Child Labor Data Methodology and Sources
A. Characteristics of Child Labor Reporting
Reliable statistics on child labor are rare and, when available, often incomplete. Because child labor is illegal below a certain age in almost every country, national government surveys often do not collect information on working children below 15 years old.
Moreover, when data are collected on working children, large differences in methodology and definition make it difficult to make comparisons across countries. The main difficulty is that each nation chooses its own definition of what constitutes a "child" as well as what it classifies as "labor." As a result, the number of working children reported by one country may be higher or lower than that reported by another simply because of which groups of children and which activities are included in the statistic.
While most countries that collect data on working children classify child laborers as workers under 15 years old, some consider those up to 17 or 19 years to be child workers. There are also different minimum age limits associated with labor data collected by governments with a majority of countries placing a lower age limit of five years.
There is even less agreement among countries on what types of activities by children constitute child labor. Some countries collect information on children working in either paid or unpaid work that takes place either inside or outside the home, while others count only full-time paid labor. Likewise, some countries do not classify students as child laborers no matter how many hours they work outside the home. Others, however, count students working at least one hour a week outside of class as employed.
Depending on the definitions used and the activities included, therefore, certain statistics may grossly over- or underrepresent the true number of child laborers in a country. The five to 14 year-old age group is used whenever possible in measuring child labor incidence. When data for children ages five to nine are not available, the 10 to 14 age group is reported. It should be noted, however, that since children in the upper age bracket have a much higher probability of working, countries for which only data on 10 to 14 year olds are reported will tend to have higher percentages of children in the work force than those with estimates extending below 10 years old.
In analyzing the available statistics, the following concepts must be considered:
1. Working Children
Census and labor force surveys usually gather information on working children by asking respondents if they worked at least one hour of any day during the reference period specified. This question forms the basis for the number of working children. However, depending on the country, this statistic may or may not include unpaid workers or household laborers working on family farms or enterprises. In many countries, only those children who worked more than a certain number of hours a week are included.
Because this grouping may exclude child laborers performing unpaid or household duties or children who are in the work force but did not work in the reference period specified, the estimate of working children may under represent the child labor situation in a country.
2. Economically Active Children
According to the internationally-endorsed definition,1 the term "economically active" used by the ILO, other international and regional organizations, and most countries, comprises all citizens who supply productive labor in a country. The generally accepted definition includes all those found to be working as discussed in the preceding section, paid and unpaid family laborers working in economically productive areas, and those in the work force who were unemployed but looking for jobs during the reference period specified.
Economically active children, therefore, comprise all children who are found to be:
Additionally, most countries include children in the labor force who are attending school full time but working part time either after classes or on weekends.
Because this definition includes both working children and those who are in the labor force but unemployed at the time of the survey, it provides the best estimate of the number of child laborers in a country. For this reason, the statistics cited in this report are taken from the number of economically active children whenever possible.
3. Reference Period
Census and employment surveys collect information by asking respondents whether they have worked during a certain reference period. There are generally two different reference periods used when collecting labor information: one week (current activity) and one year (usual activity).
The current activity period counts only those who were in the labor force during the seven-day period preceding the survey, while the usual activity period includes those who worked at any time during the previous year. In general, usual activity estimates are higher than those for current activity. Because many child laborers work intermittently or in jobs that are highly seasonal (agriculture), the usual activity reference period is more likely to capture the true number of child workers in a country. For this reason, usual activity reference period estimates are used in this report whenever possible.
4. Other Issues
In addition to the collecting and reporting considerations mentioned above, other factors can influence the comparability and validity of child labor statistics in various countries. Factors such as homelessness, lack of birth registration, informal sector employment, or a large number of refugees or immigrants can increase the probability of significant under reporting. Considering the prevalence of many of these factors in the developing countries studied in this report, the statistics presented here are likely to understate the true extent of child labor.
There are also issues specific to child labor reporting that can impair the accuracy of child labor estimates. Parents, children, and employers are all likely to be reluctant to report child laborers due to shame or fear of prosecution. In addition, initial experimental surveys by the ILO found that children under the age of 10 were often too young or too shy to effectively answer questions about their employment status, and the terms and parameters used to classify labor force participation were often too confusing for them.2 Young children interviewed with parents or employers present may also give inaccurate answers for fear of reprimand or losing their job.
