CAIRN-INT.INFO : International Edition

1In many African societies, children belong to a lineage and not to a couple, and may circulate between different members of the extended family (Lallemand, 1993). Under this system, children may be temporarily fostered, or given away permanently through adoption (Bledsoe, 1990; Castle, 1995; Etienne, 1979; Goody, 1982; Jonckers, 1997; Lallemand, 1988, 1993; Madhavan, 2004; Rabain, 1979). These practices are intended to create or strengthen mutual support and kinship ties within a social and familial mode of organization based on the principle that burdens must be shared across the entire family network (Antoine et al., 1995; Marie, 1997; Oppong, 1999; Pilon and Vignikin, 2006). While children frequently move around between relatives, both temporarily and permanently, such movements are much rarer outside the family circle.

2However, this practice of child transfers for purposes of social exchange is becoming more diversified today. For several decades now, children have been fostered out to urban families in exchange for some domestic duties so that they can attend school (Jonckers, 1997; Vandermeersch, 2002). In a context of rural poverty, families are now less reluctant for their children to move to town to find work, or even encourage them to do so. Such practices are becoming increasingly widespread, and when the farming calendar so permits, a growing number of rural adolescents, at ever younger ages, now set off to look for an additional source of income (Delaunay et al., 2006; Erulkar et al., 2006). Children and adolescents who move to another household, on their own initiative or otherwise, may be exposed to various forms of discrimination (access to health care, food, education) and exploitation (labour, violence, including sexual violence).

3Children generally enter the labour market via family networks. These networks are increasingly structured and organized, with limited family control (Jacquemin, 2009), and the risks of child exploitation are considerable. These aspects deserve fuller investigation. A study of young maids in Côte d’Ivoire identified various forms of child labour placement (Jacquemin, 2009). The first is an extension of the practice of child transfers; children are placed with relatives so that the family of origin has one less mouth to feed while the host family acquires an extra pair of hands in exchange for food, lodgings, health care and possibly education (“little nieces”). The second form involves the family network, in the sense that the child is placed by a relative acting as “tutor”, but in a non-kin household, in exchange for a small wage that is managed by the tutor (“rented child”). The third and most recent form involves intermediaries outside the family network (placement agencies, acquaintances). In such cases, the child is considered as a worker and receives a wage directly from the employer. These children, often referred to as “little maids”, are generally older. Similar situations are described in other countries, including Senegal (Delaunay and Enel, 2009) and Mali (Lesclingand, 2004).

4The issue of child labour has prompted intense concern over the last decade (Bonnet and Schlemmer, 2009; Invernizzi, 2003; White, 1996), and international directives have been issued to identify and eliminate the “worst forms of child labour”, including types of work that are “likely to harm the health, safety or morals of children” (International Labour Organization, Convention no. 182).

5This article proposes an indirect measure of situations of child labour based on Demographic and Health Survey (DHS) data. While these surveys do not include information on employment, they provide data on children’s residential situations. The measure proposed here is based on the postulate that a school-aged child living with non-relatives and not attending school is engaged in child labour. If this postulate is verified, it becomes possible to measure child labour indirectly in many countries and at different times.

6In the case of Madagascar, child fosterage is limited to the circle of close relatives (Dahl, 2006). The rules of mutual assistance apply solely within the family, and it is very unusual for a household to take in a child who is not a relative. So the postulate appears to be confirmed. But we also know that many Madagascan children have jobs, including within the family, and that even those who attend school may work outside school hours (Rakoto-Tiana, 2011b). The proposed indirect measure will thus inevitably underestimate the scale of the phenomenon.

7Our aim here is to test the validity of this estimation method – which has the advantage of using data that already exist and are available in many countries – and to test its usefulness in understanding the determinants of child labour.

I – Child labour in Madagascar

8The most recent census in Madagascar dates back to 1993, at which time the specific occupation rate [1] was 27.1% for all children aged 10-14, with a higher rate in rural areas (31.8%) than in urban areas (11.4%). Boys were mostly employed in agriculture (90%), with a small percentage in domestic work (4.1%). Girls also worked on the land (78%), but a larger proportion were in domestic service, notably in the towns and cities (17.2%) (Institut national de la statistique, 1997).

9A more recent survey on child labour in Madagascar (Institut national de la statistique, 2008) has produced statistical data on the employment situation of children in households. It shows that 28% of children aged 5-17 were in employment at the time of the survey, especially in rural areas (31% versus 19% in towns). The share of economically active children increases with age, from 13% at ages 5-9, to 32% at ages 10-14 and 55% at ages 15-17, with a slightly higher rate for boys than for girls (30% versus 27%). [2] More than half of all working children are employed in farming and fishing (69% of working boys and 48% of working girls), and two-thirds have the status of domestic helper. A very high proportion (87%) do work that is considered harmful [3] (Institut national de la statistique, 2008).

10One of the conclusions of this study is that “living with one’s biological parents is a precious advantage for child protection” (Institut national de la statistique, 2008). Children who live elsewhere are more likely to engage in harmful economic activities or to be exploited. These findings are less clear-cut in the most disadvantaged populations of Antananarivo, where a study of child beggars showed that the vast majority actually do live with their parents (Ballet et al., 2010).

