Monitoring International Labor Standards: Quality of Information: Summary of a Workshop (2003)

Data Analyst Amy Ritualo (2002), of the ILO International Programme on the Elimination of Child Labour, said that, unlike previous speakers, she would not claim that child labor was the most difficult to measure of the four labor standards. On the contrary, researchers have made “great strides” in collecting data on child labor. Nevertheless, she said, child labor researchers face some of the same challenges mentioned by experts on the other labor standards. These include the challenge of translating ILO standards into quantifiable measures and the difficulty of investigating the worst forms of child labor, which are illegal and often hidden.

Ritualo explained that ILO and national definitions of “child labor” differ from the definition of children’s “economic activity,” which has been used in most research. The ILO has defined child labor in two conventions: ILO Convention 138 (C. 138) obligates countries to fix a minimum age for employment that should not be less than the age for completing compulsory schooling and, in any event, should not be less than 15 years. Developing countries may set the minimum age at 14. She said that C. 138 provides flexibility for countries to establish a younger minimum age of 12 or 13 for children to partake in “light work.” The convention defines

“light work” as that which is (a) not likely to be harmful to their health or development; and (b) not such as to “prejudice their attendance at school, their participation in vocational orientation or training . . . or their capacity to benefit from the instruction received” (ILO, 1973a). Convention 138 requires countries’ minimum age laws to ensure that no person under 18 is allowed to be employed in “hazardous work.” This term is to be defined by each country, based on the guideline that “hazardous work” by its very nature or the circumstances in which it is carried out is likely to jeopardize the health, safety, or morals of young persons (ILO, 1973b).

In 1999, Ritualo said, the ILO unanimously adopted a separate convention designed to eradicate the worst forms of child labor. Convention 182 (C. 182) prohibits the following:

1. All forms of slavery or practices similar to slavery, such as the sale and trafficking of children, debt bondage and serfdom and forced or compulsory labour, including forced or compulsory recruitment of children for use in armed conflict.

2. The use, procuring or offering of a child for prostitution, for the production of pornography or for pornographic performances.

3. The use, procuring or offering of a child for illicit activities, in particular for the production and trafficking of drugs as defined in the relevant international treaties.

4. Work which, by its nature or the circumstances in which it is carried out, is likely to harm the health, safety or morals of children (ILO, 1999).

Ritualo noted that the convention does not clearly define this last type of prohibited child labor, allowing countries to establish their own definitions.

Next, Ritualo discussed “economic activity” among children, noting that this is often used as a proxy for “child labor.” Most research and reports on child labor have been based on economic activity “because it is easier . . . than translating the ILO standards into a statistical definition.” According to the ILO definition of the “economically active population” in a given country, it:

This broad definition, she said, includes children who do paid and unpaid work (including those who are temporarily unemployed), self-employed children, apprentices, and paid family workers.

According to Ritualo, the “economic activity” definition has both strengths and limitations. One strength lies in the fact that this type of data has been gathered in many labor force surveys, providing experience in design of questionnaires. Another strength is that there is a large amount of data on children’s work in many countries, based on this common definition. This means that “it is easier to compare countries” with this definition than with a “child labor” definition. However, the economic activity definition is limited in its ability to measure unpaid family work, part-time work, work such as carrying wood or water, illegal activities, begging, and household chores.

Ritualo said that, for several reasons, data on children’s work are usually gathered through household surveys. The household is the best place to identify and collect information on household members, and children work for reasons related to their family circumstances. In addition, because questions on child labor can be added to ongoing national household surveys, this method can be the most cost-efficient and sustainable way to collect data. However, household survey methods often miss street children, who do not live in households, and may have difficulty in capturing the worst forms of child labor. In addition, because children’s work is often clustered in certain locations and/or occupations (Anker, 2000), surveys of representative national samples of households may inadvertently omit some types of child labor.

Ritualo explained that she is employed in ILO’s SIMPOC. SIMPOC was created in 1998 to support the ILO International Programme on the Elimination of Child Labour by providing reliable, gender-sensitive information that can be used for research, advocacy, education, and program design. Currently, SIMPOC provides more than 40 countries with technical assistance in surveying child labor. Surveys are conducted when a country requests assistance, needs data on children and families, and has expressed a commitment to collect and disseminate such data. SIMPOC staff work closely with national statistical offices, government labor ministries, academic institutions, NGOs , and other international organizations.

