1The media have often presented child labour as one of the most striking symbols of poverty. Although it first came to attention in the western world with the advent of the industrial revolution (Schlemmer, 2006), child labour, particularly in developing countries, was thrust into the spotlight in the mid-1990s with the Convention on the Rights of the Child in 1989 and the creation of the International Programme on the Elimination of Child Labour (IPEC) by the International Labour Organization (ILO). Indeed, child labour is primarily observed in developing countries, implying that poverty is the main explanation for its continuing existence. Fallon and Tzannatos (1998) presented findings for a panel of countries which seem to confirm the poverty hypothesis overall. However, it was confirmed only for countries where the average per capita income is below USD 600. In countries where average income was above this threshold, the relationship no longer holds true, even though there are large variations in the prevalence of child labour within this group.  Many studies nonetheless corroborate the poverty hypothesis (Basu, 1999; Basu and Tzannatos, 2003; Edmonds and Pavcnik, 2005; Edmonds, 2005).
2Using data from the 2005 survey of households in Madagascar, Lachaud (2008) demonstrated that the poverty hypothesis is relatively robust for that country. According to the Madagascar National Institute of Statistics, 68.7% of the population lives on less than one dollar a day (INSTAT, 2006).
3In the wake of ILO efforts, a general consensus to abolish child labour was reached. Such an abolitionist approach, however, clouds the debate on legitimate forms of child labour and what constitutes exploitation (Bonnet and Schlemmer, 2009), although ILO Convention 182 on the worst forms of child labour details what are considered to be its most extreme manifestations (ILO, 1999).
4This study analyses a specific case of child labour, that of child begging, based on a pilot study conducted in Antananarivo. Begging can be considered an extreme form of child exploitation in certain cases.  Rohinton Mistry’s novels paint a vivid picture of how child beggars are raised in the slums of Bombay, India (Family Matters, 2003), and the forced begging of trafficked children  from Roma communities in Europe is also a worrisome phenomenon (Joppart, 2003). The topic remains marginal in academic circles, however. In fact, there are no statistics on child begging (Vargas-Barón, 2005) for several reasons. First, the distinction made by UNICEF (1984) between children “of the street” and “on the street” has contributed to splitting the analyses of street children into two categories: those on street-living children who have fled their families and sleep in public spaces, and those on street-working children who work on the streets during the day and return to their family home to sleep. Begging is generally associated with cases of children who have fled the family home, while research on child labour generally assumes that children work for the household to which they belong. Child begging is thus generally viewed as just one survival tactic among many for street-living children (Patel, 1990; Kombarakaran, 2004; Ballet, 2006, for a description of other such activities) and is not associated with child labour that occurs within a familial setting. Second, aside from cases of children who have left their family homes, child begging is often analysed from the viewpoint of highly specific cultural contexts, for instance in countries where Koranic schools make children beg as a learning activity (Loimeier, 2002; Perry, 2004). In such cases, child begging is viewed as separate from the problem of child labour. Child beggars are therefore often depicted as one of several types of children in difficulty or in danger. 
5This short paper follows the same reasoning as the literature on child labour, and thus runs contrary to most of the studies on child begging. An eyewitness account from the NGO Enda-Océan Indien, recorded by Marguerat (2003a), stated that young children begged for their parents in Madagascar. Ravololomanga and Schlemmer (1996) indicated likewise. We have sought to go beyond these accounts and to verify the hypothesis that children beg to generate income for their families. The veracity of the household poverty hypothesis does not come under scrutiny, however, even though our findings leave little doubt about this aspect. We wanted to find out if children are being exploited by their families or if they beg because they have fled their families. In cases of the former, we sought to understand the strategies used by families to induce their children to beg.
6In keeping with child labour analyses, we assumed that children are under the authority of the head of household and are forced to work to augment the family income.  In this short paper, we will explore the means of coercion employed by parents. In particular, we attempt to find out through this survey whether parents engage in violent behaviour to force their children to beg. The survey results show that families use different strategies depending on the age of the children. When age increases, coercion is centred less and less on direct violence.
