Abstract
This paper summarizes trends in the use of child domestic servants in six Latin American countries using IPUMS-International census samples for 1960 to 2000. Child domestics are among the most vulnerable of child workers, and the most invisible. They may be treated kindly and allowed to attend school, or they may be secluded in their employers’ home, overworked, verbally abused, beaten, and unable to leave or report their difficulties to kin. Estimates and imputations are based on labor force and relationship-to-head variables. We find that domestic service makes up a substantial fraction of girls’ employment in some countries. We also analyze trends in live-in versus live-out status and school enrollment of child domestic servants. While all child workers are disadvantaged in enrollment relative to non-workers, domestics are sometimes better off than non-domestic workers. In some samples, live-ins are more likely to go to school than live-out child domestics. In others, they are substantially worse off.
I. Introduction
Counting child domestic servants seems a commonplace task. In fact, it is both complex and important. Child domestic servants are among the most vulnerable of child workers, and the most invisible. They may be treated kindly and allowed to attend school, or they may be secluded in their employers’ homes, beaten, overworked, and unable to leave or report their difficulties to kin. In this article we ask how many children in six Latin American countries are employed as domestics, how many live with their employers, whether domestics make up a high proportion of child workers, and whether they are disadvantaged in school attendance, in the hope that this information will be useful to policy makers and non-governmental organizations.
We have been told that it is not possible to count child domestic servants.1 Too many of them are “invisible”: they are engaged in informal work, hidden away in residences, and sometimes identified to census and survey enumerators as relatives rather than servants. If they are in fact both relatives and servants – as in the Cinderella story, told world-wide in many versions, which recounts the tale of a step-daughter who served as a maid2 – they themselves may prefer to be identified as family. There are many reasons why counting and identifying trends in the use of child domestics may be difficult; we discuss these below. Still, under some conditions, we assert that we can make reasonable estimates of child domestic servants. Using census data made available via the IPUMS-International project3, we present estimates, time trends, and descriptive information about child domestics for six Latin American countries: Argentina, Brazil, Chile, Colombia, Costa Rica, and Mexico. We know of no other large-scale attempt to count child domestics.
Our focus on Latin America is practical: when we began this project, the IPUMSInternational census data collection did not include a critical mass of censuses from any other region that allowed us to identify domestic servants.4
We examine trends in the usage of child domestic servants over time. Our earliest data is from 1960, but this varies by country; in some cases we can follow trends over four decades, in other cases only three decades. We expect to see changes over time for a variety of reasons, as Latin American countries have experienced a number of large-scale social and economic changes over the relevant time period. Prominent among these are demographic transition; expanding access to education; increases in women’s labor force participation; economic fluctuations and trends, such as sectoral shifts in labor force opportunities; and human rights campaigns, Some of the reasons imply decreases in child domestic service; other imply increases. Since many of the social changes we describe have happened (or are happening) more or less simultaneously, we do not attempt to attribute particular causes to the observed patterns.5
In this paper we use the term “child” as it is used in the United Nations Convention on the Rights of the Child: to indicate that a person is under 18 years of age. Although we recognize that many adolescents in poor countries are effectively adults in their mid-teenage years, we use the term “child” to refer to both younger children and adolescents. Because very few children can be useful before approximately the age of five, we initially searched for domestic workers ages five to seventeen year-olds (inclusive). Most of the child domestics we identify are between the ages of 10 and 17, so our analysis focuses on that age range.
II. Conceptual Difficulties in Defining “Domestic Servant”
Most authors writing about child domestic servants do not bother to include a definition of what, exactly, they mean by “domestic servant.” After all, it seems obvious: a domestic servant is someone who does domestic work in someone else’s home, for pay or in-kind remuneration.
But what is domestic work, and how much does one need to do it (for someone else) in order to be classified as a domestic servant? In some parts of Africa, for example, young girls labor from morning to nightfall under the direction of their mothers and female kin, yet this work is not counted as domestic service (Reynolds 1991). Girls who are fostered into a family may, similarly, spend their days in various types of domestic labor; are they servants? What about, in Brazil, when country cousins want to escape the stagnation of their rural towns and move in with distant kin in the city, in exchange for domestic services – are they servants? What about if they only do domestic work during part of each day and attend school for part of the day – do they count?
In other situations, people who are not related take care of their “patrons,” doing whatever work needs to be done, including much domestic work. The rewards for doing this may be undefined and depend on the good will, resources, and networks of the patron.
The point here is that some people are not called servants by the people for whom they are performing services, and yet they may be performing exactly the same tasks as others who are called servants.
Another difficulty is with respect to the tasks performed. Typical tasks performed by domestic servants include cleaning the home; shopping for food; preparing food for cooking; cooking meals; serving meals; washing dishes; carrying water, washing clothing and linens; drying and ironing the laundry; putting away groceries, clean dishes, laundry, and anything else needing tidying; child care, including dressing, diapering, feeding, taking to and from school, and watching children; care for the ill or disabled or elderly, including the most intimate types of care; and so forth. This is a long list, but it is by no means all-inclusive. Some domestic servants sweep the yard, water plants, care for kitchen gardens, or even spend time working in their employers’ fields. Others care for poultry, goats, pigs, or other farm animals. How can we tell whether someone is more of a domestic servant or more of a farm hand?
Clearly there is a continuum, with one end denoting people who are very clearly domestic servants, and the other end denoting people who are very clearly not domestic servants. In between, it gets fuzzier. Yet, we have to draw an arbitrary line if we are to count child domestic servants, since we have to define each child in our census samples as either a domestic servant or not. To a great extent, the location of this line is determined by the data that was collected in the censuses of the countries we study and, more importantly, by the people responding to census enumerators.
III. The IPUMS-International Data
The availability of integrated public use samples of census microdata makes it possible for us to investigate the presence of child domestic servants in a number of low income countries. The international samples of the Integrated Public Use Microdata Series (IPUMS-International) are freely available to researchers at www.ipums.org. We have chosen to investigate child domestics in countries for which samples are available via IPUMS-International, and for which particular information was collected that helps us to identify domestic servants. In this paper, we focus on Latin American countries: Argentina, Portuguese-speaking Brazil, Chile, and Colombia in South America; Costa Rica in Central America; and Mexico in North America.
Data from multiple censuses per country was used in order to examine trends over time. While the years available differed depending on the country, all the censuses (and the one survey) in our sample were conducted between 1960 and 2002. Appendix Table A1 describes the years in which these censuses were conducted, as well as other characteristics of the samples.
Samples
We initially included individuals between the ages of five and 17 in our samples, excluding individuals in group quarters who were not living with relatives.6 In general, they are unlikely to be available for domestic service. In any case, they comprise a small portion of the population in our age range (in general, less than one percent). Finding that 5–9 year olds working in domestic service were represented in our sample in such small numbers as to make estimates problematic, our final sample focuses on 10–17 year-olds.
Table A1 describes the samples used in the analysis. For each country and census year, we indicate the sample density, enumeration rule, and ages for which labor force information – for occupation and industry in particular – was collected. The earliest age at which labor force activity was recorded has important implications for our estimates. The table also indicates how many 10–17 year-olds were included in the sample, as well as our conservative estimate of the total number of 10–17 year-old domestic servants in the sample, including only those employed in the reference period. These are sample sizes; they are not the estimated number of domestics in the population, which are shown in Table 1 and appendix Table A3. The small number of male domestics in these samples is one reason that much of the paper focuses on female domestics.
Table 1.