Because the SIMPOC program is designed to encourage countries to

routinely collect data on child labor, Ritualo said, countries are allowed flexibility in designing questionnaires, survey samples, and sampling methods. Based on experience and pilot studies (ILO, 2002a), more recent SIMPOC surveys include more detailed, probing questions, which have been found to increase the estimated percentage of economically active children. These include questions on the children’s occupations, hours and schedule of work (including household chores), injuries and illnesses, schooling, and parents’ views of child labor. The surveys also ask basic demographic, social, and economic questions. Although survey methods vary somewhat among countries, SIMPOC requires participating countries to update their master sampling frame and to oversample households in urban areas, lower income areas, and areas where child labor is concentrated. Large samples of more than 10,000 households are usually surveyed.

The World Bank’s LSMS household surveys provide another important source of data on child labor, Ritualo said. The World Bank conducts these surveys in partnership with national agencies, academic experts, and other groups in order to assess how government policies and programs affect welfare and living conditions. Currently, data are available from more than 40 LSMS surveys conducted in various countries. Along with detailed questions on family income, expenditures, and welfare, LSMS surveys ask some questions about children’s work. However, Ritualo cautioned that the age of children about whom questions are asked varies among surveys, from age 7 in some countries to age 12 or 14 in others. Another problem is that the sample sizes of between 1,500 and 5,000 households “can be quite low when you are trying to measure detailed information on the work activities of children.”

UNICEF launched its MICS in response to the 1990 World Summit for Children, creating another useful source of data on children’s employment. In order to measure progress towards the goals of this summit, the MICS were designed to identify “the proportion of children in households aged 5 to 14 who are currently working” (UNICEF, 2000). To date, more than 100 of these surveys have been conducted in partnership with national agencies, international organizations, and research institutions. Although implemented differently in different countries, all MICS are based on a battery of survey modules. The child labor survey module includes seven questions related to work; the questions address paid and unpaid employment, hours of work, and household chores. However, the module does not include questions about children’s occupations or the industries in which they are employed.

Ritualo briefly mentioned two other sources of data on child labor— national labor force surveys and censuses and the ILO Database on the Economically Active Population. She said that national labor force surveys are designed primarily to measure adult activities, and many do not collect any data on children under the age of 15. Other problems include interviewers’ and respondents’ attitudes towards children’s work and the difficulty of measuring unpaid, part-time, and seasonal work. The ILO database presents economic activity for 10-year periods, including information on children age 10 to 14. However, SIMPOC surveys have shown that these estimates of children’s economic activity are too low.

In order to capture the full extent of children’s work, the ILO and UNICEF have developed a “rapid assessment” methodology. Research institutions, universities, and NGOs conduct these short-term studies to gather information on the worst forms of child labor. After collecting background information, researchers use a participatory approach to observe and interview key informants, working children, other children, and adults.

Ritualo suggested using the following key indicators to measure the extent of child labor:

1. number and proportion of economically active children;

2. number and proportion of children who are child laborers;

3. ratio of children not in school;

4. proportion of children involved in household chores in their own homes, organized by ranges of hours worked;

5. proportion of children working in industry sectors (agriculture, mining, construction, transportation, etc.);

6. proportion of children working by range of hours worked; and

7. other indicators that help put child labor within the national context.

She cautioned that, although data on child labor, as defined by the ILO, would be most useful to measure compliance with ILO child labor standards, these data are hard to find. She observed that the more plentiful data on economically active children were more useful for three reasons. First, national efforts to reduce child labor would affect the ratio of economically active children. Second, information using this definition is

available for more countries, making cross-country comparisons “more feasible.” Finally, ILO global estimates using the two different definitions indicate that about 88 percent of economically active children were also considered to be involved in child labor.