B. Sources of Child Labor Data
Child labor statistics in this report were obtained from four main sources.
Since 1992, the ILO's International Program for the Elimination of Child Labor (IPEC) and Bureau of Statistics (STAT) have worked in close collaboration with national statistical institutions in several countries in designing and implementing specialized surveys on child labor. These surveys, administered either independently or as part of a country's existing labor force surveys, have followed standard design and methodological approaches devised by IPEC/STAT after extensive experimentation using different statistical approaches for quantifying child labor.
At the time of this report's publication, 15 countries had completed or were in the process of completing child labor surveys based on the recently developed methodologies for collecting comprehensive information on characteristics of child labor such as reasons for working, working hours, employment of children in specific industries or sectors, and number of children laboring under hazardous conditions.3 While using the same newly-developed methodology by the ILO, each of these surveys was carried out by national statistical agencies with slight modifications for country differences (see individual country sections for specific survey results in some of these countries).
New child labor surveys are either planned or already begun in 18 additional countries. These countries are: Cambodia (second round), Colombia, Ethiopia, Georgia, Indonesia (second round), Madagascar, Mozambique, Namibia, Nicaragua, Nigeria, Pakistan (second round), South Africa, Trinidad and Tobago, Turkey (second round), Ukraine, Venezuela, Vietnam, and Zimbabwe. Several other countries are also preparing to launch surveys in the near future.4 The ILO hopes to use the survey results to compile a comprehensive database on child labor within a couple of years.
Until recently, few countries collected statistical information on working children.5 Most government data on child labor are gathered through decennial population censuses or periodic labor force surveys (LFS) conducted by national statistical agencies in individual countries. While some countries include work force participation by children under 10 years old in their survey programs, most report national labor force rates for five-year age groups beginning at 10 or 15 years old.
Census and labor force surveys, however, have been found to be less than effective tools for measuring and understanding child labor. Since a large proportion of child labor occurs in the informal sector or within household-based enterprises performed mainly as unpaid work, child labor incidence can remain largely undocumented. Lastly, neither census nor labor force surveys collect data on the working and living conditions of children, especially health and safety aspects, which are important factors for determining the consequences of child labor.6
The ILO's decennial publication on labor statistics, the Economically Active Population (EAP), includes estimates of the number of economically active children ages 10 to 14 years for every country and territory with a population of 200,000 or more.7 These data are compiled from national censuses and labor force surveys, with data or methodological gaps filled by statistical projections and estimates by the ILO Bureau of Statistics.
The only available source for comprehensive and comparable data on working children 10 to 14 years old, the EAP reports labor force estimates from 1950 to 1990 at 10-year intervals in five-year age groups. Projections are also included for the years 1995, 2000, and 2010.
While useful as a general indicator of labor force participation rates in different countries, the EAP is less accurate in estimating the number of children working below the age of 15. In addition to the absence of data for the five to nine-year-old age group, several estimates in the 10 to 14-year-old age group are well below the estimates from other data sources.8
One of the main weaknesses in the EAP is the use of statistical modeling to fill in gaps where data are not available. To provide estimates in every age category for every year, the ILO uses estimates from neighboring countries in calculating labor participation rates. In addition, to make the data comparable, models are sometimes used to adjust figures that have been collected using different methodology or age groupings. As a result, the EAP contains estimates produced solely by statistical techniques and not by national surveys or sampling methods.9
While most NGOs cite ILO data (child labor surveys or the EAP) or official government surveys of the number of working children, some have produced their own estimates of child laborers in individual countries. These estimates vary greatly in coverage and methodology and must be assessed for accuracy and comparability on a case-by-case basis.
In cases where official child labor estimates are considered to be grossunder estimates, nongovernmental statistics may provide a more accurate picture of the child labor situation and have been noted in the country sections below for this reason.