7In the first section, we present the survey method as well as descriptive statistics. In the second section, we propose a typology of child beggars based on a correspondence factor analysis. In the third and final section, we discuss our results and conclusions.
I – Survey method and descriptive statistics
8This was a pilot survey designed to identify the main issues related to child begging in order to pave the way for a larger-scale study. It does not claim to capture all aspects of child begging.
9The survey was conducted from early September to late October 2009 in two major phases. The first phase involved identifying the areas in the city where child beggars worked and contacting children from the principal begging sites, which covered an area of 67 hectares.  A total of 100 children from all sites were selected. This number was chosen to obtain an adequate statistical base while avoiding a lengthy survey procedure.
10The second phase comprised a series of interviews of the children. To earn their trust, the survey was carried out in several stages. First, researchers identified child beggars and made contact through an initial conversation. They were then approached again on several occasions to establish a climate of confidence. The children opened up very quickly, however. Once they were given attention, they were very eager to talk and to provide the information requested. No child refused to speak. This survey method was possible because the children generally have well-defined begging territories. As pointed out by Ravololomanga and Schlemmer (1996), begging is very territorial in Antananarivo. Even if some children work an area rather than a fixed location, their patch is generally very small and finding them is easy. Examples of these areas are the bus terminal near the university, the market, downtown arcades, etc. Children are often found in places where their activities can be structured (Morelle, 2006).
11Interviews consisted of open discussions with the children carried out in Malagasy.  Once the child had left, the information collected during the discussion was written down so that details would not be forgotten. To minimize mistrust and to avoid mechanical answers, there was no written questionnaire during the interview (the children could have guessed at answers expected by the interviewer). Moreover, discussions with the children were possible because there were no adults with them. To avoid bias, we never conducted interviews if an adult was present. This eliminated children who always begged in the presence of an adult, primarily the youngest ones, aged below 4, who are supervised by their parents or another adult. At age 4 and above, as we highlight in the data, many children are obliged to fend for themselves while begging. 
12The respondents were chosen at random, as soon as a child or group of children was spotted. In other words, we did not select a specific group. Once a group was identified, all of the children present were involved in the discussion so that the interviews covered them all. This choice entailed a risk of a cluster effect that could lead to overrepresentation of children from the same family, thus leading to selection bias. However, it would have been difficult to interview the children separately, for reasons of both efficiency and ethics. Interviewing a lone child would have created an “interrogation” effect and would have potentially weakened the reliability of the information gathered. 
13The survey cannot claim to be perfectly representative, as no information was available to employ a selection method other than that of a random survey. Moreover, there is no information on the prevalence of begging in Antananarivo. Owing to our choice of method, we only interviewed children who always begged in the same area. Children who wandered around the city to beg were not identified or interviewed. This constitutes an undeniable limitation to our findings.
Characteristics of child beggars and organization of begging
14Three categories of information were collected from child beggars. The first covered personal characteristics, age, sex, and whether or not they lived with their parents. The term “parents” is used in the broad sense to mean all the adults of the household, and is not limited to the biological parents. The second category covers the situation of the parents, specifically their employment. Finally, the third category concerns the organization of begging.
15Regarding the characteristics of surveyed children (Table 1), note that no “child” above the age of 14 was interviewed during the survey. This is hardly surprising, since children turn to other activities beyond a certain age. Marguerat (2003b) pointed out that begging in Antananarivo is the preserve of children aged below 10. Beyond that age, children turn to “more profitable” activities such as washing cars, carrying parcels, etc. According to Marguerat, the critical age is around 10 years, and could be tied to the fact that younger children inspire more pity than older ones, especially since Christianity has a strong presence in Madagascar. That said, our survey shows that some children beg up to the age of 14, a phenomenon highlighted by Bousquet in 2003. Children start begging on their own at age 4 or 5. This does not mean that children below 4 do not beg, however; rather, they are accompanied by relatives or an adult as they are generally not sufficiently autonomous to be left unattended. Our choice of survey method excluded these children.