11An analysis of fostered children showed that 39% are fostered so that they can attend school. However, 12% were not in school at the time of the survey (Ballet et al., 2011), perhaps because they had performed poorly in school, or perhaps because they were prevented from attending by the host family. In group interviews with social workers involved in looking for lost children in Antananarivo, frequent cases were mentioned of children fostered to relatives so that they could attend school but who were exploited by their foster family and ended up running away (Bhukuth and Ballet, 2009).

Research framework and hypothesis

12Children are considered to be in a situation of child labour if the three following conditions are satisfied: they do not live with either of their parents; they live in a household headed by a person who is not a relative; they do not attend school.

13The Demographic and Health Survey (DHS) provides representative national data that can be used to identify children who are fostered or placed. Children living in a household without their parents can be identified, along with their relationship to the household head. Children living with non-relatives and not attending school are considered to be in a situation of child labour. For this reason, the analyses focus on children aged 6-17, ages at which children normally attend school.

14The proportion of children engaged in child labour at national and regional levels can thus be calculated. It is also possible to calculate this proportion for sub-groups defined according to the household’s socioeconomic status, the educational level of the household head, or the place of residence (urban/rural), and hence to estimate the prevalence of child labour by selected characteristics.

15The hypotheses tested here concern the existence of different models of child labour among children placed with non-relatives. These models appear to differ by sex. Demand for farm labour, concentrated in certain zones of the Central Highlands (rice growing and livestock breeding), and greater demand for domestic workers in urban areas suggest that placement models are geographically oriented towards active rural areas and towards towns and cities.

II – Method

16The data used are those of the 2008-2009 Demographic and Health Survey conducted on a sample of 17,857 households. [4] Only de jure residents are considered, i.e. habitual members of the household, excluding persons who simply spent the night preceding the survey in the household (de facto residents).

17The survey recorded family relationships, and the response categories included “fostered or adopted child”. We did not use this information as it is incomplete and often unreliable. For example, a fostered child can also be the nephew of the household head, in which case the interviewer can either record him as a nephew or as a fostered child. We therefore preferred to define a fostered child in terms of the presence or absence of his or her parents in the household. Household members are analysed using a method developed by Christine Tichit to define household structures (Tichit, 2009). Under this method, children living in a household without their father or mother (including orphans) are considered as fostered.

18Our study concerns a total of 43,727 de jure residents aged 6-17, excluding married children, parents or household heads.

19The analysis focuses both on fostered children and on children engaged in child labour, measuring their prevalence by certain characteristics relating to their place of residence, their parents’ vital status, the characteristics of the household head and of the household of residence.

III – Results

Fostered children

20By analysing the children’s situation in terms of the parents’ presence in the household, it is possible to distinguish between children of the household’s principal nucleus (children of the household head or of his/her spouse), and children of other adults in the household (secondary nucleus) or children whose parents are not resident in the household. It is these last children who form the “fostered” category.

21There are 5,912 fostered children in our sample, i.e. 13.5% of the total. The proportion is low for very young children and increases with age: 7.7% at ages 0-5, 11.5% at ages 6-9, 18.8% at ages 10-14 and 22.5% at ages 15-17 (Table 1).

Table 1

Distribution of children by position in the household and age (weighted numbers and estimated proportions)*

Table 1
Nucleus Age 0-5 Age 6-9 Age 10-14 Age 15-17 Total N % N % N % N % N % Principal 12,539 80.2 9,068 79.8 9,542 77.5 3,302 74.6 34,451 78.8 Secondary 2,077 13.3 695 6.1 461 3.7 130 2.9 3,363 7.7 Fostered 1,012 6.5 1,594 14.1 2,309 18.8 996 22.5 5,911 13.5 Total 15,628 100.0 11,357 100.0 12,312 100.0 4,428 100.0 43,725 100.0

Distribution of children by position in the household and age (weighted numbers and estimated proportions)*

* The principal nucleus is that of the household head; the secondary nuclei are those of other adults in the household (husband/wife couples or parent(s)/child(ren)).
Source: DHS 2008-2009, author’s calculations.

22The fostered children category can be sub-divided by relationship with the household head. More than half of all fostered children – and more than 80% of those aged 0-4 – live with their grandparents (Table 2). They are generally young, with a mean age of 8.6 years. This finding attests to the role played by grandparents in raising small children (Andiamaro, 2013; Dahl, 2006; Delaunay et al., 2010). The proportion decreases with age, down to 26.5% at ages 15-17, as the grandparents grow older and their mortality increases.

Table 2

Distribution of children by place of residence and age (weighted numbers and estimated proportions)

Table 2
Place of residence Age 0-5 Age 6-9 Age 10-14 Age 15-17 Total Mean age (years) N % N % N % N % N % Grandparents 833 82.3 1,114 69.9 1,089 47.2 264 26.4 3,300 55.8 8.6 Uncle/Aunt 97 9.6 243 15.2 398 17.2 180 18.1 917 15.5 10.8 Sibling 13 1.2 79 4.9 283 12.2 161 16.2 536 9.1 12.4 Other relative 33 3.3 81 5.1 209 9.1 170 17.1 493 8.3 12.3 Other non-relative 37 4.3 77 4.8 330 14.3 221 22.2 666 11.3 12.5 Total 1,013 100.0 1,594 100.0 2,309 100.0 996 100.0 5,912 100.0 10.0

Distribution of children by place of residence and age (weighted numbers and estimated proportions)

Source : DHS 2008-2009, author’s calculations.