Numbers of child domestics - "best guess" estimate - girls and boys by age group (weighted) Estimates for 10–14 year olds are based partly/only on household information for some ages in some samples
| Ages 10–14 |
Ages 15–17 |
|||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Country/Year |
Total in Age Group (000s) |
Number in Labor Force (000s) |
% of (4) Female |
Number of Live- out Domestics (000s) |
% of (6) Female |
Number of Live- in Domestics (000s) |
% of (8) Female |
Total in Age Group (000s) |
Number in Labor Force (000s) |
% of (11) Female |
Number of Live- out Domestics (000s) |
%of (13) Female |
Number of Live- in Domestics (000s) |
%of (15) Female |
||
| (1) | (2) | (3) | (4) | (5) | (6) | (7) | (8) | (9) | (10) | (11) | (12) | (13) | (14) | (15) | ||
| Argentina | ||||||||||||||||
| 1970 | 2,167.4 | 179.6 | 36.8 | 20.6 | 92.7 | 26.4 | 87.7 | 1,241.1 | 463.3 | 33.4 | 39.7 | 96.2 | 43.15 | 96.6 | ||
| 1980 | 2,437.0 | 65.2 | 32.5 | 8.3 | 96.8 | 4.9 | 97.6 | 1,398.9 | 453.6 | 32.5 | 44.6 | 98.1 | 26.342 | 97.2 | ||
| 1991 | 3,338.2 | 80.8 | 38.4 | 6.2 | 89.2 | 5.2 | 89.8 | 1,757.2 | 442.4 | 35.7 | 40.2 | 92.6 | 13.864 | 97.4 | ||
| 2001 | 3416.3 | 20.0 | 32.2 | 2.3 | 72.9 | 0.8 | 57.7 | 1913.4 | 132.7 | 30.6 | 16.3 | 81.7 | 2.0 | 89.3 | ||
| Brazil | ||||||||||||||||
| 1960 | 7,297.6 | 1,132.7 | 24.3 | 77.9 | 91.4 | -- | -- | 3,734.1 | 1,649.9 | 27.5 | 144.4 | 94.4 | -- | -- | ||
| 1970 | 11,859.4 | 1,464.2 | 25.1 | 80.1 | 95.5 | 71.0 | 95.4 | 6,392.5 | 2,310.7 | 30.7 | 163.6 | 97.3 | 173.53 | 96.6 | ||
| 1980 | 14,206.7 | 1,833.3 | 29.5 | 174.7 | 93.5 | 79.4 | 96.1 | 8,462.8 | 3,397.1 | 33.4 | 286.8 | 96.1 | 165.67 | 97.2 | ||
| 1991 | 17,037.2 | 1,455.8 | 29.3 | 164.7 | 93.9 | 48.2 | 97.4 | 9,223.1 | 3,306.9 | 33.4 | 341.5 | 94.8 | 130.404 | 97.3 | ||
| 2000 | 17337.8 | 1137.0 | 32.5 | 89.7 | 93.6 | 11.6 | 95.7 | 10716.9 | 2789.1 | 35.7 | 292.6 | 94.7 | 46.2 | 95.5 | ||
| Chile | ||||||||||||||||
| 1960 | 807.0 | 29.5 | 30.1 | 1.9 | 82.6 | 6.6 | 83.8 | 431.4 | 126.2 | 30.8 | 5.6 | 94.0 | 23.987 | 91.7 | ||
| 1970 | 1,116.3 | 22.2 | 30.4 | 4.2 | 88.6 | -- | -- | 573.0 | 113.1 | 29.7 | 22.7 | 92.0 | -- | -- | ||
| 1982 | 1,215.4 | 2.7 | 45.9 | -- | -- | -- | -- | 773.0 | 100.5 | 30.8 | 21.6 | 93.0 | -- | -- | ||
| 1992 | 1,205.8 | 8.2 | 25.9 | 0.9 | 76.6 | 0.7 | 92.8 | 692.6 | 78.3 | 28.2 | 7.3 | 84.3 | 7.02 | 95.6 | ||
| 2002 | 1398.2 | -- | -- | -- | -- | -- | -- | 756.0 | 66.2 | 38.3 | 1.5 | 86.8 | 0.8 | 97.4 | ||
| Colombia | ||||||||||||||||
| 1964 | 2,269.6 | 243.4 | 25.5 | 5.1 | 76.2 | 55.6 | 78.1 | 1,089.8 | 396.8 | 26.6 | 8.4 | 88.1 | 61.55 | 88.5 | ||
| 1973 | 2,891.7 | 233.3 | 26.6 | 6.5 | 87.2 | 39.8 | 89.5 | 1,448.2 | 407.3 | 30.2 | 10.8 | 94.6 | 67.29 | 94.5 | ||
| 1985 | 3,169.5 | 288.2 | 33.9 | 19.0 | 87.5 | 26.1 | 77.3 | 1,920.0 | 504.7 | 34.9 | 35.1 | 94.4 | 67.017 | 84.6 | ||
| 1993 | 3722.0 | 286.5 | 25.4 | 5.6 | 91.0 | 21.3 | 85.3 | 1969.6 | 482.7 | 28.1 | 13.7 | 94.5 | 53.9 | 90.6 | ||
| Costa Rica | ||||||||||||||||
| 1963 | 174.8 | 13.9 | 18.1 | 0.8 | 75.6 | 1.4 | 83.5 | 81.0 | 31.4 | 22.3 | 1.8 | 90.7 | 2.72 | 99.4 | ||
| 1973 | 273.8 | 15.2 | 19.6 | 0.9 | 97.8 | 1.5 | 87.7 | 136.3 | 44.3 | 26.1 | 3.0 | 99.0 | 4.35 | 99.3 | ||
| 1984 | 270.8 | 15.8 | 14.7 | 0.9 | 98.9 | 0.6 | 90.3 | 166.8 | 47.5 | 19.8 | 2.9 | 100.0 | 2.17 | 98.2 | ||
| 2000 | 429.9 | 10.5 | 17.8 | 0.6 | 90.0 | 0.3 | 51.9 | 241.1 | 41.0 | 21.9 | 2.6 | 92.7 | 0.7 | 95.5 | ||
| Mexico | ||||||||||||||||
| 1960 | 4,278.4 | -- | -- | 31.4 | 87.4 | -- | -- | 2,113.6 | -- | -- | 52.4 | 91.9 | -- | -- | ||
| 1970 | 6,395.4 | 424.3 | 32.8 | 41.7 | 95.9 | -- | -- | 3,159.7 | 978.5 | 32.7 | 99.8 | 95.6 | -- | -- | ||
| 1990 | 10,438.3 | 431.8 | 24.7 | 25.1 | 94.4 | 25.8 | 81.5 | 6,028.2 | 1,493.0 | 28.1 | 73.8 | 93.6 | 59.51 | 92.4 | ||
| 1995 | 10,636.7 | 896.4 | 28.0 | 48.7 | 96.1 | 8.1 | 91.1 | 6,122.3 | 2,045.0 | 33.1 | 129.8 | 94.3 | 48.144 | 94.0 | ||
| 2000 | 10,883.6 | 566.7 | 30.3 | 36.0 | 92.2 | 7.6 | 87.9 | 6,209.1 | 1,826.1 | 34.5 | 99.7 | 94.7 | 44.088 | 93.9 | ||
Defining Child Domestics
The measures of child domestic service used here are based on information from two sources within Latin American census data: employment-related information, in particular the child’s occupation and industry, and relationship to head of household or family (that is, to the reference person). Countries must usually include these two sources of information on domestic servants in their censuses in order to be included in this analysis.
Occupation and industry data are collected for individuals who are recognized as members of the labor force. (Definitions of the census universes for labor force questions are given in appendix Table A2.) In most Latin American censuses, it has been standard for some decades to recognize domestic service as a distinct occupational category or set of categories.
The second source of information comes from a description of how an individual is related either to the head of the household or the head of a subfamily. In Latin America, it is standard to include “domestic servant” as one possible relationship to the reference person.
In some ways, both of these sources are inadequate, insofar as the respondent may not be well-informed about the activities of the child in question or may wish to mislead the enumerator. While all the censuses in our study are based on questions posed to respondents by enumerators (rather than mail-in questionnaires), census procedures almost always rely upon a principal respondent for each household or family. Adults home at the time of the enumerator’s visit typically respond on behalf of children, especially younger children. (Older adolescents, if present, may or may not be allowed to self-report.) Responding adults may not be well-informed about, for example, the number of hours worked in a week by any particular child. They may wish to stretch the truth: a distant relative living-in and doing the household’s domestic work may be described as a relative rather than a servant, for example. This is especially likely for younger children. The stigma of domestic service contributes to the invisibility of child domestics. Publicity campaigns about child labor increase misreporting, as respondents learn to be fearful of repercussions for the use of child servants.
