Can cash transfers reduce child labor?

Defining child labor

Child labor (a subset of child work that is not allowed according to national and/or international regulations) is a legal rather than a statistical concept. Translating broad legal norms established in international agreements into statistical terms for measurement purposes is not straightforward. The international legal standards contain a number of flexibility clauses that are left to the discretion of national authorities. Hence there is no global legal definition and is no standard statistical measure of child labor.

Consequently, the terminology and concepts used to categorize children’s work and child labor (and to distinguish between the two) are at times inconsistent in published studies [1]. Similarly, there is substantial variation in the productive activities covered by the studies considered here. Some focus on specific activities (such as work in agriculture), whereas others use a broader definition of work (such as work in economic activities). There is also variation in reference periods. Finally, some studies focus on how many children work and some on how much children work.

Household vulnerability and child labor

By the latest International Labour Organization estimates, more than 120 million children aged 5–14 were engaged in labor in 2012; this is about 10% of children in this age group. There are many causes of child labor, with household economic vulnerability being the principal one [2]. Poor households with inadequate resources and no access to credit markets are likely to make inefficiently low investments in their children’s education and to let their children work at an early age. Lack of access to financial and insurance markets also makes households more vulnerable to the effects of income losses. One of the coping strategies is to withdraw children from school and send them to work.

Two figures illustrate these points both across and within countries. Figure 1 presents correlations between income per capita and child labor for children ages 7–14 (as proxied by participation in economic activities) for a large group of countries. Figure 2 presents the difference in the incidence of child labor across five income groups (quintiles) in selected countries. Both across and within countries, the link between poverty and child labor emerges clearly. Countries with higher GDP per capita tend to have a lower incidence of child labor (Figure 1), while within countries children in the poorest households engage in economic activities at a substantially higher rate than children in higher income groups (Figure 2). It is also clear, however, that poverty is not the only cause of child labor, as illustrated by the high variation in the incidence of child labor for similar income levels across countries, and by some incidence of child labor even in the top part of the income distribution within countries.

Social protection policies are thus an obvious potential means of addressing some of the factors driving child labor. By alleviating the economic vulnerability of households, social protection policies may remove some of the reasons that families send their children to work.

Transfers to families with children

Transfer programs aim to relieve economic vulnerability by increasing household income. The cash transfers are often accompanied by certain behavioral requirements, or conditions (conditional cash transfer programs). For example, some conditional cash transfer programs require that beneficiaries send their children to school or take them regularly to health clinics.

Cash transfers can have complex effects on household behavior through two mechanisms. They can affect child labor by modifying children’s likelihood of attending school or by changing the returns to child labor [3]. First, cash transfers may change how households value children’s use of time, encouraging them to send children to school and thus to work less or not at all. However, school attendance requires the commitment of a set amount of a child’s time and also increases household spending on education [3]. Thus, deciding to send a child to school is likely to affect the household budget (more costs for school and less income from child labor). Second, if the household uses part of the cash transfer to invest in assets that make child work more productive (and thus more profitable), the transfer could increase the value of children’s work to the household. Both these mechanisms make the likely effect of cash transfer programs on child labor uncertain.

Because school attendance generally requires a fixed minimum time investment (it is not usually possible to choose the number of hours a child will spend in school), a child who begins to attend school following a cash transfer will have less time for leisure and work. Moreover, the household faces some fixed costs to send a child to school (school fees, uniform, travel costs). The final impact on child labor depends on whether the amount of the cash transfer exceeds the cost of attending school. If the transfer falls short of the monetary cost of sending a child to school, the resulting change in child labor is ambiguous because both consumption and leisure would be reduced by sending a child to school. If the transfer exceeds the cost of sending the child to school, then consumption can also increase and child labor should decrease as the household can improve its condition without relying on the child’s earnings.

This conclusion holds only if the household uses the cash transfer to finance consumption or education. However, because money can be spent in many different ways, the additional resources could be used instead to invest in productive assets. Such investments could increase the returns to child work directly if the child works in the household business or farm and productivity in these activities increases as a result of the investment. They could also increase the returns to child work indirectly, through changes in the adult labor supply—for example, if household members devote more time to productive activities and less time to household chores, children’s time will become more valuable in household chores. While there is evidence that cash transfers are used in this way, to invest in productive assets, there has been only limited analysis of the consequences of this choice for child labor (see, for example, [4], [5], [6], [7] and the literature cited therein).

The impact of cash transfers on child labor

In general, impact evaluations find that both unconditional and conditional cash transfers can reduce child labor (Figure 3 and Figure 4). There are, however, large variations in the effects of different cash transfer programs, and for several programs, no significant impact could be identified.