23Around 15% of fostered children live with an uncle or aunt, with a proportion ranging from 9% for the youngest children to 18% for the older ones. It seems that maternal uncles and aunts play an important role in child care, in certain social groups at least (Goedefroit, 1998), and represent an alternative to parental care, whatever the child’s age (Dahl, 2006 ; Rakotonarivo, 2010). In Madagascar, it is not unusual for a young person with a job to support and care for a younger sibling (Dahl, 2006), and 9% of fostered children live with a brother or sister, most often after reaching school age. A further 8% of fostered children live with another relative, with a proportion ranging from 3% for the youngest to 17% for the oldest. Note that this type of fosterage may be for reasons of schooling, or for employment as a domestic helper.

24Last, 11% of fostered children live with a person who is not a relative. The proportion ranges by age from 4% to 22% and represents 666 children in total. [5] Here too, children may be fostered for reasons of schooling or employment as a domestic helper, although the rules of solidarity with regard to schooling rarely extend beyond the family circle (Rakoto-Tiana, 2011a).

25For each child, the survey provides information on their school attendance during the year in which the survey was conducted. We sorted the fostered children by place of residence and school attendance (Table 3). Information is missing for 31 children. The highest proportion of fostered children not attending school are those living with a non-relative (61%), and the percentage is also high among those living with an “other relative” (45%). Children aged 6 and above who live with a grandparent, an uncle/aunt or a sibling more often attend school (73%, 67% and 74%, respectively), confirming that in these cases fosterage is most often for schooling purposes.

Table 3

Distribution of fostered children aged 6-17 by place of residence and school attendance at the time of the survey (weighted numbers and estimated proportions)

Table 3
Place of residence Child’s situation Not in school In school Unknown Total N % N % N % N % Grandparents 660 26.7 1 797 72.8 10 0.4 2 467 100.0 Uncle/Aunt 269 32.8 547 66.7 4 0.5 820 100.0 Sibling 132 25.3 387 73.9 4 0.9 523 100.0 Other relative 208 45.2 248 54.0 4 0.9 460 100.0 Other non-relative 384 61.0 237 37.7 8 1.3 629 100.0 Total 1,653 33.7 3,216 65.6 30 0.6 4,899 100.0

Distribution of fostered children aged 6-17 by place of residence and school attendance at the time of the survey (weighted numbers and estimated proportions)

Source : DHS 2008-2009, author’s calculations.

26The DHS sample thus provides us with 384 cases corresponding to our definition of children engaged in child labour (child not living with parents, residing with a non-relative, aged 6 or above and not enrolled in school).

Children engaged in child labour

27Some 1.4% of all children aged 6-17 in the sample are potentially engaged in child labour (Table 4), a proportion that is low but not negligible. It corresponds to a possible national total of 100,000 working children. [6]

28There is no distinction by sex (1.4% for boys and 1.3% for girls), but gender differences appear when place of residence and region are taken into account. Female child labour is more prevalent in the capital, Antananarivo (4,6 %), and in other cities (2,5 %), while for boys the prevalence varies little between urban and rural areas. Female child labour tends to be an urban phenomenon which corresponds to the employment of girls as domestic helpers.

29The prevalence by region reflects these findings: it is higher in Analamanga and Atsinana, the regions where Madagascar’s two largest cities, Antananarivo and Toamasina (Tamatave), are located. The highest prevalences of male child labour are found in the regions of Alaotra Mangoro, Itasy, Betsiboka and Bongolava. The first three are major rice-growing regions, notably Alaotra Mangoro, known as Madagascar’s rice granary, while Bongolava is a region of intensive livestock production. Male child labour appears to be primarily concentrated in farming.

30Prevalences by characteristic show some variations. They increase with age (3.9% at ages 15-17), are higher among orphans (or children whose parents’ vital status is unknown, 11.0%), in households whose head is educated (4.3% when the head has secondary education or higher), in wealthier households (3.5%), in childless households (5%) and in complex households (3.1% for boys) (Table 4).

Table 4

Prevalence of children potentially engaged in child labour (aged 6-17, living with a non-relative and not enrolled in school) by sex

Table 4
Characteristic Total Boys Girls % N % N % N Place of residence *** ns *** Capital 3.25 1,096 1.94 556 4.60 540 Other towns 1.74 2,429 1.00 1,197 2.46 1,232 Rural areas 1.24 24,741 1.40 12,881 1.06 11,859 Province *** ns *** Antananarivo 1.78 8,725 1.59 4,440 1.98 4,284 Finarantsoa 1.19 5,546 1.36 2,844 1.02 2,701 Toamasina 1.52 4,730 1.63 2,520 1.40 2,210 Mahajunga 1.30 3,436 1.56 1,792 1.02 1,644 Toliara 0.79 3,879 0.90 2,004 0.67 1,875 Antsiranana 0.79 1,950 0.67 1,033 0.92 917 Region *** *** *** Analamanga 1.79 3,752 1.11 1,900 2.50 1,852 Vakinankaratra 1.42 2,757 1.22 1,410 1.63 1,347 Itasy 2.62 1,320 2.97 676 2.25 644 Bongolava 1.61 895 2.66 455 0.53 441 Haute Matsiatra 1.31 1,481 1.33 739 1.30 742 Amoron’i Mania 1.93 1,129 1.91 581 1.96 548 Vatovavy Fitovinany 0.67 1,468 1.05 767 0.25 701 Ihorombe 0.81 499 0.37 259 1.28 240 Atsimo Atsinanana 1.12 969 1.74 499 0.48 470 Atsinanana 2.05 1,451 1.44 743 2.68 707 Analanjirofo 0.32 1,696 0.34 929 0.30 767 Alaotra Mangoro 2.32 1,583 3.20 848 1.31 735 Boeny 1.93 934 2.06 453 1.80 481 Sofia 0.92 1,752 1.18 951 0.61 801 Betsiboka 1.89 420 2.67 221 1.04 199 Melaky 0.83 330 0.96 167 0.69 163 Atsimo Andrefana 0.18 1,411 0.28 726 0.07 685 Androy 1.25 879 1.42 456 1.07 422 Anosy 0.73 886 0.48 460 1.00 426 Menabe 1.49 704 1.99 363 0.95 342 Diana 0.49 693 0.16 356 0.84 336 Sava 0.95 1,258 0.93 677 0.97 581 Age *** *** *** 6-10 0.20 14,313 0.14 6,932 0.25 7,381 11-14 1.90 9,408 1.80 4,618 1.99 4,790 15-17 3.89 4,544 4.22 2,081 3.61 2,463 Parents’ vital status *** *** *** Father and mother alive 0.94 25,058 0.96 12,055 0.92 13,003 Mother alive + father dead / don’t know 3.97 917 3.28 474 4.71 443 Father alive + mother dead / don’t know 3.55 1,875 2.77 894 4.25 982 Father dead / don’t know + mother dead / don’t know 11.06 415 11.92 207 10.20 207