C. Data Sources for Table II-1
This section provides the specific sources of child labor data reported in Table II-1. It also provides brief descriptions of the surveys and/or methodologies used to collect the data cited in Table II-1 and presents unofficial child labor estimates that are not reported in the table.
In 1995-96, the Government of Bangladesh conducted an ILO-sponsored child labor survey. The National Sample Survey of Child Labour was a household survey that measured child labor based on responses to several interview questions. Based on the interviews, all children ages five to 14 within a household were classified into one of three categories: "working," "not working," or "inactive."
Working children were defined as those who had worked for one or more hours during the reference period (either one week or one year). Work was defined as employment for pay or profit or non-remunerated economic activity on a family farm or enterprise. Children who were full-time students and engaged in household economic activity during leisure hours were not counted as working.
Not-working children were defined as those who were involuntarily out of employment during the reference period but who had been actively seeking a job or prevented from seeking one because of illness or lack of opportunity. Inactive children included full-time students and child household workers (not including children who worked unpaid for a family farm or other family-owned/operated business).
Children in the first two categories made up the total child labor force in Bangladesh. The official government estimate for the number of child laborers, however, counted only children working, not those who were looking for work or involuntarily unemployed.
Two reference periods were used to classify child laborers based on the "usual activity" and "current activity" approaches. The current activity period asked whether children had worked at least one hour in the previous seven days. It led to an estimate of 6,455,000 for the total number of children in the labor force, including 6,304,000 working children and 151,000 not-working children.
The usual activity reference period, which was used to capture seasonal or sporadic employment, asked whether or not a child had worked at least one hour in the past 12 months. This approach yielded a total child labor force of 6,584,000, which included 6,298,000 working children and 286,000 not-working children.
One major limitation of this survey is that children attending school were considered ineligible for inclusion in the work force no matter how many hours they worked during the week. As a result, the number of economically active children estimated is likely to be underreported.
The Government of Brazil has not conducted a specialized child labor survey but has collected information on working children through its annual national household sampling surveys (PNAD). Conducted by the Brazilian Institute of Geography and Statistics (IBGE) every year except 1994, the surveys collect information on population and work force participation rates for children ages five to 14 years old. The data tables, available through the IBGE website at http://www.ibge.org, break down labor participation rates by gender, location (urban/rural), and number of hours worked.
The annual PNAD surveys use both "current" and "usual" reference periods and collect data on the number of working children from five to nine and 10 to 14 years old. The estimates in this report come from the 1995 PNAD and reflect the one-year (usual activity) reference period for the number of working children five to 14 years old (the IBGE does not report statistics on the economically active population below 10 years of age). The most recent available PNAD (1996) did not collect information on working children from five to nine years old.
The Government of Egypt is currently considering a special child labor survey in conjunction with the ILO. The best current sources for child labor data are the official census and labor force sample surveys (LFSS). Most of these surveys collect data on children as young as six years old, but show considerable confusion about the true number of child workers. Data until 1986 show working children (six to 14 years old) accounting for approximately 10 percent of the labor force (see Table B-1). After this date, however, the surveys begin to show a dramatic decrease in the number of child workers--from 1.3 million in 1988 to a low of 361,300 in 1995--and a corresponding drop in the percentage of the work force which they represent.
Table B-1 Egypt: Census and Labor Force Sample Surveys
The precipitous drop in the number of child laborers and their percentage of the work force has been attributed to changes in data collection methodology and accuracy. Until 1988, the LFSS included a specific module for collecting in-depth information on child labor. Its discontinuance since that year has affected the accuracy of child labor reporting and calls into question the reliability of all child labor force data collected in the past decade.10 The data information reported in Table II-1 come from the 1988 LFSS, which was the last national survey to utilize the special child labor module.
Other estimates of working children range from 2 million (age range unknown), made by the Egyptian Minister of Manpower at a 1995 Cairo seminar on child labor, to 3 million (including 2.25 million working boys between six and 12 years old).11
The 1994 national census in Guatemala was the first to collect data on working children as young as seven years old. It found 151,494 economically active children between seven and 14 years old out of a child population of approximately 3.7 million. The reference period was the week immediately before the survey. Persons doing unpaid work in family enterprises and students with part-time jobs were all included in the economically active population.