Using the first source of information, labor force variables relating to industry and occupation of employment, we identified whether or not an individual was reported as being included in domestic service industries and/or occupations. In some cases industry and occupation variables provided identical information. In other cases, both were needed. For example, in the 1991 Argentina census, individuals were identified as domestic servants if they were labeled as being in an “other service activities” industry and a “workers in domestic services, non-specialized” occupation.7
One issue that arose with the use of labor force data was whether or not to include domestic workers who were unemployed at the time of the census. Because children tend to move in and out of employment more frequently than do adults, on any particular census day we would expect to find unemployed child domestics who had been employed the previous month and who would be employed again shortly (Levison et al, 2007). We do not include unemployed domestics in our conservative estimates, shown in appendix Table A3. However, because our goal is to count the numbers of children who usually work as domestics, we include the unemployed in our “best guess” counts of child domestics, shown in Table 1. In any case, they make up a very small fraction of all domestics, as discussed below.
Using the second source of information, we identified whether or not an individual was categorized as a domestic servant by her household relationship. In a substantial number of cases, individuals recognized as being servants of the household or family head were not reported to be members of the labor force.8 A sensitivity analysis, discussed below, includes estimates of the extent to which child domestics would be undercounted based only on labor force or only on household relationship information.
We are unable to distinguish between full-time and part-time work in this analysis; hours of work are not known for part of our sample. Given the sporadic nature of domestic labor, estimated hours of work would, in any case, be especially likely to be mis-measured (by the child) or misreported (by employers or parents) due to ignorance, carelessness, or shame. If data were available, the distinction between full- and part-time work would be important to the extent that outcomes differed between children working different numbers of hours. Children working part-time, for instance, may be able to attend school more easily than those working full-time.
IV. Estimated levels and trends
Our goal is to provide estimates useful to researchers, policy-makers, and activists. While one purpose might call for the most careful, conservative estimates, another might reasonably want to include invisible domestics, using an estimate designed to do this. We thus present several different estimates of the numbers of child domestics discernable using samples of census data. Some users will prefer our most conservative estimates of the numbers of child domestics, presented in appendix Table A3: it excludes unemployed children, and it only includes ages for which we have labor force information. The estimates presented in Table 1, however, are our “best guess” estimates of the numbers of 10–17 year-old child domestic servants in these six countries. In Table 1 we have included all the bits of information we have. In Table A3, when we had labor force information starting at age 12 (as in Costa Rica), then the entire estimate was for ages 12–14. In Table 1, we added information for 10- and 11-year-olds based on the relationship-to-head variable to the more complete data for 12–14 year-olds. Since unemployed domestic servants are likely to work again as domestics, they have also been included in these estimates. The Table 1 estimates are not substantially larger than the conservative estimates, and most of the differences are for the younger age group.
Table 1 is divided into two sets of columns, with columns (2) through (9) referring to ages 10–14, while columns (10) through (15) refer to ages 15–17. We include estimates of numbers for labor force employment and numbers of live-in and live-out domestics, also providing the total number of children in the age group so that readers may calculate any of a number of percentages.9 Additional columns show the percentage female for each estimated number. It is important to keep demographic trends in mind when using Table 1. Column (10) most clearly reflects the increase in youth due to mortality decline. Brazil had 3.7 million 15–17 year-olds in 1960, but 10.7 million in 2000. It is noticeable that employment growth rates rarely keep pace with population growth rates for this age group, indicating declining labor force employment rates
Overall child employment
In some countries, the percentages of girls and boys who work in the labor force have been declining since the 1960s (see appendix Table A4). In Argentina, declines have been monotonic since the first available sample for 1970. In Brazil, the same can be said for boys (since 1960), but for girls declines have been steady only since 1980. In Chile and Costa Rica declines are monotonic on the whole, with minor exceptions Colombia’s patterns are more complicated. The employment percentages do not vary substantially between 1964 and 1993 for three of the four age-sex groups. For 15–17 year-old boys, however, employment declined from 57 percent in 1964 to about 36 percent in the 1980s and 1990s. Employment also fell for 15–17 year-old girls in 1993. Mexico’s employment rates bounce around a bit but do not show any substantial declines over the four decades presented here.
A higher share of boys than girls is employed, in both age groups and in every country and sample. Generally, girls make up fewer than one-third of youth employed in the labor force (Table 1, columns (4) and (11)).
Employment as domestics
In every country, sample, and both age groups, boys who are working in the labor force are substantially less likely than girls to be employed as domestic servants (Table 1, columns 6, 8, 13, 15). In fact, fewer than five percent of employed boys are typically domestics. For this reason, most of our discussion focuses on girls. Appendix Table A4 shows, for six countries and two age groups, the percentages of girls who are employed, the percentage of employed girls who are employed as domestic servants, and the percentages of all girls who are employed as domestics,. A key finding of this paper is that domestic service accounts for a substantial fraction of girls’ labor force employment – a remarkably high fraction in some samples. Among employed girls, at least 20 percent are domestic servants in most samples. In the older samples, sometimes 60 to 80 percent of employed girls were domestics. In more recent samples, 30 percent is more common.
In Argentina and Colombia, there are close-to-monotonic declines in the percentage of employed girls who are domestic servants between the mid-1960s and the end of the century. In Brazil, Costa Rica and Mexico, the percentage of employed girls who are domestics first increases, then declines. A similar pattern can be seen when examining the estimated numbers of children who are employed as domestics – in this case, for all countries except Chile and Argentina. It should be noted that data for the 1960s is not available for Argentina; perhaps we would find a similar pattern if we had that data. It is likely that as social norms changed in Latin America, the demand for child domestics was less elastic than the demand for other kinds of child workers. That is, as more and more children began to spend more and more time in school, the reduced availability of a family’s own children for domestic work combined with the increase in women’s labor force participation meant that many households must have felt an increased need for help in accomplishing essential household tasks and caring labor. Thus, even while the total rate of labor force employment was falling for children, the demand for child domestics was strong enough to cause increases in employment in both absolute numbers and as a percentage of all employed children. Eventually, however, a combination of other social factors – including the normative understandings that children should be in school and should not be full-time workers, smaller numbers of available children due to fertility decline, and increased labor-saving devices among the middle class – led to declines in both absolute numbers of children who are domestics and percentages of employed girls working as domestics.
While a relatively small percentage of girls were employed as domestics in the most recent census years – at or below five percent in most countries (with Brazil slightly higher) – Table 1 shows that the absolute numbers of child domestics were substantial in the larger countries: over 400 thousand in Brazil, over 180 thousand in Mexico, and over 90 thousand in Colombia.
Comparing the percentages of younger and older employed girls who are domestics, we find that in almost every case, a higher percentage of the older girls are domestics. This could, in part, be due to parents moving girls out of work on the street as they reach puberty. Madsian (2004: 130) writes about Brazilian children who work as peanut vendors:
Parents, above all, are concerned that their daughters maintain their virginity and live up to the image of the ideal woman. The street is a constant source of danger, and may even lure girls to prostitution. Regularly the girls, and occasionally also the boys, are solicited for sexual services [as they sell peanuts]. Hence, around the age of 15, girls tend to stop working on the street.
Other girls may enter domestic service simply because it is an obvious first employment, given the training in domestic skills that most receive in their own homes.
Living-in vs. living-out
We are able to discern whether a domestic servant is “living in” – that is, residing with her employers – or “living out” by the way in which she was enumerated. Those children who were identified as domestic servants by the household relationship variable must have been “living in” (if they were correctly enumerated). Children who were enumerated with their own families but were identified by their labor force information as domestic servants were assumed to be “living out.” We pay attention to this status because of its implications for the relative power and privileges of a domestic servant. Children who live with their own kin may report abuses of employers, while live-in domestics may have much more limited access to kin or others who could assist them. The place of residence of a child domestic may also have implications for her/his schooling.
School enrollment
Based on the literature, we expected domestic workers to be disadvantaged in terms of education related to other workers in their age groups as well as to non-workers, even if such service provided some children with educational advantages. Oyaide (2000), for example, found that among 159 child domestics identified in various parts of Lusaka, Zambia, only one was attending school. We do not make a causal argument here, because any association between enrollment and domestic service may be due to who becomes a servant (a selection issue). Ainsworth (1992) found that in Côte d’Ivoire, children who left rural areas to become domestics in urban areas were less likely to be in school than other children of the households in which they worked, but the child domestics were more likely to be in school than the siblings they left behind.