Prevalence of children potentially engaged in child labour (aged 6-17, living with a non-relative and not enrolled in school) by sex

Table 4 (cont’d)

Prevalence of children potentially engaged in child labour (aged 6-17, living with a non-relative and not enrolled in school) by sex

Table 4 (cont’d)
Characteristic Total Boys Girls % N % N % N Sex of household head ns ns ns Male 1.36 23,046 1.39 12,080 1.32 10,966 Female 1.16 5,219 1.14 2,554 1.19 2,665 Educational level of household head *** ** *** No schooling 0.63 5,835 0.77 3,021 0.49 2,814 Incomplete primary 0.91 12,499 1.27 6,454 0.53 6,045 Complete primary 1.10 1,805 0.66 961 1.59 844 Incomplete secondary 2.24 6,212 1.82 3,181 2.69 3,031 Complete secondary 4.36 432 2.40 230 6.59 203 Higher education 4.25 639 2.99 329 5.58 311 Don’t know 2.11 843 2.69 460 1.42 384 Household income *** *** *** Poorest 0.69 5,673 0.99 2,916 0.38 2,757 Poor 0.35 5,821 0.42 3,092 0.27 2,730 Moderate 0.83 5,837 0.98 3,069 0.66 2,768 Rich 1.52 5,874 2.03 3,042 0.98 2,832 Richest 3.50 5,061 2.53 2,517 4.45 2,544 Type of nucleus of household head *** *** *** Isolated 2.06 2,271 2.42 1,131 1.70 1,139 Childless couple 4.96 1,734 4.95 875 4.98 859 Parental 1.08 18,259 1.08 9,486 1.09 8,773 Reconstituted 0.75 2,390 0.81 1,337 0.67 1,053 Lone-parent 0.70 3,611 0.75 1,805 0.65 1,806 Number of secondary nuclei ns ** ns 0 1.29 24,085 1.22 12,519 1.37 11,567 1 1.47 3,490 1.89 1,781 1.03 1,709 2 or more 1.71 690 3.10 335 0.39 355 Total 1.36 28,265 1.39 14,635 1.33 13,631

Prevalence of children potentially engaged in child labour (aged 6-17, living with a non-relative and not enrolled in school) by sex

Interpretation: 3.25% of children aged 6-17 in the capital live with a non-relative and do not attend school.
Significance levels: *** p 0.01; ** p 0.05; * p 0.1; ns: non-significant.
Source: DHS 2008-2009, author’s calculations.

31A logistic regression (Table 5) confirms these relations and distinguishes the male and female models more clearly. The children at highest risk are those aged 11-14 and 15-17 with a parent who is deceased or whose vital status is unknown. The male model is characterized by a much higher risk of child labour in rural areas, while for girls the risk is higher in the capital and in the countryside than in small towns. The risk increases considerably with the educational level of the household head, and is especially high in the richest households, for both boys and girls; it is low in lone-parent households.

Table 5

Logistic regression of children potentially engaged in child labour (aged 6-17, living with a non-relative and not enrolled in school) by sex (odds ratios)