Characteristics of child beggars interviewed
Characteristics of child beggars interviewed
16Girls are as involved in begging as boys, but girls are less likely to beg in the street beyond age 10 because they find domestic work as housemaids (Ravololomanga and Schlemmer, 1996). A cross-tabulation by age and sex (Table 1) shows that girls are represented in all age categories and that there is no significant difference between girls and boys in the structure by age. The fact that we observed children – both boys and girls – who are older than in previous studies may be a sign of increased poverty in some households and of the inability to find more lucrative activities for the children.
17Finally, the majority of child beggars live with their parents. Rasamimanana and Revelonanosy (2010) point out that in Madagascar, 11.9% of children below 15 do not live with at least one biological parent. The family situation of child beggars is therefore comparable to that of Malagasy children in general. This result indicates that begging is not an activity exclusive to street-living children (under the UNICEF definition, 1984). Bousquet (2003) described the importance of begging among street-living children in Antananarivo and highlighted the role of family crisis and the disorganization of the family unit in this phenomenon, but our survey does not confirm this link. On the contrary, it gives strength to the hypothesis whereby begging is an integral component of family activity.
18We cannot explain this divergence in findings, however. Three hypotheses may be put forward. First, Bousquet’s study is quite different from ours. It focused on an older, largely male population. Second, since 2003, worsening poverty may have pushed entire families into the streets. In this case, the children in our survey may well live with their parents, but in makeshift lodgings that they consider as their “house”. They cannot be considered as street-living children. Third, because of our choice of survey method, children who do not have a fixed begging territory were not interviewed. It is possible that the street children identified in Bousquet’s study, who were generally older, are constantly on the move. In our case, the children begged during the day and went home at night. While begging appears to be a means of survival, this finding supports our hypothesis that children are being exploited by their families.
19The primary activity of these children is begging. They do not go to school, but this does not prove that a dichotomy exists between schooling and begging. Our interviews took place during the day, in other words, at a time when children are supposed to be in school. As a result, our survey can only include unschooled children. However, many accounts by NGO workers suggest that some children beg after school. The 2007 national survey on child labour carried out in Madagascar (INSTAT, 2007) indicated, moreover, that 12.9% of children between 5 and 17 years combined school with work. However, these accounts, gathered on the sidelines of our survey, also indicate that children begging after school hours do so voluntarily, primarily to acquire consumer goods (such as toys) like those of their more well-off peers. This category of children was excluded from our survey, but their situation is outside the scope of our study which focuses on the possible exploitation of children by their families through begging.  For the same reason, our survey did not cover children who work during the day and beg at night. It is limited to those whose sole activity is begging.
20The children in our survey are from families who are unable to pay for schooling. Descriptive statistics indicate, moreover, that their parents are in situations of extreme poverty.
21Data on the situation of parents shed light on the family context of begging (Table 2). Of all the parents (including those of 5 children no longer living with their family), only 23 mothers and 16 fathers have no occupation, and only 2 of the 100 children interviewed belong to households where both parents are inactive. Generally, child beggars belong to households where at least one adult has an occupation, and the majority of children have two active parents. The concept of “occupation” is used broadly and includes begging, collecting garbage, etc. For all parents, the occupation may be considered as informal.
Situation of parents of interviewed children
Situation of parents of interviewed children
22Among mothers, 23 out of 100 are also beggars, and in all cases, they work in casual employment (street vendor, washerwoman, etc.). Fathers also hold jobs that are typical of poor households (porter, coalman, garbage collector, etc.). Finally, some parents do not work owing to illness or disability. All families of child beggars live in extreme poverty.
23The information collected on the organization of begging covers travel to the begging area, begging itself, the way the money is spent and the use of physical violence by parents against children to ensure that they bring back money (Table 3).
24Most of the children go to the begging area on their own, but the youngest ones are more frequently accompanied by their parents. Above age 9, this is rarely the case (Table 4),  a finding which reveals a certain degree of independence among the children, even if it stems from parental negligence. As highlighted by several studies (on Brazil: De Oliviera, 2000; on Peru: Horna Padrón and Ballet, 2011), children acquire skills through their street activity, such as spatial orientation, ability to count, etc.