Other estimates of child labor are much higher but difficult to document. A 1989 study reported 901,800 working children from 10 to 17 years old.12 In a 1995 interview, Carlos Mancilla García, Secretary of Social Welfare for the Confederation of United Unions of Guatemala, estimated the number of child workers (six-17) at 1.5 to two million.13
India is one of the original four countries (the others being Ghana, Indonesia, and Senegal) in which the ILO sponsored experimental surveys in 1992-93 to develop and pretest the methodology and survey instruments for child data collection. The 1993 ILO-sponsored survey, conducted in only two districts of Gujarat state, found approximately 23 percent of the child population ages five to 14 working either full or part time. Because it was only a regional survey, however, no estimate was made for the total number of child workers in the country and there are currently no plans to conduct a national child labor survey at this time.
Official government estimates of the percentage of children working in India are much lower than those in the ILO-sponsored regional study. The most recent national census (1991) recorded approximately 11.3 million working children out of a five to 14 year-old population of 210 million, a child work-participation rate of only 5.4 percent (estimates reported in Table II-1). The census also showed, however, that only half of all children ages five to 14 were attending school. Of the 105 million children who were not going to school, the census recorded less than 11 million of them as working. A 1996 ILO report classified 74 million of those children who were neither at school nor economically active as "nowhere" children. It calculated the combined total of full-time child workers, marginal child workers, and "nowhere" children as over 97 million, or almost 40 percent of the five to 14 age group.14
Several NGOs have produced or cited higher estimates of child labor, which are based on unknown methodology. For example, the Operations Research Group, which completed an all-India child labor sample survey in 1980-81, found 44 million working children (age range unknown).15 An Indian nonprofit called Centre for Concern for Working Children cites a 1994 estimate of 100 million working children (age range unknown) apparently based on the number of children out of school.16 Lastly, the Balai Data Bank of Manila is also credited with an estimate of 100 million child laborers (age range unknown).17
The actual number of child workers is likely somewhere between the official estimate and the highest unofficial figures, with many NGOs and international organizations using 44 million to 55 million as a working figure.18
Kenya is currently conducting an ILO-sponsored survey on child labor which should be completed by early 1999. At the present time there are no other reliable data for the number of children working in Kenya besides the ILO's Economically Active Population database. The 1995 estimates published by the ILO are based on the 1989 national population census, which used only a "one week" reference period in classifying economically active citizens ages 10 years old and above.
The census used a concept of "economically active" which included all those who were working for pay or doing unpaid work on a family farm or business during the week immediately preceding the census.19 Working students with a part-time job and students seeking work were excluded from this category.
Other estimates are based on the number of children who are not attending primary school. The Kenyan Central Organization of Trade Unions estimates 3.5 million working children (ages six to 15) using this method.20
Currently, no comprehensive statistics are available on working children in Mexico. The best source for child labor statistics is the government's National Employment Survey (ENE). This annual household survey, however, collects employment data only on persons 12 years of age and older. For this reason, official child labor estimates in Mexico are for the 12 to 14-year age group only.
The most recent ENE survey, conducted from April to June 1996, found 1.1 million working children between 12 and 14 years old in Mexico. This represents 17.3 percent of the child population at that age and roughly 3.1 percent of the total national work force during that year.
Other estimates of child labor in Mexico are higher. For example, the United Nations' Economic Commission for Latin America and the Caribbean (ECLAC) reported 29 percent of school-age children in Mexico working (approximately 2 million child laborers in the 12 to 14-year age group alone).21 Another estimate, based on the number of children who do not finish school, puts the number of working children at 2.8 million.22
Nepal completed an ILO-sponsored survey on child labor as part of its 1995-96 national Migration and Employment Survey. The child labor component of the study, which used the 13,000 households in the sample with at least one child ages five to 14, found close to 2.6 million working children in Nepal, or 41.7 percent of the child population.
The survey used answers to a series of questions and a one-year reference period to group children into three categories: only attending school, working, and idle. Only 36.7 percent of the children were attending school full-time, 41.7 percent were working (either full-or part-time while attending school), and 15 percent were idle. Working children were further broken down into those who were attending school while working (1.6 million) and those working full time (one million).