Table 5
Total Boys Girls Place of residence Capital Ref. Ref. Ref. Other towns 0.624 ** 0.942 ns 0.509 * Rural areas 1.432 * 3.269 *** 0.933 ns Province Antananarivo Ref. Ref. Ref. Finarantsoa 0.944 ns 0.758 ns 1.145 ns Toamasina 1.128 ns 0.965 ns 1.344 ns Mahajunga 0.983 ns 0.997 ns 0.965 ns Toliara 0.659 * 0.630 ns 0.697 ns Antsiranana 0.569 * 0.397 ** 0.846 ns Age 6-10 Ref. Ref. Ref. 11-14 5.864 *** 5.350 *** 6.604 *** 15-17 9.933 *** 8.036 ** 12.420 *** Parents’ vital status Father and mother alive Ref. Ref. Ref. Mother alive + father dead / don’t know 3.149 *** 4.307 *** 2.277 ** Father alive + mother dead / don’t know 3.500 *** 3.740 *** 2.859 *** Father dead / don’t know + mother dead / don’t know 6.053 *** 6.106 *** 6.437 *** Sex of household head Male Ref. Ref. Ref. Female 1.223 ns 1.088 ns 1.354 ns Educational level of household head No schooling 0.861 ns 0.697 ns 1.265 ns Incomplete primary Ref. Ref. Ref. Complete primary 0.947 ns 0.489 * 1.979 ** Incomplete secondary 1.289 * 0.847 ns 2.231 *** Complete secondary 1.668 * 1.191 ns 2.589 ** Higher education 1.982 ** 0.897 ns 3.498 *** Don’t know 1.365 ns 0.934 ns 2.423 ** Household income Poorest Ref. Ref. Ref. Poor 0.507 ** 0.514 ** 0.502 * Moderate 0.837 ns 0.921 ns 0.734 ns Rich 1.356 ns 1.567 * 1.163 ns Richest 3.212 *** 3.244 *** 3.455 ***

Logistic regression of children potentially engaged in child labour (aged 6-17, living with a non-relative and not enrolled in school) by sex (odds ratios)

Tableau 5 (cont’d)

Logistic regression of children potentially engaged in child labour (aged 6-17, living with a non-relative and not enrolled in school) by sex (odds ratios)

Tableau 5 (cont’d)
Total Garçons Filles Type of nucleus of household head Isolated 1.106 ns 1.172 ns 1.070 ns Childless couple 3.342 *** 3.235 *** 3.697 *** Parental Ref. Ref. Ref. Reconstituted 0.748 ns 0.751 ns 0.727 ns Lone-parent 0.296 *** 0.331 *** 0.272 *** Number of secondary nuclei 0 Ref. Ref. Ref. 1 0.855 ns 0.899 ns 0.791 ns 2 or more 0.595 ns 1.282 ns 0.110 * Constant -?4.290 *** -?4.861 *** -?4.329 ***

Logistic regression of children potentially engaged in child labour (aged 6-17, living with a non-relative and not enrolled in school) by sex (odds ratios)

Interpretation: Children aged 15-17 are 10 times more likely to live with non-relatives and not attend school than children aged 6-10. Boys in rural areas are 3 times more likely to live with non-relatives and not attend school than boys in the capital.
Significance levels: *** p 0.01; ** p 0.05; * p 0.1; ns: non-significant.
Source: DHS 2008-2009, author’s calculations.

IV – Discussion

Indicator quality

32Estimates of the prevalence of child labour are based on a definition that is plausible but needs to be discussed. Children are considered to be engaged in child labour if they live with a non-relative without their parents and are not enrolled in school. This category may include children who are adopted and who, like the adoptive family’s biological children, cannot afford to attend school, but who are not necessarily exploited. This would lead to an overestimation of children engaged in child labour whose amplitude is difficult to determine, but which, in the context of Madagascar where non-family adoption is rare (Dahl, 2006), is likely to be small. Moreover, attending school does not protect against economic exploitation, and many school-children also work outside school hours (Rakoto-Tiana, 2011b). This suggests that the estimated prevalence of child labour obtained with our indicator is actually below the true level.

33Moreover, certain testimonies (Bhukuth and Ballet, 2009) suggest that children living with distant relatives (“other relative”) are also at risk of economic exploitation. Indeed, there may be a tendency to report a child maid as a distant relative in order to avoid any suspicion in this respect. This category was not included in our analyses. It is clear that our measures are underestimated and represent minimum prevalences at best. They can be compared with those obtained by the ENTE child labour survey (Enquête nationale sur le travail des enfants), which obtained a prevalence of 4.5% for child trafficking victims versus 1.4% for children engaged in child labour in our study (Bhukuth and Ballet, 2009). This gives further grounds for believing that our findings are underestimated.

Main findings

34Child labour mainly concerns children aged 11 and above. The educational level of the household head appears to be an important factor, for girls especially; these households have fewer children (they are the groups with the lowest fertility) and their female members are more often in employment, so their demand for domestic child labour is high. Level of wealth also appears to be a factor in demand for labour, both domestic and agricultural. Non-family child workers in farming or domestic service provide a replacement for children who are unborn, absent or deceased. These findings are consistent with a study performed in a rural area which shows that the fewer the children under age 14 in a household, the more likely it is to include a fostered child (Rakoto-Tiana, 2011a).

35We have seen that the prevalence of child labour varies by place of residence and region, and that the pattern of variation is different for boys and girls. The highest prevalences are found in the Central Highland regions, which are also the richest regions with the strongest demand for farming and domestic labour.

36This reveals patterns of placement that differ by sex. Girls tend to be placed in socially advantaged urban households to work as maids, while boys are placed both in the cities but also in rural areas with strong demand for farm labour. They are also found in the capital, and the types of work performed by boys are probably very varied. These hypotheses need to be confirmed, however.

37It is difficult to determine the extent to which these situations are harmful to the children concerned. Half of the children engaged in child labour are aged 15-17. The data gives no indication of the type of work they do. If they enjoy decent wages and working conditions, these adolescents may be better off than in their household of origin. The 10-14 age group is of greater concern. At these ages, children are growing fast, and they have little power to negotiate their wages or working conditions. It is these children who should receive the closest attention, as they are probably the group most exposed to the risk of exploitation.

38These results can be compared with those of the ENTE survey which show that children not living with either parent are more frequently employed in jobs considered as harmful (Institut national de la statistique, 2008).