Distribution of children by age and type of travel to the begging area
Distribution of children by age and type of travel to the begging area
25The survey reveals that children do not beg in the neighbourhood where they live but go to other areas deemed more lucrative. Nonetheless, the movement of children between neighbourhoods is relatively limited, and the great majority of children do not beg in areas that are more than a twenty-minute walk from their lodgings.
26Children have a tendency to beg in groups.  Membership of a group provides company and friendship. It is also a means of self-protection and reassurance against the dangers of the street (particularly assaults and theft perpetrated by adults or adolescents). However, more than a third of children always beg on their own.
27In terms of reasons for begging, 58 children reported being forced by their parents to beg and then hand over the money collected,  while 42 said that they beg to eat and spend the money directly on food. This result leads to a dichotomy between the categories of children. Parents may well force children to beg so that they can feed themselves. However, the above result shows that some parents force their children to beg and then collect the money. Meanwhile, other children appear to be neglected by their parents, which forces them to beg in order to eat. While the final result is the same – children beg and, with the money from begging, are either fed or feed themselves – the rationale behind the activity is different. There is coercion exerted by parents in the first case but not in the second.
28Given the very small sums collected daily, in most cases less than 500 ariary (0.18 euro), the money is used solely for food. None of the children stated that the money was spent on other things when not given to the parents. As with the parents’ occupation, this fact reveals the extreme poverty of the families and the children.
29Finally, 17 children reported that their parents used physical violence to force them to beg and bring back money. This result corroborates the results of several studies that directly or indirectly address the question of violence against children in Madagascar (ODEROI, 2006). Moreover, 16 of the 17 children who said they were beaten if they did not beg also stated that they were forced to beg by their parents. The dichotomy between those who are forced – including through the use of physical violence – and those who are neglected is reinforced by this finding.
30At this stage, descriptive statistics suggest that there are at least two different rationales behind child begging: either the children are coerced and exploited by parents (who collect the money) or they are neglected and obliged to fend for themselves. To complete this analysis, the following section offers a typology of children.
II – A typology of child groups reflecting different family begging strategies
31Our typology is based on a correspondence factor analysis.  Table 5 shows all the significant modalities used to construct the typology, which made use of all the variables described in the previous section. Two problems were encountered. First, given the diversity of parents’ occupations, they were grouped into three categories: inactivity, begging and informal occupations. The occupation variable is not significant in our typology, however. Second, the “reason for begging” and “use of money” variables are strongly overlapping, given that the children who said they were forced to beg by their parents also reported giving the money they earned to their parents. However, we chose to keep both these variables as they are quite different: children forced to beg by their parents sometimes also reported using the money to buy food. In other words, the close link between the modalities of the two variables highlights the fact that two rationales co-exist, and it is the different rationales behind begging that our typology aims to reveal.
32Table 5 shows that three distinct types of children exist. Category A concerns 43 children who are forced to beg by their parents. The parents generally accompany these children to the begging area and oblige them to hand over the money that they earn. This category includes the children whose parents use physical violence to force them to beg and to bring home more money. Their daily earnings are very low. They are practically all aged 8 or less.
33Category B includes 39 children who go alone to their begging area, generally use the money directly to buy food and who are not beaten by their parents. They earn more than children in category A and are most often aged between 9 and 11.
34Category C concerns the oldest children and includes 18 individuals. It has two main features: it is the group with the highest daily earnings and includes the children with at least one ill or disabled parent who cannot work for this reason. A more detailed analysis reveals that out of the 7 children with at least one ill parent, 6 are collecting money for the family. Of the 11 children with no ill parent, 10 share the money with their family.
35This typology shows that the groups are strongly structured by the coercion exerted by parents and by the use of physical violence against the children, which depend to a large extent on the child’s age. It thus suggests that the family rationale for begging can be explored in terms of the type of coercion and the children’s age. Of course, the family rationale we discuss is not confirmed by the topology; rather, it is the typology which enables us to extrapolate, in hypothetical terms, a certain family rationale.