The official report of the survey has not yet been published, but the data tables have been made available. Nepal is planning a second ILO-sponsored child labor survey in 1999.
Nicaragua is currently working with the ILO on a specialized child labor survey. Until it is completed, the best data on working children is the 1995 national census which found 60,247 children between 10 and 14 years old who were economically active--almost 10 percent of the total 10 to 14 year old child population.
While this is the most recent official source of statistics on child labor in Nicaragua, other estimates show a slightly higher number of children working. The United Nation's Economic Commission for Latin America and the Caribbean (ECLAC) reports 71,922 children working 10 to 14 years of age.23 Other estimates have put the number of children working in Nicaragua at over 100,000 (age range unknown).24
Although the government of Pakistan completed an ILO-sponsored child labor survey in 1996, estimates of the number of child laborers are still being debated. The ILO-sponsored child labor survey found just over 3 million working children five to 14 years old--about 8 percent of the child population.
This estimate was obtained from a national household survey using both a one-week and a one-year reference period. But critics have questioned the reliability of this estimate, and the Pakistani Federal Bureau of Statistics (FBS), which administered the survey, has acknowledged some imprecision in its findings. The FBS noted that the child participation rate of the 10 to 14 age group had been found to be much higher in various annual labor force surveys (LFS). The 1993-94 LFS, for example, found over 14.5 percent of children ages 10 to 14 working in the rural provinces of Sindh and Balochistan, while the child labor study estimated only eight percent and 1.2 percent respectively. The FBS acknowledges that up to 300,000 children may be underreported in the official estimate in these two provinces.25
Another problem encountered in the survey was the difficulty of collecting information on working females. It was not possible to interview many girls ages 10 to 14 because they were not attending school, and male surveyors were not able to question them at home due to cultural constraints. Also, the survey did not include in its measure of working children those who were working in private homes as domestic servants and girls assisting their mothers at home in domestic labor. Because of these constraints, the economic participation rate for females is likely to be highly underreported.26
The accuracy of the survey can also be called into question because of a discrepancy between the estimates from the two different reference periods used. Despite the opposite expected outcome, the estimate for children working in the one-week reference period (3.3 million) was greater than that for the one-year period (3.2 million). This discrepancy is apparently due to the fact that only those who listed work as their principal activity during the 12-month period were counted as working children. Thus, some children who attended school and also worked were included as child workers in the one-week period but not in the one-year estimate. All of the above-mentioned discrepancies and constraints are expected to be resolved in a second child labor survey planned for 1999.
Nongovernmental estimates of working children in Pakistan are much higher than the 1996 survey. A 1990 UNICEF and Government of Pakistan publication estimated the number of child workers under 15 to be "not below 8 million." This estimate is purportedly based on an "analysis of all data available, including school participation rates."27 Another estimate using data on the number of children not enrolled in school in 1989-90, puts the number of child laborers five to 14 years old in Pakistan at 19 million, including 12 million working children ages 10 to 14 and 7 million five to nine years old.28
Because the Government of Peru has not requested assistance from the ILO in completing a child labor survey, and because it does not report labor survey statistics for children below 15 years of age, the only source of official data on child labor are national census data collected once a decade.
In the latest national censuses, completed in 1961, 1972, 1981, and 1993, the economically active population estimates included children six to 14 years old who were working for pay, temporarily absent from work, or working as a helper to a family member for no pay. Students with part-time jobs were also included in the estimates.
Table B-2 shows the child population and economically active estimates for children six to 14 years old from the last four population censuses.
Table B-2: Perú: Economically Active Children (6-14 years old)
Data from the 1993 census show approximately 196,000 economically active children six to 14 years old out of a total child population of 4.8 million, an increase of 72,000 children from 1981. The percentage of children who are economically active also increased during the same time period, from 3.0 percent in 1981 to 4.1 percent in 1993.