We use a very basic measure of education: school enrollment. Most census questions on this topic translate to something like, “Is [this person] going to school?” (See Table A2.) The analysis confirms that domestic workers10 are disadvantaged in enrollment in comparison to non-working individuals in their age group was confirmed by the analysis. Our evidence shows, however, that domestic workers are not always disadvantaged in comparison to workers in other industries.
The pattern for Brazil is typical of Latin America: Brazil’s school enrollment-by-age figures shift upwards over the decades, with peak enrollment about ages 10–11 (not shown).11 Enrollment typically declines with age throughout the teen years, as children move into more adult roles.
In Brazil, live-out domestics have an advantage in education over live-in domestics until the late teen years in decades prior to 2000. Moreover, in earlier decades, younger live-ins clearly have an advantage over other (non-domestic) workers. In 2000, however, older (age 13+) live-in domestics are substantially less likely to be enrolled in school than live-out domestics or other workers. Similarly, in Mexico, a clear live-in enrollment advantage in school enrollment in 1990 changed to a clear live-in disadvantage by 2000.
In Colombia, in 1973 and 1993, domestic workers who live with their employers are more likely to be enrolled in school than live-out domestics or other workers. This advantage does not exist in 1985, a year in which all child workers’ enrollment levels were high relative to other years.
Costa Rica shows the most decided advantage for younger live-in domestics with respect to school attendance, with a slight advantage continuing at some older ages. Child domestics in Costa Rica are more likely to attend school than other child workers.
It is interesting to note that where there has been a shift over time in the degree of educational advantage of live-in versus live-out domestics, it has become more of a disadvantage to live in. Overall, however, live-in child domestic servants were not necessarily disadvantaged in enrollment compared to live-out servants.
Without a better understanding of the direction of causality or potential selection issues, we cannot explain the reasons underlying these patterns. For example, it seems likely that as fewer families have felt the need to place children in live-in situations in order to ensure them regular meals, those who have remained in live-in service are from the most destitute families – that is, increasingly selected – and least able to leave employers who keep them out of school. On the other hand, it could be that as school became accessible to more children, only those children who did very poorly (for example, because of reading disorders) dropped out and became live-in domestics. In this scenario, children are selected on educational success rather than poverty (although the two are highly correlated) and causality runs from school to domestic service rather than vice-versa. Both patterns may occur within one population of child domestics.
V. Sensitivity Analysis
Our first type of sensitivity analysis, in which we compare conservative estimates of child domestic servants (appendix Table A3) with our “best guess” estimates (Table 1) has already been discussed. The more conservative estimates were produced only for samples and/or age groups where both our primary sources of information were available. The “best guess” estimates took into account all available data that might inform us about the children’s work.
Other studies of child domestic workers may have to rely only on household status information or only on labor force information. In either case, the result will be an undercount of child domestic workers. Appendix Table A5 reports the degree to which the number of child domestics estimated is sensitive to labor force or household relationship information availability, focusing on recent samples for which labor force information is available for the entire age group in question.
If we only had one source of information, how much would we underestimate the number of child domestic servants, compared to using both sources of information (conservative estimate, as in Table A3)?
Inadequate labor force measures
The labor force status of some child domestic servants is not acknowledged by the adults responding to census enumerators. How big a problem is this? According to Table A5, column (3), it is a relatively small problem in most of the countries included in this analysis. In Brazil and Chile, for example, 99 percent of domestics are identifiable using labor force information. The greatest degree of misreporting is found in Costa Rica: seven percent of 15–17 year old domestics were not identifiable using only labor force data.
Inadequate household relationship measures?
Household relationship measures also did not catch some domestic servants that labor force measures did. In many cases, this is to be expected, given that live-out domestic servants are not enumerated with the households of their employers. For example, if a live-out domestic servant lives with her parents, she should be enumerated as “child” according to the household relationship variable. Table A5 (column 4) shows the percentages of child domestics that were captured using only the household relationship variables. In Colombia 80 percent of domestics were identifiable via the relationship variables only, but in the other five countries much smaller proportions were identifiable in this way. In Argentina and Brazil, over 85 percent of domestics would be overlooked if one were using only household relationship information.
Imputed live-in domestic servant status
Given the high potential for non-reporting of the true work status of relatives acting as servants (“Cinderellas”) – especially in countries with highly-visible anti-child-labor campaigns – we consider the extent to which we might have undercounted live-in child domestics. We do this using the household and/or family relationship variables. It is possible to identify people in the household who do not have a clearly-identified relationship to the head. Thus, “other relatives,” “agregados” (in Brazil), and “non-relatives” have Cinderella potential. Clearly, some of them are not domestic servants. For example, Latin American censuses do not have a “girlfriend / boyfriend” category, so an unidentified adolescent may be the live-in partner of a family member. A few censuses have a category for “child of a servant.” Because we suspect that children of servants are treated more like servants than like the sons and daughters of the head in households, since their parent is a live-in domestic, we consider this small group to have Cinderella potential. Since most domestics servants are girls (see Table 1), we considered only females in the arriving at an estimated number of unreported domestics. The total number of girls in this “Cinderella” group gives an upper bound for the number of hidden live-in female domestics, so it is reported in column (6) of Table A5.
The numbers of co-resident youth with Cinderella potential seem much too large – surely not all of these youth are domestic servants, even on a part-time basis. However, they put an upper bound on the number of child domestics we could be missing. In column (6) we assume that an arbitrary percentage of them – currently 25 percent – are in fact domestic servants. (We continue to search for qualitative evidence on which to base this percentage.) We expected these imputations to substantially increase our estimates of the numbers of child domestic servants. Oddly enough, they do in some countries but not in others. For Mexico, they increase the original estimate by only six percent.
Advocacy organizations could play a role in identifying hidden child domestic servants. In terms of absolute numbers, this issue is most pressing in Brazil and Colombia.
VI. The Welfare of Child Domestic Servants
While many differences exist in the conditions under which children perform domestic work, child domestic servants are vulnerable to several characteristic threats. For one, they may not be allowed to take breaks or may be required to work long hours. Child domestic workers may also suffer from a lack of access to education, which can contribute to social isolation and a lack of future opportunity for the child (Oyaide, 2000). UNICEF considers domestic work to be among the lowest status, least regulated, and most poorly remunerated of all occupations, for either adults or children, and reports that most child domestics are live-in workers and under the round-the-clock control of their employer (Innocenti Digest, 1999).
When exploitation of the child worker is extreme, or conditions are akin to slavery, the ILO considers domestic service to be a “worst form” of child labor (Black, 2005). Stories of beatings and sexual abuse are not uncommon among qualitative studies of child domestic servants.12 Due to the fact that they frequently live with their employers out of others’ view, child domestic workers may be particularly vulnerable to this type of exploitation. This summary of study findings from Haiti includes features found in other studies from around the world:
“…[servant] children living outside the home tended to have a heightened risk of treatment as second class citizens and also a heightened risk of physical and sexual abuse – though neither is inevitable. According to field interviews, the living conditions of servant children tend to be distinctly different from other children in the same household. They sleep in the least desirable places, e.g., on a section of carpet in the outside kitchen or on the floor at the foot of a bed. They eat different food. They do significantly more work than other children in the household. They may well carry the workload of adult domestic servants and more. According to direct observation by informants, such children are subject to public humiliation and corporal punishment including beatings with cooking pots, shoes, whips, or fists. They may well not go to school, or if they do, it is an inferior school and in any case a different school from those attended by other children in the household. They are subject to sexual abuse by other children in the household and sometimes by adults, yet they would not be likely to be allowed to marry the sons or daughters of the household served” (Smucker and Murray 2004: 35–36).
In this paper we differentiate between live-in and live-out domestic servants as a way of taking account of these heightened risks. We are, however, unable to compare actual conditions of employment, either within the domestic service occupation or in comparison to other sectors of employment. What evidence exists comes from small-scale studies and appears to be context-specific. Moreover, negative reports are likely to be highlighted in the reports of non-governmental organizations (NGOs) – for fund-raising purposes – while positive reports may be overlooked. It is not possible to calculate the proportion of live-in child domestics who are abused based on media or NGO reports.