It is not easy to identify the factors that are associated with the observed differences in the impact of cash transfer programs on child labor. A large part of the variability could be due to program characteristics that are difficult to capture in a comparative analysis. Nonetheless, some general points can be derived from the analysis that could inform program design so as to make cash transfers more effective in reducing child labor.

First, the impact on child labor does not precisely mirror the increase in school attendance generated by cash transfer programs: the increase in school attendance is generally much greater than the reduction in child labor. Within the set of conditional cash transfer programs considered in Figure 4, a one percentage point increase in school attendance is accompanied by only a 0.3 percentage point reduction in children’s work [3]. This is not surprising, since school and work are not mutually exclusive (for example, children can work after school or during school holidays, or miss some classes for work). Nonetheless, working while attending school is potentially detrimental to educational achievement and grade completion.

There are other reasons why increases in school attendance are not fully matched by reductions in child work. A key reason is that cash transfers can increase household investment in productive assets used in economic activities, such as in farming and small businesses. While there has been little direct examination of the impact of this mechanism on child labor [7], impact evaluations of microcredit programs find that households’ increased involvement in productive activities can lead to increases in child labor, as has been the case with microcredit programs in Bosnia, Thailand, and Bangladesh [8], [9], [10].

A second important point is that additional resources beyond the amount of the cash transfers might be needed to help households pay for the increased costs associated with sending a child to school. Without such additional resources, children who were working and begin to attend school might not stop working, or children who begin to attend school might also begin to work to help the household meet the additional expenses. Evidence of these types of effects have been found for the Bright program in Burkina Faso (designed to improve access to quality education for girls) and the Pantawid Pamilyang Pilipino Program (4Ps) in the Philippines (a conditional cash transfer program to eradicate extreme poverty by investing in children’s health and education). In both programs, school attendance increased but was not associated with a reduction in child labor. In fact, involvement in child labor increased, especially among children who began to attend school following household participation in the program.

The impacts of conditional cash transfer schemes

Conditions were introduced to transfer schemes with the intention of improving their effectiveness and ensuring that the benefits were used, at least in part, to improve children’s human capital. Common conditions in cash transfer programs require that participating households send their children to school regularly and take them in for regular health checks. Conditions have not addressed child labor participation because reducing child labor has not been a direct objective of most cash transfer programs (although participation of children in work has been a criterion for identifying households to receive benefits).

There are obvious reasons why conditions have not been imposed on child work. For instance, school and health clinic attendance are easy to monitor using school or clinic records. Objective verification of child work, by contrast, would be very difficult since most children who work are engaged in family businesses or in the informal sector, often in violation of national law. Monitoring would have to rely on statements by parents and children, which could be unreliable in these circumstances.

While evidence suggests that conditional cash transfers have a stronger impact on school participation than unconditional cash transfers do, whether the conditions related to child education reduce child labor has been much more difficult to assess. First, the decision to include an education condition in a program might depend on the expected impact of the program on the target population (meaning that the condition is endogenous, making assessment of impact difficult). Most studies do not specify the precise conditions attached to the delivery of the services or provide information on the degree of enforcement of the conditions, thus making straight comparisons of effectiveness difficult.

A few studies, however, offer some evidence in this area [11], [12]. The studies use variations in implementation of Ecuador’s Bono de Desarrollo Humano cash transfer program that, according to official reports, led some participating households to believe incorrectly that the cash transfers were conditional on children’s school attendance. The effect of the program on child labor was similar in households that believed that the program was conditional on school participation and in households that did not [3]. In both groups of households, child labor decreased.

Recent discussions of cash transfer programs have focused on whether conditions make a difference in the effectiveness of the programs or whether other approaches could achieve better results. One study of the Tayssir program in Morocco examined whether a cash transfer program open to all poor households without conditions could be as effective in increasing school enrollment as traditional targeted and conditional programs. Enrollment for the unconditional program was school-based, thus conferring an implicit endorsement of education. The study asked whether “a ‘nudge’ may be sufficient to significantly increase human capital investment, while [conditional cash transfer programs] as currently designed provide a big shove” [13]. The results show that a rural cash transfer program simply “labeled” as supporting education had a large impact on school participation even though the transfer was not conditional on school attendance. This has relevance for the case of child labor, where imposing explicit conditions might not be feasible.

Integrating cash transfer programs with other interventions

The impacts of cash transfer programs on child labor appear to be greater when cash transfers are complemented by interventions that reduce the costs of health care and education services or improve their quality [1]. Some other cash transfer programs have focused on both safety net provision and active poverty reduction by including support for household income-generating activities through grants or loans to buy investment goods. These programs seem to have a smaller impact on child labor, confirming that interventions that increase household economic activity tend to generate greater demand for children’s work.