Although time series data would be most useful to measure a country’s progress toward the elimination of child labor, Ritualo said such data are lacking. For example, within the SIMPOC program there are only two countries for which data have been gathered for more than one year. She cautioned that “you must know something about the country” to correctly analyze and interpret data on child labor. Finally, given the lack of data to measure child labor over time, she suggested that key indicators might include a country’s actions, such as ratifying ILO conventions on child labor and developing laws aimed at reducing child labor. She referred to Thomas’s presentation, which described how the ILO obtains, analyzes, and disseminates extensive information on various countries’ actions (or inactions) related to compliance with international labor standards.

World Bank economist Harry Patrinos said that he had enjoyed reading Ritualo’s “very rich” and “very useful” paper. He was pleased by the many new surveys that have produced better data that enable analysis of the factors that cause child labor and help policy makers design programs that will reduce child labor. Noting the explosion of research on child labor in recent years (see, for example, Basu, 1999), Patrinos said that, with better data, researchers can evaluate different types of interventions (see, for example, Ravallion and Wodon, 2000).

Patrinos was also happy to learn that surveys are beginning to include questions about children’s education as well as their work. He noted that Ritualo’s paper calls for more information on how children’s work affects their performance in school, and he said that TIMSS could provide this information. The TIMSS not only provides information on academic performance but also includes a background questionnaire that includes a question on child work.

Patrinos noted that Ritualo’s paper cites ILO Convention 138, which requires countries to establish a minimum working age that should not be less than the age for compulsory schooling. He questioned whether countries were implementing this aspect of the convention and asked Ritualo to compare the minimum employment age and compulsory schooling age for

a number of countries, using updated data. He said he had assembled data on this a few years ago (Siddiqi and Patrinos, 1995; Sinclair and Trah, 1991; UNESCO, 1993). In some countries, he said, there is an overlap “so that you can legally start working before compulsory education comes up.” In other countries, compulsory education ends at a much younger age than the minimum working age. He asked what signals countries with these overlaps and gaps were sending to young people.

Patrinos called for more information on children who work at home, taking care of siblings and doing household chores. He expressed concern for these children, who receive no education and do not learn any skill or trade, and called for more data on the hours they work.

He also expressed concern at the idea expressed in Ritualo’s paper that more detailed, probing survey questions led to higher estimates of child labor. Although such questions are “good in the sense of getting a better picture,” he wondered whether activities such as taking out the trash should be considered child labor. He warned that looking at activities without considering the hours devoted to them could result in overestimated counts of child labor, and he suggested adding survey questions on hours worked.

Although Patrinos agreed with Ritualo that we need more information on child labor on a regular basis, he asked about the costs of large-scale surveys, noting that larger samples provide more information on child labor but also cost more. He asked for information on whether governments and national agencies made use of the large, expensive, and irregular SIMPOC, LSMS, and MICS surveys to formulate policies or programs. If child labor data are indeed useful, he expressed the hope that national statistical agencies would begin to integrate questions from these surveys into their regular national surveys. Today, he said, “It’s clear that the only regular and timely surveys are the ones that give us the least information about child labor.”

Patrinos agreed with Ritualo that it is important to collect information on the worst forms of child labor, although it is very difficult to do so. He suggested that she add a summary discussion of the benefits of having better, and timely, regular data on child labor.

He suggested adding information on programs to reduce child labor to the list of indicators in Ritualo’s paper. He noted that the paper mentions only ratification of ILO conventions and enactment of minimum working age laws. Yet his research suggests a five-point strategy for reducing child labor as follows: a gradual policy approach; home business support and school enrollment incentives; targeting children of parents with low educa

tion; targeting locations where child labor is concentrated; and poverty programs that target characteristics of poverty that contribute to child labor (Grootaert and Patrinos, 1999).

As part of this gradual approach, Patrinos said, countries could allow a flexible school schedule so that children could do some work and attend school at night or on weekends. Because school incentive programs have proven very effective at reducing child labor, Patrinos suggested including more information on Mexico’s Progresa (now known as Opurtunidades) program. In this program, the government provides cash payments to low-income families of children who regularly attend school. Rigorous evaluations (see, for example, Krueger, 2002; Coady and Parker, 2002) have demonstrated that the program reduces child labor and increases educational levels among the poor.