Observers are careful to note that one should not automatically assume that child domestic work is exploitative or worse than what a child would experience if he or she were not a domestic worker. For many families, placing a child in a stable household that has a higher standard of living than the parents’ household is seen as beneficial (Innocenti Digest, 1999); for the most vulnerable families and for orphans, it may be a way to ensure that a child is fed, clothed, and sheltered. Some children are sent from rural homes to be domestics in urban areas, in order to enable them to further their education. We address these issues at greater length below.
VII. Policy Implications
Whether or not census data can be used to accurately identify child domestic servants depends to a great extent on the census tradition in particular countries. We encourage ministries of statistics to (1) include a servant category among the relationship-to-reference-person options; (2) specify domestic service as an occupation, unmixed with other occupations; and (3) collect labor force statistics starting no later than age 10. Higher cut-offs simply assume away child workers.
A key finding of this paper is that domestic service accounts for a substantial fraction of girls’ labor force employment. Combined with information from qualitative studies about the poor conditions under which many children work as servants, this finding points to a need for a substantial emphasis on domestic service in programs aimed at reducing the negative effects of work on children.
One aim of this paper is to allow activists to determine whether many child domestics are employed in particular Latin American countries. To the extent that child domestic service does exist on a substantial basis, we hope that any activism around this issue will consider children’s alternatives. It is important to listen to the children themselves, such as this girl:
“Do you understand how you insult me, when you talk of ‘combating’ and ‘eradicating’ the work that I do? I have worked as a domestic servant since I was eight. Because of this, I have been able to go to school (which my parents in the village could not afford). I also help my parents with the money that I earn. I am proud of the work that I do! I joined the movement of working children, and I know what the Convention says about children’s rights – it says that you must listen to me !”
– 13-year old Senegalese girl, Urban Childhoods Conference, Norway 199713
While domestic service is unlikely to be ideal for any given child, it may be better than the child’s alternatives. We have shown that sometimes children who are live-in domestic servants are more likely to be enrolled in school than other domestics or children doing other kinds of labor force work. Other times, child domestics are clearly disadvantaged relative to other child workers with respect to enrollment.
Our results indicate a need for future research to determine what percentage of “potential Cinderellas” are really hidden domestic servants. For example, Smucker and Murray (2004) document a variety of arrangements of children who live or work away from their biological parents in Haiti. A restavèk is “a person who lives with others and serves them, an unpaid domestic servant”; labeling someone a restavèk relegates him or her “to the lowest possible servile status” (page 21). Other categories are identified via terms indicating adoptive kinship (pitit), living with an extended family (pitit kay), or less pejorative terms for unpaid servant children (timoun). However, in all of these cases, children living away from their parents are expected to perform some domestic tasks, and unpaid servants are expected to work much harder than the children of the house. If this study had identified what proportion of children living with non-parents fall into the different categories, its results could have been used in conjunction with census data to produce better estimates of the numbers of child domestics.
Similarly, Jacquemin (2004: 384–5) describes three types of child domestics in Abidjan, Côte d’Ivoire: the little niece, who works for kin; the hired help, who works for strangers and whose payment goes directly to her guardian; and the little paid maid who also works for strangers but is paid directly, in cash. The author notes that “some paid maids consider that they only started to ‘work’ when they had their first [employment] placement where they received a monthly salary, while during the months or even years before that, they had been carrying out exactly the same tasks” but in the role of little niece or hired help (p. 392). This implies that standard labor force questions designed for adults may not capture all of the Cinderellas, even if they speak on their own behalf.
Maggie Black (1997) has written a handbook about how to identify and interview potential child domestic servants. She points out there is “pressure in numbers”: “without estimates of numbers [of child domestics], we cannot make the point that this is a large group of child workers and deserves serious attention” (page 41). Michael Bourdillon (2009: 1) states that support for child domestic workers “should be a matter of urgency.” He writes that that such support means respecting children who “have tried to overcome adversity by working for themselves and their families, often in painful situations” (p.13) and argues convincingly that a ban on child domestic work will not do this.14 Programs to improve the working conditions of child domestics, such as Shoishab in Dhaka, Bangladesh (described in Black 2002: 47–49), will need funding based, to some extent, on the numbers of child domestics in the community.
We have shown that in some countries, there is a large number of potential Cinderellas, in addition to the more identifiable child domestic servants. The fact that the great majority of identifiable domestics are girls, and that they are engaged in something as seemingly mundane as housework, may render all of them invisible to policy makers. The status and well-being of all of these children, however, deserves further attention.
Acknowledgements
We appreciate helpful suggestions from Ragui Assaad, Michael Bourdillon, Misty Heggeness, and Christopher McKelvey. Thanks also to Owen Thompson-Ferguson, who provided useful input at an early stage in this project.
Table A1. Characteristics of census samples included in the analysis (unweighted sample sizes)
| Country/ Year |
Sample Density (%) |
Enumeration Rule |
Labor Force Data Collected |
Total Ages 10–17 |
Domestics Ages 10–17 |
Female Domestics Ages 10–17 |
Male Domestics Ages 10–17 |
|
|---|---|---|---|---|---|---|---|---|
| (1) | (2) | (3) | (4) | (5) | (6) | (7) | (8) | |
| Argentina | ||||||||
| 1970 | 2.0 | de facto | age 10+ | 68,169 | 2,584 | 2,430 | 154 | |
| 1980 | 10.0 | de facto | age 14+ | 405,850 | 9,046 | 8,805 | 241 | |
| 1991 | 10.0 | de facto | age 14+ | 678,252 | 7,789 | 7,175 | 614 | |
| 2001 | 10.0 | de facto | age 14+ | 532,968 | 2,132 | 1,718 | 414 | |
| Brazil | ||||||||
| 1960 | 5.0 | de facto | age 10+ | 551,588 | 11,048 | 10,320 | 728 | |
| 1970 | 5.0 | de facto | age 10+ | 947,460 | 24,874 | 23,980 | 894 | |
| 1980 | 5.0 | de jure & de facto | age 10+ | 1,075,606 | 34,145 | 32,691 | 1,454 | |
| 1991 | 5.8 | de jure | age 10+ | 1,551,439 | 39,840 | 37,997 | 1,843 | |
| 2000 | 6.0 | de jure | age 10+ | 1,713,976 | 28,010 | 26,510 | 1,500 | |
| Chile | ||||||||
| 1960 | 1.2 | de facto | age 12+ | 14,922 | 452 | 408 | 44 | |
| 1970 | 10.0 | de facto | age 12+ | 168,929 | 2,641 | 2,420 | 221 | |
| 1982 | 10.0 | de facto | age 15+ | 198,837 | 1,988 | 1,865 | 123 | |
| 1992 | 10.0 | de facto | age 14+ | 189,836 | 1,524 | 1,358 | 166 | |
| 2002 | 10.0 | de facto | age 15+ | 215,419 | 196 | 179 | 17 | |
| Colombia | ||||||||
| 1964 | 2.0 | de facto | age 12+ | 67,187 | 2,604 | 2,177 | 427 | |
| 1973 | 10.0 | de facto | age 10+ | 433,982 | 12,384 | 11,463 | 921 | |
| 1985 | 10.0 | de jure | age 10+ | 492,643 | 14,082 | 12,173 | 1,909 | |
| 1993 | 10.0 | de jure | age 10+ | 569,168 | 9,374 | 8,433 | 941 | |
| Costa Rica | ||||||||
| 1963 | 6.0 | de jure | age 12+ | 15,049 | 381 | 357 | 24 | |
| 1973 | 10.0 | de jure | age 12+ | 41,011 | 962 | 936 | 26 | |
| 1984 | 10.0 | de jure | age 12+ | 43,760 | 641 | 630 | 11 | |
| 2000 | 10.0 | de jure | age 12+ | 67,093 | 415 | 374 | 41 | |
| Mexico | ||||||||
| 1960 | 1.5 | dejure | all | 96,499 | 1,250 | 1,128 | 122 | |
| 1970 | 1.0 | de jure | age 12+ | 112,870 | 1,091 | 1,037 | 54 | |
| 1990 | 10.0 | de jure | age 12+ | 1,629,126 | 18,410 | 16,871 | 1,539 | |
| 1995 | 0.4 | de jure | age 12+ | 83,790 | 885 | 834 | 51 | |
| 2000 | 10.6 | de jure | age 12+ | 1,829,769 | 21,271 | 19,991 | 1,280 | |
Notes: Columns (6) – (8) use the conservative definition of employment, like that used in Table 2. Counts of domestics for Mexico 1960 include umemployed workers.