Typology of child beggars
Typology of child beggars
36As suggested by type A, starting at age 4, children are taken to the begging area by their parents and left on their own to collect money. The children are forced to beg, in some cases through physical violence, and the money they earn is handed over to their parents at the end of the day. However, as pointed out earlier, our survey does not include children below age 4 who generally beg with their parents, under their supervision. Above age 4, children become more independent and streetwise. Their parents leave them to beg alone, but use coercion, including physical violence, to ensure that they concentrate on earning money by begging rather than playing with other children in the street.
37For the children of Type B, between the approximate ages of 8 and 11, the nature of the pressure exerted by parents appears to change. At these ages, it becomes more difficult to maintain a physical hold over children and to control their activities in the street. Instead, the parents apply coercion by depriving the children of food. Indeed, if the parents continue to beat the children at these ages, they are liable to run away; family violence is known to be a factor in children’s desertion of the family home (Lalor, 1999). While the youngest children give their earnings to their parents, who then spend the money to buy food for the family, the children aged approximately 8 to 11 are no longer fed by their parents and are obliged to fend for themselves. These children travel alone to the begging area and use the money earned directly to buy food. Their increased independence thus reflects the nature of the coercion exerted upon them. Rather than resorting directly to physical violence, the parents force their children to beg through neglect.
38Last, type C suggests that certain children become the family breadwinner. The fact that a significant variable in this category is the existence of an ill parent suggests that the child’s begging activity may be the family’s primary source of income. Two hypotheses can be put forward regarding this category of children. First, an ill parent can be used as a strategy for placing moral pressure on the child to earn money for the household. However, this does not explain why the children continue to beg rather than taking up a potentially more lucrative activity. The second hypothesis, therefore, is that begging provides a means to exert parental control over children’s earnings, since the children are less autonomous and less likely to leave the family home than if they earned money by a different means.
III – Discussion and conclusion
39The place of children in Madagascan society has changed considerably since the economic crises of the 1980s and 1990s (Ravololomanga and Schlemmer, 1994). Children who were once valued as an asset are now seen as a burden for poor families. They are expected to earn their living and to support their relatives, and begging is one form of child activity that benefits the parents.
40Although our study is no more than exploratory, it seems quite clear that begging is not the preserve of street-living children. Unlike most studies in the literature, but in line with the analysis of Ravololomanga and Schlemmer (1996), our findings show that begging provides a means for poor households to exploit child labour. Only 5% of the child beggars interviewed do not live with their parents. This result confirms the hypothesis, generally accepted in the literature on child labour, that the head of household holds power of decision over other family members.
41Our study findings also raise questions on the family rationales which lead to child begging. Given its exploratory nature, however, we can only give some tentative interpretations. The typology presented above suggests that from ages 8-9 there is a transition for many children from exploitation to abandonment or neglect. Of course, this is merely a hypothesis since our study was static and no definite dynamic interpretations are possible. Nonetheless, one way to analyse the typology is to posit that above a certain age it is difficult to exploit children through coercion, physical violence in particular, and that negligence predominates. The children continue to live at home, but must fend for themselves and carry on begging in order to survive. Other children remain under parental control, with some continuing to beg for their parents up to age 14. These children are not neglected, but are coerced using means other than beating, since physical violence might cause them to leave the family home. In this case, coercion tends to be psychological. In particular, the fact that children with an ill parent continue to give money to the household rather than use it to buy food and gain independence suggests that parents use strong moral pressure to convince the children that without them the family would not survive. From this point of view, begging, unlike other potentially more lucrative activities, may provide an effective means for parents to maintain control over their children by limiting the development of their autonomy.
42The existence of two different family strategies – on the one hand, children who remain under family control and, on the other, those who are forced to become independent through neglect – raises new questions about the trajectories of child beggars. Do some children follow a path towards a career in the street (Hanssen, 1996; Lucchini, 1996), and others not? Do the children who continue to beg under parental authority go on to become beggars in adulthood, so that begging becomes a long-term livelihood? Conversely, do those who leave home because they are rejected by their family manage to develop other activities that give them a certain financial autonomy? The study by Bousquet (2003) highlighted the existence of street-living child beggars. Neglected children may thus become street-living children with no ties to their family, for whom begging is one activity among many and not their main source of income.