39This method has the advantage of being applicable to other contexts where households provide homes to unrelated children not through altruism, but rather as a source of labour. Many countries have data from DHS or household surveys that can be used to identify children living away from their family. Here, the relationship of the children to the head of household is a crucial item of information that deserves to be recorded more fully in the DHS surveys.

Recommendations for data collection and analysis

40Further to our analyses, we would like to make a number of recommendations. The first are of a methodological nature and concern improvements in DHS data. The second are scientific and concern potential new avenues of research.

41In the DHS surveys, it would be preferable to separate the “relationship to head of household” question from the “adopted/fostered child” question, so that the mechanisms whereby a child is taken into the household can be studied. Children generally tend to be recorded as adopted/fostered when they do not correspond to any other category, i.e. when they are not related to the household head. But this may lead to confusion. It would be more useful to ask a specific question about each child, to determine whether they are either adopted or fostered, distinguishing clearly between the two categories. A child adopted by the household head is often reported as a “son/daughter”. Adopted children are thus more likely to be identified if a specific question is asked. This is especially true in Madagascar, where many child adoptions are invisible in surveys. A second methodological suggestion concerns the need to identify in surveys those children who do not live with their parents but in the household of a non-relative, so that information on the household of origin can be collected (place of residence, socioeconomic status, household structure), along with the reasons why the child is fostered.

42Regarding potential avenues of research, we believe it is very important to characterize the different types of occupation held by children and to identify those most harmful to them. According to a recent study, employment as a maid in a household of non-relatives is a key factor of vulnerability among migrants to the city: “the poorest and most vulnerable female migrants in Antananarivo (including prostitutes, garbage collectors and beggars) are those who report having been employed as maids” (Freeman et al., 2010, pp. 30-31). Likewise, we have little information on children’s participation in farm labour. Certain types of farming jobs are not necessarily harmful to children. It is important to describe more fully the types of activity and the conditions of the informal contracts binding children and their families (debt repayment, wage paid to the child or a third party, provision of food, etc.). Qualitative studies are needed to shed light on these situations and to provide pointers for the implementation of awareness, prevention and protection initiatives.

43It is equally important to understand the processes that cause children to engage in harmful types of employment, the circumstances leading to child placement, the resources and the actors involved. The causes of placement are linked to the family situation. Placement may be the result of family crisis (death, separation) or economic hardship (job loss, poor harvest, debt) which oblige the family to send the child away to work. Children may also leave on their own initiative, to escape a violent relative or spouse for example.

44The type of placement and the conditions in which children find themselves depend on the way the placement is organized. The formalization of placement networks via agencies which actively solicit rural families is an issue of growing concern.

45Last, it is important to examine how child labour affects the outcomes of these children, who are often placed in harmful employment at very young ages. Employers who use children see them as a cheap and malleable source of labour. As they grow up and become less pliant, they are replaced by younger new recruits. It is important to understand what happens to these children, whether they maintain links with their families or are drawn into other harmful labour networks for which older workers are needed, such as prostitution or modern slavery abroad…


  • [*]
    Institut de recherche pour le développement (IRD), Aix-Marseille université (AMU).
    Correspondence: Valérie Delaunay, campus international UCAD-IRD de Hann, BP 1386, CP 18524, Dakar, Senegal, e-mail:
  • [1]
    The occupation rate is the ratio of the working-age population in employment to the total population.
  • [2]
    In the definitions used, domestic labour, often accomplished by women, is not considered as an economic activity.
  • [3]
    Under decree no. 2007-563 relative to child labour, this concept refers to all types of labour by children aged under 18 that must be abolished in accordance with the national texts governing the labour market and international labour conventions. It includes the worst forms of labour: immoral labour (sex, alcohol), labour beyond the child’s physical strength, forced labour, dangerous or unhealthy labour, all work performed by children aged below 15 without the authorization of a labour inspector (Institut national de la statistique, 2008).
  • [4]
    The preceding surveys concerned a smaller number of households and the samples were too small to conduct this type of analysis.
  • [5]
    Of the 666 children fostered to a non-relative, 157 are coded as “adopted/fostered” in the question on relationship to head of household. It is therefore impossible to know whether the child has any kin relationship with the household head. The data would be more precise if the status of the adopted or fostered child was entered separately from the relationship with head of household so that a distinction could be made between temporary and permanent fosterage.
  • [6]
    Extrapolation based on an estimation of 7,000,000 children aged 6-17 living in Madagascar in 2009.

In Africa today, children may be sent to live with non-relative for employment reasons. This short paper proposes a measure based on data from the 2008-2009 Demographic and Health Survey in Madagascar to identify children living with non-relatives and not attending school who are liable to be engaged in child labour. In Madagascar, children may move between family households, but may also be sent to live outside the family for economic reasons. We postulate that the models of child placement for economic reasons are gender-based and linked to geographic location. The results show that 1.4% of children aged 6-17 are in this situation. Two profiles can be identified: girls engaged in child labour tend to work in towns in households with high socioeconomic status, while boys work in areas with strong demand for farm labour. The authors make recommendations for improving data collection and suggest that surveys of child maids and farm labourers be conducted to identify harmful situations, to understand the processes whereby children become engaged in child labour (origin, means deployed), and to find out more about these children’s outcomes.