The most recent available household survey, conducted by the National Institute of Statistics and Information (INEI) in the first trimester of 1996, reported 1.9 million economically active children between six and 17 years old. Since previous surveys have shown 15 to 30 percent of economically active six to 17-year olds to be in the six to 11 age group,29 this estimate suggests 300,000 to 600,000 working children below 12 years old.
Other estimates include a 1995-96 survey on urban employment, which found 4.3 million urban children from six to 17 years old working, approximately 600,000 of them in the 6 to 11 age group.30
In 1995, the Philippine National Statistics Office, in collaboration with the Philippine Department of Labor and Employment and ILO/IPEC, conducted a survey of children five to 17 years old. This study, which was part of the annual July labor force survey, was the first national survey of working children in the Philippines.
Although results from the survey and all related data tables were released in 1996, a comprehensive official report has not yet been published. The survey found approximately 3.7 million working children in the country, including about 1.9 million children from five to 14 years old (11 percent of the child population in this age group). Twice as many males in the five to 14-year age group (1.2 million) were working than females (0.6 million), and child labor participation rates were over three times higher in rural areas (24 percent) than in urban areas (7 percent).
A second child labor component will be included in an upcoming labor force survey.
South Africa recently signed an agreement to conduct an ILO/IPEC-sponsored child labor survey. Preliminary statistics from this study should be available by early 1999.
The best statistics currently available on child labor participation rates are from the 1994 October Household Survey (OHS) conducted by the South African Central Statistics Service, which only collected data on children 10 to 14 years old.
The Government of Tanzania is currently in discussions with the ILO about conducting a specialized child labor survey. At the time of this report, however, the best available statistics on child labor in Tanzania are from the ILO's Economically Active Population. The 1995 estimates published by the ILO are based on 1988 national census figures which used a 12 month reference period in classifying the economically active population.
The 1988 census counted as economically active all citizens 10 years and older who were employed or looking for work during the year preceding the census day, including all those doing unpaid work in family firms or businesses, working students with a part-time job, and seasonal workers. The census classified 395,372 working children 10 to 14 years old and another 10,948 in the same age group looking for work, out of a total 10 to 14 population of 2.9 million.31
Taking the 1988 census figures, the ILO used population projections and other statistical modeling techniques to produce updated estimates conforming to the internationally-accepted definition of economically active. Their 1995 estimates include 1.5 million working children 10 to 14 years old with an activity rate of 39.5 percent.
Because the Government of Thailand only collects information on individuals 13 years of age and older in its national census and labor force surveys,32 reliable statistics on working children in Thailand are lacking. A regional child labor study sponsored by the ILO was recently undertaken in two provinces (Karnchanaburi and Ubonratchathani) with the purpose of testing methodology and definitions for a national survey of child labor to be undertaken in the near future.33
In the meantime, the most recent available data on working children are from the 1996 labor force survey, which found 196,000 children ages 13 and 14 in the labor force (8.6 percent of children in this age group).34 The ILO's Economically Active Population database estimates that about 905,000 children from 10 to 14 years of age, or 16 percent of children in this age group, are working.
Under the sponsorship and technical assistance of ILO/IPEC, the Turkish State Institute of Statistics introduced a Child Labour Survey (CLS) as part of its October 1994 Household Labour Force Survey (HLFS). While previous HLFSs only collected data from age 12 (the legal working age) and older, the 1994 CLS/HLFS was expanded to include children as young as six years old. This survey collected information on a total of 13,537 rural and urban households from every region of the country and interviewed a total of 10,327 children between the ages of six and 14. A second round is expected to be included in the 1998 HLFS.
The reference period used in the survey was the last week of October 1994 (October 24-30). Labor participation was measured by asking interviewees whether or not they had worked during that seven-day period. Working children included all those active for at least one hour during the reference period as a regular employee, casual employee, employer, self-employed persons, or unpaid family worker. Children who were not engaged in economic activity but helped the family in household work such as shopping, cooking, and cleaning were also included.
The estimate of 1.5 million reported in Table II-1 includes all working children, as defined above, except those employed in domestic chores while attending school.
This report was produced by the staff of the International Child Labor Program and is published by the U.S. Department of Labor, Bureau of International Labor Affairs.