during peak seasons (Yeboah 2019).
From children's perspectives, however, contributing to the wellbeing of the household and the family is ingrained in the habitus of rural childhoods.
Work on the farm and in the household is seen as an everyday activity carried out when not in school, and sometimes at the expense of revising schoolwork at home and of attending school. Systemic failures in the school system of providing continuous and quality teaching in rural areas may also impact adults' and children's choices about balancing work and schooling, as we discuss in more detail below (Yeboah 2019).
From an institutional perspective, there is often a general assumption that poverty is the main driver of child labour. For example, according to the World Cocoa Foundation: We now know that child labour is both a symptom and a self-perpetuating cause of poverty. Households in cocoa growing areas face the realities of rural poverty, and some parents have little choice but to put their children to work, and keep them out of school, to reduce Table 9. Prevalence of all children in cocoa-growing households exposed to the most frequent types of hazardous work Studies of children's work in cocoa are, however, often ingrained in two different positions on the relationship between child labour and poverty. One, often upheld by advocates for banning child labour, argues that the employment of children is due to poor farmers' inability to pay adult wages and thus their inability to attract young men over the age of 18 to work on the farms. The other, which heralds a child-and family-centred perspective, emphasises the use of family labour and argues that farmers who lack money to hire workers need to balance the desire to increase production with the desire for their children to concentrate on schooling and for their out-of-school children to engage in their own economic activities (Thorsen 2012).
Poverty is clearly an important dimension that shapes whether or not children end up in child labour and, in many cases, children are faced with limited alternatives to working -if they do not work, they do not eat. 10 In situations of desperate poverty, extra demands may be put on children to contribute to farming activities, either on the cocoa farm or plots for food production. However, as the cocoa sector has become liberalised in recent years, and globally transformed into a more intensive production system that relies increasingly on the use of inputs, poverty has been exacerbated for many small cocoa farmers. As Owusu and Kwarteye argue, the 'invisible labour' of children has become even more vital, as cocoa farmers are forced to reduce labour costs: 11 Farmers tend to use the cheapest form of available labour due to the influence of producer prices, yields and excessive taxation. Coupled with this, the unstable global price index as well as high wages of hired farm labour compels cocoa farmers to reduce labour cost on the farm by employing children. (Owusu and Kwarteye 2008: 2) In extreme cases, there have also been accounts of children being sold to traffickers to work on cocoa farms (Ould et al. 2004). It is therefore likely that children's ability to exhibit agency and make meaningful choices in their lives is constrained by structural poverty and household dynamics (Berlan 2013;Boas and Huser 2006), which limits the options available to them (see also Chapter 7 in Bourdillon et al. 2010).
Luckstead, Tsiboe and Nalley (2019) argue that the number of children working in hazardous work in cocoa production in West Africa has increased in recent years due to the introduction of high yield, disease-resistant varieties of cocoa that demand more labour to harvest and process. 12 But at the same time, Amanda Berlan points out that children on family cocoa farms in Ghana constantly negotiate these challenges. For example, from an early age, children are taught to use machetes -a potentially dangerous tool -and are not provided with protective clothing. Yet they work under family supervision, learn to handle their tools competently, and show pride in their work and skills (Berlan 2009: 142-4).

Discourses and policy
13 A study published by IITA (2002) pointed to significant differences between the four main cocoa-producing countries in West and Central Africa, highlighting the distinct histories of mobility that should inform the analysis of social relations and migrant labourers.
The Harkin-Engel Protocol is emblematic of interventions to abolish child labour in cocoa production. Negotiated and signed in 2001, the protocol encouraged child protection work that was already underway in the ILO's International Programme on the Elimination of Child Labour (IPEC) and ILO Convention 182 on the elimination of the worst forms of child labour. International policies in the late 1990s and early 2000s were based on the assumption that children working in cocoa were either family labour and not really in the target group of interventions at that time, or migrants who were employed as farm hands, exploited, exposed to hazardous work, and often victims of trafficking (Dottridge 2002;ILO-IPEC 2007c). This perspective on children's work was informed by studies and media reports capturing the negative experiences of child and youth migrants in Côte d'Ivoire (Ould et al. 2004;Sackett 2008), sometimes without paying attention to country-specific data. 13 Moreover, this position did not take into consideration the many children who worked on the family farm and migrated to work without being deceived by recruiters or exploited by farm owners (Dottridge 2011). As a consequence, early policy and programme responses were preoccupied with preventing migrant child workers in the cocoa-growing regions.
Legislation in Ghana that stipulates compulsory education and regulation of the hours that children work pre-dates the Harkin-Engel Protocol. As research with Ghanaian cocoa producers did not find substantial evidence of child labour migrants employed in cocoa farms, early policy discourses focused primarily on hazardous work and the work-school nexus (MMYE 2008;Mull and Kirkhorn 2005). The Harkin-Engel Protocol instigated a cocoa certification process in 2001, coordinated by the Cocoa Verification Board (CVB), which aimed to stamp out the worst forms of child labour. However, the CVB failed to unite the various stakeholders to meet the goals set out in the protocol. Simultaneously, certification programmes -led by the International Cocoa Initiative (ICI), the Fairtrade Foundation, the Rainforest Alliance and international cocoa buyers (among others)proliferated and started training and sensitisation programmes for producers (Clark and Gow 2011).