  • child labour
  • Madagascar
  • fosterage
  • children’s rights
  • demographic and Health Survey (DHS)

L’exploitation économique des enfants à Madagascar à partir de l’Enquête démographique et de santé 2008

L’exploitation économique des enfants à Madagascar à partir de l’Enquête démographique et de santé 2008

En Afrique, l’accès à l’emploi est devenu une cause de placement des enfants à l’extérieur de la parenté. L’objectif de cette note de recherche est de proposer une mesure des enfants résidant hors de la parenté et non scolarisés, susceptibles d’être en situation d’exploitation, en utilisant les données de l’Enquête démographique et de santé 2008-2009 à Madagascar. Dans le contexte malgache, la circulation des enfants reste interne à la parenté, mais les enfants peuvent être placés hors de la parenté pour des raisons économiques. Notre hypothèse est que les modèles de placement des enfants pour des raisons économiques sont genrés et géographiquement répartis. Les résultats montrent que cette situation concerne 1,4 % des enfants de 6 à 17 ans. Parmi eux, deux profils se dégagent : les filles se situent plutôt en ville dans des ménages à niveau socio-économique élevé et les garçons dans les zones à forte demande de main d’œuvre agricole. Les recommandations portent sur l’amélioration des données et la réalisation d’enquêtes auprès d’enfants domestiques et travailleurs agricoles afin d’identifier les situations dommageables, et d’enquêtes sur le processus de mise au travail des enfants (origine, moyens utilisés) ainsi que sur le devenir des enfants travailleurs.


La explotación económica de los niños en Madagascar a partir de las Encuestas demográficas y de salud

La explotación económica de los niños en Madagascar a partir de las Encuestas demográficas y de salud

En África, el acceso al empleo ha llegado a ser una causa de colocación de los niños al exterior de la parentela. Utilizando la Encuesta demográfica y de salud de 2008-2009 hecha en Madagascar, esta nota se propone estimar el número de niños que residen fuera de la parentela, no escolarizados, y susceptibles de ser explotados. En el contexto malgache, la circulación de los niños esta circunscrita a la parentela, pero los niños pueden ser colocados al exterior por razones económicas. Nuestra hipótesis es que la colocación de niños por razones económicas está influida por el género y varia geográficamente. Los resultados muestran que esta situación concierne 1,4 % de los niños de 6 a 17 años. Dos perfiles principales se destacan: las niñas se sitúan más bien en las ciudades en hogares de nivel socioeconómico elevado, y los niños en las zonas con fuerte demanda de mano de obra agrícola. Se hacen recomendaciones sobre la mejora de los datos y la realización de encuestas sobre los niños empleados domésticos y los trabajadores agrícolas con el fin de identificar las situaciones perjudiciales. Se recomienda también hacer encuestas sobre el proceso que conduce a este tipo de explotación económica (origen de los niños, medios utilizados) así como sobre el devenir de los niños trabajadores.
Translated by Catriona Dutreuilh.