Table A2. Definitions of the census universe for labor force and current school questions, and other notes
| Country/Year |
Who is considered to be a member of the labor force? |
Who is in the universe for the current schooling question? (approx. question in italics) |
Other Notes |
|
|---|---|---|---|---|
| (1) | (2) | (3) | (4) | |
| Argentina | ¿Asiste a algún establecimienta educacional? | |||
| 1970 | Employed or experienced unemployed | Persons age 5+ | ||
| 1980 | Employed or experienced unemployed | Persons age 5+ | ||
| 1991 | Had a job last week | Persons age 3+ | ||
| 2001 | Had a job last week | Persons age 3+ | ||
| Brazil | Freqüenta escola (ou creche)? | |||
| 1960 | In the labor force | Persons age 5+ registered for school (even if temporarily absent) | No relationship-to-head code for domestic servant | |
| 1970 | In the labor force | Persons age 5+ registered for school (even if temporarily absent) | ||
| 1980 | Persons who were employed | Persons age 5+ registered for school (even if temporarily absent) | ||
| 1991 | Persons who were employed | Persons age 5+ registered for school (even if temporarily absent) | ||
| 2000 | Persons who were employed | Persons age 5+ registered for school (even if temporarily absent) | ||
| Chile | ¿Asiste actualmente a un establecimienta de enseñanza regular? | |||
| 1960 | Persons who ever worked | Only heads of households | ||
| 1970 | Either worked, did not work but had a job, or seeking work between April 13 and 18 | Persons age 5+ | No relationship-to-head code for domestic servant | |
| 1982 | Persons who ever worked | Persons age 5+ | No relationship-to-head code for domestic servant | |
| 1992 | Persons who ever worked | Question not asked | ||
| 2002 | Working or seeking employment | Question not asked | ||
| Colombia | ¿Asiste actualmente a algún establecimienta (centro) de enseñanza…? | |||
| 1964 | In the labor force | Question not asked | ||
| 1973 | With a job or experienced unemployed | Persons enrolled in school age 5+ | ||
| 1985 | Persons age 5+ (imposed by IPUMSi) | No occupation or industry data | ||
| 1993 | In the labor force; not new workers | Persons enrolled in school age 5+ | ||
| Costa Rica | ¿Asiste a la escuela… / Está matriculado en algún centro de enseñanza regular? | |||
| 1963 | Employed or unemployed | Persons age 7+ | ||
| 1973 | Persons who ever workerd | Persons age 5+ | ||
| 1984 | Persons who ever workerd | Persons age 6+ | ||
| 2000 | Employed week prior to census | Persons age 6+ | ||
| Mexico | ¿(Actualmente) Vá a la escuela? | |||
| 1960 | Question(s) not asked | Question not asked | No relationship-to-head code for domestic servant | |
| 1970 | Worked the previous year | Persons age 6+ | No relationship-to-head code for domestic servant | |
| 1990 | Persons who were employed | Persons age 5+ | ||
| 1995 | Worked the week before the census or did not work but had a job | Persons age 5+ | ||
| 2000 | Persons who were employed | Persons age 5+ | ||
Table A3. Numbers of child domestics - conservative estimate - girls and boys by age group (weighted) Estimates are reported here only for ages with labor force information available
| Younger Girls and Boys (ages depend on data availability) |
Older Girls and Boys (ages 15–17) |
|||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Country/Year |
Ages Included |
Total in Age Group (000s) |
Number Employed in Labor Force* (000s) |
% of (4) Female |
Number of Live- out Domestics |
% of (6) Female |
Number of Live- in Domestics (000s) |
% of (8) Female |
Total in Age Group (000s) |
Number Employed in Labor Force (000s) |
% of (11) Female |
Number of Live- out Domestics (000s) |
% of (13) Female |
Number of Live- in Domestics (000s) |
% of (15) Female |
|
| (1) | (2) | (3) | (4) | (5) | (6) | (7) | (8) | (9) | (10) | (11) | (12) | (13) | (14) | (15) | (16) | |
| Argentina | ||||||||||||||||
| 1970 | 10–14 | 2,167.4 | 179.6 | 36.8 | 20.3 | 92.6 | 26.4 | 87.7 | 1,241.1 | 463.0 | 33.3 | 39.4 | 96.2 | 43.2 | 96.6 | |
| 1980 | 14 | 472.0 | 65.2 | 32.5 | 8.2 | 96.8 | 4.9 | 97.6 | 1,398.9 | 452.8 | 32.4 | 43.8 | 98.1 | 26.3 | 97.2 | |
| 1991 | 14 | 518.9 | 78.0 | 36.8 | 6.2 | 89.2 | 2.5 | 97.3 | 1,757.2 | 442.4 | 35.7 | 40.2 | 92.6 | 13.9 | 97.4 | |
| 2001 | 14 | 616.1 | 19.4 | 31.5 | 2.3 | 72.9 | 0.2 | 63.2 | 1,913.4 | 132.7 | 30.6 | 16.3 | 81.7 | 2.0 | 89.3 | |
| Brazil | ||||||||||||||||
| 1960 | 10–14 | 7,297.6 | 1,132.2 | 24.3 | 77.4 | 91.5 | -- | -- | 3,734.1 | 1,649.1 | 27.5 | 143.6 | 94.4 | -- | -- | |
| 1970 | 10–14 | 11,859.4 | 1,463.1 | 25.0 | 79.0 | 95.4 | 71.0 | 95.4 | 6,392.5 | 2,308.7 | 30.6 | 161.6 | 97.3 | 173.5 | 96.6 | |
| 1980 | 10–14 | 14,206.7 | 1,833.3 | 29.5 | 174.7 | 93.5 | 79.4 | 96.1 | 8,462.8 | 3,397.1 | 33.4 | 286.8 | 96.1 | 165.7 | 97.2 | |
| 1991 | 10–14 | 17,037.2 | 1,455.8 | 29.3 | 164.7 | 93.9 | 48.2 | 97.4 | 9,223.1 | 3,306.9 | 33.4 | 341.5 | 94.8 | 130.4 | 97.3 | |