Interventions involved both 'upstream' and 'downstream' actions of creating awareness of what kind of work was hazardous to children. Upstream actions targeted district and central governments to encourage a change in attitude in the enforcement of regulations. Moreover, early upstream actions built institutional capacities to address the problems and prepare the ground for developing child labour monitoring systems as part of the certification of cocoa (Baah 2010). The process of establishing child labour monitoring systems prompted rigorous fact-finding schemes, which have been very effective in mapping the use of child labour in cocoa production (Sadhu et al. 2020;Tulane University 2015).
Downstream actions were mainly aimed at creating awareness about child labour and hazardous work among cocoa producers, trade unions and non-governmental agencies. The West Africa Cocoa and Commercial Agriculture Project (WACAP) was central to establishing community child labour committees, whose members received training and were actively involved in sensitisation activities at the local level (ILO-IPEC 2007c). Although the approach was framed in terms of discussions or consultations with various stakeholders, due to time constraints and the need to start interventions before a change in attitudes had happened, the process was one of sensitising stakeholders about universalised notions of what was right and wrong for children (Thorsen 2012). At this time, ICI worked to raise awareness among radio and television broadcasters, which in turn stimulated discussions on air with the participation of radio listeners (ILO-IPEC 2007b).
The mobilisation of villagers to monitor children's work and help disseminate information was perceived by programme leaders as a sign of changing attitudes among cocoa producers. A labour survey in 2007-08 suggested that sensitisation programmes had revealed that 76 per cent of the people sampled in six cocoa-producing regions of Ghana were aware of the prohibition of the worst forms of child labour, although awareness was generally higher concerning the application of agro-chemicals and the dangers of transporting heavy loads (MMYE 2008). However, Yeboah's (2019) research with child protection NGOs revealed a disconnect between the NGO field officers' views on what constituted an ideal childhood and how villagers understood ideas about child rights and child protection. This disconnect rendered the committees ineffective and calls for in-depth research with parents and leaders at the village level to understand their concerns and viewpoints. Moreover, given the plethora of reports of monitoring programmes run by Fairtrade and the Rainforest Alliance, a detailed analysis of their evaluations should also be undertaken.
The emphasis on the work-education nexus resulted in programmes focusing on education or technical training as a means to prevent children from being engaged in hazardous work, and as alternatives for children who had been removed from such work (MMYE 2007;Tulane University 2010). However, an important lesson learned from early WACAP interventions is the need to carefully assess and map out educational resources at the outset. Lack of access to formal education in rural areas, due to inadequate infrastructure, absence of teachers, or the inability of parents to pay formal and informal fees, implies that education does not always constitute an alternative to agricultural work (Thorsen 2012).
Sadhu et al.'s (2020) study shows that improved accessibility and affordability of schools facilitated the enrolment of children and reduced the time they would work in cocoa production, whereas assistance with the provision of school materials and uniforms had no significant effect on children's exposure to child labour or hazardous work. Local vocational training programmes for children who were no longer in school were mostly pursued by girls, because they provided a foundation for economic activities outside farming. The most effective means to address exposure to hazardous work appeared to be occupational safety and health training, which children and youth appreciated, but which was most effective if the parents had also received training (Sadhu et al. 2020: 9).

Conclusion
Over the past two decades, with the rise of traceability and ethical sourcing agendas, considerable debate has emerged over what role, if any, children should be allowed play in cocoa supply chains. At the heart of this debate has been the issue of how harmful and hazardous work is defined and understood, and by whom. A clear-cut division exists between those critics who view all children's work as harmful and unacceptable, and those who favour a more nuanced and textured response to understanding the role of work in children's lives. However, in cocoa production, and agriculture in Africa more broadly, the distinction between the two is sometimes difficult to discern, as the hazards, risks, and benefits of work are often intertwined. Children and their families on cocoa farms find themselves weighing up both the costs and benefits, making trade-offs, and situating this assessment within the realistic alternatives that exist.
The review provided in this background paper highlights the fact that there have been numerous research projects, policies, and programmes designed to address child labour in cocoa production over the years, but many have had limited success. Moreover, different standards and criteria are often applied in assessing child labour (e.g. ILO standards for hazardous work versus Ghanaian legal standards), which can lead to very different outcomes. Likewise, in some cases, studies that have assessed children's work, hazards or potential harm in cocoa production have lacked sensitivity to both age and gender. But perhaps more serious is that many 'objective' sets of criteria or data lack understanding of the context in which children contribute to cocoa production. This can lead to a misdiagnosis of the situation as well as the development of inappropriate and, at times, harmful policies and action plans to address the issues at hand.
The abstraction of labour practices and judgements about the role that children play in cocoa production, in isolation from their wider social, cultural, and economic contexts, is problematic. As became evident while undertaking this review, while poverty may be a clear driver of children's work, in cocoa communities there are numerous other reasons why children may choose to labour on farms. Work on the cocoa farm may allow a child to attend school, gain practical skills, or build social capital. In many cases, it is also clear that children understand and experience the work they do as a pathway through which to attain maturity, selfesteem, and social responsibility before entering adulthood.
The ACHA programme aims to deepen understanding and provide a more dynamic assessment of children's work in Ghana's cocoa sector. As we have seen, an understanding of the specific conditions or social relations that surround the work being undertaken is crucial. Ultimately, whether or not children's work on cocoa farms is experienced as hazardous or harmful may be more closely connected to its social context and the relationships in which it takes place, than to the nature of the work itself.