43Our study does not answer these questions. Generally speaking, the literature on child labour pays scant attention to children’s type of activity by age, or to family modes of exploitation. In the case of Indian brick kiln workers, however, Bhukuth et Ballet (2006), observed a specialization of children by age in the various production activities. The oldest were assigned to jobs requiring physical strength, while the youngest were given simpler tasks to do. Trajectory analyses could provide more detailed information on the conditions under which activities becomes acceptable to children at different ages. Static assessment of the effects of an activity could thus be considered in a new light if the dynamic effects of this activity on the overall trajectory were taken into account. A further survey to analyse trajectories in more detail by interviewing former child beggars and parents would certainly offer new insight into the many questions that our study leaves unanswered.
AcknowledgementsWe are very grateful to the three referees for their helpful comments on a previous version of this paper. We also wish to thank the participants at the Enfance vulnérable seminar held on 8 June 2010 at the Centre de recherche pour le développement, Université catholique de Madagascar for their many useful reactions and remarks. Our thanks also to Valérie Delaunay for her suggestions. We of course remain solely responsible for any errors or omissions that may remain.
Institut de recherche pour le développement.
University of Antananarivo, Centre d’économie et d’éthique pour l’environnement et le développement à Madagascar.
Fonds pour la recherche en éthique économique.
Correspondence: Jérôme Ballet, Institut de recherche pour le développement, UMI Résiliences, 32 avenue Henri Varagnat, 93143 Bondy cedex, e-mail: firstname.lastname@example.org
In countries where the average per capita income is higher than USD 600, other factors explain the differences in prevalence of child labour: failures in the educational system, cost of transport, schooling costs, unemployed parents, etc. ( Jensen and Neilsen, 1997; Grootaert, 1998). These factors are, of course, also present in countries where average income is below USD 600.
Bhukuth (2006; 2009) makes a distinction between “weak” and “strong” exploitation. Begging is a form of “strong” exploitation from this viewpoint.
Child victims of trafficking, under the definition given by the Palermo Protocol to the 2000 United Nations Convention against Transnational Organized Crime.
Assi (2003) suggests a functional typology (child guides for the blind and disabled, children from Koranic schools, child porters, etc.), while Douville (2003-2004) suggests a typology based on status (trafficked children, child refugees, etc.).
Economic models of child labour are based on a unitary household model. In the 2005 regular survey of households, a household is defined as a group of individuals, related or not, usually living together and recognizing the authority of a single individual referred to as the “head of household”.
Children from the following neighbourhoods were interviewed: Ambanidihy, Ambatonakanga, Ambohijatovo, Ankadifotsy, Anosy, Ampasampito, Ampefiloha, Analakely, Andohalo, Andravoahangy, Andrainarivo, Ankatso, Antaninarenina, Antsahavola, Behoririka, Besarety, Mahamasina, Tsaralalana.
Interviews were conducted by two of the authors.
In a study of street children in Antananarivo (Ravololomanga, 2003), the survey was carried out on children aged six and up for the same reason, i.e. to prevent parents from controlling the answers.
Ideally, we should have talked to groups of children and only retained one interview per group. However, such an approach would have significantly increased the duration and cost of the study.
Clearly, for a broader survey involving the categorization of all forms of begging, this category of children must be taken into account.
While age is a discriminating factor in whether children are accompanied or not, sex is not significant.
A cross-tabulation of the variable “organization of begging” with the characteristics of the children (age and sex) produced no significant results.
We asked the children “why do you beg?”, which allowed a wide range of answers. In practice, two types of answers were given: “because my parents make me beg” or “because my parents don’t give me money for food”. We considered that there was coercion in the first case and negligence in the second. The strong correlation between reason for begging and use of the money tends to show that our interpretation was correct.
In principle, correspondence factor analysis can only be used to compare modalities and not to produce a typology. However, given that certain significant modalities concern a large proportion of individuals, modalities can be grouped together to form a classification by type. By definition, the typology aims to define types of individuals with similar characteristics. By “type” we mean a standard profile of individuals concerned by the survey.