| 2000 | 10–14 | 17,337.8 | 1,137.0 | 32.5 | 89.7 | 93.6 | 11.6 | 95.7 | 10,716.9 | 2,789.1 | 35.7 | 292.6 | 94.7 | 46.2 | 95.5 | |
| Chile | ||||||||||||||||
| 1960 | 12–14 | 484.6 | 28.2 | 30.3 | 1.7 | 81.0 | 6.5 | 83.3 | 431.4 | 125.8 | 30.7 | 5.1 | 95.2 | 24.0 | 91.7 | |
| 1970 | 12–14 | 656.8 | 22.1 | 30.2 | 4.1 | 88.6 | -- | -- | 573.0 | 112.7 | 29.5 | 22.3 | 92.2 | -- | -- | |
| 1982 | none | -- | -- | -- | -- | -- | -- | -- | 773.0 | 98.8 | 29.9 | 19.9 | 93.8 | -- | -- | |
| 1992 | 14 | 232.8 | 8.1 | 25.2 | 0.9 | 74.7 | 0.7 | 92.8 | 692.6 | 77.6 | 27.7 | 6.7 | 83.8 | 7.0 | 95.6 | |
| 2002 | none | -- | -- | -- | -- | -- | -- | -- | 756.0 | 65.9 | 38.1 | 1.2 | 87.5 | 0.8 | 97.4 | |
| Colombia | ||||||||||||||||
| 1964 | 12–14 | 1,327.7 | 226.4 | 22.0 | 4.9 | 76.5 | 43.7 | 80.5 | 1,089.8 | 396.6 | 26.5 | 8.2 | 87.7 | 61.6 | 88.5 | |
| 1973 | 10–14 | 2,891.7 | 233.1 | 26.5 | 6.3 | 87.3 | 39.8 | 89.5 | 1,448.2 | 407.1 | 30.1 | 10.5 | 94.8 | 67.3 | 94.5 | |
| 1985 | 10–14 | 3,169.5 | 287.8 | 33.8 | 18.6 | 87.3 | 26.1 | 77.3 | 1,920.0 | 503.5 | 34.7 | 33.9 | 94.4 | 67.0 | 84.6 | |
| 1993 | 10–14 | 3,722.0 | 286.3 | 25.4 | 5.4 | 90.9 | 21.3 | 85.3 | 1,969.6 | 482.2 | 28.0 | 13.2 | 94.4 | 53.9 | 90.6 | |
| Costa Rica | ||||||||||||||||
| 1963 | 12–14 | 100.6 | 13.7 | 17.7 | 0.6 | 85.7 | 1.4 | 86.3 | 81.0 | 31.3 | 22.4 | 1.7 | 96.0 | 2.7 | 99.4 | |
| 1973 | 12–14 | 161.1 | 15.1 | 19.4 | 0.9 | 97.7 | 1.4 | 90.4 | 136.3 | 44.3 | 26.0 | 2.9 | 99.0 | 4.4 | 99.3 | |
| 1984 | 12–14 | 162.5 | 15.7 | 14.2 | 0.9 | 98.9 | 0.5 | 94.4 | 166.8 | 47.4 | 19.6 | 2.7 | -- | 2.2 | 98.2 | |
| 2000 | 12–14 | 254.7 | 10.5 | 17.8 | 0.6 | 90.0 | 0.2 | 60.0 | 241.1 | 41.0 | 21.9 | 2.6 | 92.7 | 0.7 | 95.5 | |
| Mexico | ||||||||||||||||
| 1960 | 12–14 | 2,530.4 | -- | -- | -- | -- | -- | -- | 2,113.6 | -- | -- | -- | -- | -- | -- | |
| 1970 | 12–14 | 3,761.5 | 413.8 | 31.1 | 31.2 | 94.9 | -- | -- | 3,159.7 | 956.6 | 31.2 | 77.9 | 95.1 | -- | -- | |
| 1990 | 12–14 | 6,340.5 | 427.8 | 24.3 | 25.1 | 94.4 | 21.8 | 84.9 | 6,028.2 | 1,493.0 | 28.1 | 73.8 | 93.6 | 59.5 | 92.4 | |
| 1995 | 12–14 | 6,388.3 | 895.4 | 27.9 | 48.7 | 96.1 | 7.1 | 90.2 | 6,122.3 | 2,045.0 | 33.1 | 129.8 | 94.3 | 48.1 | 94.0 | |
| 2000 | 12–14 | 6,466.2 | 566.2 | 30.3 | 36.0 | 92.2 | 7.1 | 89.3 | 6,209.1 | 1,826.1 | 34.5 | 99.7 | 94.7 | 44.1 | 93.9 | |
Note: Includes domestic servants who were not enumerated as economically active. Dashes indicate categories for which data is not available.
Note: For Mexico 1960, currently employment is not known. Table 3 estimates are based on reported occupation and industry.
Table A4. Selected Employment Characteristics of Samples (weighted)
| Country & Year | % Employed | % of Employed Girls who are Domestics |
% of All Girls who are Domestics |
||||||
|---|---|---|---|---|---|---|---|---|---|
| 10–14 | 15–17 | 10–14 | 15–17 | 10–14 | 15–17 | ||||
| F | M | F | M | ||||||
| (1) | (2) | (3) | (4) | (5) | (6) | (7) | (8) | ||
| Argentina | |||||||||
| 1970 | 6.2 | 10.3 | 24.9 | 49.7 | 63.4 | 51.6 | 3.9 | 12.8 | |
| 1980 | 1.8 | 3.6 | 21.0 | 43.7 | 60.2 | 46.7 | 1.1 | 9.8 | |
| 1991 | 1.9 | 3.0 | 17.9 | 32.5 | 32.9 | 32.1 | 0.6 | 5.8 | |
| 2001 | 0.4 | 0.8 | 4.3 | 9.5 | 32.5 | 37.1 | 0.1 | 1.6 | |
| Brazil | |||||||||
| 1960 | 7.5 | 23.5 | 23.3 | 66.9 | 25.7 | 29.9 | 1.9 | 7.0 | |
| 1970 | 6.2 | 18.5 | 21.3 | 52.0 | 39.1 | 46.0 | 2.4 | 9.8 | |
| 1980 | 7.7 | 18.1 | 26.4 | 54.4 | 44.2 | 38.5 | 3.4 | 10.2 | |
| 1991 | 5.0 | 12.0 | 23.9 | 47.9 | 47.2 | 40.9 | 2.4 | 9.8 | |
| 2000 | 4.3 | 8.7 | 18.7 | 33.3 | 25.7 | 32.2 | 1.1 | 6.0 | |
| Chile | |||||||||
| 1960 | 2.2 | 5.1 | 17.5 | 41.3 | 80.0 | 69.7 | 1.7 | 12.2 | |
| 1970 | 1.2 | 2.8 | 11.4 | 28.2 | 54.9 | 61.8 | 0.7 | 7.0 | |
| 1982 | 0.2 | 0.2 | 7.7 | 19.8 | -- | 63.2 | 0.0 | 4.8 | |
| 1992 | 0.3 | 1.0 | 6.3 | 14.6 | 63.2 | 57.1 | 0.2 | 3.6 | |
| 2002 | 0.0 | 0.0 | 6.7 | 10.6 | -- | 7.1 | 0.0 | 0.5 | |
| Colombia | |||||||||
| 1964 | 5.3 | 15.9 | 18.9 | 57.2 | 78.6 | 56.8 | 4.2 | 10.7 | |
| 1973 | 4.3 | 11.8 | 16.1 | 41.3 | 66.3 | 59.9 | 2.9 | 9.7 | |
| 1985 | 6.6 | 12.1 | 19.1 | 35.9 | 35.5 | 47.4 | 2.3 | 9.0 | |
| 1993 | 4.0 | 11.3 | 13.5 | 35.9 | 31.7 | 45.4 | 1.3 | 6.1 | |
| Costa Rica | |||||||||
| 1963 | 2.8 | 13.0 | 17.2 | 60.3 | 70.1 | 62.1 | 2.0 | 10.7 | |
| 1973 | 2.2 | 8.8 | 16.7 | 48.7 | 72.1 | 62.8 | 1.6 | 10.5 | |
| 1984 | 1.7 | 9.9 | 11.3 | 44.9 | 63.2 | 52.3 | 1.1 | 6.1 | |
| 2000 | 0.9 | 3.9 | 7.6 | 26.2 | 36.2 | 34.0 | 0.3 | 2.6 | |
| Mexico | |||||||||
| 1960 | -- | -- | -- | -- | -- | -- | -- | -- | |
| 1970 | 4.1 | 8.7 | 18.8 | 42.0 | 23.0 | 24.8 | 0.9 | 4.7 | |
| 1990 | 2.1 | 6.2 | 13.8 | 35.3 | 41.9 | 29.5 | 0.9 | 4.1 | |
| 1995 | 4.8 | 12.0 | 21.9 | 44.4 | 21.6 | 24.8 | 1.0 | 5.4 | |
| 2000 | 3.2 | 7.2 | 20.1 | 38.8 | 23.2 | 21.6 | 0.7 | 4.3 | |
Dashes indicate categories for which data is not available.