  • Andiamaro F., 2013, Ruptures familiales et prise en charge de l’enfant à Madagascar, Université Paris X-Ouest Nanterre La Défense, U.F.R. Sciences sociales et administration, PhD thesis, 260 p.
  • Antoine P., Bocquier P., Fall A., Guisse Y., Nanitelamio J., 1995, Les familles dakaroises face à la crise, Dakar, IFAN, ORSTOM, CEPED, 209 p.
  • OnlineBallet J., Bhukuth A., Rakotonirinjanahary F., Rakotonirinjanahary M., 2010, “Family rationales behind child begging in Antananarivo”, Population, English Edition, 65(4), pp. 695-712.
  • Ballet J., Bhukuth A., Rabevohitra N. et al., 2011, “Situation des enfants confiés pour scolarisation à Madagascar”, Antananarivo, Document de travail n° 12.
  • Bhukuth A., Ballet J., 2009, La traite des enfants à Madagascar, Unicef, Antananarivo, 20 p.
  • OnlineBledsoe C., 1990, “The politics of children: Fosterage and the social management of fertility among the Mende of Sierra Leone”, in Handwerker W. (ed.), Births and Power: Social Change and the Politics of Reproduction, San Francisco, Westview Press, pp. 81-100.
  • OnlineBonnet M., Schlemmer B., 2009, “Aperçus sur le travail des enfants”, Mondes en développement, 37(146), pp. 11-25.
  • OnlineCastle S. E., 1995, “Child fostering and children’s nutritional outcomes in rural Mali: The role of female status in directing child transfers”, Social Science and Medicine, 40(5), pp. 679-693.
  • Dahl O., 2006, Signes et significations à Madagascar. Des cas de communication interculturelle, Paris, Présence africaine, 309 p.
  • Delaunay V., Enel C., 2009, “Les migrations saisonnières féminines: le cas des jeunes bonnes à Dakar”, in Vallin J. (ed.), Du genre et de l’Afrique. Hommage à Thérèse Locoh, Paris, INED, pp. 389-401.
  • OnlineDelaunay V., Adjamagbo A., Lalou R., 2006, “Questionner la transition de la fécondité en milieu rural africain : les apports d’une démarche longitudinale et institutionnelle”, Cahiers québécois de démographie, 35(1), pp. 27-50.
  • Delaunay V., Andriamaro F., Tang N. et al., 2010, “Des ménages ‘sans parents’: prévalence et différentiels régionaux à Madagascar”, Actes du Colloque AIDELF, Relations intergénérationnelles. Enjeux démographiques, Geneva, pp. 595-614.
  • OnlineErulkar A., Mekbib T.-A., Simie N., Gulema T., 2006, “Migration and vulnerability among adolescents in slum areas of Addis Ababa, Ethiopia”, Journal of Youth Studies, 9(3), pp. 361-374.
  • OnlineEtienne M., 1979, “Maternité sociale, rapports d’adoption et pouvoir des femmes chez les Baoulé (Côte d’Ivoire)”, L’Homme, 19(3-4), pp. 63-107.
  • Freeman L., Rasolofohery S., Randriantovomanana E., 2010, Tendances, caractéristiques et impacts de la migration rurale-urbaine à Antananarivo, Madagascar, UNICEF, Antananarivo, 60 p.
  • Goedefroit S., 1998, Á l’Ouest de Madagascar. Les Sakalava du Menabe, Paris, IRD, Karthala, 529 p.
  • Goody E., 1982, Parenthood and Social Reproduction. Fostering and Occupational Roles in West Africa, Cambridge, Cambridge University Press, 348 p.
  • Institut National de la Statistique, 1997, Recensement général de la population et de l’habitat. Volume 2, tome VI. Activité économique, Organisation internationale du travail, République de Madagascar, Antananarivo, 184 p.
  • Institut National de la Statistique, 2008, Enquête nationale sur le travail des enfants – 2007, Organisation internationale du travail, République de Madagascar, Antananarivo, 108 p.
  • OnlineInvernizzi A., 2003, “Des enfants libérés de l’exploitation ou des enfants travailleurs doublement discriminés?”, Déviance et société, 27(4), pp. 459-481.
  • OnlineJacquemin M., 2009, “‘Petites nièces’ et ‘petites bonnes’ à Abidjan. Les mutations de la domesticité juvénile”, Travail, genre et sociétés, 22(2) pp. 53-74.
  • Jonckers D., 1997, “Les enfants confiés”, in Pilon M., Locoh T., Vignikin E. Vimard P. (eds.), Ménages et familles en Afrique. Approche des dynamiques contemporaines, Paris, CEPED, p. 193-208.
  • Lallemand S., 1988, “Un bien qui circule beaucoup”, Autrement (Abandon et adoption), 96, pp. 128-134.
  • Lallemand S., 1993, La circulation des enfants en société traditionnelle. Prêt, don, échange, Paris, L’Harmattan, 220 p.
  • OnlineLesclingand M., 2004, “Nouvelles stratégies migratoires des jeunes femmes rurales au Mali : de la valorisation individuelle à une reconnaissance sociale”, Sociétés contemporaines, 55(3), pp. 21-42.
  • OnlineMadhavan S., 2004, “Fosterage patterns in the age of AIDS: continuity and change”, Social Science and Medicine, 58(7), pp. 1443-1454.
  • Marie A., 1997, “Les structures familiales à l’épreuve de l’individualisation”, in Pilon M., Locoh T., Vignikin E., Vimard P. (eds.), Ménages et familles en Afrique. Approche des dynamiques contemporaines, Paris, CEPED, pp. 279-300.
  • OnlineOppong C., 1999, “Les systèmes familiaux et la crise économique”, in Adepodju A. (ed.), La famille africaine. Politiques démographiques et développement, Paris, Karthala, pp. 221-254.
  • Pilon M., Vignikin K., 2006, Ménages et familles en Afrique subsaharienne, Paris, Éditions des archives contemporaines/AUF, 131 p.
  • Rabain J., 1979, L’enfant du lignage. Du sevrage à la classe d’âge chez les Wolof du Sénégal, Paris, Payot, 237 p.
  • OnlineRakotonarivo A., 2010, “La solidarité intergénérationnelle en milieu rural malgache : le rôle des personnes âgées dans la migration”, Autrepart, 53(1), pp. 111-130.
  • Rakoto-Tiana N., 2011a, “Confiage et scolarisation des enfants en milieu rural à Madagascar”, Actes du 51e Congrès annuel de la Société canadienne de sciences économiques, 11-13 mai 2011, Sherbrooke, Canada.
  • Rakoto-Tiana N., 2011b, L’école primaire pour tous? Politiques éducatives, demande d’éducation et travail des enfants à Madagascar, Université Paris 13, PhD thesis, 203 p.
  • Tichit C., 2009, Understanding the Family Composition of Households in Demographic and Health Surveys, Paris, CEPED, 89 p.
  • OnlineVandermeersch C., 2002, “Child fostering under six in Senegal in 1992-1993”, Population, English Edition, 57(4-5), pp. 659-686.
  • OnlineWhite B., 1996, “Globalization and the child labor problem”, Journal of International Development, 8(6), pp. 829-839.
Valérie Delaunay [*]
  • [*]
    Institut de recherche pour le développement (IRD), Aix-Marseille université (AMU).
    Correspondence: Valérie Delaunay, campus international UCAD-IRD de Hann, BP 1386, CP 18524, Dakar, Senegal, e-mail:
Translated by
Catriona Dutreuilh
Latest publication on cairn or another partner portal
Uploaded on on 03/03/2014
Distribution électronique pour I.N.E.D © I.N.E.D. Tous droits réservés pour tous pays. Il est interdit, sauf accord préalable et écrit de l’éditeur, de reproduire (notamment par photocopie) partiellement ou totalement le présent article, de le stocker dans une banque de données ou de le communiquer au public sous quelque forme et de quelque manière que ce soit.
Loading... Please wait