Table A5. Sensitivity of recent child domestics estimates to data availability and definitions, for 15–17 year olds. Also for 10–14 year olds when labor force information is available.
| Country/Year and Age group |
Conservative estimate of domestics in age group |
Percent of col (2) using Emp'd LF data only |
Percent of col (2) using HH data only |
Additional # if using "best guess" |
Other potential live-in female "Cinderellas" |
Additional (imputed) # of female live-ins using 25% of col (6) |
Estimate of domestics including imputed in col (7) |
|
|---|---|---|---|---|---|---|---|---|
| (1) | (2) | (3) | (4) | (5) | (6) | (7) | (8) | |
| Argentina 2001 | ||||||||
| 15–17 | 18,290 | 0.96 | 0.11 | 100 | 7,030 | 1,720 | 20,010 | |
| Brazil 2000 | ||||||||
| 10–14 | 101,282 | 0.99 | 0.11 | 0 | 266,556 | 66,626 | 167,908 | |
| 15–17 | 338,860 | 0.99 | 0.14 | 0 | 355,924 | 88,354 | 427,214 | |
| Chile 2002 | ||||||||
| 15–17 | 1,960 | 0.99 | 0.39 | 310 | 25,590 | 6,370 | 8,330 | |
| Colombia 1993 | ||||||||
| 10–14 | 26,650 | 1.00 | 0.80 | 170 | 45,890 | 11,620 | 38,270 | |
| 15–17 | 67,090 | 1.00 | 0.80 | 500 | 39,280 | 9,830 | 76,920 | |
| Costa Rica 2000 | ||||||||
| 15–17 | 3,280 | 0.93 | 0.20 | 0 | 6,900 | 1,710 | 4,990 | |
| Mexico 2000 | ||||||||
| 15–17 | 143,796 | 0.95 | 0.31 | 0 | 19,831 | 5,059 | 148,855 | |
Notes: Conservative estimates in (2), (3), and (4) do not include unemployed child domestics.
Column (3) identifies domestics via their employment in labor force work.
Best guess estimates in (5) include unemployed children whose last job was as a domestic servant.
Column (6) uses the relationship variable to identify all young household members who might be domestics, including "other relatives" and "non-relatives" who are not boarders/lodgers.
Imputations in column (7) include only 25% of potential "Cinderellas" identified in column (6).
Footnotes
Oloko (1997: 5–6) indirectly estimates the number of young domestics in Nigeria using the number of women who were government employees; she assumes that every such woman employed one young domestic. She considers her estimate conservative, since some women working in the informal sector would also employ young domestics.
As the story goes: “Now began a bad time for the poor step-child….They took her pretty clothes away from her, put an old grey bedgown on her, and gave her wooden shoes….and led her into the kitchen. There she had to do hard work from morning til night, get up before daybreak, carry water, light fires, cook and wash. Besides this, the [step]sisters did her every imaginable injury…. “ (The Grimm Brothers, Cinderella)
Every year, more census samples are added to the IPUMS-International online data collection.
A more extended discussion of these large-scale changes and their potential effects on child domestic servants can be found in Levison and Langer (2009).
In most cases, we exclude only those youth who are living in group quarters apart from family, such as those living in institutions or at boarding schools. Some children and youth living in group quarters were living with their families. For example, an entire family might live in a military barracks. These children were included in our sample. In some cases we could not discern whether or not children in group quarters were living with their families; in this case we excluded all group quarters children.
Some individuals indentified in Brazil as domestic workers by household variables are categorized as being employed in agricultural industries. In these cases, the individuals were counted as domestic servants, but the rest of the individuals in that agricultural industry were not.
When we compute employment trends, we count such individuals as employed members of the labor force. That is, they are included in the denominator.
Excel spreadsheets of the tables in this paper are available upon request to dlevison@umn.edu.
This analysis follows the conventions for Table 1 and our “best guess” estimates: unemployed domestics are counted among domestic workers.
Figures showing the results of this analysis can be found in Levison and Langer (2009).
For example, Bourdillon (2007:60) interviewed child domestics in Zimbabwe, some of whom had been beaten. Oyaide (2000: 54) documents substantial verbal abuse and humiliation among child servants in Lusaka, Zambia. Kielland and Tovo (2006: 98) cite Onyango (1991), who interviewed prostitutes in Nairobi and Cotonou and found that the majority had been sent into domestic service at an early age. Among these, most had been sexually abused, many by a member of the employer’s household.
Cited in Bourdillon (2009: 6).
Bourdillon writes, “child protection is meaningless if removing children from a harmful situation results in driving them into something worse; and a ban on its own does not guarantee that the [former child domestics] will be better placed. Besides, a ban will remove from many disadvantaged children opportunities to travel, learn, prepare for the future, and escape abuse, exploitation, and poverty at home” (p.11).
Contributor Information
Deborah Levison, Hubert H. Humphrey Institute of Public Affairs, University of Minnesota, 301 – 19th Avenue South, Minneapolis MN 55455 USA, Tel. 612-624-3540, dlevison@umn.edu.
Anna Langer, 4831 36th Street N.W., #510, Washington, D.C. 20008 USA, Tel. 612-964-3961, aklanger@gmail.com.
Bibliography
- Ainsworth Martha. Economic Aspects of Child Fostering in Côte d’Ivoire. Washington D.C.: The World Bank; 1992. [PubMed] [Google Scholar]
- Black Maggie. Child Domestic Workers: A Handbook on Good Practice in Programme Interventions. London: Anti-Slavery International; 2005. [Google Scholar]
- Black Maggie. Child Domestic Workers: Finding a Voice. A Handbook on Advocacy. London: Anti-Slavery International; 2002. [Google Scholar]
- Bourdillon Michael. “Child Domestic Workers in Zimbabwe: Children’s Perspectives,” Chapter 5. In: Hunderford Beatrice, Manfred Liebel, Brian Milne, Ann Wihstutz., editors. Working to Be Someone: Child Focused Research and Practice with Working Children. Philadelphia: Jessica Kingsley; 2007. pp. 55–65. [Google Scholar]
- Bourdillon Michael. Children as Domestic Employees: Problems and Promises. Journal of Children and Poverty. 2009 Mar;15(1):1–18. [Google Scholar]
- Jacquemin Mélanie Y. Children’s Domestic Work in Abidjan, Côte d’Ivoire: the ‘petit bonnes’ have the floor. Childhood. 2004;11(3):383–397. [Google Scholar]
- Kielland Anne, Maurizia Tovo. Children at Work: Child Labor Practices in Africa. Boulder CO: Lynne Rienner; 2006. [Google Scholar]
- Levison Deborah, Jasper Hoek, David Lam, Suzanne Duryea. Intermittent Child Employment and Its Implications for Child Labour Estimates. International Labour Review. 2007;146(3–4):217–251. doi: 10.1111/j.1564-913X.2007.00014.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Levison Deborah, Anna Langer. Child Domestic Servants in Latin America: Numbers, Trends, and Education. Minnesota Population Center Working Paper; University of Minnesota, Minneapolis. 2009. http://www.pop.umn.edu/research/mpc-working-papers-series. [Google Scholar]
- Madsian Carla. Not for Bread Alone: Peanut Vendors in Brazil. In: Lieten GK, editor. Working Children Around the World: Child Rights and Child Reality. New Delhi: Institute for Human Development; 2004. pp. 128–139. [Google Scholar]
- Minnesota Population Center. Integrated Public Use Microdata Series — International: Version 4.0. Minneapolis: University of Minnesota; 2008. [Google Scholar]
- Okolo Sarah BA. Child Labour: What We Know: The Nigeria Experience, 1986–1996. Presented at a UNICEF Workshop; Turin, Italy. June–July.1997. [Google Scholar]
- Onyango P, et al. A Report on the Nairobi Case Study on Children in Especially Difficult Circumstances. Nairobi: African Network for the Protection Against Child Abuse and Neglect; 1991. [Google Scholar]
- Oyaide Omolara Dakore. Child Domestic Labour in Lusaka: A Gender Perspective. University of Zambia, manuscript; 2000. [Google Scholar]
- Psacharopoulos George, Zafiris Tzannatos. Women’s Employment and Pay in Latin America: Overview and Methodology. Washington, D.C.: The World Bank; 1992. [Google Scholar]
- Reynolds Pamela. Dance Civet Cat: Child Labour in the Zambezi Valley. Athens OH: Ohio University Press; 1991. [Google Scholar]
- Smucker Glenn R, Murray Gerald F. The Uses of Children: A Study of Trafficking in Haitian Children. Port-au-Prince, Haiti: Report to USAID/Haiti Mission; 2004. [Google Scholar]
- UNICEF. Child Domestic Work: Innocenti Digest No. 5. Florence: UNICEF International Child Development Centre; 1999. [Google Scholar]